MATH0
STAT.
Edward
Bright
Mathematics Dept
r
A
COURSE
OF
PURE MATHEMATICS
CAMBRIDGE UNIVERSITY PRESS
C.
F.
:
CLAY, MANAGER
LONDON
FETTER LANE,
E.G. 4
NEW YORK THE MACMILLAN
:
CO.
BOMBAY CALCUTTA L MACMILLAN AND CO., LTD. MADRAS TORONTO THE MACMILLAN CO. OF
\
J
:
CANADA,
:
LTD.
TOKYO MARUZENKABUSHIKIKAISHA
ALL RIGHTS RESERVED
A COURSE
OF
PURE MATHEMATICS
)
BY
G.
H.
HARDY,
M.A., F.R.S.
FELLOW OF NEW COLLEGE SAVILIAN PROFESSOR OF GEOMETRY IN THE UNIVERSITY OF OXFORD LATE FELLOW OF TRINITY COLLEGE, CAMBRIDGE
THIRD EDITION
Cambridge
at
the University Press
1921
VU First Edition 1908 Second Edition 1914 Edition 1921 .
character of the criticisms I have was too meticulous and pedantic for pupils of fifteen years ago: I am altogether too popular for the Trinity scholar of today. I need hardly say that I find such criticisms very gratifying. In order to find space for these insertions I have deleted a good deal of the analytical geometry and formal trigonometry contained in Chapters II and III of the first edition. I have made some attempt to genesis meet this objection in 158 and Appendix III. as the best evidence that the book has my to some extent August 1921 fulfilled the purpose with which it was written. I a sketch of Dedekind s theory of real numbers. in general principle of convergence a proof of the HeineBorel Theorem Heine s theorem Chapter concerning uniform continuity. and a proof of Weierstrass s theorem concerning points of condensation in Chapter IV an account of limits of . Appendix IV is also The an addition. and the fundamental theorem . The mosfc 8082. It is curious to note how the I had to meet has changed. These changes have naturally involved a large number of minor alterations. which I have rewritten in accord important are in ance with suggestions made by Mr S. H. H. EXTRACT FROM THE PREFACE TO THE SECOND EDITION principal changes THE have inserted in Chapter I made in this edition are as follows. October 1914 781474 . H. G. earlier editions contained no satisfactory account of the of the circular functions. H. in Chapter VI some additional matter concerning the integration of algebraical functions and . G. concerning implicit functions. in Chapter VII a section on differentials. indetermination and the V .PREFACE TO THE THIRD EDITION NO extensive changes have been made in this edition. Pollard. I have also rewritten in a more general form the sections which deal with the defini tion of the definite integral.
but I have confined myself to the very simplest cases and given a special discussion in each instance. I make no use of the principle of convergence uniform convergence. anything that involves really difficult ideas. September 1908 . wherever space permitted. double series. dydx d*f In the last two chapters I have occasion once or twice to integrate a powerseries. case a book for mathematicians: I have nowhere any any attempt to meet the needs of students of engineering it is but or indeed any class of students whose interests are not primarily mathematical. It is in made this class whose wants I have considered first. where a full and adequate discussion of all these points will be found. I regard the book as being really There are elementary. Anyone who has read this book will be in a position to read with profit Dr Bromwich s Infinite Series.EXTEACT FEOM THE PEEFACE TO THE FIEST EDITION book has been designed primarily for the use of first year students at the Universities whose abilities reach or approach something like what is usually described as scholarship I hope that it may be useful to other classes of standard . For instance. are never alluded to and I prove no general theorems whatever concerning the inversion of limit: : operations d*f I never even define 5% cxdy and =4. of hard examples (mainly at the ends of the chapters) to plenty : these I have added. readers. an outline of the But I have done my best to avoid the inclusion of solution. infinite products.
. .. ..~ . Polar coordinates 2425. * 13 Relations of magnitude between real numbers Algebraical operations with real numbers 1011. .. Sections of the real numbers. Weierstrass s Theorem . . .. . Points of condensation . 31. CHAPTER 20. The number x/2 .. 22. . . 12. means.. 23. Polynomials Rational functions .. . .... 2627... II FUNCTIONS OF REAL VARIABLES 21. V . . . 30 31 Miscellaneous Examples Gauss s Theorem.. 19.. . Algebraical functions Transcendental functions . 36... . Arithmetical and geometrical equations.. 6.. 18... . Important inequalities. numbers Real numbers . 9. . ". Dedekind . ..*.. 20. . their graphical 49 Graphical solution of equations Functions of two variables and sentation 58 repre ^ .. Coordinates 41 djfc &. . . 33. 15.. . Algebraical numbers. Quadratic surds The continuum The continuous .. 17... Cubic and other surds. . 32. . . 12..CONTENTS CHAPTER SECT. I REAL VARIABLES *AGE Rational numbers Irrational . 32. 19 19 1334. 16.. Schwarz s Inequality.. ... 1 37. Graphical solution of quadratic Decimals.. . . 34.".. . The The idea of a function 38 graphical representation of functions.. s Theorem .. .15 17 . . 23 26 27 real variable . 8. 2829. 1. . 4 ". * . 30.
. . 86 88 98 101 46. 78. 97. Infinite series . 44. . surfaces.. 62. . 51. 93. 66.Vlll CONTENTS PAG!* SECT. . . Surfaces of revolution. 60 61 33. . 6970. . 53. . . 94. . . Complex numbers The quadratic equation with . CHAPTER 3438. Ruled 62. . Condition that four points should be concyclic...? . . . . . . . . 64. Coaxal circles. The limit of xn 1 34 The limit of A +V Some algebraical lemmas The limit of n (#071). 5861. 103. 84 . Equations with complex coefficients. . Rational functions of a complex variable .. . 75. 6368. Complex functions of a real variable. Cylinders. . . 3942.140 143 1 The infinite geometrical series .. 52. Cones. Cross ratios. il 1. .. 63. Roots of complex numbers Miscellaneous Examples . 34&. 72 73. 97. . CHAPTER IV LIMITS OF FUNCTIONS OF 50. 101. . . 65 Arithmetical functions. Argand s diagram de Moivre s Theorem . . .. 68. * . * . . .. . 45. .. . 91.. Imaginary points and lines. Bilinear and other transformations. . Contour maps. 78 81 real coefficients . . . 32. . General theorems concerning limits functions Steadily increasing or decreasing Alternative proof of Weierstrass s Theorem .. Ib S 139 7677. . III COMPLEX NUMBERS Displacements * 69 . 96.. 90.. . . 55. 62. ..106 107 108 5357. .. A POSITIVE INTEGRAL VARIABLE .116 125 l.. Miscellaneous Examples Trigonometrical functions. . n 109 121 . . values of Properties possessed by a function of n for large Definition of a limit and other definitions . Curves in a plane Loci in space . . Properties of a triangle.137 . 97. 4749. 104.. Oscillating functions . .lt. Quadrature of the circle. 100. . Geometrical constructions for irrational numbers.. 71. . . . . 74.. 63. _ .. Construction of regular polygons by Euclidean methods.. Functions of a positive integral variable Interpolation Finite and infinite classes . 43. ..
CHAPTER V LIMITS OF FUNCTIONS OF A CONTINUOUS VARIABLE. Harmonical series. . 113. 123. continuous functions. Arithmetical series. . 159. The HeincBorel Theorem Continuous functions of several variables Implicit and inverse functions 108109. 197 203 205 . 82. series .. . 177* Classification of discontinuities. . %/x. . . 147. Decimals. . 158. 177.136. . 151.. CHAPTER VI DERIVATIVES AND INTEGRALS 110112.. . . IX PAGE 80.. Limit mean value. General rules for differentiation 114. ... Differentiation of polynomials Differentiation of rational functions . Continuity of cos x and sinrr. Limits as #soo or x*~ Limits as x+a Continuous functions of a real variable Properties of . The representation of functions of a continuous real variable 147 by means of limits The bounds of a bounded aggregate 149 The bounds of a bounded function 149 The limits of indetermination of a bounded function 150 . \n) } a. 156 157 Miscellaneous Examples Oscillation of sinn07r. .... 176. .... . Derivatives . 179 185 Sets of intervals on a line. ... 173. .. . .. 172.. . .. The general principle of convergence . .. Limits and continuity of polynomials and rational functions. ... . .210. Infinity of a function... .. . oscillation of a function in an interval ... .. . 117.. 160. Limit of 171.. Bounded functions. .. ^ * 190 191 . . ..194 Limit of ... n . 79. ... . The 105106. 139.. 9899. 8788.. ... .. . 81. .. 146. .. .. . . 8586. f/n. CONTINUOUS 162 165 AND DISCONTINUOUS FUNCTIONS 8992. 118. 121.151 Limits of complex functions and series of complex terms 153  . . 107. The notation Derivatives of complex functions of the differential calculus 116.. . Equation x n+l =f(x n ).. co 174 100104. Applications to zn and the geometrical ..CONTENTS SECT. 178. .. 9397. 115. Limits of n kx n . ^m _ gin Orders of smallness and greatness.. . . 169. . of a Expansions of rational functions. n! t . 205 207 209 Differentiation of algebraical functions . . 8384..j^ Miscellaneous Examples . 143. .. .
Derivatives of cosz and siux.. 164. 249 251 146. k . Lengths and areas in polar coordinates. 254.. 149. 219 The logarithmic function .. Axes of a conic. . Applications of Taylor s Theorem to maxima and minima . 208. 205. Integration of polynomials Integration of rational functions Integration of algebraical functions. ... 132139. . 258. 253 Derivative of 201. Lengths of plane curves ... 300. . 163. Newton Binomial series.. 268 270 274 280 283 298 299 299 152154.. The second mean value theorem. Leibniz Theorem. 125126.. 253. 150.. Tangent and normal to a curve. 223. 156161. Miscellaneous Examples s ... 255 Rolle s Theorem for polynomials.*... 302. rationalisation.. . Maxima and minima of the quotient of two quadratics.245 by \ . 259. Formulae of reduction. Series for cosx and smx. . 268 The contact Differentials Applications of Taylor s Theorem to the calculation of limits of plane curves . .. 130131. Jacobiaus. I I Curvature. Differentiation of functions of several variables f . . 162.. Miscellaneous Examples xm . 256. . 209. 214. Differentiation of a determinant. . 303. . CHAPTER 147. 303. Maxima and minima The Mean Value Theorem Integration... 129. 226. . .:... 290. 226 228 232 .. Repeated differentiation General theorems concerning derivatives. 272.. . 307. 294. VII ADDITIONAL THEOREMS IX THE DIFFERENTIAL AND INTEGRAL CALCULUS Taylor s Theorem Taylor s Series . 121.:". Simpson s Eule. Euler s Theorem.. 145.. 201... . 119. . 302. Approximate values of definite integrals.. . 273. 274. 233 236 Integration 140144. Homogeneous integrals. Areas of curves Alternative proof of Taylor s Theorem Application to the binomial series Integrals of complex functions .. . * * . * equality for integrals. 283. Integration by parts Integration of transcendental functions Areas of plane curves . 296.CONTENTS SECT. .. Multiple roots of equations. 302. Differentiation of implicit functions. 262 266 . 267. . 148. 302. Tangent to a curve. . . Extensions of the Mean Value Theorem... Fourier s conies. ^ . Schwarz s in functions.. . ^ . Rolle s PAGE 212 214 217 Theorem 122124.. 155. 127128. Differentiation of transcendental functions 120... 267. . Definite Integrals. 307. 399 method of approximation to the roots of equations. 151. 215. 283. Points of inflexion. . . 201. Osculating 272.
. Dirichlet s Theorem . .321 335 336 338 184185. Logarithmic 348. 188. 210. 369 374 . Transformation of infinite The integrals by substitution and integration by parts. 357 360 behaviour of log# as x tends to infinity or to zero . logarithmic function functional equation satisfied logarithmic scale of infinity by log. . . Binomial series. 354. 352. The 204206. Cauchy s condensation Infinite integrals .) CONTENTS XI CHAPTER SECT. . . series. . Multiplication of series of positive terms Further tests of convergence. Abel s and Dirichlet s tests of convergence Series of complex terms . Common logarithms . 338. . 327. 349. . Multiplication of conditionally convergent series. .. VIII THE CONVERGENCE OP INFINITE SERIES AND INFINITE INTEGRALS PAGE Series of positive terms. The The 202. Conditionally convergent series Alternating series 189. 340 342 344 345 191194.. exponential limit logarithmic limit 364 . 175. .. * The series Sn~ 8 test . 350. 190.. number e 363 exponential function general power a* . . ... . . 203. . 341. 328. 3GO 362 .nk rn and allied series. The The 207. Schwarz s inequality for infinite integrals. Abel s Theorem..T . .. . Becurring series. ~ and negative terms Absolutely convergent series Series of positive 186187.319 320 176. 196197. 177182. 356. . series by rearrangement. . . . of convergence 165168. .366 368 369 . . 195. Definite integrals. 344. . .. . CHAPTER IX THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS OF A REAL VARIABLE The The 199201. . 311. . 171174. 333. Power series Multiplication of series in general 349 350 Miscellaneous Examples The series I. 183. Logarithmic tests of convergence. 211. 353. 198. 308 169. The The 209. 343. . integral test 313 315 Maclaurin s . Difference equations. 355. Sa n sinw0. 170. 208. Alteration of the sum of a series Sa n cosn0. Cauchy s and d Alembert s tests . 348.313 . .
. . 222224. aad tangent of a complex number. 393. Irrationality of Iogi n. Irrationality of e 380. 215. 372. . The exponential function . The function e 407. 373. The inverse cosine.. The 234235. functions. 420. 431.. . sine. . . 432. and geometry . 232. . . . . AND CIRCULAR FUNCTIONS 217218. 4 . Logarithms to any base. 219. 216. . 373 .v xii SECT.. 387. by the binomial theorem. Transformations. Approximation to surds constant. 412. . Trigonometrical series. 417. . . . The The 236. Functions of a complex variable . . 403 404 409 413 414 . . . 408.. . . 429. 402. Definite integrals. . 370. Boots of projection. 214. . 416 417 421 422 logarithmic series exponential limit binomial series Miscellaneous Examples 425 3 . The hyperbolic Integrals containing the exponential function. inverse trigonometrical functions The 233. integrals. 428. Euler s 389. Integrals of certain algebraical functions. APPENDIX IV. . Level curves. . . The functional equation satisfied by Log 2. . 428. . 225226. The general power a* . 433 II. 385. . . . 332 384 . Definite APPENDIX APPENDIX APPENDIX I. The 237. . . 445 . exponential series series for coaz and sin z . Curvilinear integrals . . 427. . . 213..381 . . . 221. Alternative development of the theory . . The trigonometrical and hyperbolic functions The connection between the logarithmic and 231. . 425. . transcendental equations. Stereographic Mercator s projection. 395 397 399 The values of the logarithmic function . 227230. EXPONENTIAL. 377. . CONTENTS The exponential series The logarithmic series The series for arc tan x The binomial series 212. . . * . 386 337 Miscellaneous Examples . A note on double limit problems The The circular functions infinite in analysis 439 443 III. 426. CHAPTER X THE GENERAL THEORY OF THE LOGARITHMIC. . . Definition of the logarithmic function . 395 220. The proof that every equation has a root .
w. IV. f .. then 2 2 2 2 Hence show X(?n ft ). 1. Ifr and s rjs are rational numbers. We this. and in one way only. then r + s. Conversely. [In this series every rational number is repeated indefinitely. 2.] 4. unless in the are rational numbers. [The general theory of decimals will be considered in Ch. To the rational numbers thus denned we may add the obtained by taking p = 0. occurs as i.2) + q]ih term of the series. and n are positive rational numbers. I. can suppose (i) that p and q have no common factor. and (ii) We that q is positive. since (p)/(q)=p/q.*. and n. rational assume that the reader is familiar with the ordinary arithmetical rules for the manipulation of rational numbers... can of course avoid this by omitting every We Thus 1 number 1 H. &. and last case s = (when r/s is of course m 3. and \(m + n ) are positive rational numbers. A fraction r=p/q.. rs. where p and q are positive or negative integers. The examples which follow demand no knowledge beyond Examples meaningless).. . Any terminated decimal represents a rational number whose denomi nator contains no factors other than 2 or 5. number pl(q) = (p)lq.CHAPTER I REAL VARIABLES 1. is called a rational number. r . as if they have a common factor we can divide each of them by it.gt. 2Xwm. how to determine any number of rightangled triangles the lengths of all of whose sides are rational. any such rational number can be expressed. If X. as a terminated decimal. Rational numbers. The positive rational : numbers may be arranged in the form of a simple series as follows Show that plq is the [ (p + q  1) (p + q .
] 2. but then the problem of deter mining the precise position of pjq becomes more complicated. it is natural to regard length as a magnitude capable of sign. the line is drawn horizontally In order to obtain a point to represent a across the paper. We call A the origin. . and A the point 1. 1. representation of rational numbers by points It is convenient. then. produced indefinitely in both directions. as in Fig. the numbers l rational In order to obtain a point which shall represent a positive number r=p/q. of imply that analysis has any sort of dependence upon geometry they are illustrations and nothing more. we choose the point A r such that A Ar being a stretch of the line extending in the same direction A A l} a direction which we shall suppose to be from left to right when. we maybe content to suppose. to make a good deal of use of geometrical illustrations. along the line as ~ATT". This being so. ~AT~ number r = A Ax A. that we know what is meant by a straight line. of geometrical illustrations in this way does not. course. in on a many branches of mathematical analysis. or tJte point 0. REAL VARIABLES [l which has already occurred in a simpler form. far it may be from the truth. and a segment A A l of any length. negative rational s. and are em : The use ployed merely for the sake of clearness of exposition. it is not necessary that we should attempt any logical analysis of the ordinary notions of elementary geometry. that we know Assuming. positive if the length is measured in one direction (that of A^A^. Fig. and the length of a segment. however what they mean. 1. r the point A s such that .1 ^ . The line. let us take a straight line A. and we regard these points as representing and 1. a segment of a line. and negative if measured in the = BA and to take as the point representing other. so that AB .
we can 1 . We may express this by saying that BC includes infinitely many rational points.. It is Suppose. } an large..13] REAL VARIABLES thus obtain a point r. included in one of the k parts into which A A 2 has been divided. himself that he can cover the line with rational points as closely We can state this more precisely as follows if we : take any segment EG on A. is natural.. l as 1. 3 We Ar on the line corresponding to every rational value of positive or negative. we can find as many rational points as we please on BO.. we take A A 1 as our unit of length. for example. evident that if we choose a positive integer k so that k..BC&. he will readily convince mark off on the numbers whose as he likes.... another between B and Q.... that BC falls within the segment A! A. without coinciding with For if this were not so. as we asserted above. 3.. But P obviously corre sponds to a rational number whose denominator is k. Thus at But then we least one rational point P lies between B and C. We shall call the points 3. (1). and write A A = then we have A A r = r. is * The assumption that this is possible known as the Axiom of Archimedes. IV..e. infinity of positive integers means given any positive integer n.l .. 2. and such that and if. can find another such point Q between B and P.. line all the points corresponding to the rational denominators are 1. j find as many as we please.gt. which contradicts the supposition (1).... BC would be entirely either B or C. is equivalent to the assumption of what 12 .. and so on indefinitely i.. The meaning such sentences as of such phrases as * infinitely many or an infinity of\ in includes infinitely many rational points or there are BC an C or there are an infinity of positive infinity of rational points on The assertion there are integers .* and divide A^A^ into k equal parts... Ar the rational points of the If the reader will line.. in succession.. then at least one of the points of division (say P) must fall inside BC. will be considered more closely in Ch.. ... however This is plainly true we can find more than n positive integers .. Irrational numbers.
shows that this view difficulties. And it is certainly the case that if we points imagine the line to be made up solely of the rational points. means same as we can will is many positive integers as we please The reader assertion. that is to say by less than l/?i. by means of addition or multipli cation.000. and any segment lie between its end With by points. given any two rational numbers From infer that these considerations the reader might be tempted to an adequate view of the nature of the line could be it to be formed simply by the rational which lie on it. according tothe ordinary rules of algebra. where n is any positive integer assigned beforehand. and all other points (if there are any such) to be eliminated. and these lengths must be capable of combination with one another. A little further consideration.4 whatever n exactly the REAL VARIABLES may be. which must be a quantity capable of numerical measure ment in terms of any standard or unit length. Again. however. the same straight line. the figure which remained would possess most of the propertieswhich common sense attributes to the straight line.g. look and behave very much like obtained by imagining a line. and would. must be composed of points. for [l n = 100. to put the matter roughly. must be associated a certain entity called its any such segment length. is a. it must be possible to construct a line whose The straight line all of it the points which If the or product of any two given lengths. The assertion . e. we can interpolate between them a chain of rational numbers in which any two consecutive terms differ by as little as we please. would involve us in serious common Let us look at the matter for a moment with the eye of sense. easily convince himself of the truth of the follo\ving substantially equivalent to what was proved in the second paragraph of this section given any rational number r. along length PQ along must be a + 6.000 find as or 100. and any positive integer ?i. a given line. the length length is the t sum PR . r and s.000. is b. It is say that we can find a rational number from r by as little as we please. and the length QR. merely to express this differently to lying on either side of r and differing Again. and consider some of the properties which we may reasonably expect a straight line to possess if it is to satisfy the idea which we have formed of it in elementary geometry. we can find another rational number lying on either side of r and : which differing from r by less than l/n.
a + (b 4. a (b + c) = ab + ac. OQ. are 1 and a. C are three points lying along A from left to right. on our fundamental line A.c) = (a + b) + c. OQ. we must have AB&. B. Our requirements therefore involve the existence of a length measured by a number x. There are it is Now various elementary geometrical constructions. each corresponding to a rational composed number. cannot possibly satisfy all these requirements. which = 2. 13) for a mean proportional to 1 and 2. #2 = 2. a + b = b + a. For instance. 12) for a fourth proportional to the lines OP. vi. OR. and more.lt. VI. very easy to see that the idea of a straight line as of a series of points. and a point P on A such that AB = AC=1. presuppositions of our elementary geometry. AC. 2. . b.3] REAL VARIABLES if the lengths OP. along one straight line. may construct an isosceles rightangled triangle ABC such that Then if BC=oc. And length must be ab. a. for example. Or we may determine the length x by means of Euclid s construction (Euc. a (be) = (ab) c. ab = ba. viz. and so on. Moreover it must be possible. are involved in the line. the algebraical fourth proportional to 1. we purport to construct a length x such that x* M Fig. as indicated in the figure. and if we determine the length OS by Euclid s construction (Euc. it is hardly necessary to remark that the sums and products thus defined must obey the ordinary laws of algebra . this Moreover. and the length OR along another straight line is b. The lengths laws of our lines must also obey a number of obvious as equalities : concerning inequalities as well thus if A. to find a point P such that A Q P is equal to any segment whatever taken along A or along any other straight All these properties of a line.
then that our requirements involve the existence of a number x and a point P. .lt. as was to be proved.f having no factor in common with q.lt. if possible. which contradicts the assumption that p/q is in its lowest terms. that p must divide np 2 and as p and q have no common factor. such that A P = x. by taking n = 1. which precede as particular cases. In fact we For suppose. cannot have a rational but nonintegral root. unless m and n are both perfect squares. Show that no rational number can have its cube equal Prove generally that a rational fraction pjq in its lowest terms cannot 2. .?) .lt. Then np2 = mq z Every factor of m no factor in common : . root a/b. remember from elementary algebra) we write x The following alternative proof that no rational is number can have its square equal to 2 interesting. &. II.lt. and so pq&. Examples to 2.2q. which is due to Gauss and includes those 3. where m/n is any positive fraction in its lowest terms. be the cube of a rational number unless p and q are both perfect cubes. not one of the rational points already It appears constructed. is no rational number such that square may go further and say that there no rational number whose square is m/n. Hence n = \q*. where a and b are integers [For suppose that the equation has a . #2 = 2. Suppose. But clearly q&. &. in its lowest terms. In particular it follows. X must be unity. every factor of (f must divide n. is the following an algebraical equation : with integral coefficients. and with n.p&. Hence there is another fraction equal to pfq and having a smaller denominator. 1.6 EEAL VARIABLES it is [l But its is easy to see that there is 2. m Thus in = p2 n q*. A more general proposition. same property. It is easy to see that this involves such that (p/20 2 = 2 or p 2 = 2q 2 2 2 and so (2q p)/(p q) is another fraction having the (2g j) =2(p.lt.q. But this involves = \p 2 and as m and n have no common factor. that an integer cannot be the square of a rational number. unless that rational number is itself integral. if possible. that pjq is a positive fraction. and (as the reader will \/2. where X is an integer.
7 #.] The result of our geometrical representation of the rational numbers is therefore to suggest the desirability of enlarging our conception of number by the introduction of further numbers of a new kind. or rational numbers into the lower class. : that we can 4. It is clear possibilities ". But. and the other those whose squares are greater than 2. and we shall call these two classes the class L.T. Find the rational roots (if any) of 10 = 0. without a ing by 6* 1 ". + 2. and the class R. or the upper . 5. one containing the numbers whose squares are less than 2. if The These facts are plainly sufficient to if it make some generalisation of our idea of number desirable. 4] REAL VARIABLES common 1 . 5. One of the central problems of algebra is that of the solution of equations. first equation has the two rational roots 1 and 1. lowest terms equal to an integer. a? = 2. It is evident that a must be a divisor of p n . is a. at = 7. +10 are the only [The roots can only be integral. then the equation cannot have a rational root. our conception of number is to be limited to the rational numbers. can therefore divide the We two classes. Irrational (continued). Writing a/b for and multiply we obtain a fraction in its and the root 4. and the same is the case with such equations as ot? = 2. We shall confine our attention to the positive rational numbers. and so whether these are roots can be determined by trial. factor. or the lefthand class. c numbers The same conclusion might have been reached without the use of geometrical language. should prove to be possible. such as #2=1. in this way determine the rational roots of any such equation. We satisfies have already seen that there is no rational number x which this equation. The square of any rational number is either less than or greater than 2.3. Let us consider more closely the equation # 2 = 2. which is absurd. and b is positive.\ if Thus 6 Show that p n =l and neither of is zero. we can only say that the second equation has no roots.
are all less than but approach nearer and nearer to and by taking a sufficient number of the figures given by the process can obtain as close an approximation as we want.lt. d being any we please. say less than 8. and in which any two consecutive numbers differ by as little as we please. 1. can supply a formal proof as follows.Z . than all the members of L. 1414. though differs less we can find a member of the class little as L whose please.\\ and we may plainly suppose that both x and y are less than 2. given any a chain of rational numbers.. 141. by unity. if we carry out R from 2 by as we root of the ordinary arithmetical process for the extraction of the square 2.lt. whose squares 1.. such as 01 or 0001 or 000001. In the first place. 199996164. 20164. also from 2 by as little as we please. differs and a member of whose square. we can find We a two rational numbers a and b. 1415. in each of the approximations given above. The reasoning which precedes. is hardly of the precise character required by modern mathematics. For we can construct and last. Substituting j for 8 in the argument which precedes.. 15.... differing by as little as we please. 1999396. though greater than 2. 200024449. 14143. can now prove that an x can be found in L and a y in R such that x l and y 2 2 are as small as we please.. 225. we see that we can choose x and y so that y x&. or the REAL VARIABLES [l R is It is obvious that every member of righthand class. we obtain a series of rational numbers 2. and* interpolate between them a chain of rational numbers of which x is the first and y the last. than 2. viz. 14. positive rational number as small as In this chain there must be a last which belongs to L and a first which belongs to R. we obtain a series of rational numbers. although it will probably convince the reader.. and these two numbers differ by less than d.. Thus We &. of which a and b are the first member we saw in of L and a member of 21. are all greater than 2 but approximate to 2 as closely as we please. it . whose squares 4. Let us then take a member x of L and a member y of J?.8 class. 2*002225. 196. 19881. 14142. Moreover it is easy to greater convince ourselves that square. 2.. j^that.. 142. And if we we increase the last figure.. In fact. and in which any two consecutive numbers differ by less than S.
Our commonsense notion of the attributes of a straight line. For suppose.gt. L and R. such that than every member of L. and similarly R no smallest. 5. is o^ has member L can be larger than the rest.2 it follows a fortiori that 2 x* and 2 ?/ 2 are each than d. for 2 example. Hence L has no largest member. cannot be either less than or greater than 2. if x is 2 any member of Z. (ii) we (i) every member of R is greater can find a member of L and a member of R whose difference is as small as we please. 5] HEAL VARIABLES 2 . Irrational numbers (continued}. and as x and of so ^ of L x. # has a definite meaning.lt.4.gt. it larger follows that no members L # 2 #2 than #. so that. that so is less than 2. Then it follows from what pre cedes that we can find a positive rational number f such that f 2 lies . or ^ &.lt. It follows also that there can be no largest member of L &. Fig. (iii) L has no greatest and R no least member. and that it may be operated upon in accordance with the laws of 2 Then # 2 algebra.gt. the requirements of our elementary geometry and our elementary algebra. smallest member of R. for example. 2 Suppose that # = 2 2 For S. &.?. then # Then we can less find a member x &. Let us suppose for a moment that there is such a number x. such that #! differs from 2 by than of 8. alike demand the existence of a number x greater than all the members of L and a corresponding point P correspond to members of L from than all the members of R.2 and less since and ^ 2 &. and of divides the points which on A such that less P those which correspond to members ofR. 3. Thus there are member all of L. We have thus divided the positive rational numbers into two classes. or 2.
= 3. N t where N is any integer which is not a perfect square. and similarly it cannot be greater than 2.cannot be less than We are therefore 2. for no rational number has its square equal to 2. ^3. and that x is the number which in algebra we denote by \/2. and may pass on at once to If. he is not disposed to adopt so naive an first * This is the point of view which was adopted in the edition of this book. or to 3 a 3 = 7.10 REAL VARIABLES [l between #2 and 2. if he pleases. exist and are amenable to irrational numbers such as \/2. . And of course this number \/2 is not rational. as we shall see later on. . Thus x. That is to say. on the other hand. If he does this . a 4 = 23. The reader will no doubt remember that the root of such an equation as xftn meaning is attached to such symbols as is in treatises on elementary algebra denoted by fln or n 1 ^. x 3 = 3x or. almost word for word . driven to the conclusion that #2 = 2. he will be able to avoid the more abstract discussions of the next 13 et seq. to + 8. even when these lengths cannot (as \/2 can) be constructed by means of elementary believe in the existence of irrational geometrical methods. few sections. the algebraical laws with which he is familiar*. The reader may now follow one or other of two alternative He may. we can find a member of L greater than x\ and this contradicts the supposition that # divides the members of L from those of R. the laws of indices such as nr xn*=nr + 8 . be content to assume that courses. a. in virtue of these definitions. It is the simplest example of what is called an irrational number. (n r 8 } =nri 5 are extended so as to cover the case in which r and are any rational numbers whatever. and that a by means of the equations And he will remember how. But the preceding argument may be applied to equations 2 for example to # = other than x* 2. We are thus led to numbers x and points P on A such that x satisfies equations such as these. .
n r^ 1 * a good approximation to v/2. since 2 2 is neither rational numbers.gt. by defect and by excess respec The equation #2 =4 is satisfied by #=2. or #4 7 and XA 7. &. &. Suppose that P and Q stand for two properties which are mutually exclusive and one of which must be possessed by every Further.gt. &. he will sections which follow.1 4.lt. Find the differences between 2 and the squares of i. be well advised to pay careful attention to the in which these questions receive fuller consideration *. Find the difference between 2 and the squares of the decimals given in 2. 2.#2 If &. tively.gt.5. u. &. ti. number which possesses P is less than any such number which Thus P might be the property x 2 2 and Q the possesses Q. 4. 4 as approximations to x/2.2 S. numbers in quite general terms. 2 for those of the others. 5. Let us now attempt to state the principles of the positive rational sections of the construction of such classes &. .lt. In #2 &.] 4 we discussed a special mode of division of the positive rational numbers x into two classes. Then we call the numbers which possess P the property x* lower or lefthand class L and those which possess Q the upper or &. 1. they do not include all throughout). then yx&. * In these sections I have borrowed freely from Appendix I of Bromwich s Infinite Series. x and y and 2 . ment R less than or greater than 6. 6] REAL VARIABLES 11 attitude. s. if&. . It is plain that we could equally well construct a section in which the numbers of the two 2 and x 3 were characterised by the inequalities a? 2..lt. Examples III. Examine how far the argu of the preceding sections applies to this equation (writing 4 for 2 as before. such that a? 2. .lt. (continued}. Show that if mjn is example.for the members of one class and Irrational numbers &.lt.lt. [If we define the classes Z. The rational number 2 is an exception. .lt. Such a mode of division is called a section of the numbers in question. 8.gt. Apply this result to continue the series of approximations in the last 3. i. are approximations to x/2. then (mf2)/(flf n) r is a better one. and that the errors in the two cases are in opposite directions. suppose that every such positive rational number.
of L and then it and a member . Ex. number 7. denoted by an English by the corresponding Greek .12 righthand class R.lt. then + T) would be a positive rational number lying between I and i (I portance. in. Conversely. 1.gt. 1) Irrational numbers (continued). it is clear that to we shall denote by a*. we shaft confine ourselves to cases in which both classes do exist follows. . and so could belong neither to L nor to and this contradicts our assumption that every such number belongs to one class or to the other. 5). 1 and Q second possibility. here Z=l. &. (i) R The has a least least. An example 1 and Q to be *# 2 to be a? 2 P &.1 . ^ a.lt. with r\. member or (iii) L L has a greatest has no greatest member or member and I.gt. This question of the existence greatest member and of greatest or least members of the classes is of the utmost im R We observe first that it is impossible in any case that should have a greatest member and a least. If P is #2 &. which is I section corresponds to a positive rational number in the one case and r in the other. R are mutually exclusive. any such number a corresponds a section which For we might take P and Q to be the properties expressed by x a. that we can is find a of R whose difference as small as member we please. there are but three possibilities. y&. x &. This would obviously happen. In the particular case which we considered in 4.gt. for were the property of being rational. In the first two cases we say that the a. a and x the greatest * member of L. P (or Q) being positive. This being so. REAL VARIABLES . &. or of For the present. L R &. 1 escapes classification (cf. corresponding to a rational letter. and r the least of R. however. so that I r.gt. or by sc &. &. L had no no least. which . In the first case a would be and in the second case the least member number letter.lt. a respectively.gt. [l happen in special cases that In general both classes will exist but it may one is nonexistent and that every number belongs example.lt. For if I were the greatest member of L. It will be convenient to denote a section. as in 4. (ii) Either r. if to the other. R no of the section of first is obtained by taking 4 gives an example of the last possibility. a section at all by taking P to we have an example of the It should be observed that we do not obtain for the special 1 and Q to be #2 1 be x* is x2 &. r.lt.
We shall consider sections. * ". For there are sections as that of which do not correspond to any rational (such 4) number. and must therefore be regarded as temporary and provisional. for mathematical purposes. In other words.68] REAL VARIABLES 13 There are in fact just two sections corresponding to any of R. we must alter our point of view a classes. There being this correspondence numbers and the sections between the positive rational denned by means of them. frame our little. final definitions. it includes sections corresponding to all these numbers. 7.gt. is called a positive real number. which will however be modified in the next section. . if cc a would mean the same as Thus. merely omitting the word positive occasionally.gt. It is this fact which we make the basis of our generalisation of the idea of number. and more besides. in which both classes no greatest member. it would be numbers by the numbers. In order to avoid ambiguity we select positive rational number. but of all rational numbers. let A ~ only sections in which the lower class us agree that we will consider L has no greatest number. is called A 8. ". The aggregate of sections is a larger aggregate than that of the positive rational numbers. for example. We have confined ourselves so far to certain sections of the positive rational numbers. . we are almost forced to a generalisation of our number system. Real numbers. including zero. a and a are the sections which correspond to a and a &. A exist section and the lower class has of the positive rational numbers. which we have Before we agreed provisionally to call positive real numbers. and to regard the symbols which occur in our formulae as standing for the sections instead of for the a &. We accordingly frame the following definitions. perfectly legitimate. positive real rational number number which does not correspond to a positive a positive irrational number. But when we have in this way substituted sections of rational numbers for the rational numbers themselves. one of them let us select that in which the number itself belongs . a. to the upper class. or divisions into two not merely of the positive rational numbers. to replace the sections. We may then repeat all that we have said about sections of the positive rational numbers in 6.
Mr s Bertrand Russell has said that mathematics is the science in which we do not know what we are talking about. so far as mathematics is concerned. as being a section. From infer J . and are perfectly harmless. indeed it would be easy to define an infinity of classes of entities each of which would possess the properties of the class of real real We defined a number We numbers.gt. or the corresponding real number. that the symbols of mathe matics are capable of varying interpretations. It . be ambiguous. since the relations between different propositions are exactly the same whichever interpretation is attached to the and we can propositions themselves. .gt. one a proposition of elementary arithmetic. Sometimes. What is essential in mathematics is that its symbols should be capable of some interpretation . class . concerning sections of the rational numbers. and do not care whether what we say about it is true a remark which is expressed in the form of a paradox but which in reality embodies a number of important truths. It may happen that all negative rational numbers belong to the lower class and zero and all positive rational numbers to the upper. the real the context in which occurs V(i). may happen Such a section . Or again it that the lower class includes some positive numbers. or simply a number. i. and then. of course. and  are arithmetical fractions or real numbers. moreover. to a rational number If the real shall use the number does correspond rational to a rational number. we also. might equally well have defined it as being the lower.gt. that no particular logical importance to be attached to the precise form of definition of a real number that we The reader should have adopted. observe. it may mean the rational number of 1. When we is say (see 9) that we must mean by number 3 . There are now three cases to distinguish. We describe this section as the real number zero. &. is sufficient to fix its interpretation. generally they are capable of many. number The term rational number &. but one at any rate of its implications is this. or the upper.lt. Ambiguities of this kind are common in mathematics. J.e.j &. would take too long to analyse the meaning of Mr Russell s epigram in detail.} J&.gt. If we say that we may be asserting either of two different propositions. a pair of classes.REAL VARIABLES DEFINITIONS. (e. as a result of our definitions. the other a proposition .g. it does not matter which we adopt. the inference is in no way affected by any doubt as to whether i&. in which both is and the lower class has no greatest member. called a real number. classes exist [l A section of the rational numbers. term as applying to the real will. and that we are in general at liberty to adopt whichever we prefer. A is real called number which does not correspond an irrational number.
with real numbers in general in exactly the same way operate To do all this systematically 1. inequality. The most natural definitions of addition and multiplication do not work in a satisfactory way. Moreover. we should obtain a negative real 2 and Q to be took P to be #3 &. we are bound to make corresponding extensions of our and so conceptions of equality. when this extension of them is made.. belong to may happen class. and that.gt.1. The difference and that of 7 amounts between our present definition of a positive real number a to the addition to the lower class of zero and all the negative rational numbers. . addition. by taking the property An of P example of a negative real number is given and Q to be #+1^0. The reader may be tempted to ask why we do not regard these sections also as defining numbers. It we have extended our conception of plain that. which we might call the real numbers positive and negative infinity. If we This section plainly corresponds to the negative rational number .lt. . .. We have to show that these ideas can be applied to the new on. for a beginner. sponding English letters a. . /3. that of learning to attach precise senses to phrases containing to show that he is likely to be confused by . that some negative numbers a section we describe as a negative real Such number*. . .0 3?&. (A). there are three possibilities : (i) every a is a b and every with (b) and (A) with A&B\ in this case (a) is identical * There are also sections in which every number belongs to the lower or to the upper class. and we shall be content to summarily how a more systematic discussion would proceed. The classes them If a and /3 are two real numbers. its lower and upper classes by the corre . the chief difficulty in the elements of analysis the word infinity to .2. 9] REAL VARIABLES 15 Finally it the upper we describe as a positive real number..lt. C. now that number. multiplication. all the so that we can ordinary laws of algebra retain their validity. numbers. is Relations of magnitude between real numbers. 7. B\ selves we denote by (a). 6 to be x + I&..8. A 6. is any addition and experience seems their number. We the rational numbers of denote a real number by a Greek letter such as a. number which 9. c. as with the rational numbers of indicate would occupy a considerable space. is not rational. . but it proves to be incon venient in practice. There is no logical objection to such a procedure..
and in case a and ft are both rational. A in the lower class and all numbers a putting all numbers The real number thus defined. (A) are the classes which consti tute a. or a&.lt. 9. = /3 &. and that any positive number is greater than any negative number.lt.gt.lt. negative.gt.0. in when case (ii) a &. Examples IV. which is clearly in the upper. according as a=ft a&. we can always numbers lying between a arid /3. if a find an infinity of rational and /3 are two different real numbers.a. and (B) a proper part of (A) .lt. is positive. 2. v/2 &. and/3 = y. 8. but not all a s are b s. &. then a=y.16 REAL VARIABLES [i s are B s in this case (a) (ii) every a is a b. 5. It is plain Of the two numbers a and a one is always a = 0). immediate consequences of our definitions. ft. Prove that = /3 a&. we can define another section of the rational numbers by we denote by .0 a /3&. Prove that a Prove that Prove that. is It will be convenient to define at this stage the a negative a positive number a.0 if a is negative.gt. The one which is positive we denote by a.lt.lt./3. (i) a &.e.a or a&.lt.gt.lt. rational ot ? Fi g . ft or /3 ^ y. v/3 &. a  and call the modulus of 1. 6. a. a&. Prove that if a . or /3&. /3&.lt. .y. a  1&.gt. It is clear that. ft./3. we write a = ft. (iii) every A is a B. then a&.lt. is positive. In case (iii) 4.ft 4. A is .gt. a&. according as a = /3.y. a. 2. Prove that j3&. 3. but not all a proper part of (6)*.] [All these results are * I. These three cases may be indicated graphically as in Fig.lt.lt. is included in but not identical with (&). 1 (iii) 4. 7..3.lt. a. these + ? (i) definitions agree with the ideas of equality and inequality between (ii) numbers which we began by taking for granted. or /3&. and a&.ft Prove If a If a that /3 = a.lt. a. = . If (a). Similarly we can a define a when a negative or zero = also that a) ( positive (unless if a is negative.
(ii) the class (C) formed by all sums C= C in all cases.. a and b are rational. and write In any case 7=a If both a classes + ft. Prove that a + (  a) = 0. e. say r and s. there cannot be more than one rational number which does not belong either to (c) or to (C). 10] 10. as applied to real numbers in general. In order to define the sum of two numbers : consider the following two classes (i) the class (c) formed by all sums c = a + b. the fact that the classes a6 = 6f a when JI. and plainly hypothesis. and so C c and we can choose are as a. same. Examples V. (ii) Subtraction. since.] 4. B so that both this A a and contradicts Bb our small as we like.lt. Prove that a + (0 + y ) = (a + 0) + y. (i) Addition. In this case it is clear that af/3 is the least member (A) of (6*). we add it to (C). &. the classes (c). REAL VARIABLES Algebraical operations with real numbers. We define a ft by the equation The idea of subtraction accordingly presents no fresh difficulties. so that our definition agrees with our previous ideas of addition. we call 7 the sum of a and ft. they are the least members of the upper and (B}. ft. cannot be less than s less than every (7. 17 We now proceed to define the meaning of the elementary algebraical opera tions such as addition. 3. that is to say. A. b. 1. which must clearly be rational.g. 7. 2. (0) form a section of the rational numbers. since it corresponds to the least member of (C). Again. [This follows at once from or (A+) and (B+A}. 2 . Plainly c a and we A+B. a number have If there is one which does not.9. are the (a+b) and (& + ). and let s be the greater. Prove that a + /3 = /3 + a. must be greater than every c and But r. We now a section or real number 7. If every rational number belongs to (c) or to (C). For suppose there were Then both r and s two. Prove that a + 0=0 + a = a. and /3 are rational.
and to go back for a moment to the sections of positive rational numbers is operations with real numbers (con When we come to multiplication.a) B + a (B . [l 6. 7. except when we rational numbers with at most one exception This depends. we define the reciprocal of a positive number a by means class (I/a). From the definition of subtraction. and Exs. /3j:ga/3&. 1. if a and /3 are positive.18 5. Finally we define a//3 by . definition of addition. a. 10. In order to define we begin by de Con fining the reciprocal I/a of a number a (other than zero). REAL VARIABLES Prove that a . then (a)/3 (iv) = a& a(/3) = a. 11. and 2 above.] Prove that II". (C) for which that which we adopted It is easy to show that this definition is equivalent to in the text. A. it Algebraical (iii) most (among which we may include 0) in the first instance. tion. convenient to confine ourselves to positive numbers 4 7. We may then follow practi only which we considered in the same road as in the case of addition. taking (c) to be (ab) cally and (C) to be (AB). tinued).a).a = 0.c = AB Finally definition ab = (A . form the classes (c). 9. The argument is the same.lt. We of the lower class (1 /A) and the upper then define the reciprocal of a negative number a by the equation l/( the equation )= (I/a). define the difference a/3 = y by the equation y+/3 = a. as in the case of addi so that (7 on showing that we can choose as small as we please. = a. (a) (0) division. we include negative numbers within the scope of our by agreeing that.C=Ab. are proving that all must belong to (c) or (G). Prove that a (/3y) = a. Multiplication./3 + y. Give a definition of subtraction which does not depend upon a previous [To define y = a ft. c = aB. fining ourselves in the first instance to positive numbers and sections of positive rational numbers. Prove that a  = . b. 4. Division.(/3 . and B Here we use the identity c is G . it follows that We might therefore 8.b).
1013] REAL VARIABLES 19 are then in a position to apply to all real numbers. We there constructed a section by means of the inequalities y? 2.. This was a section of the positive rational numbers but we replace it (as was explained in 8) by a section of only all the rational numbers. a 3. 13. rational or irrational. where and are defined are A A A positive rational numbers whose squares are greater than 2. classes of irrational numbers. where a and a are positive rational numbers whose squares are less than 2. (ii) (A ). the form a \/b. the whole of the ideas and methods of elementary algebra. where number which is not the square of another A number of rational number.. &. aXl = lXa = a. where a is rational. detail. which can only be 2 itself.gt. is called a pure quadratic surd. These classes exhaust all positive rational numbers save one. 2.V2) ( A/2) = V2 V2 = (V2) = 2 2. a/3 (a o. and the corre \/3. x3 = 7. = (a/3) y.. Thus the equation x* = 2 has the two roots \/2 and A/2. Examples VI. a is a positive rational 29  .V2) 2 = (. ax(l/a) = /3a. 4. $7.. + /3) = + . a? . is sometimes called a mixed quadratic surd.. (/3y) 8. The by means of which the product of \/2 by itself is (aa).lt.. 2. The number */2. Quadratic surds. but particularly important. We denote the section or number thus &. . Thus Again (. sponding irrational numbers V3. and ^/b is a pure quadratic surd. formulae : Prove the theorems expressed by the following OXa=0. We Naturally It will be we do not propose to carry out this task in our attention to more profitable and more interesting to turn some special. 12. 6. 7. Similarly we could discuss the equations # 2 = 3. defined by the symbol classes (i) \/2. A number of the form + \fa. Let us now return for a moment to the particular irrational number which we discussed in 4 5.
the equation by the least common A will remember that the roots are {. 22. but this is a mere . and z has as its roots the two quadratic surds p*J(p 2 q). It is easy to construct geometrically a line whose length is equal to any number of this form. Examples VII. construction is the The reader 2 following.20 The two numbers a&.0 there to be are no such values of The reader will remember that in books on elementary algebra the equation is said to have two complex roots. the equation x + 2px + q=0. 1. and the more complicated quadratic which may be expressed in a form extraction of square roots. aa. by Euclidean methods (i. where p and q are rational. Misc. I. b c rational.lt. The meaning When it is 6 2 rtc the equation has only one root. This property of quadratic surds makes them especially interesting. The only kind of irrational numbers whose existence was 3 are these suggested by the geometrical considerations of surds.e. by taking three to be integers.e. q&. II. For the sake of uniformity two equal roots. such as repeated irrationals involving the V2 + V(2 + V2) + V{2 + V(2 + V2)). III.b^/(b ac)}/a. 2bx + c = 0. attached to this statement will be explained in Ch.gt.0. easy much more elegant. Give geometrical constructions for 2. * See Ch. by geometrical constructions with ruler and compasses) is a point the proof of which must structed be deferred for the present*. pure and mixed. 2 f f there are two values of x for which x.gt. first constructing *J(b 2 ac). though less straightforward. Exs. for we can multiply multiple of their denominators.i2 + 26#+c Nothing is has lost two real roots t if all b2 ac&. The quadratic equation Suppose a. If 6 2 ac&.Jb REAL VARIABLES are the roots of the quadratic equation [l p 2 Conversely. as the reader will easily see for That irrational numbers of these kinds only can be con himself. cu.0.lt. It is to construct these lengths geometrically. generally said in this case to have convention.
P 4.c. having regard to sign*. Q Q Fig.lt. t I have taken this construction from Klein s Lemons sur certaines questions di geometric elementaire (French translation by J. cutting the circle in are the roots of the equation with their proper signsi. Two pure quadratic surds are said to be similar if they can be ex of the same surd. Some theorems concerning quadratic surds. Draw BC= 2b/a perpendicular to AB. Verify also that if 6 2 = ac the circle in the second construction will touch BC. c/26. 2 2 Q will not meet the circle if 6 &. ^^f. perhaps even simpler. 14] REAL VARIABLES circle 21 Draw a of unit radius. and otherwise to be pressed as rational multiples dissimilar. The reader should draw figures for other cases. if M and N no common factor. Take a line Another.13. Za/b and QQ and PN. 5. while if b = ac it will be Verify that a tangent. Griess. Paris. if possible. Vir* q V A u which b where * all the letters denote integers. For suppose. and CD=c/a perpendicular to BC and in the same direction as BA.are similar surds. and neither of which *JN are dissimilar surds.JM On the other hand. AB of unit length. and the tangents at the ends of the diameters. On AD as diameter The proof describe 3. a diameter PQ.gt. are integers which have and is a perfect square. &. Then BX and BY are the roots. . cutting QQ in and N. positive PP and QQ will be drawn in the same direction. and Then PP = = P M PM X QX QY is simple and we leave it as an exercise to the reader. Join Take Q. drawn to suit the case in The figure is and c have the same and a the opposite sign. Draw and Y. construction is the following. 1896). a circle cutting If ac is BG in X and Y. Thus and so \/8. Prove that 14.
. Examples VIII. integers. (Ex. // A + *JB = C+&. and therefore either (1) and divisible by a2 or (3) . . . If A. where a is rational. Hence squares. 2a is a.D is rational. \ where P is an integer. .gt. For B.22 REAL VARIABLES [l integral. case may be ruled MN divisible by M or (2) N by a M N are both divisible by The last out. 1. which is contrary to our hypothesis. b. . Then is divisible is . unless a = b. C. . c. If B is D (in which case it is obvious that A is also* equal to follows that is also rational.Ja + *Jb cannot be rational unless Ja and /& are rational. . ". where P^ denotes the product of some of the factors a 2a bP. then &. c2y and are both perfect and P22 the product of the rest. then (unless *JB and *JD are both rational). 3) a.lt. . B=D or (ii) B and D are loth squares of rational numbers. c be divisible by some of these factors and by . since M and N have no common factor. . . The same is true of *Ja Jb. and therefore 2 . so that are positive where a.gt. If a and b are rational. Hence \/B and */D are rational. .. . 3. Let be the prime factors of P. y3. . Then .. . and so is not equal to it (7). This argument so that M must may be Thus 2a 2 2y applied to each of the factors a b ?. .JD. a2 *. n. 2. D are rational and then either (i) A G. COROLLARY. y. N the remainder. are similar surds (unless both are rational). Ax/3 are not similar Prove ab initio that ^2 and Prove that &. surds. JfPA NPf.JMN Thus is MN = P evidently rational. M N THEOREM. B.Ja and *J(I/a).
15. If a2  b is positive. then # can be expressed in the form P+QJq. expressed in the form P + QJq. arithmetical continuum. of 2 It is convenient to suppose that the straight line to all the numbers of the of corresponding A The points of the * This supposition is merely a hypothesis adopted (i) because it suffices for the us with convenient geometrical purposes of our geometry and (ii) because it provides As we use geometrical language only for illustrations of analytical processes.or (c) JA.] 5. then a*Jb equation. a n are rational numbers) can be expressed in the form v/&&. where P and Q any integer. . is points composed arithmetical continuum. q. 15] 4. .Jq.14. or (6) A =D and J3 = C. is is equation with rational coefficients. business to study the foundations purposes of illustration. any expression . and of no others*. where p and q are rational. [Square the given equation and apply the theorem above. we can form express J(p+*Jq) in the where Determine the conditions that it may be possible to express J(p^ + *Jq\ 11 where p and q are rational.7. t If p. . in the form V^ + Vy. in is m where if x=p + *Jq. Express l!(p+Jq) in the form prescribed in numerator and denominator by p . it is not part of our of geometry. where P and Q 10.e. For example. the root of an algebraical another root of the same J 8. [Multiply Deduce from Exs. and p 2 q are positive. 12.. REAL VARIABLES 23 If (a) then either A = C and B=D. ^/^) are all rational or all similar surds.] Ex.. in x with rational coefficients (i. If a + where b is not a perfect square. can be are rational. where G(x] and H(x) are polynomials in x with rational coefficients. where x and y are rational. the necessary and sufficient conditions that 2 2 should be rational are that a b and ^{a + N/(a 6)j should both be squares of rational numbers. 6.gt. Deduce that any polynomial of the form where a 7.lt. ^&. The continuum. Neither (a + Jb) 3 nor (a  3 */b} can be rational unless Jb is rational. are rational. is called the of all real numbers. 6. and The aggregate rational irrational. *JB. Prove that . 6 and 8 that any expression of the form G (x)IH (x\ 9.
10. i * ^* zz^z^. we . + 8&.lt. and so 8/&.gt. For y is any number greater than #.2.lt. : The righthand =4+ similarly we can define 2 2 =4/(4 ^15) and then. 4.gt. can 2 is positive and negative and equal to r. of the We have considered in some detail the chief properties of a few classes of real numbers. \ $?* given a direct proof of the existence of a unique number 3 \z Buch that z =3z+8.lt.3?/8. rational. and how. side of this equation is not rational but exactly the same reasoning which leads us to suppose that there is a real number x such that #3=2 (or any other rational number) also leads us to the conclusion that there is a number z l such that 2 1 3 We thus define ^ = ^(4+^15).v/15. &. &. 3+8 = 0. and so the equation &.3# &. then also y 3 3 + 8 we obtain on sub suppose if possible y &.lt.gt.gt.gt. such. 2= Hence 3. 4. 22 Zi + z 1 z 2 + z2 12. It is easy to see that there cannot be two such numbers. REAL VARIABLES the aggregate of which [l may be said to constitute the linear continuum. And it cannot be For any rational root of this equation must be integral and a factor of 8 (Ex. that there is a number ?y = x/15 such that 2 and we can then.0. for example.gt. &.gt. For if z i 3 = 3z 1 + 8 and 223 = 3%f8. or y + xy + x &.gt. 8 is a root.2.8 And we might have &. if it exists. then supply us with a convenient image arithmetical continuum. 10.gt. Thus z 3 = 32+ 8 has at most one root and that root. Let us consider a more complicated surd expression such as supposing that the expression for z has a meaning might be show.lt. which is impossible. Now V it is easy to verify that 23 i + v c .2 which is less than 8/ 3 But then . as rational numbers or quadratic surds. to put it roughly. as in .gt. And is be negative.gt. 3. define * rv v&. &.8. &. is positive and not rational.4.2 and so 8/. and it is easy to verify that no one of 1. as in 12.gt. We can now divide the positive rational numbers x into two classes Z.gt. &. &. as in 2/ =15. cannot be equal to 3. . &. define the numbers 4 + ^/15. = 3. 3 2 2 z2 and therefore Zj&.2. We first . A according as x3 3x + 8 or x 3 &v + 8. For or if zi it is easy to see that neither z l nor z. We add a few further examples to show how very special these particular classes of numbers are. 2. &. &. &. Now consider the equation in z l for Our argument as follows. and v/li3. we find on subtracting and 2 2 But if z l and z2 are positive ^ 3 dividing by Zi~z2 that Zi +z 1 z2 +z2 =3. for y is tracting y tf &.lt.24 line. It is easy to see that 3 3.3(y#). Then since # 2 2 3 3 3. they comprise only a minute fraction of the infinite variety of (i) numbers which constitute the continuum. + 8 and if a? 3y + 8. n. just found 3 is impossible. Hence there is at most one z such that z 3 = 32 + 8.gt.gt. 3).
found. 3y f 8. we have not exhausted the different explicitly expressed kinds of irrational numbers contained in the continuum. proof will be found in Ch. for example. 7T 3 = H. And it has been shown f (though the proof is un TT. there are others which are roots of It is only in algebraical equations but cannot be so expressed. how ever.gt. In this case. and they form a section of the positive rational numbers or positive real number z which satisfies the equation z3 = 3z + 8.gt. as 7T 2 = ??. 1913). Finally.lt. natural to suppose* that the circumference of such a circle has a This length is usually length capable of numerical measurement. . such. Thus. x then also y z &.e. it not possible to obtain a simple explicit expression It can in fact for x composed of any combination of surds. The reader who knows how to solve cubic equations by Cardan s method will be able to obtain the explicit expression of z directly from the equation. VII.. or combinations of such surds. et seq. denoted x? = cc\ by that this number TT is not the fortunately long and difficult) root of any algebraical equation with integral coefficients. Let us draw a It is circle whose diameter is equal to A A l} i. besides irrational numbers which can be expressed as pure or mixed quadratic or other surds. pp. to unity.2 25 i (since ^3 &. (ii) The direct argument applied above to the equation y? 3x little +8 a could be applied (though the application would be more difficult) to the equation a? =x+ 16. is be proved (though the proof is difficult) that it is generally impossible to find such an expression for the root of an equation of higher degree than 4. 7T 5 = 7T + U. See Hobson s Trigonometry (3rd edition). and would lead us to the conclusion that a unique positive real number exists which satisfies this equation.8). 305 f * A or the same writer s Squaring the Circle (Cambridge. Similarly we can show that and y &. very special cases that such expressions can be  (iii) But even when we have added to our list of irrational numbers roots of equations (such as 16) which cannot be as surds.15] positive REAL VARIABLES and #&. it is evident that the classes L and It both exist .lt.
need not necessarily be continuous. We may think of may them as an aggregate. 7. The reader should . Any number of other examples can be constructed. 0. Let us take the last case. a and n are only examples of variables. 5. In fact it is only special classes of irrational numbers which are roots of equations of this kind. we may think either of a particular number (such as 1. such as n is either odd or even . These are the most important examples. 2. n is called the variable. In the theory of decimals. The continuous real variable. the whose field of variation is formed by all the real numbers. we denote by x any figure in the expression of any number as a may decimal. the arithmetical continuum defined in 16. for instance. 8. or the aggregate of positive integers. 3. just as it is only a still smaller class which can be expressed by means of surds. Instead of considering the aggregate of all real numbers. the number x. \/2. the preceding section. but we have often to Naturally variable consider other cases. And this number TT is no isolated or exceptional case. or TT) or we may think of any an unspecified number. This last is our view when we make such assertions as x is a number measure of a length # may be rational or irrational ( . A aggregate. or an unspecified positive integer. ( a positive integral variable. and that whose field is formed by the positive integers. think of specified number. variable however. but a variable which has only ten different values. : the variable. or individually. number which is not rational nor yet belongs to any of the classes of irrational numbers which we have so far considered.26 REAL VARIABLES [l where n is an In this way it is possible to define a integer. 9. And when we them individually. 1. . viz. The real numbers be regarded from two points of view. Then in statements about any positive integer. l x point of is the The x which occurs in propositions such as these is called the continuous real variable and the individual numbers are called the values of . Then # is a variable. 4. J. such as the aggregate of rational numbers. we might consider some partial aggregate contained in the former . and the individual positive integers are its values. 6.
even if he be one of those who prefer to omit the discussion of the notion of an irrational number contained in 6 12.gt. Either r.e. with the pair x 2 2 and x2 to positive numbers) with #2 Every rational number in possesses one or other of the properties. the xth day of the week. It is plainly possible to apply the same idea to the aggregate of all real numbers. &. There are now two possibilities!.gt. 17. and one of which is possessed by every real number. the driver of cab examples in ordinary life policeman the year x. but not every real number. Let us then suppose * that P and Q are two properties which are mutually exclusive.1517] REAL VARIABLES 27 think of other examples of variables with different fields of varia tion.lt. sections In 4 7 we con sidered of the rational numbers. is one which can. II is important to observe that a pair of properties which suffice to define a section of the rational numbers may not suffice to define one of the real numbers. as the reader will find in later chapters. or R has a least member Both of these events similar to that of 6. The discussion which follows is in many ways We have not attempted to avoid a certain amount of repetition. and only one of the two (ii) (iii) classes that both classes exist . since &. * &. R properties: (i) that every number of the type considered belongs to one . This is so. that any member of L is less than any member of R. &. or righthand class R. either case v/2 escapes classification. Sections of the real numbers. X) He will find interesting x. b / ^ . j t There were three in 6. the rational numbers (or of the positive into two classes L and possessing the following characteristic modes of division of rational numbers only) i. for s/2 or (if we confine ourselves v/2 and x example. . and and Q be x ^2. The idea of a section. first brought into prominence in Dedekind s famous pamphlet Stctigkett und irrationale Zahlen. be grasped by every reader of this book.lt. L has a greatest member * I. .lt. We call the numbers which possess P the those which possess Q the upper Thus lower or lefthand class L. and the process is. Further let us suppose that any number which possesses P is less than any which possesses Q. and indeed must. The : values of these variables are naturally not numbers. P might be x N /2 c &.gt. of very great importance. &.
HEAL VARIABLES For if [l had a greatest member I. Then the classes There are now L^ and 1 form a section of the rational numbers. and these. precisely as before. has a kind of completeness which the aggregate of the rational numbers lacked. R R two cases to distinguish. it is it is that to R. In this case a must be also the greatest member of L. or the continuum. Starting from the rational numbers. This conclusion is rational of very great importance for it shows that the consideration of sections of all the real numbers does not lead to any further generalisation of our idea of number. l On the other hand it may happen that L has no greatest member. and that the aggregate of real numbers. L member r. have been expected that the idea of a section of the and it might real numbers would have led us to a conception more general still. For if not. 6. Thus least.28 cannot occur. In this case the section of the rational numbers formed by L and R! is a real number a. It may happen that L l has a greatest member a. R a real number in the sense in which a section of the rational numbers sometimes. and so could not belong to either class. that more general than that of a rational number. and R a least would be greater than all members of L and less than all members of R. if it belongs the least member of R. the number \(l\r) For let LI and Rj_ denote the classes formed from L and by taking only the rational members of L and R. but not always. . new conception of a number. a completeness which * This is expressed in technical language closed. a in any case either L has a greatest member or section of the real numbers therefore corresponds to Any to a . corresponds number. being less than ft. If it belongs to L we can shew. and therefore to L^ and this is plainly a contradiction. we could find a greater. Hence a is the greatest member of L. This number a must belong 1 to L or to R. by saying that the continuum is was not the case in . On the other hand one event must occur*. belong to L. say (S. There are rational numbers lying between a and ft. The discussion which precedes shows that this is not the case. the greatest member of L and similarly. we found that the idea of a section of the rational numbers led us to a of a real number.
we can take S to be any number less than 1. Suppose then that we are given a set of points. if is halfway between two On the other positive integers. or. L each class contains at least one number. If the real numbers are divided into and R in such a way that every number belongs to one or other of the two classes. any member of L is less than any member of R. A system of real numbers. or (ii) this is not possible. which we will It is Take any point f which may or may not belong to S. (iii) then there is a In applications we have often to consider sections not of all numbers but of all those contained in an interval (0. hand. Then there are two possibilities. which has the property that all the numbers less than it belong to L and all the numbers greater than it to R. we can take 8 to be any number less than \. number a in this case satisfies the inequalities /3 &. most convenient here to use the language of geometry*. It is clear that the conclusion stated in Dedekind s Theorem still holds if we substitute the and that the real numbers of the interval (ft y} for the real numbers numbers than . The number a. (ii) is true . 18. denned in any way whatever. that S consists of the points corresponding to all the positive integers.17. of all the positive integers. other than f itself f. number a. and (iii). itself may belong to either class. y\ that is to say of all numbers section of such numbers is of course a division of x such that /3 x y. whatever the value of . t This clause is of course unnecessary if belong to S. (ii). % + S) does not con denote by S.lt. . or of all the rational points.a^y. then. Points of accumulation. and (i) will be true. Suppose. The set might consist. If is itself a positive integer. be converted into a section of all numbers by adding to /3 Such L all and to R all numbers greater than y. 18] REAL VARIABLES result 29 The which we have just proved may be stated as follows: Dedekind two classes (i) (ii) s Theorem. for any interval whatever contains an will hardly require to be infinity of rational points. * The reader reminded that does not itself this course is adopted solely for reasons of linguistic convenience. ^ A them into two a section may less classes possessing the properties (i). OP of the points on a straight line corresponding to them. for example. Either (i) it is possible to choose a positive number 8 so that the interval (f S. is called an aggregate or set of numbers or points. for example. . tain any point of S. if S consists of all the rational points.
as we please. f+ 81) But this interval also surrounding f but not reaching as far as flt must contain a point. point of accumulation of $ may or may not be itself a point The examples which 1. consists of the points corresponding to the no points of accumulation. the points of accumula positive points of the line. THEOREM. 3. and all lying inside the interval A \ of S. 1 19. or all the integers. follow illustrate the various possibilities. and is situated in an interval (a. is Examples IX. If S consists of the points 1. For. If S positive integers. 2. 4. . If S consists of all and all tion are the origin the positive rational points. contains at least one point fj which and this whether f belongs to 8 and does not coincide with itself be a member of S or not. there are If S consists of all the rational points. J..S). Then any interval (f 8. The general theory of sets r S and importance in the higher branches of analysis but it is for the most part too difficult to be included in a book such as this. We divide the points of the line A into two classes in the following manner. f+S) must contain. It is easy to see that the interval (?~8. when we have determined f1} we can take an .30 REAL VARIABLES [l Let us suppose that (ii) is true. viz. nojj_merely one. and which we shall require later.. f 4. j .. The point P belongs to L if there are an . every point of the line a point of accumulation. member of $ and does not coincide with f. then at least one point of the entirely interval is a point of accumulation of 8. There is however one funda mental theorem which is easily deduced from Dedekind s Theorem of points of the utmost interest . . repeat this Obviously we may and so on with f2 in the place of f argument. the origin. j3). If a set S contains infinitely many points. all belonging to S. but infinitely many points of 8. say fa which is a . however small its length. In this way we can determine as many points indefinitely. there is one point of accumula tion. In this case we shall say that f is a point of accumulation of 8. interval (f 81. Weierstrass is s Theorem.
18. is $ = 1. y. In this case the sole point of accumulation.] (if 3 any) of 9# Find the rational roots . /3. Hence f is a point of accumu contains an infinity of points of S. * 1+ r J 2 + rTT3~f 2 ". pp. In what circumstances can cA + d  s Infinite Series. 0^ 3 &. where a. 2 ". y. This point may of course coincide with a or 1..lt. 19] REAL VARIABLES and to (i) and 31 infinity of points of S to the right of P.lt. z\ (2) for all values of x. to L and /3 to R. be rational? J . (ii) is satisfied a point f such that. A line AB is divided at AB .e. * 5. a.lt. i . see Hobson. . Show that the ratio A CjAB is irrational. . z subject to both 2. 11) i.lt.2. I.. so that the interval (f5. however small be 8. f S f+S to R. and lt 0&. are positive integers. + i7273~7^I where aj. II. as for instance when a=0. MISCELLANEOUS EXAMPLES ON CHAPTER 1. and S consists of the points .a fc &. [A direct geometrical proof will be found in Bromwich 143.3. Theory of Functions of a Real Variable. p.&v 2 f 1 5x . is irrational.. case. AC=C A C in aurea sectione (Euc. Then it is evident that conditions R in the contrary (iii) of Dedekind s Theorem are condition satisfied .. For further information as to modes of representation of rational and irrational numbers. ^. is Hence there belongs to L and +) lation of S.lt.lt. c. 363.. &. y. and since a belongs also. so that 2 . What are the conditions that ax + by + cz=0. 4549. b. 4.0&. 3.10 = 0. d are rational.gt. ax+$y + yz = Q and ax+py+yz=0 in one (3) for all Any positive rational number can be expressed and only one way in the form &. Any positive rational number can be expressed in one 7 only as a simple continued fraction and one way where a 1} 2 &. [Accounts of the theory of such continued fractions will be found in text books of algebra.] 6. of which the first only may be zero. subject to (1) for all values of . z values of x.j % Oga are integers..
a 2 ) *&.. de or Since &.. An inequality of a slightly different character is geometric means..... as the truth of the proposition obvious when G= 0. Some elementary inequalities. we have (af .. .... or forms of the inequality arithmetic p = 2 in (4).. Suppose that a r and a are the greatest and ? least of the as (if of them &.lt....2 have the same sign.gt.... The general form of the theorem concerning the arithmetic and .gt... at most n repetitions value of the arithmetic mean G.....) . positive integers.. As the final value cannot have been less. #... and..... that which asserts that the arithmetic mean of a lt a 2 .? = 2=1 in (1).... we obtain the inequality .. and in particular the inequality (7)... indifferently).. ajPa2P and !.. If we write down the \n(nl} which can be formed with n numbers a a 2 lt ..gt..... / + a 8 a r as = (a r . and add the results.. (6). ...... In what follows a 1? a 2 note positive numbers (including zero) and jt?. Hence we can deduce an obvious extension of (3) which the reader may formulate for himself. 8.. .. an inequality which may also be written in the form + 2 2 p+g &..(?)/ (a... there are several greatest or least a s we and let O be their geometric mean.. .. ..0. inequalities of the type (1) Generalisations for n numbers. 9........ is may choose any replace We may suppose now we 0....gt. If a r and as by we do not alter the value of the geometric mean . When ........ which expresses the fact numbers is different that the mean of two positive not less than their geometric mean. 0. (4)&. since &.a 2 (a^ ... [l 7.gt. It is clear that we may repeat this argument until will is of a l .G) ... a n is not less than their geometric mean. we certainly do not increase the arithmetic mean. the initial we have replaced each be necessary. a n by G. .. /V + 2 =\ By repeated application of this formula we obtain and in particular &. the inequalities are merely af+af^Za&z..32 REAL VARIABLES .gt.. 2 ......... (1).
where then a = 0.gt. c = 0. where 17.a 8 8 where r and s assume the values 1. 16.. and where a. If a.. 2 . If !. & and ^4..  . then (Math. a+b *Jp+c */q+d J(pq}**0. etc. /3. n. and apply a. a are rational numbers which can easily be found.gt. .Jpq. Any polynomial in Jp and V^ with rational coefficients sum of a finite number are integers. .. 6. and A (Jp} rational). then 13. d=Q. 1909. c. a.2 (a r b .. 2 . in the form where A. and all of the first set are positive.. 6. c. The required 3 n. 2. etc. 33 a n and 6 1? 6 2 .lt..Jq} where m any and n a+b Jp + c V q + d &. Schwarz s inequality.. b. . .. .. Z?..REAL VARIABLES 10. C&. Express**.. c=0. K are two sets of numbers. d are rational. where a.) 12. c.. S.. 6. D are rational. 6 n are two sets of positive numbers.. 14. bn It is easy to verify the b r )\ identity (2a A) 2 =2 2 r 2a. It follows that an inequality usually known as Schwarz s (though due 11. . 2. then lies between the algebraically least and great ast of A. __ [Evidently Jq _ (a + b Jp + c Jq) (d+e Jpfjq) where a. then a = 0.e. If Jp Jq t are dissimilar surds. n and 6 15 b%. K.. 6 = 0. a n are all positive. originally to Cauchy). b. . d are rational.... [Express *Jp in the form the theorem of 1 5. ... can be expressed in the form of terms of the form A m (i. ... Trip. (7.] if Show that a *JZ + b ^/3 + c J5 = 0. c 14. arranged in descending order of magnitude. n ^ (&. + . a + e ~f .. /?. are rational. and *n = a + o2 + 1 . 6 = 0.. are rational numbers. . . M+ N Jq t where M and N are rational. If alf a 2 . Suppose that a lt are any two sets of numbers positive or negative...
c = 0.] a=0. Trip. (where a. Trip. c are rational. (Math.4/2 + 1. can be expressed in the form where a.. is . with rational coefficients.a &) 2 + 4 . this is already of the form required. 6 I) and (#2l)/(#2 + l) in the form [Multiply numerator and denominator of the second expression by 24. are rational and a = %p. b.gt.gt. 1899. then _ (a . j. c where are rational.1 and a r =a xm+s =p as of all powers of a higher than the (m l)th. Hence we can Express a.) If all the values of x and y given by a x* + 2k xy + b y2 = I ax 2 + Zhxy + by 2 = 1.b h] are both squares of rational numbers. x. where X get rid 23. 2wi 2 .v/3. with rational and that ^2 ^6 + 3 is the ratio of two linear functions of coefficients. 6.6 ).a h] (bh (h A )2 (6 (ah . (a& . A. 1. 21. (4/2 where c are rational. 1905. Show that any polynomial in #2. a &. ar be any power of a higher than the (m l)th. b.. m. and to 2 v/(a . b. 20.gt. 1 .REAL VARIABLES reduction [l may now denominator by eVjo. Trip. then .lt. be easily completed by multiplication of numerator and For example. y are rational numbers such that then either (i) x=a y=b t or (ii) lab and \xy are squares of rational numbers. . 19. For any such polynomial is of the where the Jc&.wi 2 ) if a &. an integer and 0^s&. (Math. Show x/2 + .gt. x . a h b are rational) are rational. ^4. if p is any rational number. m 2 . More generally. 5 = 0. 1903. 5. any polynomial m ^p with rational coefficients can be expressed in the form where a form . Then r=\m + s.) that N/2 and N /3 are cubic functions of N ^2 + x/3. (Math. prove that 18. let If Jc^ m If 1.a ) .) The expression is equal to 22. 2m if 2m 2 &. If a. If a.m 6 s are rational. .
b = e. /3 are rational. . or and D are both cubes of rational numbers. If neither a nor b is zero. j/B. are rational. Hence a6 = 0. we can divide the second equation by the first. then a =ai = . 22.REAL VARIABLES 35 = [Let y f/2. then either A = C. and it follows from the original equation that a. or JfA surd j/X. [We can where express the polynomial (x say) in the form l it m ly . 963 + a v//a6*\\ + v r~ v". 26.36)/ /a6*\ sf s/f 96 +^ is [Each of the numbers under a cube root of the form where a and 29. : Hence a6 = 2c2 a4 =46 2 c 2 As a It corollary. if then a = d.v/4 = d+e^/2+/&.26e = 0. as in Ex. with rational coefficients. If (aZ)3)6&. B=D. C is zero. which a 2 .(a2 . and c are all zero. 3 such that 28. Hence 2c 2 Then # 3 = 2 and + 26+a 3 =0 or we Multiplying these two quadratic equations by a and c and subtracting.] 25. B.2c 2 ).gt. = a m _ 1 = j but the proof is less simple.lt. that p not being a perfect mth power. 29 . If %A + $B + $C= 0. if a+&^2=Hc&. c=f. c = 0./4. and the other are rational multiples of the same $C Find rational numbers a. which gives a 3 =26 3 and this is impossible. 27. a rational is number. If A + $B=C+j/D. obtain (a62c 2 )y+a 2 26c=0. either one of A. . or y= . may be proved... any polynomial in a is the root of an equation of degree n.. The only alternative is that ab 2c 2 =0..] If a = Z/p. then 3f is rational. impossible. since ^/2 cannot be equal to the rational number a/6.o.26c)/(a6 .gt. then t two equal and opposite. 6. more generally.
. are called transcendental numbers. . a lt . n We have seen that some irrational numbers (such as x/2) are roots of equations of the type a^x \ } + a^xn ~ + l .xy 33. and 3 deduce the theorem of the equation 14. .6 p .. Show that 3 y = a + bp 1 + cp 2/3 satisfies 2 3 3 2 3 # . then x equations of the type l is an algebraical [We have n + l a 0&.... and eliminate x... 1. the minors of this process to l lt 1 2 .gt.c ^ + Zabcp = 0.a ... . (r = 0. x\y&. an Apply this process to the equation is x2 . + an = 0...+a mrir = r&. Such irrational numbers are called algenumbers: all other irrational numbers. 1 1 where a number.. such as TT ( 15).. a n are any algebraical numbers. Write x+y=z. which the coefficients between these and the original equation for in 35. .r m a r mr + ai tr a r r~ .gt. a n are integers. 31.+a n = 0.... + a m = 0. . then so are . then so are and x\y.. r&. by If . + L nxn = A. z.gt.y = zx in the second. a i. a 1} . ...] a n + a #*. If x and y are algebraical numbers.+ .6cjo) . Hence where A is L x + L%x2 + v . 30.36 Similarly REAL VARIABLES x2 =l2 + m 2 a+. 32. [The result ^ . +r 2 a (n ~ 1) .....xy.} Eliminate a . i m n .. + . Similarly for the other cases. n Z2 . where braical a^ . ..] Apply x=p+Jq.3a/+ 3y (a . [We have equations m m~ a$x + aiX 1 + .gt. the determinant mi . Algebraical numbers. ri). where the a s and 6 s are integers. ... We thus get an equation of similar form satisfied 34. .2x V2 + V3 0. kx where k t is any rational number. and x m n where . are integers x. m and n are any integers.r and Zi. Show that if x is an algebraical number.
(Afat/i.) be found that y 1 + y + = 0.] . where . 1903. satisfied 37. with rational coefficients. 38. 4 . 37 by Find equations. If # 3 = x + 1 then #3n = a Hx + bn + cjx. If #G +^5 2.r3 +tf 2 + l=0 1 and y = A4 ^2 4^ 1. Trip. then y satisfies a quadratic equation with rational [It will coefficients.REAL VARIABLES 36.
This notion of functional dependence of one variable upon another is perhaps the most important in the whole range of higher mathe matics. we proceed to do this. . which we may suppose to be repre sented geometrically by distances A^P = x. M. Q and y are also known. In these circumstances y is said to be a function of x. for example. In order to enable the reader to be certain that he understands it clearly. but by means of a geometrical construction which enables us to determine Q when P is known. the relation between x and y is expressed by means of (3) an analytical formula. in this chapter. or # 2 + l. but connected by a relation which we can imagine to be expressed as a relation between x and y: so that. when P and x are known. or y=2x. from which the value of y corresponding to a given value of x can be calculated by direct substitution of the latter. (1) (2) is viz. We might. And let us suppose that the positions of the points P and Q are not independent. or #. Or again. suppose that y = x. Suppose that x and y are 20. we might suppose that the relation between x and y is given. not by means of an explicit formula for y in terms of as. two continuous real variables.CHAPTER II FUNCTIONS OF HEAL VARIABLES The idea of a function. illustrate it by of examples.: determined for every value of x\ y to each value of x for which y is given corresponds one and only one value ofy. In all of these cases the value of x determines that of y. B^Q^y measured from fixed points A 0) B along two straight lines A. we must point out that the simple examples of functions mentioned above possess three characteristics which are by no means involved in the general But before idea of a function. means of a large we number shall.
In this case the functional relation makes the same value of y correspond to all values of x. be said at present Examples X. negative there is no value of y which satisfies the equation. but neither (1) nor (2). Boyle s experimental law asserts that the product of p and v is very nearly constant. only gives a reasonable approximation to the the gas is not compressed too much. Let v be the volume of the gas when the system is thus in equilibrium. That is to say. Then if x is positive this equation defines two values of y 3. All that is essential is that there should be some relation between x and y such that to some values of x at any rate correspond values of y. When v is decreased and p provided increased beyond a certain point. The same would be true were y equal to Such a function of x is called a constant. and the corresponding value of y (viz. viz. 0) is known. are sideration of the following examples will make it clear that they by no means essential to a function. This function therefore possesses the characteristic (3). The volume upon exact. S. per unit of area on the piston. for we can give x any value. corresponding to each value of #. Let #=0 whatever be the value of x. Nothing further need 2. Hence to the Jx. of 0. The p its weight. Let y =x or 2# or \x or x2 + 1 about cases such as these. a correspondence which. be the area of the cross section of the piston and held in a state of compression by the piston. It is known that a borrow this instructive example from Prof. would be represented by an equation of the type if pv=a where a is (i). 1. y0.of x corresponds one and only one value of y. upon the piston the latter is forced downwards. so that Let A W W=A Po If additional weight is placed . the function y is not defined for negative values of x. Then y is a function of x. which balances the weight TF. H. 4. exerts a certain pressure gas. a number which can be determined approximately by experiment. If #=0. But if x is particular value 0. which it exerts (v) of the gas diminishes . Carslaw s Introduction to the Calculus. however. Consider a volume of gas maintained at a constant temperature and contained in a cylinder closed by a sliding piston*. 1 or \ or V2 instead Let y = x. the relation between them is no longer expressed with tolerable exactness * I by the equation (i). .20] FUNCTIONS OF REAL VARIABLES 39 It is indeed the case that these particular characteristics are But the con possessed by many of the most important functions. facts Boyle s law. the pressure (p) unit area of the piston increases.
For p is not denned as . Thus y = 7 if x . And value of This function possesses the characteristic (2) to any value of v only one p corresponds but it does not possess (1). rt &. (i) do not give anything This relation is no form changes. but not terms of x. viz. where a. h being the depth of the ball. y are numbers which can also be determined approximately by experiment. and rebounds continually. from a mathematical point of view. there is nothing to prevent us from con templating an ideal state of things in which. the denominator of ^2 being a meaningless form of words.3r. as v varies. ** is 6. show that h = gt if h=\g (2rt) 2 if &.r.(2w original position.lt. if x is rational.t &. and so our defining relation fails to define for such values of x as these. Suppose that y is denned as being the largest prime factor of x. (3). generally if (2n l)r&. at time t. integral values. 5. for all values of v not less than a certain value V. + l)T. values of v less than V. and so negative values of v do not present themselves for considera tion at all.11/7 but y is not defined for # =^2.40 FUNCTIONS OF HEAL VARIABLES [ll much better approximation to the true relation can then be found of what is known as van der Waals law the . a function The ordinary formulae of elementary dynamics. with which the reader is 2 and probably familiar. complete account of the relation between p and doubt in reality much more complicated. below its Obviously A is a function of t which is only denned for positive values of t. and (ii) exactly true for all then we might regard the two equations as It is an example of a function which together denning p as a function of for some values of v is denned by one formula and for other values of v is denned by another.lt. : Let y be defined as the denominator of x when x is expressed in its This is an example of a function which is defined if and only .lt.lt. But. as there is no simple formula which expresses y in lowest terms. Of course the two equations. Suppose that a perfectly elastic ball is dropped (without rotation) from a height \gr l on to a fixed horizontal plane.lt.&. /3. even taken like a together. and its from a for ^ nearly equivalent to (i) to a form nearly equivalent to (ii). would be exactly true. This an instance of a definition which only applies to a particular class of values * of x. It does * possess (2). 7. The largest prime factor of J3 or of N/2 or of TT means nothing. expressed by means by equation (ii). . : of v for negative values of vj a negative volume means nothing. v. Thus this function does not possess the characteristic (1).
m. we shall express it by writing not y/() (or F (x}. 2. to sign. The nature illustrated and of particular functions may. 1. at 5. Then y is defined for a certain number of integral values of x. Y at right angles to one another and produced indefinitely in both direc tions.. OY respectively. Metropolitan Police. P P". . and complete the OAPB. a and b the coordinates of a the abscissa. as the case may be). If to the value a of x correspond several values of y (say b. 21] FUNCTIONS OF REAL VARIABLES 41 Let y be defined as the height in inches of policeman Cx. Take OA = a. Imagine the point rectangle P marked on the diagram. . or together the OY P P . regard being paid. .30 p. OB = b. in the 8. we have. where is the total number of policemen in division C at that particular moment of time. OX Y . in virtue of the functional relation between x and y. be Draw two lines X.lt. N 21. b". &. Fig.. 1907.. But for the present we shall shall look at this relation from the first point of view. b) . . and the positive directions of measure ment being in Fig.20. b the ordinate of referred to the axes OX. instead of the single point P. 6. We can represent values of x and y by distances measured from along the lines OX. This marking of the point P may be regarded as showing that the value of y for x = a is b. many cases. . . those indicated by arrows Let a be any value of x for which y is defined and has (let us suppose) the single value b. viz. generally be open to us also to regard x as a function of y. number of points P. in very made easily intelligible as follows. i/r (x). call x the independent variable and y the dependent variable relation is . of course. when the particular form of the functional specified. N. a . and the axis of x and the axis of y. b . is Sup It will pose that the variable y a function of the variable x. The graphical representation of functions. We shall call P the point (a.j) (x).}. We then and. . on 8 Aug.
. C are any fixed numbers*. It is equally easy to prove the converse sents a straight line. function of x defined for one value only of x.. the locus of the point (x. for which y is and the corre ..C/A. The equation Ax + By + G = degree. (1).. and may be applied whenever the relation between x and y is capable of being represented by an ... b".. The reader is to show that the graph of y is a straight line.. We call the aggregate of all these points the graph is of the function y.. To take a very simple example. y) is a straight line. the origin of coordinates.) . is the general equation of the first the most general polynomial in x and y which does not involve any terms of degree higher than the first Hence the general equation of the first degree repre in x and y..... We in such a way that equation (1) is always true.... and the origin. P P".. and say that the equation We shall shall say that when x and y vary This use of the terms locus equation of represents the locus.. .. and we shall call (1) the equation of the Zocws. P (or points P....). defined. x * We must then regard y as a .. have been determined.42 FUNCTIONS OF REAL VARIABLES [ll axes of coordinates. We may metrical loci defined mention a few further examples of interesting geo by equations.. where A. a of x of y.. (2). for Ax + By + C is proposition that the equation of any straight line is of the first degree. easy in all probability familiar with one or other of the various proofs of this proposition which are given in textbooks of Analytical Geometry. analytical formula. the locus is quite general. the value b (or values b t . sometimes use another mode of expression.. suppose that y a function of x by the equation defined as Ax + By + C = . viz. y does not occur in the equation. or simply Let us now suppose that sponding point for all values b. (3) of 20. and then having all values.. An equation of the form If Z? = 0. Then y is a function of x It is which possesses all the characteristics (1). B...
21. (the general equation of the second degree) represents.2rc cos (6 . MOP = 0. a conic section.gt. TT) correspond to the same point. Polar coordinates. where According to our definitions r and therefore cos0&. the other having the equation l\r = \e cos 0. If o is * Fi g7. an ellipse.a) = A where A. .0) and (1. e&. The distinction between the two systems may be illustrated by means of the equation lfr = l must be of of ecos0. these loci 22. the equation represents the whole Z&. it should be observed. essentially positive*. 22] FUNCTIONS OF REAL VARIABLES #2 + 2 43 or 2/ + 2Gx+ 2%f (7 = 0. (1. P moves on a locus there will be some relation between r and =/(0) or = F(r). MP = y. 0. The polar equation may be deduced from the (x. circle passing circle is of through the origin and the general equation of 2 . we must refer to books For further discussion of on Analytical Geometry. In this case two pairs of coordinates e. y) equation (or vice versa) by means of the formulae above. 6. Thus the polar equation where a a of a straight line is of the form rcos(# p and a are constants.l/e: the equation represents one branch only a hyperbola. positive hyperbola. 6.l. Polar coordinates are sometimes defined so that r may be positive or negative. * c. + c 2 . The equation where G +P 2 2 C &. or hyperbola. it is evident that = r cos 0.0.gt. M The say r former. represents a circle. r = \/(# 2 + x y 2 2/ ). i. With the system coordinates which admits negative values of r.gt. assuming that the coefficients satisfy certain inequalities. 6 being an r sin : : and 2?r (measured in angle between the positive direction). In what precedes we have determined the position of If OP = r and P by the lengths of its coordinates OM=x. the form r2 a)=p. and that the position of P is equally well determined by a knowledge of r and 6. This we call the polar equation of the locus. The equation r = 2a cos 6 represents . parabola.lt. We call r and 6 the polar coordinates of P.g.e. : : cos 6 sin 6 1 x y : : r. and a are constants.
. 8. in fact.. give = J. 4. a m are constants. A polynomial in x 1 is a function of the form a Q xm + a^. however. . .+ .+ am .0) values. the simple powers y x. 3. according as First let (1... Polynomials. where a . If the reader will plot off a fair number of points on the graph.. .i. 3 9. ( The simplest polynomials are The graph of the function m is even or odd. Then three Any number of may be found by values points on the graph are (0. . he will be led to conjecture that the form of the graph like that is something If shown in Fig. There is..44 23. 0). m .. reader a better notion of the infinite variety of possible types of functions. 1).9. additional points on the graph other special values to x\ thus the assigning tf =i 2. 8. and calculating the cor responding values of y. he draws a curve through the special points which he has proved to lie on the graph its accuracy and then tests by giving x new (0. of course a parabola. 2. x xm is of two distinct types. 1). he will find that they lie as near to the curve as Fig. 8. a 1} . 1. a?. one fundamental question which we cannot answer adequately at present. This cannot be proved by merely y . The reader has no doubt some notion as to what is meant by a continuous curvej_a_c_urve without breaks or jumps such a curve. m = 2. . The inevitable inaccuracies of drawing are considered. when the curve is it is reasonable to expect. a?. = #2 is in fact such a curve. 4.. A. as is roughly represented The question is whether the graph of the function in Fig. FUNCTIONS OF KEAL VARIABLES [n Further examples of functions and their graphical The examples which follow will give the representation..
and y xm where m is even and greater than 4. In that chapter we shall consider in detail what our common sense idea of continuity really means. until the curve is practically indis the figure. And so that the chord lies entirely above the curve. The reader should next consider the curves given by y=xm when m is odd. The fundamental difference between the two m so that the cases is that whereas when m is even ( x) m = x xm so curve is symmetrical about OF. V)a point on the chord PQ Pi are # X# +/i#i.gt. PI (Fig. are really continuous curves. V. Then the coordinates of Let PO. As m gets larger and larger the flatness and steepness grow more and more pronounced. is similar to o Fig. . This question cannot be discussed properly until Ch. . A (Fig. For the present the reader may be content to draw his curves as common sense dictates.23] FUNCTIONS OF REAL VARIABLES 45 constructing any number of isolated points on the curve. y=X# 2 f/*#i 2 where X and p. positive numbers whose sum is 1. and steeper beyond the points A. 8) be the points (xQt (a?i. Fig. tinguishable from the thick line in The curve y = x4 . although the more such points we construct the more probable it will appear. 9. and how we can prove that such graphs as the one now considered. and others which we shall consider later on in this chapter. is everywhere convex to the axis of x. y = x* in general appearance. 9). 10. are It is easy to see that the curve y=# 2 x\ &. when m is odd ( x) m = y . is still more so. but flatter near 0.
we can find that of f(x) to be the + F(x) by taking the ordinate of every point of the ordinates of the corresponding points on the two original curves. 10 shows the curves and the form to which y larger odd values of m x m approximates for It is now easy to see how (theoretically at any rate) the graph of any polynomial may be constructed. &.000. as x gets larger and polynomial as realise larger. [The reader should draw the curves carefully.000. where C is a constant. from the graph of y = xm we can at once derive that of Cxm. in order to prevent the figure becoming of an awkward size. Thus even when #=4. by multiplying the ordinate of every point of the curve by C. 1. Draw the graph of ax* + Zbx + c {(ac [Here y old . of y a be found convenient to take the scale of measurement along the axis good deal smaller than that along the axis of x. If we take new axes b 2 )/a.000.000. it importance when x is fairly is the first term which if is of really preponderant a? large. in such a 10 (or even # f SO^+VOCte*). 1. = (ac b/a. 12 .000.] #=10 the preponderance Compare the relative magnitudes of x when #=1.000 and 700^?4 180.62 )/a} =a {x + (bja}} 2 .] 3. He will then of x increase. which will be later on. 2.lt. In the first place. sum The drawing of graphs of polynomials is however so much by the use of more advanced methods. This idea of the relative rate of growth of different functions of x is one with which we shall often be concerned in the following chapters.000^ [The reader should make up a number of examples of this type for himself.000. 100. the parallel to the and passing through the point x= 2 The curve is a parabola. 1.000^. It will y=#2 (#l). 10. 1.lt. Trace the curves 3/=7#4 . that we shall not pursue the subject further explained facilitated here. 10 &. y=x(x\} z . and all three should be drawn in one figure*.gt. while 30^&. how and rapidly the higher powers will see that. Fig. * . negative. etc. l . y=3#5 y=x . y = a?. y new equation Trace the curves y =#3 3# + l. And if we know the graphs of f(x) and F (#). . Examples XI. 35. while of the first term is still more marked.000.46 that y is FUNCTIONS OF REAL VARIABLES negative [ll when x is y = x.] isy = ax 4.
The following points concerning the definition should be noticed. to the form a rational function of the standard form..+k. 2 all lead to a meaningless expression.e.. 47 which Rational Functions. The same is true of the reduced form. and so the function is not defined for these values. is which noticed that the reduction But. In order to calculate the value of a function for a given value of x we must substitute the value for x in the function in the form in which it is In the case of this function given.1 and 1 are concerned. 1 if 1. On removing the common factor x we obtain It should a But the equal to therefore differs from the function is only so long as # =f=0. we may denote the general rational function by A : P reduces to unity or any other reduces to a polynomial x). (4) there will generally be a certain number of values of x for which the function is not defined even when it has been reduced to a rational function of the standard form.23. (2) We P does as which 1/1 = 1. thus the class of rational functions includes that of polynomials as a subclass. . x^Q and is is undefined when #=0. Thus (^2 7)/(^ 2 3^ + 2) is not defined when # = 1 2. 0. The class of functions ranks next to that of polynomials in simplicity and importance is that of rational functions. as appears from the particular example considered under (3). 1. Consider for example the function #/#. Thus the function xjx original function is not always equal to 1 : it is equal to 1 If # it takes the form 0/0. But x = and x = 2 give the value 0. the values x= 1. is a rational function. by the ordinary rules of algebra. which is = meaningless. (x} In the particular case when Q (i. so far as the values . But here again it must be is not always legitimate. 24] 24. all such factors being removed by division. however be observed that this removal of common factors rule change the function. Thus once more the two functions are not the same. It which always equal to (3) Such a function as may be reduced. constant does not involve R (x) : (1) usually suppose that (x} and Q (x) have no common factor x +a or xv + axv~ l + bxv* + . or These are the values of x (if any) for which the de nominator vanishes. FUNCTIONS OF REAL VARIABLES B. rational function is the quotient of one polynomial by another thus if (x). Q (x) are polynomials. 1.
including^on^mjiltiplic^tipn of The rational function x by a finite number of operations upon x by itself or a constant. . and to reduce our function to the standard form of rational function. . 1. which can be obtained in this way is sufficiently illustrated by the example 2 which can obviously be reduced to the standard type of rational function. by another.e. any function which can be deduced from x by the elementary functions operations mentioned above. obtained by such multiplications and additions. y = I/a It . a procedure exemplified in the equation 3~ 1+1+1 Again. or quotient of two rational functions may themselves be reduced to rational functions of the standard type. where P and Q are polynomials. to disregard the exceptional values of x for which such processes of simplification as were used there are illegitimate. and division of one function. can be reduced to The most general kind of function the standard type of rational function. if in z = P(y)IQ(y}&. even more 25. In so far as the variable x is concerned. (6) It is in no way presupposed in the definition of a rational function that the constants which occur as coefficients should be rational numbers. in dealing with expressions such as those con sidered in (2) and (3). Draw the graphs ofy=l/x t y= I/a8 . should be # = 0. And generally a rational function of a rational function is itself a rational function: i.. shall therefore content ourselves at present with a very few examples.48 FUNCTIONS OF REAL VARIABLES [ll (5) Generally we agree. The drawing of graphs of rational functions. The word rational has reference solely to the way in which the variable x appears in the function.] . this pro cedure is very much like that by which all rational numbers can be obtained from unity. Thus is a rational function The use of the word rational P (x)IQ(x] may be generated from a. We 3 . is immensely facilitated by the use of methods depending upon the Examples XII. since 1/0. are meaningless expressions. The reader will easily verify that (on this understanding) the sum. using at each stage of the process which have already been obtained from x in the same way.. ajddition of terms arises as follows. . we substitute y PiWIQiW. than that of polynomials. we obtain on simplification an = equation of the form z P2 (x)/Q2 (x ). . thus obtained. I/O 2.gt. differential calculus. product. [The figures show the graphs of the first observed that. these functions are not defined for two curves.
26.lt. an ambiguity of notation We have. Thus 7T are explicit algebraical functions. 5. Trace y = \l(xa)(xb\ !/(#) (xb) (xc\ where a&. for a and 3. Fig.2426] 2.e. Explicit Algebraical Functions. #(!/*). positive and negative. 12. and n are any integers.) \/2 as denoting the positive square root of 2. where x is any n. number These are functions which can be generated from # by a finite of operations such as those used in generating rational functions. yat m where ) f m be noticed that there is involved in such an equation as y = Jx. FUNCTIONS OF REAL VARIABLES 2 2 y =#+(!/#).b&.lt. up to the present. Trace 4. The next im portant class of functions is that of explicit algebraical functions. ^gether^jwSB) a finite number of operations of root ]] extraction. taking various values. 49 Trace #2 (l/#2 ) and b. and it would be natural to denote by \/#. 4 . m is Fig. Sketch the general form assumed by the curves y = \lxm as becomes larger and larger. 11.g. # + (l/# ). regarded (e. It should and so is xmln (i. considering separately the cases in which m odd or even.c. C.
. negative values). Rm and the reader will easily verify that.. has generally tyx.. when this course is adopted.(#2 /a 2 )}. viz. The reader will observe that. Examples XIII. If 2. all Secondly the function.e. easy to verify that if J or if y 4 = V# 2 a? then 2/ .. are rational functions of x\ where R R 1} 2..b it V{(#..+R m =Q . a&..)(&#)}.... It is 27.. is defined only for has two values if x = a or b only one.x&. Each of these equations y may be expressed in the form + E iy ^r. if y is in the last set of examples. for all values of But for \jx is x with a certain number of isolated exceptions.. (1).. onevalued and defined for all values of x.lt.b. Draw D.. the positive square root of a?. when x has a value defined. any one of the functions considered y satisfies an equation of this form.. . x b. which it is two values of opposite is signs.(4i/ + 4y + 1) = 0...lt. the graphs of the functions y= i *j(&.#x \ y=bj{l .. 0.lt.lt.lt. undefined for a whole range of values of x (i. : Consider similarly 3. It is however more^ convenient to regard *Jx as standing for the twovalued function whose two values are the positive and negative square often roots of x. 1... 4.. a &. Implicit Algebraical Functions. the function \jx differs fundamentally from rational functions in two In the first place a rational function is always defined respects. The function on the other hand. Trace the curves y*=x. y =&. 2 y 2 =x3 .. . in which case would be a onevalued function of x. where a&.50 positive FUNCTIONS OF HEAL VARIABLES [ll y ^Jx number.
gt. known always possible in special cases for higher values of m.* (. w between these equations in order to obtain an equation of the form desired. 1. 2. 27] FUNCTIONS OF REAL VARIABLES 51 is naturally suggested that the same is true of any explicit And this is in fact true.gt. Examples XIV. difficult to prove.e. It has two 2 1 &. Exs. though such a solution is if m m= 1. 32). _x \u\v\w x u +v w v* = x + u. and we have only to eliminate u v. though we shall not delay to write out a formal It proof here. the root of an equation of the th x. whose coefficients are rational functions : We yf( x m is ) of plainly no loss of generality in supposing the first coefficient to be unity. is the equation m = l. and we can generally study the properties of the function better by means of the original equation. we can use the methods explained in treatises on Algebra for the solution of cubic and biquadratic But as a rule the process is equations. 2. i. This class of functions includes functions considered in 26. But it all the explicit algebraical also includes other functions which cannot be expressed as it is explicit algebraical functions. are therefore led to give the following definition a. or 4 and The definition of an algebraical function should be compared with that of an algebraical number given in the last chapter (Misc. For that in general such an equation as (1) cannot be solved explicitly for y in terms of x.# 1 2 =4 ft2 J . complicated and the results inconvenient in form.4:R 2 and one if . is f + R^ + R = 0. y is a rational function. and indeed not algebraic function. function will be said to be an algebraical function of x if it is the root of an equation such as (1). when is greater than 4. An example should make clear to the reader the on which such a proof would proceed.4/? 2 . If If m = 2. 42 . ". 3. If 7/1 = 3 or 4.26. Let x + \/x lines Then we have the equations y u^ x. \. so that 2 This function values if defined for all values of x for which R R? &. There degree in y. t w =1+ 3 oo.
52
3.
FUNCTIONS OF HEAL VARIABLES
Consider the functions denned by the equations
[ll
in each case obtaining
y
as an explicit function of
x and
t
stating for
what
values of
4.
x
it is
denned.
coefficients rational in #, satisfied
Find algebraical equations, with
by
each of the functions
5.
Consider the equation
y*=x z
all
.
[Here
y*=x.
If
a?
is positive,
y=\/x\
if
negative,
?/
=V #
(
).
Thus the
an
function has two values for
values of
x save #=0.]
is itself
An algebraical function of 6. algebraical function of x.
[For
an algebraical function of x
we have
where
Eliminating
we
find
(^)^i +...+,$; an equation of the form
*+,
Here
all
the capital letters denote rational functions.]
An example should perhaps be given of an algebraical function which 7. cannot be expressed in an explicit algebraical form. Such an example is the function y defined by the equation yytf = 0.
But the proof that we cannot find an explicit algebraical expression terms of x is difficult, and cannot be attempted here.
28.
for
y
in
Transcendental functions. All functions of x which not rational or even algebraical are called transcendental This class of functions, being defined in so functions. purely a manner, naturally includes an infinite variety of \vhole negative
are
Among
E.
kinds of functions of varying degrees of simplicity and importance. these we can at present distinguish two kinds which are
particularly interesting.
The
direct
and inverse trigonometrical or circular
functions.
These are the sine and cosine functions of elementary
trigonometry, and their inverses, and the functions derived from them. may assume provisionally that the reader is familiar
We
with their most important properties *.
*
The
suppose that any sector of a circle has associated with
area.
definitions of the circular functions given in elementary trigonometry pre it a definite number called its
this
How
assumption
is justified
will appear in Ch. VII.
27, 28]
FUNCTIONS OF REAL VARIABLES
1.
53
Examples XV.
Draw
the graphs of cos x, sin #, and a cos x + b sin x.
)&gt;
2 2 = and a is an angle [Since a cos x + b sin x /3 cos (x a), where /3=V(a + & whose cosine and sine are a/*J(a? + b 2 ) and 6/v/(a2 + & 2 ), the graphs of these three functions are similar in character.]
2.
Draw
the graphs of cos 2 x, sin 2 x, a cos 2 x + b sin 2 x.
3.
Suppose the graphs of f(x] and /*(#) drawn.
Then the graph
of
/ (x) cos 2 x + F(x) sin 2 #
is
a wavy curve which oscillates between the curves # =/(#), y = F(x).
Draw
the graph
4.
when / (x) = #,
the
.F (a?)
= x\
of
&lt;?)
2 cos % (p
Show that  x and q)
graph
is
cospx+cosqx
#,
lies
2 cos ^ (p +
touching each in turn.
between those of Sketch the
graph when (pq}l(p + q)
5.
small.
sin x,
(Math. Trip. 1908.)
Draw Draw
the graphs of
x+
(l/#)+sm#, #sin#, (sin#)/#.
6.
the graph of sin
?/
(I/a;)
= sin (l/#), then = when # = I/WTT, where m is any integer. Similarly = \ when d?=l/(2fH4*^)* an(i .y = ~l when #=l/(2m ) ?r. The curve is y 1 and y=l (Fig. 13). It oscillates entirely comprised between the lines y=
[If y
x approaches small*. The
half.]
7.
up and down, the rapidity of the oscillations becoming greater and greater as 0. For #=0 the function is undefined. When x is large y is
negative half of the curve
is
similar in character to the positive
Draw
the graph of
x sin (l/.
V.H V
y=
1
[This curve is comprised between the lines last curve is comprised between the lines y=
 x and and y=l (Fig.
V*
yx just as the
14).]
Fig. 13.
Fig. 14.
* See Chs.
IV and
V
for explanations as to the precise
meaning
of this phrase.
FUNCTIONS OF REAL VARIABLES
8.
[ll
Draw
the graphs of
^2 sin(l/^),
(1/j;),
a cos 2 (I/a?) + 6 sin 2 (l/#), sin^ + sin
9.
(l/#)sin(l/#), sin*(l/#), sin # sin (I/a?).
, .
{asm
(lfx)}*9
Draw
Draw
the graphs of cos #2
,
sin
# 2 a cos x2 + b sin #2
10.
the graphs of arc cos a? and arc sin x.
This enables us to draw the [If #=arccostf, # = cos#. graph of a?, con sidered as a function of y, and the same curve shows y as a function of x. It is clear that y is only denned for 1 and is
&lt;*&lt;!,
valued for these values of
infinitely
x.
when of
!&lt;#&lt;!,
y
are given
a value of y by the formula 2^7ra, where n
the reader no doubt remembers, there is, between and TT, say a, and the other values
is
As
many
any
integer, positive or
negative.]
11.
Draw
tan
a?,
the graphs of
cot
x
t
sec
a;,
coseca?,
tan 2 a?,
cot 2 a?,
sec 2 a?,
cosec a a?.
12. Draw the graphs of arc tan a?, arc cot x, arc sec a?, arccosec^. Give formulae (as in Ex. 10) expressing all the values of each of these functions in terms of any particular value.
13.
Draw
Show
the graphs of tan
(I/a?),
cot
(I/a?),
sec
(I/a?),
cosec
(I/a?)
14.
that cos a? and sin a? are not rational functions of
is
x.
said to be periodic, with period a, if f(x)=f(x + a) for all values of a? for which f(x) is defined. Thus cos^ and sin x have the period 2 TT. It is easy to see that no periodic function can be a rational function, unless it is a constant. For suppose that
/(*)&gt;(*)#(*),
[A function
where
is
P and Q are polynomials, and that /(a?) =/(#+), each of these equations holding for all values of x. Let /(O) =y&. Then the equation P (^)  @ (#) = o
satisfied
by an
infinite
therefore for all values of x.
number of values of Thus f(x) = k for
x, viz.
all
#=0,
a, 2a, etc.,
i.e.
and
is
values of x,
/(#)
a
constant
15.
]
V\lft**V
H^
Show, more generally, that no function with a period can be an
algebraical function of x.
[Let the equation which defines the algebraical function bo
*/
+ ,%
i
+ ...+ m =0 ........................... (1)
This
where
Rlt
...
are rational functions of x.
may
be put in the form
where
P P
,
lt
...
are polynomials in x.
Arguing as above, we see that
28, 29]
for all values of x.
FUNCTIONS OF REAL VARIABLES
Hence y
55
and one
set of values of
k satisfies the equation (1) for all values of #, our algebraical function reduces to a constant.
Now divide (1) by yk and repeat the argument. Our final conclusion is that our algebraical function has, for any value of x, the same set of values ... i.e. it is composed of a certain number of constants.] &,
,
;
16.
The
inverse sine and inverse cosine are not rational or algebraical
functions.
and +
1,
[This follows from the fact that, for any value of arc sin x and arc cos x have infinitely many values.]
x between 
1
in
Other classes of transcendental functions. Next importance to the trigonometrical functions come the expo
29.
F.
nential
and logarithmic
functions,
which
will
be discussed in
IX and X. But these functions are beyond our range at And most of the other classes of transcendental func present.
Chs.
tions
whose properties have been studied, such as the elliptic functions, Bessel s and Legendre s functions, Gammafunctions,
and
so forth, lie altogether beyond the scope of this book. There are however some elementary types of functions which, though of much less importance theoretically than the rational,
are particularly instruc algebraical, or trigonometrical functions, varieties of the functional tive as illustrations of the possible
relation.
Examples XVI. 1. Let y \x\ where \x\ denotes the greatest integer not greater than x. The graph is shown in Fig. 15 a. The lefthand end the righthand ones, belong to the graph. points of the thick lines, but not
2.
y=x[x\.
(Fig. 156.)
y

\

I
,cVc
K
;
b*
Fig. 15 a.
Fig. 15
&.
FUNCTIONS OF REAL VARIABLES
}.
[n
(Fig. 15 d.)
(Fig. 15
e.)
4.
y=
y
*l

i
\
\^
6.
Fig. 15 c.
7.
Fig. 15 d.
Let y be defined as the largest prime factor of x Then y is denned only for integral values of x. .If
tf=l,
2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, ...
,
(cf.
Exs.
x. 6).
then
y = 1)2j3)2)5&gt;3)7)2j3)
consists of a
the
5jHj
3&gt;13&gt;
The graph
8.
number
of isolated points.
and there are no points corresponding
denominator of x (Exs. x. 7). In this case y is defined x. We can mark off as many points on the graph as we please, but the result is not in any ordinary sense of the word a
only for rational values of
curve,
to
Let y be
any
irrational values of x.
straight joining points (N 1, JV) t (tft ^V), where is a Show that the number of points of the locus which lie on positive integer. this line is equal to the number of positive integers less than and prime to N.
D.
Draw
the
line
the
N
Let y =
is
when x
is
an
integer,
graph
derived from the straight line
y=x when x is not an integer. y = x by taking but the points
(1,1),
The
...(1, 1),
(0,0),
(2,2),...
and adding the points (1,
0), (0, 0), (1, 0), ...
on the axis of x.
The reader may possibly regard this as an unreasonable function. Why, he may ask, if y is equal to x for all values of x save integral values, should it not be equal to x for integral values too I The answer is simply, why should
itl
The function y does
:
function
there
is
known. We are perfectly at liberty to take this relation to be what we please, however arbitrary and apparently futile. This function y is, of course, a quite different function from that one which is always equal to #, whatever value,
integral or otherwise,
in point of fact answer to the definition of a a relation between x and y such that when x is known y is
x may
have.
29]
10.
FUNCTIONS OF REAL VARIABLES
Let y = 1 when x
is rational,
57
but y when x is irrational. The graph arranged upon the lines y=l and y Q. To the eye it is not distinguishable from two continuous straight lines, but in reality an infinite number of points are missing from each line.
consists of
two
series of points
11.
Let
y=#
when x
is
irrational
and y = J{(l
2
fp )/(l
+ g 2 )} when x is
a
rational fraction pfq.
Fig. 16.
The
irrational values of
x
contribute to the graph a curve in reality dis
continuous, but apparently not to be distinguished from the straight line
y=&.
Now
2
consider the rational values of x.
First let
x be
i.e.
positive.
\/{(lfp )/(l
+
2
&lt;?
)}
cannot be equal to p/q unless p = q,
x=\.
Then Thus all
except
the points which correspond to rational values of x lie off the the one point (1, 1). Again, if p&lt;q, \/{(l+p 2 )/(l+? 2 )}
&gt;
line,
5
if
pl&lt;l
P
&gt;
&
V/{(1
+p 2 )/(l + q 2 )}
&gt;
&lt;p[q.
Thus the points
lie
above the line
2 2
y=#ifO&lt;#&lt;l,
below if x
1. If p and q are large, /{(! +P )/(1 + 2 )} i s nearly equal to p[q. Near any value of x we can find any number of rational fractions with large numerators and denominators. Hence the graph contains a large number of
points which crowd round the line y =x. values of x) is that of a line surrounded
by a swarm
Its general appearance (for positive of isolated points which
gets denser and denser as the points approach the line.
The part of the graph which corresponds to negative values of x consists of the rest of the discontinuous line together with the reflections of all these
isolated points in the axis of y.
Thus
to the left of the axis of
x,
y the swarm
part of the
of points is not
round y=oc but round
16.
y=
which
is
not
itself
graph.
See Fig.
58
30.
FUNCTIONS OF REAL VARIABLES
[ll
Graphical solution of equations containing a single
unknown number.
form
Many
equations can be expressed in the
/()*(*) ........................... a).
where/ (x) and
&lt;f&gt;
(x) are functions
whose graphs are easy
to draw.
And
if
the curves
intersect in a point
P whose abscissa
1.
is
,
then f
is
a root of the
equation
(1).
Examples XVII.
may be
solved graphically
The quadratic equation ax 2 + 2bx+c=0. This For instance we may draw in a variety of ways.
the graphs of
whose
intersections, if any, give the roots.
Or we may take
y = a?
t
y=(2bx + c)/a.
is
But the most elementary method
probably to draw the circle
whose centre
The abscissae of is (6/a, 0) and radius y(b*ac)}/a. with the axis of x are the roots of the equation. intersections
2.
its
Solve by any of these methods
3.
the
The equation xm +ax+b=0. This may be m curves Verify the following
yx
^
y=axb.
solved by constructing table for the number of
roots of
b positive, two or none,
(0
/TA (b)
jj m odd
(
I
a
, positive, one,
,
.
7
or
Construct numerical examples to illustrate all possible cases.
4.
Show
that the equation tan
#=# + & has always an
sin^ = ^,
(e.g.
infinite
number
of roots.
N^
5.
Determine the number of roots of
sin#:=tf,
sin
v
^\ir^^6.
x = $x
t
Binx = ^x.
a = 01), the equation
Show
that
if
a
is
small and positive
x
has three roots.
Explain how
r
a=
7r
sn 2
is
Consider also the case in which a
small and negative.
the
number

of roots varies as a varies.
K
^V
S, X
(\
IA
",
X
11
*
^
^.
w.
K
*j
\
^
A
30, 31]
31.
FUNCTIONS OF REAL VARIABLES
59
Functions of two variables and their graphical representation. In 20 we considered two variables connected
by a relation. We may similarly consider three variables (x, y, and z) connected by a relation such that when the values of x and known. In this case y are both given, the value or values of z are the two variables x and y x and y the we call z a function of and we express independent variables, z the dependent variable;
;
this
dependence of z upon x and y by writing
*/(*. *)
The remarks of 20 may more complicated case. The method
is
all
be applied, mutatis mutandis, to this
graphically must take three axes, OX, OF, functions of a single variable. OZ in space of three dimensions, each axis being perpendicular The point (a, b, c) is the point whose distances to the other two.
of representing such functions of two variables of exactly the same in principle as in the case
We
from the planes YOZ, OZ, are a, 6, and c.
lengths measured in
parallel to OX, OF, must of course be paid to sign, Regard the directions OX, OF, OZ being regarded
ZOX, XOY, measured
The definitions of coordinates, axes, origin are the as positive. same as before.
Now
let
z=f(x,y).
As x and y
The will move in space. vary, the point (x, y, z) of all the positions it assumes is called the locus of the aggregate point (x, y, z) or the graph of the function z =/(#, y). When the
relation
between
x, y,
and z which defines z can be expressed in an
the equation of the locus. analytical formula, this formula is called is easy to show, for example, that the equation It
(the general equation of the first degree) represents a plane, and The equation that the equation of any plane is of this form.
or
x+
F + 6r
2 2
7/
z
x
&gt;
Gy + 2Hz + (7=0,
;
where
f
H
2
C
0,
represents a sphere
and so
on.
For
proofs of these propositions
we must again
refer to textbooks of
Analytical Geometry.
60
32.
FUNCTIONS OF REAL VARIABLES
[ll
Curves in a plane.
We have hitherto used
/(&gt;
the notation
a)
It is evident that to express functional dependence of y upon x. this notation is most appropriate in the case in which y is ex
pressed explicitly in terms of x by means of a formula, as
for
when
example
y
= x~,
sin x,
a cos 2 x
+
2 b sin x.
We
which
If,
have however very often
it is
to deal
with functional relations
5
impossible or inconvenient to express in this form.
5
to
ay 0, it is known be impossible to express y explicitly as an algebraical function
for
example, y
y
x=
or
a?
+y
of x.
If
a?
+
2 7/
+ 2Gx + 2Fy + C = 0,
by the formula
C)
is
;
y can indeed be
so expressed, viz.
y=
F+ V(^
2
 &  2 Gx 
but the functional dependence of y upon x
simply expressed by the original equation.
It will be observed that in these
is
better and more
two cases the functional fully expressed by equating a function of the two variables x and y to zero, i.e. by means of an equation
relation
/(*,y)
=o
(2).
adopt this equation as the standard method of the functional relation. It includes the equation (1) expressing as a special case, since yf(%) is a special form of a function of x
shall
We
and
y.
We
to /(a?, y)
can then speak of the locus of the point (x y) subject 0, 0, the graph of the function y denned by f(x, y)
t
the curve or locus
locus.
f(x
y
y}
0,
and the equation of
this curve or
There
useful.
another method of representing curves which is often Suppose that x and y are both functions of a third
is
variable
t,
which
is
to be regarded as essentially auxiliary
and
devoid of any particular geometrical significance.
=/(*),
We may
write
(3).
V
= F(t)
x and
If a particular value is assigned to t, the corresponding values of Each pair of such values defines a point of y are known.
whose centre We a direct relation between x and y by squaring and adding: find that x* + y 2 = a2 t being now eliminated. t/) = the points of intersection of/=0 and = 0. Then (x. (#. The solution generally consists of a finite number of pairs of &. The points of intersection of the two curves whose = 0. . y)+A$(#. 3.32. has two degrees of freedom its position can only be fixed by the determination of two coordinates. e. y = ct + d. we construct all way to different values of the points which correspond in this obtain the graph of the locus t. our fundamental line is . and is It may or may not (in the obvious simple cases called a surface. by If we take the line as its distance from a fixed point on the line.gt.lt. What loci are represented by (ft) ? (a) x=at + b. for example x Let t a cos y= it is a sin t. we defined by the equations (3). its position! degree of freedom. which moves along a straight line has only one I Its direction of motion is fixed. Trace the curves (#+y) 2 =l. the position of any of its points determined by a single coordinate x. #)=0. on the other hand. 33] (sc. I. y) describes the can in this case at once obtain circle over and over again. If t varies beyond these limits. when t varies through all real values Loci in space. easy to see that the point is the origin and whose vary from to 2?r. .j) values of x and y. we Examples XVIII.  A locus represented by a single * =/O&. 1. where / and are polynomials. represents a curve passing through 4. xy = l. FUNCTIONS OF REAL VARIABLES If 61 y). y) describes the circle radius is a. can be equations are /(#. y/a=(l* 2 )/(lH2 )&. two fundamentally different kinds of loci. can be completely fixed by one measurement of position. a?/a = */(! + **). &. The two equations therefore generally represent a finite number 2.gt. A particle which moves in a plane. equation y) plainly belongs to the second of these two classes of loci. Suppose t. #) determined if these equations can be solved as a pair of simultaneous equations in x and y. particle A A of Chap. x2 y^ = l. The curve /(#. of which the simplest examples are the plane and the straight line. of isolated points. In space of three dimensions there are 33. (x.lt.g.
The [All points on the surface satisfy / (x. y.] XOY. The locus is the surface formed by drawing lines parallel to OZ through all is called a cylinder. Actually the curve (a) is the projection on the plane . considerations of The = as the standard form of the equation agreed to adopt f(x.62 it FUNCTIONS OF REAL VARIABLES will) [ll satisfy our commonsense notion of what a surface should be.jr. locus represented by two equations or f(sc. y) of a plane curve. the exceptional cases ? 3. which we may regard as determining a plane curve on the paper. so now we shall agree to adxypt as the standard form of equation of a surface. #)=0. 1.7 ) = /r.f)01 What is represented by three equations of the type Three linear equations in general represent a single point. regarded =/(#. y} =0 in the plane as a curve in space ? [/(#. nature of the surface may often be obtained as follows. Ax + By+Cz + D= 0. as the intersection of a sphere and a plane. y} curve /(#. Contour Maps. Such a surface 5 Graphical representation of a surface on a plane. 2. 4. y)=0j ? = 0. y) = a. of the form z Ax D A may be it may therefore be represented by two equations of the forms 2 (a )2 + ( y . We trace this curve and mark it (a). might seem to be impossible to represent a surface adequately by a and so indeed it is but a very fair notion of the drawing on a plane It : . The a curve.What are What when regarded are the equations of a plane curve f(x. y\ If we give z a particular value a. considered as a locus in space of three dimensions ? = 0. /(*. y)=0. Examples XIX. y.] points of this curve. z=0 is the curve in which the locus cuts the plane XOY. 31 may evidently be generalised so of a function / (x. y) and is called may be represented by two equations loci. What is the meaning of a single equation /(#. z) of three variables (or as to give definitions And as in 32 we of functions of any number of variables)./3)2 + (z. we have an equation /(#. whatever be the value of z. Cylinders. z) = belongs to the first class of Thus a straight line circle in space of the type + By + Cz + = Q. ". Let the equation of the surface be z=f(x.
17. above it*. of course. and let a be the semivertical angle of the cone. for a selection of values of a). The cone of Ex. That is to say. Ay. A Show that the particular case is given by the right circular cone of Ex. of the section of the surface of the land plane of the sea level and 1000 ft. xf(z\ round . Draw a series of contour lines to illustrate the form of the surface 7. Take the origin of coordinates at the Right circular cones. is surface formed by straight lines passing a cone: the point is called the vertex. y. The equation of the cone (which must be regarded as extending both ways from its vertex) is # 2 +y 2 . #2 =2 2 tan 2 a round the axis of z is derived from the second of 8. [If (x. 17. the axis of z. Cones in general. obtain some such figure as is shown in Fig. \z\ for * any value of X. these equations by changing # 2 into x2 +y 2 Show generally that the equation of the surface generated by the revolution of the curve #=0.33] FUNCTIONS OF REAL VARIABLES 63 XOY of the section of the surface by the plane z = a. 7 cuts ZOX in two lines whose equations may be combined in the equation # 2 =2 2 tan 2 a.2 2 tan 2 a = 0. 7. vertex of the cone and the axis of z along the axis of the cone . so must 9. is A through a fixed point called (X#. is of the form/(z/#. equation of a cone whose vertex is z/y)=0. on the plane of the sea level. 6. map and : The contour in fact this is the principle on which such maps line 1000 is the projection.] We assume that the effects of the earth s curvature may be neglected. It will at once suggest a contoured We We Ordnance Survey are constructed. and that any equation of this form represents a cone. by the plane parallel to the 3000 TOGO Fig. do this for all values of a (practically. . the equation of the surface generated by the revolution of the curve y = 0. 7 ^ Surfaces of revolution in general. z) lies on the cone.
7 the eliminating t between the two equations (1). where t is 2/ the angle between the plane z. Cylinders and cones are special cases of surfaces Such surfaces are called ruled surfaces. the ends of the rods being fastened to two circular rods of radius a. JTOZand a plane through the line and the axis of Another simple example of a ruled surface may be constructed as follows.e. The two equations . bj c. composed of straight lines. The angle qOQ =a in the figure is evidently given by Let all same way. 18 b. Fig. We can imagine the surface of the cylinder of a number of thin parallel rigid rods of length I. such as PQ. whose equation may be found by For instance. and is the other rods of which the cylinder was composed be treated in the obtain a ruled surface whose form is indicated in Fig. where Pq=h}. this line moves and generates a surface. 18 c. d instead of being fixed are functions of an auxiliary variable t. in general shape not unlike certain forms of tablenapkin rings (Fig. [II Ruled surfaces. Fig. but the surface is curved everywhere. We It is entirely built up of straight lines . 18 a). of the line which generates the cone are equations #=2 tan a cos. . i. 18 a. As t varies. 18 c). Unfasten the end Q of the rod PQ and turn PQ about P until Q can be fastened to the third circular rod in the position Q . in Ex.(1) represent the intersection of two planes. a straight line. 18 &. For any particular value of t the equations (1) give a line. Now suppose that a.FUNCTIONS OF REAL VARIABLES 10. I made up Now let us take a third circular rod of the same radius and place it round the surface of the cylinder at a distance h from one of the first two rods (see Fig. Take two sections of a right circular cylinder perpendicular to the axis and at a distance to be apart (Fig.
The roots of x* + nx* \pxz + intersection of the parabola x2 =y 10. CD parallel to OY to meet y=f(x) in D. 6./(#) is called called is 2.2 arc cos x.) / 4.gt. . Draw the graphs of the functions 3 sin .j&. arc tan a denote. ax*~bxc. /{&. : &.arc tan a?. Trip.27 + 4 cos #. and sec + cosec = c II has two roots between 2rr if c 2 &.4 =# along = in 5. (x)} by Verify the following method of constructing the graph of means of the line y = x and the graphs of /(#) and (a?) take 0. Draw Draw the graphs of the functions the graphs of the functions (i) [l/#]. the sum of an even and an odd function of x.gt. Show that any function of x. /(#)=/(#) it is lif(x) =/(. IL Show If that if y=f(x) = (ax + b )/(cxd) then x=f(y\ for all values of #. an even function. 1896.lt. the least 7.8 and show that the equation and four if c2 &. Show that the roots of x 5 +px + q = 2 intersection (other than the origin) of the parabola are the abscissae of the points of 2 y = x and the circle + (p qx + r = are the abscissae of the points of $nx and the circle 9. 5.FUNCTIONS OF HEAL VARIABLES 65 MISCELLANEOUS EXAMPLES ON CHAPTER 1. [#]/#. a. Draw up a table of the various possible numbers of roots. and DP parallel to meet OX to meet AB in P.x\ denned for all values of x. draw ^5 parallel to OF to meet y in (7. for any value of whose cosine or tangent is a. 11. ) (Math. arc cos (2# 2 arc tan 1)  .lt.8. arc tan a (ii)  .(x) yx 8.J . 5(7 parallel to OJT to OX. = [Use the identity/(^) H/(^)+/(^)} + H/(^)/(^)}] 3. where the symbols arc cos a. sin ( ^ W* sin x . 1 CM? positive (or zero) angle. Draw sin the graphs of the functions sin 2 x). an odd function. then P is a point on the graph required. Discuss the graphical solution of the equation by means of the curves y=xm y= . f j. Solve the equation sec + cosec d = 2*j2. x (a cos2 x + b ^^ (a cos2 # + b sin a? 2 a?).lt.
[Take #=0. ?/ an algebraical function of x. a n . 13. it being supposed that we can construct lines whose lengths are equal to any of the ratios of the coefficients a. 1911. /3. then 18. when x=a... Verify that the equation cos is ^ approximately true for all values of x between and 1.gt. What is the form of the graph of the functions. Find a polynomial in x of the second degree which for the values 2 of x takes the values 1/c. we indicated one N/2. c are rational.. c. c (5) (lcos#)tana# + sin#=0. The polynomial a.) 17. 6. 15. sin## + #3 = 0.siny.gt. . an d use tables. 6.. (2) + sin 2 # = l. 5 ) 1&.. then x is an algebraical function of y. and show that when its value is l/(c + l). 2^ + 3 roots and no more. when 16. p  y Give a similar formula for the polynomial of the (nl)th degree which x=a^ 2 .66 12. FUNCTIONS OF REAL VARIABLES Show that the equation [ll where n is a positive integer.gt. In Chapter I or two simple geometrical constructions for a length equal to We also showed how to construct starting from a given unit length. a rational function of y. and y is (1/a^. cot#+#fTr = 0. Show If is that if x is a rational function of x. l/(c + 2). .r&. as is structions were what may be on the ruler and compasses certainly the case if a. of the second degree which assumes. 19. J. 21. 1. they depended . a n the values a 1} a 2 . has their localities roughly. .) Show that the equation $xsmx = l has four roots between  TT and TT. 22. For which of these values is the formula exact?] f . 0. All these con . the values y is (ab)(ac) assumes. l/(c + l). indicating (Math. (3) tan ^=2^(1+^ b. called Euclidean constructions only. 1896. 20. (1) (4) Discuss the number and values of the roots of the equations . Tn&. Trip. Geometrical constructions for irrational numbers. What is the form of the graph of the functions z=sinx\. the roots of any quadratic equation ao^ + 26a?+c = 0. 14.
however. And this remains true however often our constructions are repeated.lt. length measured means of pp. values of x and y of the form m + n*Jp. n. roots only. though our surd expressions may now be of a more complicated form. where m. only irrationals of this kind can be constructed by Euclidean Starting from a unit length we can construct any rational length. But if the coefficients are rational. provided that a. With the irrational distances thus constructed circles a number of lines and whose coefficients we may proceed to construct may now themselves involve quadratic surds.gt. p are for if we substitute for x in terms of y in the second equation we obtain a quadratic in y with rational coefficients. C 0. that of the famous problems of antiquity was that of the duplication of is to say of the construction by Euclidean methods of a by ^2. And so the same #2 ) 2 f(yi3/2 ) 2} between any two points so is true of the distance */{(#! obtained. It is evident. a condition which implies that Now figure is in any Euclidean construction each new point introduced into the determined as the intersection of two lines or circles. c are rational. This expression contains a fourth root. provided that the ratios (or . or a line and a circle. e.~ V// V 3&. that all the lengths which we can construct by the use of such lines and circles are still expressible by square roots only. 24). however complicated. f. 47 et seq. will construct any surd expression involving square One the cube. Conversely. as the mean between 1 and 11 : Or these two mixed surds might be constructed ^2_34# + 190 = 0. We should begin by constructing then 17 + and 17 3^/11) and A/11. Exs. directly as the roots of methods. on solution./ll a case in point. given in Ch. such a pair of equations as give. was . I (Misc. and the circle 2jfy+c = 0). /3. ** . See Hobson. Hence Euclidean methods and no others.g. Squaring the Circle. and so that the problem is an impossible one.7#f a. And of A. hence we can construct the line Ax + By + J3.FUNCTIONS OF REAL VARIABLES It is fairly obvious that 67 we can construct is measured by any irrational number which square roots. the first stage of the proof.r 2 t2/ 2 H2&. 5 _o . but this is of course the square root of a square root. C are rational. Thus by these methods the length denned by any combination of V is V7 // 1 7 + 3 X/11\ IV Vl73 N/li. so on. the proof that ^2 cannot be a root of a quadratic equation ax2 + 2bx+c = with rational coefficients. p are rational. It can be shown that ^2 cannot be expressed by any finite combination of rational numbers and square roots. Hence the coordinates of all points obtained by means of lines and circles with rational coefficients rational : are expressible by rational numbers and quadratic surds. viz.
is locus of M a diameter OA and the tangent at A. to be its circumference radius. the error in supposing A R is the earth s M is le. AD cut the curve in that P and OP axis of y a length cut the tangent to the circle Show AQ=f/2.68 23. parabola # 2 = 2#. in the same direction. starting 25. On AO take AN=OP and draw JVM parallel to OQ and cutting AP in M. OZ&. Take a circle of unit diameter. show that the Taking B the curve (the Cissoid of Diodes}. Draw a chord OBC line take OM=BC. Show that a circle of radius R. are rational lengths. FUNCTIONS OF REAL VARIABLES Approximate quadrature of the circle. 26. 2= ?/ 4#. Sketch the curve. Show only. and that to take AM as being equal to ?r the circumference of the circle would If lead to a value of correct to five places of decimals. A in Let Q.gt. and P is one of its points of intersection with the parabola Show that OP meets the latus rectum of the first parabola in a point Q such that SQ=j/2. the circle at and the tangent at C.ss than 11 yards. 24. . [ll Let be the centre of On the tangent at A take AP=^R and AQ = ^R. that the only lengths which can be constructed with the ruler from a given unit length. is the vertex and S the focus of the Constructions for 4/2. Take along the at = 2. On this cutting as origin and OA as axis of x.
J . the length A P along Q the A of Chap. things are needed. and we may write PQ \x\. length. the sense of the displacement being We may therefore. II (Exs.e. of view. To its magnitude. The number x. But. viz. destitute of may geometrical significance. when we are of the change of position produced by the dis thinking merely placement. fixed by the sign of x. garded as the measure of a line viz. to consider all senses are the same. position which is We on the line particle placed at shall call the displacement or P A and then to change of needed to transfer the particle from P Q the displacement PQ. measure of a displacement or change of position It is on this last point of view that we shall now I. Imagine a small displaced to Q. symbol hardly necessary to caution the reader against confusing this use of the and that of Chap. specify a displacement completely three its sense forwards or backwards along the line.CHAPTER III COMPLEX NUMBERS 34. or a geometrical significance may be attached to it in at least three different ways. concentrate our attention. the original position called its point of application. It may be re preceding chapters.). it is natural to disregard the point of application and may be P displacements as equivalent whose lengths and Then the displacement is completely speci fied by the length PQ = x. and Misc. the point P regarded as the on the line A. without ambiguity. with which we have been concerned in the two It may be regarded from many different points be regarded as a pure number. speak of the displacement * It is [x] [x\ *. Exs. xvi. real Displacements along a line and in a plane. of the particle. Or it may be whose distance from A is x. It may be regarded as the mark of a point. and what i.
if the same in magnitude. (i) the magnitude of the displacement. line OA to A A. In other words. Draw a as OX. if we consider two displacements as equivalent if they are fourth.e. * increasing experience of mathematics will reveal to him the great importance of distinguishing clearly between things which. which is . OF Fig. being the same as that from P are equivalent displacements. and write RS as Now let us take any pair of coordinate axes in the plane (such in Fig. that of Q will be a + x the displacement [x] therefore transfers a . and sense. 19. by some similar difference of notation. We call OA is completely specified OA the displacement [x. We come now to consider displacements in a plane. of if Let x and y be the coordinates Then it is evident that x and y are given. direction. PQ and RS are equal and parallel. the length of the straight line PQ (ii) the direction of the displacement. If the coordinate of P is a. equal and parallel to PQ. i. are iioj. and the sense of motion from P to Q is the same as that of motion from R to S. the sense of motion from Then PQ and OA to Q. y] and write Strictly speaking we ought. however intimately connected. 19). it would be very important to distinguish between the motion of the batsman between the wickets. and the mark which is put down in the scorebook. We may : define the displacement required in order to specify PQ it as before. particle from the point a to the point a + x. the same. to dis The reader tinguish the actual length x from the number x which measures it. . Of these requirements we may disregard the of application. PQ (iii) the sense of the displacement makes with some fixed line in and (iv) its point . determined by the angle which the plane . we regard the displacements PQ and equivalent. But will perhaps be inclined to consider such distinctions futile and pedantic. But now more data are We require to know completely. the run which he scores.70 COMPLEX NUMBERS [Ill We use the square bracket to distinguish the displacement [#] from the length or number x*. If cricket were a mathematical science.
[a?.. 36... Thus (Fig... 77 ) is therefore [rfeVHJ It is clear that if. [# .....y = y ........ equivalence of displacements... . (1)... ?/] = [a? y*] if y ] are equivalent and only if #=# reverse displacement natural to agree that .. OB = OA any real number.. it is evident that The displacement from (f. Multiplication of displacements by numbers. and f and 77 those of Q. 77  77 ].. (2)... and only if... The QP would be [{ f ... viz. and it is these equations being really definitions of the meaning of the symbols [ f . any definition We have not yet given which enables us to attach any meaning to the expressions sense at once suggests that we should define the sum of two displacements as the displacement which is the result Common of the successive application of the two given displacements.. 77 77]. ?/]. Having thus agreed that [*y] = [^yl it is natural to agree further that afcy] = [0.. If f and rj are the coordinates of P. PQ..... positive or negative...... .... where a if is #] .. 77) to (f .. y=y .. COMPLEX NUMBERS 71 Equivalence of displacements. x= a) two displacements Thus [a?.3436] 35. 19) then The equations (1) and (2) define the first two important ideas connected with displacements.. and multiplication of displacements by numbers.. In . Addition of displacements..
B are x y then those of the middle point of f .y ] = + x. we have Q OA equal and parallel to PQ.. and OB parallel to . l PQ.COMPLEX NUMBERS other words.gt. % (y+y [x. [x. If the coordinates of Let us consider the consequences of adopting this definition. 20. Fig. to it [m PQ . 2/1 commutative In other words. Hence [x + [x . y+y. law expressed in ordinary algebra by the equation This law expresses the obvious geometrical fact that if we move displacements obeys the from P first through a distance PQ 2 equal and parallel to parallel to PQ . which may be regarded as the symbolic definition of addition of displacements. = [x + x y . AB are J (x + x ). and those of C are x+x.gt. .gt. y+ y] + [&. If then we draw P equal and and complete the parallelogram OACB.y] )&. + y\ = y". suggests that if QQi be drawn equal and parallel so that the result of successive displacements PQ. we shall . y ] We [a?. observe that + y] = [x + x. and then through a distance equal and arrive at the same point Q as before.y + y ] (3).] [&. Q on P a particle at P is to transfer it first to Q and then to as being Q 1} then we should define the sum of PQ and PQ P& . . addition of a + b = b + a.
./ y } + 0".. y].] (iii). j8y] = [ax. .. (iv) ( v) (a+0) {[a?.. y] + [a/. When we there is have once defined addition of two displacements.. y". +y + y". y] + [.&. by the equation y] . 2/1 .y or as [# x.. 0] leaves the particle where it was the zero displacement. [&.0]. Pn . /].. 0] and [x.. The displacement [0.. OX. y} their resultant.gt.y][&. yl [a?..gt.. (5)..36] COMPLEX NUMBERS In particular 73 Here [#.".. y] the components of [x. y". y] 0. geometrical significance of If M M divides 2. 3/1) + [^ y 1 = [of. (i) (ii) it is 1..]. It when we were call [x.gt. then in the ratio /* X. y]+^ [a?. y*]} =a y] + a JV. a/ty]. then : ~OM=%(OP+ OQ). is in fact what we previously We considering only displacements and [0. is if PQ the middle point of PQ. then ..y [x.gt.# [*.gt.. . y. Thus. . } [&. If G is the centre of mass of equal particles at P P lt 2t . y y }) + [&. Prove that ft (iii) = [aflff. 0] = 0. Examples XX..] = + # + y] + [&. The remaining equations follow with equal The reader should in each case consider the the equation. = [^ y ] + (K y } + ([* y] + IX.. 0] denotes a displacement through a distance x in a direction parallel to denoted by along a line. = [x y] + [x + . ] ]. y + y]. 3.[^ y ]) ..y] = [0. We which define subtraction of displacements [*. ay] K ^D [&. and we agree to write [0. difficulty in the no further way of defining addition of any number.] = ([x.".. [We have already proved ease from the definitions. [a?].IX = 2/1 is the same thing as + (.] +# a". as we did above in the case of (iii). More generally. [a?..lt.. In particular &. by definition.gt. y\ + [of. .gt. y". y] + a j8a?. = y] o [ar.
is the point where TS meets AC. and then a = y and (3 = 8. AR AD Then S be denoted by a. 2. a.] DM * a parallelogram. .] in the If ~AB and ^AC are two displacements not same straight line. a/*=/3X + /z from which we deduce a/3 The symmetry of this result shows that a similar argument would also give I?/ = if a ^l Z^ Hence P 7. U. . say . [This is really only another way of stating Ex. parallel ABCD is a parallelogram. But since P lies on AC. /3 . lelogram. /3. AP is a numerical multiple of AC TP=k 1C=k lB+k ID. not and such that and conversely. Then. possible to find real numbers y. collinear points in the plane. trisects and is trisected by A C*. { . AT Fig. are collinear. 21. ~AR=$. R are /3. Complete the parallelogram A B^ P^ C^ AP = a 5J2 + AC. [Take Then A B = a A B.AB. .COMPLEX NUMBERS 4. then all zero. RQS and TQ U are drawn to the sides. intersect Through a point inside the paral Show that D U _. That is to say AD. A C = A C. Q. ABCD is Show that The two preceding examples are taken from Willard Gibbs Vector Analysis. and Jf the middle point of AB. whence the theorem follows. It is evident that AP can only be expressed 1 l /3 . 5) = (X + /i) ^.] Q. : [Let the ratios AT: AB. c RU. P and P are the same point. Let RU meet AC P in P.Al). 5. TS on AC. . AP= where /i/X is X  1 the ratio in which P divides R U. l in this 6. Hence (Ex. . form in one way. [Ill it is If P. AT=a. B since R.
and so this definition also to the second objection. It is. and the suggested definition does not give this result. it is by no means obvious what the best meaning to attach to the product [x. ay]. in other words. A But more reasonable would give definition might appear f . y\ as [ax. y} x so far [x. But there would be two serious objections to such a definition. is In fact. this product must itself be . y ]. In the first place our we might agree would be futile. for example. y] = [ax. 0] [x.36. ay]. In the second place our definition would be If a is inconvenient and misleading for the following reasons.y]=[ax. if not absolutely necessary. most desirable. the product of two displace ments must itself be a displacement. made no attempt 0. the displacement It is therefore. that the product of two displace ments was to be always equal to their sum. Now. 0]. y ] means nothing. already defined a [x. y] [x. 0] would be open . the displacement [a] a real number. as we have we saw in along the axis OX. in our later notation. and we are at liberty to define it to mean anything we like. if our definition is to be of any use. however. to be [x. All that is clear is (1) that. y + y ] . COMPLEX NUMBERS 75 So far we have Multiplication of displacements. this [a. it We might. view be regarded as a displacement. viz. that our definition should be such that [a. define [x as being equal to + x. y} [x . to attach any meaning whatever to the notion The only kind of multipli of the product of two displacements. the real number a may itself from one point of 34. we have considered is that in which a displacement cation which The expression is multiplied by a number. at any rate [or. 37] 37. fairly clear that if any definition of such a product is to be of any use. We should only be introducing a new method of expressing something which we can perfectly well definition express without it. 0][x. or. y ] = [xx yy ].
y ] = [xx xy] f .gt. . . 0] [V.. the angles We D corresponding in the order in which they are written.gt.. a? and y (2) that the must be such as to agree with the equation . y] + Ky } [&. Now let . [III a displacement whose coordinates depend on x and words that we must have or in other AY here X definition and Y are functions of x.oc = ^A . if right definition to take is suggested as follows.. y". or OB OC = OA . y] and 38.] 0.]. y } ([x. y. [*". 2/1 + K y] = y ]) [^.".. then 05/(M = OD/OC. y ] [x. . [x". (3) that the definition must obey the ordinary commutative.]) [. and associative laws of multiplication.. The know that.gt. K y". (IX. [x. 2/1 + IX ] . [*". y". y".]. y".". y] [x ]) . y\ = + [x y } y] = K y] IX. so that and [x.76 COMPLEX NUMBERS y. 2/". OD.. This suggests that we should try to way define multiplication and division of displacements in such a that WI~OA = ODI~OC. y] t [&..].]) = [of.] . 00 are two similar triangles. distributive. ([&.OD. y] ([x y [x". GAB. ~OB . y".
so that OA = [I. #+&.). If the polar coordinates of We B x and a are (/&.gt./&.OB = [1.gt. y] + IX. To construct the product 0(7. is therefore ] 2/1 IX. .. .gt...gt. We desired . y = o. and (3) that {[*. y ] = [of. OB we .] = [(a? + aO = W a. 2/". then the polar coordinates of D are evidently pa and .. cos &.(y + y ) x y"..1 .gt. . (Fig. that x p cos 0. The product UB OC is OD COD being similar to A OB. (x + a/) 7/ + (y + y } x"..} yy". T/?/ &.. [Take the commutative law for example.gt. .? + 2/1 IX /x . } .} = [* + 7 &.. In order to free this definition from ambiguity. Show directly from the geometrical definition given above Example. Then and so The product OB 0(7 is therefore to be denned as obtained by constructing on 00 a triangle similar OA # being to OAB. (6). We say that the two triangles are then similar in the same sense. F]. y] [V... it should be observed that on 00 we can describe two such triangles. OOD and OOD . y"../&.lt.. observe (1) that if then (2) that the righthand side is not altered . y = [^ . y] [&.lt.gt. Similarly are satisfied./&. 2/]} (X /x &.lt.gt. and suppose that A is the point (1.) = xy + y# 7 &.. 0). 38] COMPLEX NUMBERS 77 0]. as we we inter change x and x and y and y so that 7 0. X = ## Y = xy if .sin c/&.yy y = 0.0][Z. choose that for which the angle OOD is equal to A OB in sign as well as in magnitude.lt..lt... F] = [2T.37. 22).. y p sin 0. that multiplication of displacements obeys the commutative and distributive laws.. we can verify that all the equations at the end of 37 Thus the definition (6) fulfils all the requirements which we made of it in 37. ay + 0&. OZ. Hence X = pa cos (6 + () = ## F = pa sin (0 + The required definition l&. y] . so &. 6) and (a.lt.gt.
by putting x = x. and only if x =x . as we shall henceforth. y} in the plane correspond a point (x. We x. we obtain 2xyi. so to a displace correspond ment [x. addition. The reason for the choice of this notation will appear later. (3). The expression x + yi is called a complex number. Thus x+yi = x + y i if (1). The sum or product corresponding displacements of two complex numbers is the complex number which corresponds to the a displacement. Positive integral powers and polynomials of complex numbers are then defined as in ordinary algebra. Two complex numbers sum or product of the two corresponding displacements. x + yi = (x (x f Qi) (x + 00 + (0 f yi). (x + yi)2 = (x + yi) (x + yi) = x* . Thus. Just as to a displacement [x] along a point (x) and a real number x. y y \ (x+yi) + (x + y i) = (x + x ) + (y + y )i xx .] we want BOD AOC is . For the present the reader must regard x + yi as simply another way of writing [x.78 should have to construct on COMPLEX NUMBERS [ill OB a triangle similar to and DI coincide. y) and a pair 39. y =y m (3). or that to prove is that This is an easy piece of elementary geometry. y]. when dealing with complex numbers. y. shall find it convenient to denote this pair of real numbers y by the symbol x + yi. we write x for x + Oi and yi for + yi. To every complex number corresponds are equivalent if the are equivalent. We tion of proceed next to define equivalence.yy + (xy + yx) i (x + yi) (x + y i) f (2). + y i) = xx + xy i . In particular we have. Complex numbers.y* + . and these equations suggest that there will be no danger of confusion if. and so what D BOD similar to AOC. OX of real numbers x. as special cases of (2) and (3). and multiplica complex numbers.
+ yi) {(x + y i) f (x".. i. %=y ..i)}. (^ + /O + (^ + /O (a? 0". we obtain This solution fails if is a? and y are both zero.e.&.. as being the complex number And (x + yi)/(af f y i) + rji such that is defined or off .. Thus subtraction unless the divisor always possible... f (x + y i) (y as + yi) + {. Solving these equations for x tj+y and 77. + 2M). COMPLEX NUMBERS will easily 79 The reader verify for himself that addition multiplication of complex expressed by the equations f and numbers obey the laws of algebra yi {(x + (x + y i) = (a/ + y i) + (x + yi).y i) = number f f rji such that x)+  y )i .y + (x + y t] rj ) i = x + yi.i) (x r + yi) + {(of + y i) + (x". + yi) (x f + y i) = (x + y i) (x yi...(x + y i)} =x + yi + (x .i)} = {(a? + yi) (x + ^)) (^ + y". (4). + yi) + (of + y i)} + (x + = y". Subtraction and division of complex numbers are defined as in ordinary algebra. (x Thus we may define (x + yi) &.gt.. as the Avhich leads to the same result.. (x"..i\ the proofs of these equations being practically the same as those of the corresponding equations for the corresponding displace ments.38 39] . ^ x Zy y^x... + y". division is always possible is zero. or again.lt. . + y". if a? +yi= 0..
Thus the complex numbers i and Similarly ( i) = 1. or OC=0. One most important property of real numbers is that known as the factor theorem. . and that x+yi M ~^y 40. unless x = + yi cannot vanish 41. OY.e. which asserts that the product of two numbers cannot be zero unless one of the two is itself zero. Thus x unless either x + y i or f + rji vanishes. . The equation i 2 = 1. &. and this is COB. particular complex the number which corresponds to a unit displacement along Also (0 . .1. = 1. We + Oi agreed to simplify our notation by writing x instead of x and yi instead of + yi. +. 0.80 COMPLEX NUMBERS [III Examples. 2 i 2 The reader will now easily satisfy himself that the upshot of the rules for addition and multiplication of complex numbers is this. or x + y i = 0. the problem of the division of the displacement ~OB by ~OC is that of finding so that the triangles are similar. = x^ +y 2 . that we operate with complex numbers in exactly the same way as with real numbers. To prove that this is also true of complex numbers we put x y = in the equations (4) of the preceding section. satisfy the equation # #=H= . so that the product of two conjugate numbers is real. The It is number li we shall denote simply by i. Fig.gt. xyi Verify that (x {yi) (x yi) are said to be conjugate. treating the symbol i as itself a number. + li) (0 + li) = (0 . (1) From a geometrical point of view. and y = 0. 23. D AOD evidently possible (and the solution unique) unless C coincides with 0.1 1) + (0 1 + 1 0) i = . 77 = 0. (2) The numbers x+yi. Then = 0. These equations give f = 0.* i.
42. Finally. 6 . but it is The phrase number . The firmly * real number was introduced as an antithesis to imaginary H. and (x + yi) i would have represented the dis placement obtained by turning x + yi through a right angle. number + 1 = 0. expression is by no means a happily chosen one. we should naturally have defined (x + yi) x as xx + yx i. and OQ is drawn equal to OP and so that POQ is a positive right angle. example. = 1. (x + yi) (x + y i) = xx + xy i + yx i + yy fi = (xx . (x + yi) y i as yy + xy i. to follows that if x + yi corresponds OP. and (x + yi) (x + y i) as the sum of these displacements. it is sometimes written in the form V( Complex numbers are sometimes called imaginary*. and of i as a symbol of operation equivalent to turning x through a right angle. i the two complex numbers i and satisfy this equation.3943] COMPLEX NUMBERS ii 81 it but replacing the product for = i2 by 1 whenever occurs. = y + xi. real yy +(xy z2 The equations z such that this is expressed by saying that the equation has no real roots. multiplication of a complex turns the corresponding displacement through a right We might have developed the whole theory of complex numbers from this point of view.e. But.yy ) + (xy + yd) f i. The geometrical Since (as 4 interpretation of multiplication yi) i by it i. i. we should have been led to regard yi as a displacement of magnitude y along Y. as xx 43.e. i In other words. z* + 1 = az* + Zbz f c = 0. It would then have been natural to define x + yi as in 37 and 40. as we have just seen. . There is no We express this i by saying that the equation has i satisfies z* the two complex roots and i. i. y + xi. number by angle. Since 1). Starting with the ideas of x as representing a displacement along OX. then (x + yi) i corresponds to OQ. Thus.
than a and that it is not a number at all. be too established and has to be accepted. either we can say that b/a). 1]. what we mean is simply that we have denned a method of combining such which we call multiplica pairs of numbers (or displacements) tion . . y). when we say that i is a root of the equation z* + 1 = 0. in the sense in which the real numbers are numbers. a pair_of numbers (x. Thus is no more imaginary . 0). Now where let us consider the more general equation a. in the form a? + yi.lt. may be written in the form an equation which z is evidently satisfied if . and may be represented And geometrically by a point or by the displacement [0. \ for purposes of number a pair of numbers is no less real than any ordinary such as J. united symbolically. a quadratic equation with cases. or two distinct * complex We shall sometimes write x + iy instead x + yi for convenience in printing. or than the Solar System. . of roots. 1) with itself.82 COMPLEX NUMBERS [ill It cannot. real coefficients has two roots in all two distinct real roots. b. If 6 2 ac. but. the equation has two equal roots. Such technical convenience. It ac.s 2 + 2bz + c = 0. the reader that an imaginary number strongly impressed upon in any ordinary sense of the word.gt. real number i =Q+ li stands for the pair of numbers (0. 1). the ordinary method of solution gives two real roots {If 6 2 &.ac)}/a. and which. the equation has no real roots. when we so combine (0.b*)}/a. we substitute b*)/a *. or than the paper on which this is printed. c are real numbers. however. b V(& 2 . as should be clear from the preceding discussion. a. or two equal real roots. for + (b/a) express this either of the complex numbers by saying that the equation has the two complex roots i \J(ac We {bi ij(ac . &. as a matter of convention to say that is satisfied If we agree when 6 2 = ac (in which case the equation viz. gives the result ( 1. by one value of x only.
43]
COMPLEX NUMBERS
The question
is
83
equation
may
not,
when complex
naturally suggested whether a quadratic roots are once admitted, have
roots. It is easy to see that this is not possible. Its impossibility may in fact be proved by precisely the same chain of reasoning as is used in elementary algebra to prove that
more than two
roots.
an equation of the nth degree cannot have more than n real Let us denote the complex number sc + yi by the single
letter
z,
a
convention
which
z
= x + yi.
(1)
we may
express
z,
:
by writing
with real or
Let
/ (z)
denote any polynomial in
complex
coefficients.
Then we prove
is
in succession
is
that the remainder,
when f(z)
/(a)
;
divided by z
 a, a being
then f(z)
is
any
real or
complex number,
if
(2) that
a
a
is
;
a root of the equation f(z)
= 0,
divisible
(3)
by
z
n
that if f(z) is of the nth degree, and f(z) roots a l} a a ..., a n then
,
=
has the
,
f(z) =
A(z dj) (Z 
fife)
. . .
(Z
 an ),
where
A
is
of zn in f(z).
it follows
a constant, real or complex, in fact the coefficient From the last result, and the theorem of 40,
roots.
that/ (2) cannot have more than n
conclude that a quadratic equation with real coefficients has shall see later on that a similar theorem is exactly two roots. true for an equation of any degree and with either real or complex coefficients: an equation of the nth degree has exactly n roots.
We
We
in the proof which presents any difficulty is the the proof that any equation must have at least one root. This we must postpone for the present*. We may, however, at once call attention to one very interesting result of this theorem. In the theory of number we start from the positive integers and from the ideas of addition and multiplication and the converse We find that these operations of subtraction and division. operations are not always possible unless we admit new kinds of
first,
The only point
viz.

numbers.
can only attach a meaning to 37 if we admit negative numbers, or to f if we admit rational fractions. When we extend our list of arithmetical so as to include root
operations
extraction
We
and the solution of equations, we find that some of
*
See Appendix
I.
62
84
COMPLEX NUMBERS
[ill
them, such as that of the extraction of the square root of a number which (like 2) is not a perfect square, are not possible unless we
widen our conception of a number, and admit the irrational numbers of Chap. I.
Others, such as the extraction of the square root of 1, are we go still further, and admit the complex numbers of this chapter. And it would not be unnatural to
not possible unless
suppose that, when we come to consider equations of higher degree, some might prove to be insoluble even by the aid of
complex numbers, and that thus we might be led to the con siderations of higher and higher types of, so to say, hypercomplex
numbers. The fact that the roots of any algebraical equation whatever are ordinary complex numbers shows that this is not the
case.
to
application of any of the ordinary algebraical operations complex numbers will yield only complex numbers. In technical
The
language
operations
the field of the complex numbers
.
is
closed for algebraical
Before we pass on to other matters, let us add that all theorems of elementary algebra which are proved merely by
the application of the rules of addition and multiplication are true whether the numbers which occur in them are real or complex, real
since"
numbers.
the rules referred to apply to complex as well as For example, we know that, if a and ft are the
az&gt;
roots of
+ 2& + c = 0,
= (c/a).
.
then
Similarly, if
+
a, /3,
&lt;y
= (26/a),
are the roots of
3cz
+ d = 0,
then
a
+ 13 + 7 =  (36/a),
fa + yd +
a/3
= (3c/a),
a/3y
=  (d/a).
are
All such theorems as these are true whether
real or complex.
a, 6, ... a, (3, ...
44.
Argand
6,
s
diagram. Let
the angle
9,
P (Fig.
2
24) be the point
(x,
y\
r the length OP, and
x = r cos
y
r sin
XOP, r = *J(a? + y ),
so that
cos 6
:
sin 6
:
1
:
:
x y
:
:
r.
43, 44]
COMPLEX NUMBERS
denote the complex number
ae
85
z,
We
we
call
+ yi by
as in
43,
and
We
z the complex variable. call the point z, or
P
the point corresponding to z\ z the argument of P, x the
real part,
part,
y the imaginary modulus, and 6 the amplitude of z\ and we
r
the
write
= =
z
,
!(&gt;.
o
X
Fig. 24.
= am #.
we
say that z
is real,
When y
when x
that z
is
purely
+ yi, imaginary. yi which differ only in the signs of their imaginary parts, we call conjugate. It will be observed that the sum 2# of two conjugate numbers and their
Two numbers x
x
are both real, that they have the same modulus V(# + y } and that their product is equal to the square of the modulus of either. The roots of a quadratic with real coefficients,
product
2
ot?
+
2
7/
1
for
example, are conjugate,
It
when not
or
real.
must be observed that 6
am z
is
a manyvalued function of
x and
y, having an infinity of values, which are angles differing by line originally lying along will, if turned multiples of 2?r*. shall through any of these angles, come to lie along OP.
A
OX
We
describe that one of these angles which lies between TT and of the amplitude of z. TT as the This deprirwirjal^value finition is unambiguous except when one of the values is TT,
I
f
j
I
which case In this case we must make TT is also a value. some special provision as to which value is to be regarded as
in
of z
the principal value. In general, when we speak of the amplitude we shall, unless the contrary is stated, mean the principal value of the amplitude.
Fig. 24
* is
usually
known
as
Argand s diagram.
It is evident that \z\is identical with the polar coordinate r of P, and that the other polar coordinate 6 is one value of am z. This value is not necessarily the principal value, as defined below, for the polar coordinate of 22 lies between
and
27r,
and the principal value between  TT and
TT.
,
.
r
v?^%
86
45.
COMPLEX NUMBERS
[ill
following follow immediately from the definitions of addition
plication.
De Moivre
s
Theorem.
The
statements
and multi
(1)
The
is
numbers
real (or imaginary) part of the sum of two complex equal to the sum of their real (or imaginary) parts.
The modulus of the product (2) to the product of their moduli. equal
(3)
of two complex
numbers
is
either
The amplitude of the product of two complex numbers is equal to the sum of their amplitudes, or differs from it by 2?r.
be observed that
is
It should
^.
am (zz
z=z
1
}
the
sum
v
j
ITT.
= l+i, But zz =
The two
it is not always true that the principal value of For example, if of the principal values of am z and am/. 1 then the principal values of the amplitudes of z and z are each
2i,
and the principal value of
am (zz
}
is
\ir and not
TT.
last
theorems
may
be expressed in the equation
(/&gt;
r (cos 6
+ i sin 0)
x p (cos
+ i sin $) r*A = rp {cos (0 + 0) + i sin (6 +
&lt;f&gt;)
rr
(/&gt;)},
^H
,
which may be proved at once by multiplying out and using the and sin (6 + $). cos (0 + ordinary trigonometrical formulae for
More generally
n (cos 0! + i sin ) x r (cos + i sin ) x ... x rn (cos 6n 4 i sin n ) = rjr ... rn {cos (0j + + ... + 0) + i sin (0j + + ... + n )}X
2
2
2
a
a
a
A particularly interesting case
n=r =
2
is
that in which
...
= rn = 1,
&gt;l
0!
=
2
=
...
=
On
= 0.
We
then obtain the equation
(cos
+ i sin 0) =
cos
nO + i
result
sin w0,
where n
is
any positive integer: a
z
known
as
De Moivres
.
Theorem*.
Again,
if
 r (cos
(cos
+ i sin 0)
i
then
\\z
sin 0)/r.
Thus the modulus of the reciprocal of z is the modulus of z, and the amplitude of the reciprocal
the amplitude of
*
z.
is
reciprocal of the the negative of
We
(2)
can now state the theorems for quotients
(3).
which correspond to
It will
and
sometimes be convenient, for the sake of brevity, to denote cos 6 + i sin & De Moivre s by Cis0: in this notation, suggested by Profs. Harkness and Morley, n theorem is expressed by the equation (Cis0) =
41]
(4)
COMPLEX NUMBERS
The modulus
of the quotient of two complex
87
numbers
is
equal to the quotient of their moduli.
The amplitude of the quotient of two complex numbers (5) either is equal to the difference of their amplitudes, or differs from
it
by
2?r.
Again
(cos 6
+ i sin
6)~
n
n = (cos i sin 0) = {cos ( 0) + i sin ( 6}} n = cos ( n6) + i sin ( nd). all integral
Hence De Moivres Theorem holds for
positive or negative.
values of n,
To the theorems
which
(6)
is also
(1)
(5)
we may add the
following theorem,
of very great importance.
The modulus
is
of the
numbers
not greater than the
sum of any number of complex sum of their moduli.
pdv)
P"
Fig. 25.
be the displacements corresponding to the Draw PQ equal and parallel to OP various complex numbers. QR equal and parallel to OP", and so on. Finally we reach a
Let OP,
OP
,
...
,
point U, such that
OU is the modulus of the sum of the complex numbers, whereas the sum of their moduli is the total length of the broken line OPQR...U, which is not less than OU.
The length
A
purely arithmetical proof of this theorem
1.
is
outlined in
Exs. xxi.
88
46.
COMPLEX NUMBERS
[
m
We
add some theorems concerning rational functions of
^
is
rational function of the complex variable z defined exactly as is a rational function of a real variable x, viz. as the quotient of two polynomials in z.
complex numbers.
A
the
THEOREM form X +
In the
1.
Any
rational function
Yi, where
X
and
Y
(z) can be reduced to are rational functions of x and
R
y with real
coefficients.
first it
is evident that place any polynomial P (x + yi) can be reduced, in virtue of the definitions of addition and multi plication, to the form A + Bi, where A and B are polynomials in x and y with real coefficients. can be +
Similarly
Q(x
yi)
reduced to the form
R (X
(A
C + Di. Hence + yi) = P(X + yi)/Q (X + yi)
can be expressed in the form
+ Bi)l(G +
Di)
= (A +
Bi) (G

Di)/(C + Di) (C
 Di)
AG+BD BGAD.
"
which proves the theorem.
THEOREM
In the
2.
If R(x + yi)=X +
place this
is
Yi,
R
denoting a rational
function as before, but with real
first*
coefficients, then
R(xyi)=X Yi.
by actual expansion. true for any polynomial with
notation used above,
,
n easily verified for a power (x f yi) It follows by addition that the theorem is
real coefficients.
Hence, in the
..__ ~
~1
T
"
~
is
the reduction being the same as before except that the sign of i It is evident that results similar to those changed throughout.
of
Theorems
and 2 hold
for functions of
any number of complex
variables.
THEOREM
3.
The roots of an equation + an = a^n + a^ n +
J
. . .
0,
whose
coefficients
are real, may, in so far as they are not themselves
real, be
arranged in conjugate pairs.
46]
COMPLEX NUMBERS
For
it
89
x
Theorem 2 that if x + yi is a root then so is case of this theorem is the result ( 43) that particular yi. the roots of a quadratic equation with real coefficients are either
follows from
A
real or conjugate.
This theorem
with real
is
sometimes stated as follows
:
in
an equation
should be
in conjugate pairs. It coefficients complex roots occur with the result of Exs. vnr. 7, which may be compared
:
stated as follows
in
an equation with rational
coefficients irrational
roots occur in conjugate pairs* .
Examples XXI.
definitions
[First, to
45 directly from the 1. Prove theorem (6) of and without the aid of geometrical considerations.
prove that
z

+z
&lt;
z

\
\
+
z
\ \
is
2
to prove that
(x+xj + (y +/)
The theorem
2.
2
y (a
8
+3/
)
+ V(^ +/
2
2
2 )}
.
is
then easily extended to the general case.]
case in which
The one and only
*
is
+

+ ...!*+
z,
+.. I,
the same amplitude.
that in which the numbers
d,
...
have
all
Prove
this both geometrically
3.
and
analytically.
The modulus
of the
sum
of
any number of complex numbers
is
not
less
than the sum of their real
(or imaginary) parts.
real,
If the sum and product of two complex mimbers are both 4. the two numbers must either be real or conjugate.
5.
then
If
a + bj2 + (c + dJ2)i=A + Bj2 + (C+Dj2}i,
B, C,
^_
where
a, &, c, o?, 4,
D are real rational numbers,
b
then
a = A,
6.
= B, c=C, d=D.
A + Bi, where A and
Express the following numbers in the form
:
B
are
real
numbers
where X and
7.
\i
are real numbers.
Express the following functions of z=*x+yi in the form
real functions of
X+ Yi, where
+ /fe)/(y+8s),
Xand Tare
where
8.
a,
/3,
x andy:
22,
z\
z n, \\z, *
+ (!/*),
(a
y, 8
are real numbers.
Find the moduli of the numbers and functions in the two preceding
examples.
*
The numbers a + ^/6, aJb, where
.
a, 6 are rational, are
sometimes said to be
conjugate
90
9.
COMPLEX NUMBERS
The two
if
[ill
lines joining the points
z= a,
z
=b
and
z
= c,
zd
will
be
perpendicular
i.e.
if
(aty/(cd)
is
purely imaginary.
What
is
the condition that the lines
should be parallel ?
10.
The
a,
/3,
where
y are complex numbers.
the centre
three angular points of a triangle are given by 2 = a, z=&, z = y, Establish the following propositions
:
(i)
(ii)
of gravity
is
given by
the circumcentre is given
= % (a+/3f y) ly\za\ = z/3 =
z
;
2

 y

;
(iii)
the three perpendiculars
from
the
angular points on the opposite
sides meet in
a point given ty
B
(iv)
there is
a point
P inside
co
the triangle such that
and
[To prove
(iii)
CBP=ACP=BAP=co, = cot A f cot B + cot C. we observe that if A, E, G are the
cot
vertices,
and
is
P
should be perpendicular to point 2, then the condition that that (z a)l($y) should be purely imaginary, or that
AP
any
EG
(Ex. 9)
cyclically, are satisfied
This equation, and the two similar equations obtained by permuting o, /3, y by the same value of z, as appears from the fact that
the
sum
of the three lefthand sides
(iv),
is zero.
To prove Then*
take
EG parallel
to the positive direction of the axis of x.
&lt;7),
/3a= 
We have
to
determine
z
and
a&gt;
from the equations
)
where
yo denote the conjugates of z, a, /3, y. the numerators and denominators of the three equal fractions, Adding and using the equation
ZQ) OQ,
&&gt;
i cot
o&gt;
= (1 + Cis
2o&gt;)/(l
 Cis 2),
^
we
find that
^3yo
 j3oy 4 y
o
 yo a + a^o  a ^
From
where
*
it
is (a 2 +6 2 +c2 )/4A, this it is easily deduced that the value of cot and this is equivalent to the result given. A is the area of the triangle ;
We
suppose that as we go round the triangle in the direction
left.
ABC
we
leave
on our
46]
COMPLEX NUMBERS
91
To determine z, we multiply the numerators and denominators of the  a )/(a  y )/(y  ), equal fractions by (y /3 )/(/3a), (a 7)5 and a It will be found that to form a new fraction.
&lt;*d
(&&gt;
11. The two triangles whose vertices are the points a, respectively will be similar if
1 1
1
b, c
and
#, y, z
=0
the
[The condition required is that ABfAC^XYjXZ (large letters denoting points whose arguments are the corresponding small letters), or (ba)l(ca) = (yx}l(zx\ which is the same as the condition given.]
12. Deduce from the last example that if the points #, y, z are collinea.r then we can find real numbers a, /3, y such that af # +y = and ax + /3y +yz=0,
and conversely (cf. Exs. xx. 4). [Use the formed by #, y, z is similar to a certain
apply the result of the last example.]
13.
linetriangle
fact that in this case the triangle on the axis and
OX&gt;
The general
equation az+/3=0 has the one solution
linear equation with complex coefficients. The 2=  (/3/a), unless a = 0. If we put
and equate real and imaginary parts, we obtain two equations to determine the two real numbers x and y. The equation will have a real root if y = 0, which gives ax + b = Q, Ax + B=0, and the condition that these equations
should be consistent
14.
is
aB
bA = 0.
The general quadratic equation with complex
is
coefficients.
This
equation
Unless a and
A
are both zero
22
we can
divide through
by a + iA.
Hence
we may
consider
+ 2 (b + ffi) z + (c+ (7^ =
........................ (1)
as the standard form of our equation. Putting z=x+yi and equating real and imaginary parts, we obtain a pair of simultaneous equations for x and y,
viz.
If
we put
these equations become
2
_ ^2 = ^
2^ = k.
if = then we must also = 0. =0. (7=0.e. b + Bi and c + Ci must both be real.lt. the roots will be real even in this case. Then since the sum of the three roots is zero the third root must be . if A = 0. which obviously expresses the fact that the lefthand side of (1) is a perfect square. we arrive at the following conclusions. if k is negative unlike signs. 77 : square roots above vanish.lt. 2 . The Cubic equation. In this case there /i may be three real roots. 27Ap.lt.2#. 2 where x is then #2 + 2&. so that the coefficients of the equation are real. (c) where y is given by the equation y* 3X?/ a. x+yi and xyi. 2/to + (7=0. Hence for a pair &. and p (6) and the root is yi. From the relations between the coefficients and the roots of an equation we deduce Hence G and II must both be real.c. = then also p = 0.c. The two roots can only be equal if both the Conditions for equal roots. root. is Eliminating x we find thab the required condition C*4bBC+ 4c. Let these be Conditions for a pair of conjugate complex roots.i. and have o 3 o + 27X/iV 27/z3p = 0. i.r/3/z. root. Since the sum and the product of two conjugate complex numbers are both real. Z?=0. O and H are (a) a real root. Thus the equation (1) can have a pair of conjugate complex roots only if its coefficients are real. On the other hand. Condition for a purely imaginary 2 &. Conditions for a purely imaginary root. The reader should verify this Moreover. b2 &.fi = 0. (c) a pair of conjugate roots (b) H=\+p. real coefficients.rtc = 0. we must have =0. which has In this case there may be three purely imaginary roots.e. If /z is not zero then the purely 2 3 3 If p. If /z is not zero. or if c = 6 2 B CVbB. These conditions are equivalent to the single condition c + CY=(& + Z?i) 2. it being given that the equation has If a purely imaginary root. (7=0. . of conjugate roots 15.gt.= 0. Consider the cubic equation where complex numbers. then the real root is (a) Conditions for a real root. i. if k is positive we must take like signs. G = p + oi.7 This is easily found to be  bBC 4& 2 c=0.92 COMPLEX NUMBERS Squaring and adding we obtain [ill We must choose the signs so that has the sign of k i. In each case we can either find a root (in which case the equation can be reduced to a quadratic by dividing by a known factor) or we can reduce the solution of the equation to the solution of a cubic equation with real coefficients. p imaginary root is (p/3/i) i.e. Condition for a real real. Conditions for a pair of conjugate complex roots. If #2 + 2 (b + Bi) # + (c + CY) = 0. &.=0. 27/x o. if b 2 conclusion by means of the explicit expressions of the roots.
Then the KPA. varying supposing that from TT to +TT. have no general equation of any order with complex nor pairs of conjugate complex roots. it is easy to see that am and z b = Tr0 am 7T z1 a + 0. we obtain The system of equations obtained by is a parameter. and z . An coefficients will in Coaxal circles. us consider the equation \zb . unless J 1 7. .46] COMPLEX NUMBERS . The locus defined by the equation  am where is za = 0. is the principal value of the amplitude z l} in the figure are equal. b. . is the arc By writing 7r0. is the real part a of the complex roots of the and show that a has the same sign as G. KBP are similar. and A PB = 0. How many con ditions must be satisfied by the coefficients in order that the equation should have (a] a real root. Prove that if z3 + 3Hz + G=0 has two complex roots then the equation has one real root which original equation 18. 0. B. Prove that the remaining root is a pair of conjugate complex roots. the point in which the tangent to the circle triangles K be ABP at P meets AB. jr + for the other three arcs shown. 26. In Fig. 26. and so AP\PB = PK\BK= KA\KP = X.(1).9. 0. Examine the case in which A 3 = Q. . chosen. where a^A^ + A^i. represents the system of circles which can be drawn through the A. (6) a pair of conjugate roots ? real roots 19. has The cubic equation #3 + i# 2 + atfc + a 3 = 0. constant. let a. 93 .lt. 16. where X Let is a constant. &. J3. If the two circles shown are the arguments of P P . APB.z be the arguments of A. P Then if am^laP*. . It should however be ob points served that each circle has to be divided into two parts to which correspond different values of 20. Now let Fig.
.. Bilinear Transformations. (X....... (3).. If a = a+/3i then to the point (#. is In this particular case the relation between corresponding figures easily defined. F) describes a curve in its Thus plane.j&...gt... and of equations obtained by varying X represents a system of every circle of this system cuts at right angles eveiy circle of the 19. This gives x pX. two figures differ only in that the (#. . 2 and so positions of is constant. point kind. where z=x+yi and Z=X+Yi are two complex variables which t we may z suppose to be represented in two planes xoy XOY.. ....... very The (X.. y=pl The two figures are similar and similarly situated about their respective origins. K is a fixed point for satisfy the equation Also KP =KA...... very large or very small then the circle is a very small circle B a system of coaxal circles of the common is called a system of coaxal circles of the If X or B in 21...94) COMPLEX NUMBERS [ill all Hence which KA/KB = \Z the locus .. y) describes a curve of any kind in its plane.... containing A its interior......f&.. Such a transformation is called a translation. F). is The system of Ex.. passage of this kind from a figure in the plane xoy to a figure in the plane XO Y by means of a relation such as (1) between z and Z is called a and A transformation. Finally consider the equation 2 = (cos It is clear that = \Z\ \z\ and that + ^sin() Z . but the scale of the (#. Such a trans &... is system of Ex.....KB... y} figure.. &.. P Hence of P is a circle whose centre is K..... and a distance F) downwards.........lt.. is real..gt. y} Such a transformation is called figure is p times that of the (X. and conversely. If (#.lt... (2).... A and being the limiting points of the system. as the (#.. (1).. and orientation is shifted a distance a to the left. but figure is the same in size...... 19 The system called of Ex.. .... The system circles. y} figure is the (X.. y) corresponds the point (X. 20 limiting point kind... and therefore (1)... Y} figure turned about the origin through an angle in the positive direction...lt.. one value of am z is am Z+ and that the &. Now where p consider the equation * *=PZ .. shape. Consider the equation z=Z+a .. formation is called a rotation.. To every value of corresponds one of Z. F) figure.... The general linear transformation (4) .. = r a magnification...... to any figure in one plane corresponds a figure in the other....
.. Finally consider the transformation This is equivalent to the combination of the transformations i... be proved in a variety of ways. the general linear transformation is equivalent to the combination of a translation.... a magnification. and to pass from \z\ the y) figure to the (X. We may assume the wellknown circles theorem in pure geometry. 7} curves which correspond to (2) straight lines through the origin. the (#... (5). Z = (cos (p + ism&.. p= b * pd pc a pc a z X..lt. that inversion transforms into circles (which may of course in particular cases be straight lines).. use the results of Exs.... 95 For.. the symmetrical figure on the other side of ox}..46] is COMPLEX NUMBERS (1). (1) circles whose centre the origin. if  a combination of the three transformations a \=p and am a = 0. 19 and 20. and then construct the image of the new figure (x. = l/R and am z = 0.p)Z.. 22. then *=VZ . o = b  ad . _*6 cz a The general bilinear transformation is the most general type of trans formation for which one and only one value of z corresponds to each value of Z. If. we obtain where .. with unit radius of inversion. z = (l+Z)/(l  Z\ and draw is the (X. y] circle is Or we may and we substitute for z in terms of Z. Next let us consider the transformation If \Z\ =R and amZ^e...e.. to a certain combination of transformations of the types already con sidered.e.... and conversely. Y] figure we invert the former with respect to o. in the axis ox (i. Solving for The transformation (6) Z we obtain is called the general bilinear transformation. we Thus can replace (4) z by the three equations z=z + b. =pZ .. e. and a rotation...... (3).. 23... This may The general bilinear transformation transforms circles into circles. Consider the transformations = l/Z.. (2)......g.
. am {(z TT (cf. may be put what in the form or. If one ratio is X.96 24. then the six distinct cross ratios are X. . if 2. .266 = 0. If the points A lt A 2 are given by az 2 + 2bz + 26. 1/(1 X). . X/(X. . and the points A 3 A 4 by a z 2 + 2b z + c =Q. (X.1). ^31 24 by permuting the suffixes. The cross ratio faz Zt z3 z4 ) is denned to bo If the four points Zi. 1. But this is equivalent to OA l . z2 z 3) z are on the same line. this definition agrees with that adopted in elementary geometry. 1 X. all are real and the four points lie on a am (z 1 zi )(z2 z3 ) must have one of the three values TT. COMPLEX NUMBERS [ill make is the circle \a 25. . so that am {(z2 . l z3 )/fa . The four points are said to be harmonic or harmonically related if any one of these is equal to 1. . six groups of four equal cross ratios. TT. . A l and A 2 being separated Also A 1 A 3 IA 1 A t =A 2 A 3 jA 2 A i Let be the middle point of by A 3 and A 4 The equation AsAi. (9J 2 make 2 equal angles with A 3 A^ and OA 1 OA 2 =OA 3 =OA^. 0. 2. = \c\. There are 24 cross ratios which can be formed from 2 l5 z%. and is the middle point of A 3 A 4 and then OA lt OA 2 are equally inclined to A 3 A 4 and acf + a c . Y) plane Cross ratios. In this case the six ratios are 1.1)/X. These consist of . Z 2 Z \) 1) we have the two equations The four points A^ A 2 A 3 A 4 lie on a circle. The condition that the transformation z = (aZ+b}/(cZ+d} should #2 + ?/ 2 = l correspond to a straight line in the (JT.z^} and Ex. \. . 19). c=0. OA 2 = OA 3 2 =OA^.z3 )/fa .4)} must either be equal or differ by If (ziZ2 . is the same thing. 1/X. served that the relation between the pairs A l A 2 and A 3 A is the middle point of A 3 and ^1 are Hence. equally inclined to AiAfr and J3 O . Hence OA 1 and . AO AJ^O ^M 4 A^. If any cross ratio is real then For in this case circle. It will be ob is symmetrical. . . 2.
+ n =sec $ m sec &. Y) plane. r=R m sectors in the (X. COMPLEX NUMBERS AB. equivalent to the given condition. and 6 and me round the origin then z . (/&. rt + sec m tan . 12) we can find real numbers a /3 y such that a +/3 f/ = ~ zi) +ft /(z2 z4 ) + y /fo ~ z anc* it is easv to prove that this is 7(*i to . . The condition that four points should condition is .. If /(*). Hence and (Ex.] 30. lie on a circle.gt. 1909. y} plane corresponds to any one of To each point in the (x.lt. z3 lie on a circle through z4t the corresponding points (Ex. then CD bisects the angle A QB and QC 2 = QD* = QA.gt. (1) a complex function of curve t. Another form of the condition is that it should be possible to choose real numbers sufficient a. numbers Prove the following analogue of De Moivre s Theorem for real if fa^ fa. fay is a series of positive acute angles such that : tan $.gt. Y} The whole (#..46] 27.i=ll(zzZi\ ^3=l/(z3 z4 ) lie on a straight line. points in the (JT. (0 are two real defined for a certain range of values of t.) 28. /3. l . y) plane correspond plane..gt. Prove that. and P. 25) this condition is also necessary. n. rt +tan fa and tan fa n + sec $ m = (tan m tan m f sec fa) (/&. middle points. i)=Q&. Q their . . A that one (and therefore all) of the cross ratios should be real (Ex. each of angle 2rr/m.gt.] 29. if AB bisects the angle CPD and PA 2 = PB 2 = PC PD.n + 1 =tan &. then tan sec (p m + n = tan ?u m sec (/&. y such that 1 8 1 0. 21). 97 CD are two intersecting lines in Argaud s diagram. (/&.. If Z describes a describes a circle round the origin m times.lt. In this case circle by a multiple of 27r. differ The transformation z=Zm . y [To prove this we observe that the transformation Z= \\(zz) is equivalent an inversion with respect to the point 24 coupled with a certain reflexion If zlt z%. .QB.p m sec fa {sec fan tan fa t &. Complex functions of a t functions of a real variable real variable. we call . Trip. m m 31.lt. (Math. Z = ll(z1 zi \ Z&. [Use the method of mathematical induction.. . We can represent it graphically by drawing the H.lt.
As t varies Z describes a straight Hence z describes a circle. ^ TT. t which correspond to the two produced segments of the u where p t If and radius p. (i) we can (1) and so determine the curve Let aiid b are z=a+(ba)t. =a .gt. 6=/3 + /3% then ) t.gt. the equation z n an integer. Thus we define tya or w a1/M where n . Solution of the equation &. to the range of values of z=a The t seg from Find the values of line.lt.a The curve is the straight line joining ment between the points corresponds to 1.). and and n integers. with complex equations. The equation z=a + 2bt + ct* if b/c is real. It is. when a a complex number. #=a + (/3a). or rational function of t. B + B i=b. We present. Z=t. is 47. is positive.lt.j&./&. and a mln where t m is 48. If where a a = a + a i. represents a parabola generally. however. [Eliminate from where A+A i=a. (iv) Let z=(a + bZ)/(c + dZ).lt. TT &. &. y = a + ($ . y are real. where a. C+C i=c. as (a 1/n ) m These definitions do not prejudge the question as to whether there are or are not more than one (or any) roots of the equation. the axis of X. represents a conic section.lt. represented by coefficients. up to the mln tya. &. attributed any meaning to symbols such as have not. /3. As In general the equation z = (a + bt)j(c+dt) represents a circle. 22. (iii) This can be proved by calculating x and y and eliminating but this process is rather cumbrous. natural to adopt the definitions which are given in elementary algebra for real values of a. express it in the form the function. viz. Let where p is a = p (cos + i sin is an angle such that positive and &. line. a straight line (v) The equation t z = (a + 2bt + ct 2 )/(a + 2pt + yt*). as a number z which satisfies .lt. A simpler method is obtained by using the result of : Ex. zn = a. If .98 COMPLEX NUMBERS [ill the equation of the curve may be obtained by eliminating t between these If z is a polynomial in t. is a positive integer.] Roots of complex numbers. a . complex numbers. then the curve is the circle of centre c varies through all real values z describes the circle once. the points and z=b.
. Examples XXII. 2.) .gt. ..lt. (4&.gt.gt.gt.. n 1). roots are i.gt. is value of 9 2p?r + * matter of indifference. Aas 7i = a = p (cos + i sin 0) roofc ancZ n (wfo/. Prove that Prove that (x+ya&.lt. r = %p.lt. satisfied it is necessary and 6 sufficient that nd &. + 2k7r)/n. Driven by z = r (cos + i sin 0).gt.t&. These numbers are called the nth roots of unity. col~ .gt. 1 are t + i sin (2qir/ri) If = 0. zn where p and q are integers.c/&. the + 2q7r)/n. (l + W3). 1.lt. cube roots are i.5+z&../&. or = (&./n) + i sin ($/n)} = (q is is called the principal value of J/a. The two square roots of 1 are 1.gt. 1.gt. l. J(liV3)&. . 3 )=#2 + 2 ?/ +. . n.lt.. the equation takes the form cf&. 4. the principal value is unity itself. 1. 1.. five fifth roots are it x/5 2.gt. and in this the value of ^? is a ((/&..z 3./&. The only nth root of p is possible value of r is yp. \//3 The particular root {cos (&.p = The n roots of the equation xn = cos (ZqTr/n) of particular interest. That these n roots are in reality all distinct is easily seen by plotting them on Argand s diagram. where = + 2^7r)/. the four fourth 1.p + 2&?r... the ordinary arithmetical and in order that the last two equations should be .. + i sin &.n.gt. and Qq&. the three 1. Hence the equation (f&.. and the 1. + za%) (#+yo&. r n (cos nd so that + i sin nO) = p (cos cosn0 = coscf&. we see that the nth roots of unity are l 1. (# = 0. The nth roots of a are the products of the nth roots of unity by the principal value of %/a. rn = p.s 2 3/2 zxxy. . a&. . 72 . 1 . (1). col. The case in which a=l. .1).4648] COMPLEX NUMBERS (cos 99 we put z = r + i sin 0). we write co n for cos (2Tr/n) + i sin (27r/n).lt. sinft# = sin&.gt.lt. . where k an integer. n . If k=pn + q.gt.
4.+ a 2 ^ . is equal to *2 . Show Show that (x** . Trip. 14 that the roots of are like or unlike signs being chosen according as that this result agrees with the result of 48. possible. Misc. For this construction will be possible if and only if we can construct lengths measured by cos and this is possible (2ir/ri) and sin (Ch.100 5. by ruler and compasses. 10.lt. 8 Resolve ** 1 * + !.9 + a 2 = {# . There are other by special values of n for any power which such constructions are esting being n = 1 7. 12. 6. ".a 2 are  ^ cos The factor ffaw^ tf is #+. i. 5. The problem ^ form tho t t(l+V*X is problem of inscribing a regular by Euclidean methods. Find the roots of the equation # 2x3 + 2 = 0. and split up each all of the last two expressions into n 6  factors. the most inter . involving square roots only.e.2^a ".Zxn a n cos ". cos &. 22) if and only if these numbers are expressible in a form (/) . II. Show that x^ . 8. 1910. xxi. evident that the construction is possible for any value of n which can be found from these by multiplication of 2.2xa cos .] 9.a w (cos 6 + iam 6}} [x n a n (cos6. The factors (tfaco 2 .a 2 )/(# 2 .2m cos [Use the formula ^2 . and 15. (xa&. COMPLEX NUMBERS It follows [ m from Exs. Euclid gives constructions for n = It is 3. 8. Exs.a2) is equal to a2o* cos [The factors of tf2m . 6.] J. is positive or negative.J*) taken together give a factor 7. ^ +a + a2 J and ** + 1 + * +1 into factors n a  similar way.i sin 6)}.lt. 2 2a#cos(s7r/m) + a a.) in a numerical of finding the accurate value of involving square roots only. (Math. as in the formula * algebraical equivalent of the geometrical polygon of n sides in a circle of unit radius 10.
are described on the sides of a ABC. say. then either r=l or both triangles are equilateral.lt.z ) = 0. Express If X. Cis(f7r) = 3.gt.] 3. MISCELLANEOUS EXAMPLES ON CHAPTER 1. Show that the centres of gravity of ABC. positive negative direction.^y z 0. the values of (cos f i sin 0)^ Raising each of these expressions to the power p (where p is any we obtain the theorem that one of is cos (pO/q) a is any rational number then one of the values of (cos cos a6 4. 6.lt. V JL . are coincident. # +y&.] 3 2. III. CAY. D are four points in a plane. COMPLEX NUMBERS The general form of De Moivre s Theorem. T. (x+y+z) in terms of a.B 3 +zo&. then . through f7r an angle the =a&.lt. 49] 49.gt. If XYZ are similar. c. and 2 rr V6 . . [From the equations say. X Y Z are two triangles. or that + i sin 6)a if is sin ct&. . This is a generalised form of De Moivre s Theorem (45). AB. I T7/ r/t /j =A ~r7~V Zi V 7A YV V = A I A. be the triangle. and if ABC. we deduce 2 !/(/ . J . = 3. 4Z) &. /?are points on the sides of the triangle ABC.48.] 4. The condition thatl triangle (xyz) should be equilateral is that [Let X YZ TT l/&. is It follows from the results of the last section that if q llq is integer then one of the values of (cos 6 + ism 6) a positive cos (#/) + sin integer positive or negative). C. 5. ABZ XYZ fj [We have (^c)/(6c) = (ya)/(ca) = (06)/(a6) = X. we nave xz=(zy~] 3 or xz=(z y] w 3 Hence + za) 3 =0. 3. or 2# 2 . Now apply the result of the last example. xxn. The result follows from Exs.i + i sin (pO/q). in 3 The displacement 1ZX or is YZ turned Since Cis . If X YZ. such that BX/XC= CYI YA = AZ/ZB = r.gt. triangle Similar triangles BOX. 5.gt. BC BD CA + CD . then both triangles are equilateral. 3 =0 or ^fya&. o&.
then CD is parallel to the external 7. Another way of stating the az 2 +2/32 +y=0.14^31. . concyclic in the Argand diagram.2. [If CP. in terms of the coefficients.24^12. result is : if and z2 are the roots of then 1 K 10. z z . Then we have (Xl identically Hence 6. z where 2 2 ) (z 3 (z 1 23i 14 4 ) + fa z 3 } (z 2 are the roots of the equation. if M }. 24 [ill Z). The four roots of a a4 + 4a 1 #3 + 6a 2#2 + 4a3#+a4=0 will be harmonic 0. be the complex numbers corresponding to ^4.102 [Let j?!. then 12. its foci. [Use the same as in the last example. and has two real and two complex roots. CD are conjugate semidiameters of an ellipse and S. 8 that a + V(a 2 & 2 )MaV( ) 2 & 2 )= a + b\ + \ab\. B.] identity the points z. z z . . HP= CD\] =2{ja 2 Prove that a + 6 2 +a6 2 + 2  6 equivalent of the geometrical theorem that. If z 2 + z 2 = 1. + 1*2 Show of the equation z 2 + az + b that the necessary and sufficient conditions that both the roots = Q should be of unit modulus are a&. Deduce Ptolemy s Theorem concerning cyclic quadrilaterals from the fact that the cross ratios of four concyclic points are real.] Z =    . ally related if [Express ^23. z are ends of conjugate diameters of an the points 1.lt. 23 .345 and zi.] If ^ 4 +4ai^3 + 6a 2^2 + 4a 3^a4=0 is an equation with real coefficients 11. ax/(a 2 6 2 = 22 we ) have and so (\z :  + ^2 2 ) =2 ( 2 2 a + a 6 2 + 1 b   2 } =  a + 6 2 + a6 2 + 2 zi 2  6 2 . then foci are ellipse whose H bisector of the angle 8. and SP . is [This is the analytical the middle point of PQt Deduce from Ex. then 9. COMPLEX NUMBERS 22 . (7.1. [If +V( 2 & 2 = 2i. SPH. [The amplitudes have not necessarily their principal values. \b\ =l } am 6 = 2 am a.
If (l+x) integer. conjugate and so are the lines Verify the following assertions every real line contains infinitely many pairs of conjugate imaginary points .) tf xnl3 of 3. Solve the equations (a 3 + 1) = 0. then c the point it unless = l.Zxyz. x* . and (6) that the point of intersection of two imaginary lines should be real. an equation with complex coefficients (which of course may be real in special cases). . o are all real (or : made all real by division by a common factor). The aggregate of pairs of real or complex values of x and y which satisfy the equation is called an imaginary straight line the pairs of values are called imaginary points.. .gt. t is called the coordinates of the point (.gt. 3 (x+y+z) 15. when describes a straight line.. Prove the identities (x + y + z} (x + yo&. Find a similar formula for a +a +nxn + a 17. : 14. y}. then n n being a positive ^0^2+^4 18. 16.lt. Trip. If/ (#) = a + !# f + a kx k . ". 3) = a? + f + z . The values of x and y are x and y are real. p Pz+Ps . = l. and \n the greatest multiple of contained in k. 6. Let ax+by + c = Q be Imaginary points and straight lines. = $ n sin Inn. l Sum the series x x* 8!f2i*515I n being a multiple It is 81 #3_ _^8! + &. = $ n cos Jrarr.. When called a real point when a. . + ^^TT (Math. and are said to lie on the line. the point can be line. then n a) being any root of x =\ (except #=1).COMPLEX NUMBERS 103 13. If we give x any particular real or complex value. we can find the corre sponding value of y.8i are said to bo .t?. the line is called a real The points x=a+{ti y y + oi and xa^i.5ax? + 5a2# + (a5 + 1) = 0. x=(at + b}l(t c) describes a complex number such that a circle as t varies. y = y . 3 + 20)3) (x +3/0)3 + za&. . an imaginary line cannot contain a pair of : and find the conditions (a) that the line conjugate imaginary points joining two given imaginary points should be real. 1899. an imaginary line in general contains one and only one real point . .
J(ab). 24.lt. 6. 1912. if a.  23.   1 1  \ \ \. Trip. through the points Z= 1 and Z= from the originally selected part of the axis of x by successive applications of the transformation. = 2Z+Z then the circle \Z\ =l corresponds to a cardioid in the plane of 25. motion of the point which represents value to a large negative value. ad bc=\.^:a */3 ^/3 1 J_ za~Za + Show forms further that. and that in these two cases the transformation may be expressed in the \ Finally. 1912. i. except when (a + d) =4.) . If t varies as in the last example then the point x=^{at + (b/t)} in general describes an ellipse whose foci are given by x2 =ab.1. there will be one fixed point a unless ad. if c=4=0. If (z + l) 2 27. Show that. 2 &. (Math.26. and whose axes are a + 6 and a . to correspond to the interior of 2 . then. show that the only transformations with less than two fixed = = points are of the forms (Ifz) (1/Z)+ A z Z+K.104 COMPLEX NUMBERS [ill 20.b But if a = b then x describes the finite straight line joining the points *J(ab).gt. c. Discuss the transformation z=^{Z+(l/Z )}. Trip.I + J(t*p\ as z&. when $ 2 v (^* 2 ) denotes the positive square root of t*t2 discuss the &. z = Z+K. Prove that if t is real and z=P.gt. d are further restricted to positive integral values (in cluding zero). there are . 1911. made ci)l(Z+ci)} . points unaltered by the transformation.  \ \ \ 21.1. Show tion may be expressed in the forms 0a_ 7 . Prove that the relation z axis of x between the points z=\ and = (l+Zi}l(Z+i) z=l transforms the part of the into a semicircle passing Find all the figures that can be obtained 1. when . Assuming that. /3. and the inside of the circle to the outside of the parabola. showing in particular that to the circles JT 2 + F =a 2 2 correspond the confocal ellipses = 4/Z then the unit circle in the zplane corresponds to the 2 = parabola ^cos 0 l in the ^plane.e.) The coefficients of the transformation z=(aZ+fy/(cZ+d) are subject to the condition two fixed points 2 a.) 2 If z z. t diminishes from a large positive (Math. 22. (Math. ". z is 2 2 represented by a point which lies on the circle % +y +x=0. by means of the transformation z={(Z the upper half of the 2plane may be a certain semicircle in the Zplane. if c = 0. Trip. when there is only one fixed point a and that in these two cases the transforma that.
and ZQ denotes the conjugate of an inversion with respect to the circle a2 + d z + bc&.lt. d are real and Z. b real.". 29. p 2 are the distances of Z from the points 1. If z=Z 2 . 1. the unit circle. o&. transforms the circle  z \ =I into the boundary of a circular lune of angle ir/c.] z=aZ + n bZn . \ = C. points.or epicycloid. z where m. c. We call these the branch points in the Z and ^planes re Show that. Show that which the corresponding values of z are equal. different values of K. The transformation where c is rational and &.lt. Consider the relation az 2 + ZhzZ+ bZ 2 + 2gz + 2/Z+ c = 0. then as z describes the circle \z\ = where sponding positions of Z each describe the Cassinian oval p 1 p 2 = Trace the ovals for pi.gt.gt. take the given relation in the form the reader should satisfy himself that this is the case. the two corre l. 9) to ) + ^( 22 _ cosec 2 w +  z  _ ^2 _ CQsec 2 M ) Express this in terms of 30 as If Z. c (#2 +y 2) _ % ax 2dyb=0. and vice versa.gt.gt. c &.  2 1 . 105 &. The branch points in An ellipse of the form specified is either plane are cosec and . Z describes 31.. if z describes an ellipse whose foci are the branch spectively.lt. =  where C is a constant. then Show that the transformation _ cZ$ (a di) where a.cosec o&. when What is the geometrical interpretation of the transformation 32.COMPLEX NUMBERS 28.cosec G&.gt. given by \z + cosec is o&. 1.. is equivalent to 0. for there are two values of Z [We can. n are positive integers and a. then so does Z. describes a hypo. z This equivalent (Ex. . 6. without loss of generality.
(ii) *Jx. Now * us consider a function. (viii) a certain number of positive integral values of #. height measured in inches of convict x in Dartmoor prison. which is defined for positive integral values of x and no others. consist of (i) all all rational values of x. (iii) positive rational values of x. defined for all values of for integral values only. t) of two variables. (iii) the denominator of x. and convict x. Some were only. (viii) the or. N moments of time into consideration we have a simple example of a function y = F (x. from the time of the establishment of Dartmoor prison to the time of its abandon different ment. and for a certain number of positive integral values of varying with t. (vii) the #th prime number. defined for a certain range of values of t. (vi) the product of fjx and the largest prime factor of x. this number . (iv) the square root of the product of the numerator and the denominator of #.. (vii) all positive integral values of #. In Chapter II we discussed the notion of a function of a real variable x. (v) the largest prime factor of #. : Consider. where x has a definite Thus if we take value. is the total number of convicts at Dartmoor at a viz.. viz. some for rational values some and so on. N) where functions.. (vi). example. and illustrated the discussion by a large number of examples of such functions. as Then the aggregates of values of x for which we may say. (v) all integral values of #. such as all (vii) above. (iv) all N given moment let of time*. 2. is a different individual at different moments of time. This In the last case depends on the time. And the reader will remember that there was functions one important particular with regard to which the which we took as illustrations differed very widely. x. the fields of definition of the these functions are denned values of #. A POSITIVE INTEGRAL VARIABLE Functions of a positive integral variable. . for x.CHAPTER IV LIMITS OF FUNCTIONS OF 50. the following functions (i) x. (ii) all positive values of x.. 1.
especially when the function of n is itself &.50. 2. which would have a meaning even is a function of n. 51] FUNCTIONS OF A POSITIVE INTEGRAL VARIABLE 107 function view.lt. In this case we may write y = &. as explained above. and say that for all other values of x the definition fails.j&.. simply fill in the missing values as we pleased values of the function of : we might indeed simply regard the given the values of the function of x and say function failed for are obviously not all n as all that the definition of the latter theoretical solutions other values of x. .t&. For example assumes the value $ (n} for y x = n.lt. it would of course present no difficulty whatever..gt. corresponding to values of other than positive integral values.3. there is an obvious solution. in any way we please.2. defined by a If for example y = formula. It is called the problem of functional interpolation. It is obvious that any function of x defined for all values of x Thus from gives rise to a function of n defined for all values of n.gt. . Or we may leave values of x other than positive integral values entirely out of account. were n not a positive integer. and regard our function as a function of the positive integral variable n./&. may be regarded from two consider it.$&.lt.we deduce the function y = n z by merely the function y omitting from consideration all values of x other than positive On the other hand integers. In some cases. whose values are the positive integers 1. for all positive integral values of x. (n\ where (f) (n) 2 or cos nir. x^. of 51. could.. But such purely what is usually wanted. values agreeing with those of a given function of %.gt. &. is of extreme importance in higher mathematics. viz. as a function of the real variable x defined for some only of the values of x. x Interpolation. (x} since sin + sin xir ?i7r = 0. What is usually wanted is some formula involving x (of as simple a kind as possible) which assumes the given values for # = 1.4.lt. we naturally take our function of x to be But even in this very simple case it is easy to write down other almost equally obvious solutions of the problem.. such as n y=*&. We may slightly different points of as has so far been our custom.(x). The problem of determining a function of x which shall assume.gt.. from any function of n we can deduce any number of functions x by merely assigning values to y. Were fulfil the problem however merely that of finding some function of x to We the condition stated.. and the corresponding values of y. (n) and regard y now as a function of n defined for all values of n. positive integral values..
to transform the formula in such a x.gt.(x) may be all. but not for If for simplicity This expression has a meaning for some values of x other than Thus from y=n n we are led to y_xx some only of the remaining values of x. It is unnecessary to discuss here any logical which may be involved in the notion of a that a class is class : roughly speaking we may say the aggregate or collection of all the entities or objects which possess a certain Thus we have the class of British property. the reader is probably familiar with the class. This ! a case in which attempts to solve the problem of interpolation have led to important advances in mathematics. In the notion of difficulties first a place.lt. then x* has values of #. numbers. For mathematicians have succeeded in discovering a function (the Gammafunction) which possesses the desired property and 52. But it may be possible way that it does define for all values of x with even denominators. simple or complex. consider the case in which In this case there as is ! is x means nothing no obvious formula in x which reduces to n for x=n. ^T^\^ In other cases &. or positive integers. or members of Parliament. . In this case. for values of x other than the positive integers. . Finite and infinite classes. meaning at we are in a position to say at Thus in this case the problem moment) no We later on how the desired extension may be effected. ( !) = cos nir. Again. we all a meaning for fractional rational confine ourselves to positive values of #. or real subjects. may some values of x be defined by a formula. and the problem of interpolation solved by the function cos xn.j&. Before we proceed further necessary to make a few remarks about certain ideas of an abstract and logical nature which are of constant occurrence in Pure Mathematics. of interpolation at once leads us to consider the question of extending our definitions in such a 6 shall see way that of shall have a meaning even when x is irrational. in virtue of the definitions of j) powers adopted in elementary algebra. defined for jj^^ positive integers. such as ( l) n which (as here in the case of fractional values or irrational values). is if n is an integer.108 In other cases ceases to define for of LIMITS OF FUNCTIONS OF A (n) [iV . it is many other interesting and important properties besides. for example. But when x the present is irrational x* has (so far as all.
gt.j)(n) &. It is convenient to express this by saying that there are an infinite number of But the positive integers. or real numbers.000. Thus the class of British subjects a finite class: the aggregate of all British subjects. II. the reader has probably an idea of what is meant class. as we shall see in TEe next chapter. past.(n) not.000.lt. They have many points of difference from the functions of sc which we discussed in Chap. reader must be careful always to remember that by saying this we mean simply that the class in question has not a finite number of members such as 1000 or 1. is any function of n..gt..000.000 or n any positive if any number we like to think of.lt. and future. Similarly the class of rational numbers. present. or of real numbers. Thus. &. 3. or rational numbers. is the other hand the class of positive integers is not finite This may be expressed more precisely as follows. the number we think of is 1. such as 1000. Properties possessed by a function of n for large values of n.000. Consider. on the other hand. are three possibilities: (n) may have the property P for all values of n. We may now return to the functions of n which we were discussing in 50 51. But there is one fundamental characteristic which the two classes of func common the values of the variable for which they are defined form an infinite class. &. The class of present British subjects. apply. for each of the values n whether has the property P or 1.001 positive integers. mutatis mutandis. On but If infinite. integer. though of course we cannot tell at present the actual value of n.lt. Then there (a) (f&. to functions of x tions have in : as well.000. and that P is any Suppose that which (n) may or may not have. were the methods of the census effective enough. such as that of being property a positive integer or of being greater than 1. has a number n which could be ascertained by counting. 2.. then there are more than n positive integers. It is thisjact which forms the basis of all the considerations which follow and which. 53. has a finite number n. .5153] POSITIVE INTEGRAL VARIABLE 109 by a finite or infinite is Moreover. 1. or for : all values of n except a finite number N of such values . there are obviously at least 1.000. is infinite.f&.
2. if not for assertion (a) is true. which (n) has the property the values of n for which $ (n) &.lt.lt. number may denote these exceptional &./&.. class is finite. In the third case neither Let us consider some particular cases. P &. values of n form an Example.!.(ft)&. (ft) &.lt. in each of the following cases.gt.gt. (ft) &. (xi) (n)=(n l)/(ft + l)..lt.cyO^r . P being the property $ (n) = l . i. P If = (ft) (ft) and P is the property of being less than 1000.l) n P being the property 001. P being the property (n) = 1000 (1 + . or $ (n) may have number a finite (c) N of such values n If (a) the property for no values of : only for neither (a) nor (6) is may be is true.lt.gt. values by flj.lt. or (l) n 10000/?i. Wjy. P being either of $(ft) (n)=p n .lt. 999.  001.gt. has not the property form a finite class. denotes the property of being a positive integer If on the other hand greater than or equal to 1000.gt.. ( &. (6). W2 . (1) Then = ft. true.  &. (a). (n) is odd if is and P is the property of being odd.  the properties &. 1.lt. &. If (6) true. (/&)&.1) P being the property $ $ = I/ft.lt.gt..lt. If = n ft. whether or true : $ (n) = n. (c) is (i) (ii) Consider.l) n (I/ft). and both the odd and the even infinite class.lt. ft. (n) and has the property P.lt.2. P being the property $ = 1 . (ft) ( /rc. ( vii) (viii) (ix) (x) P being the property of being even. the values of for form a finite class.lt. (3) &. Let (ft) has the property P for all values of n. are such that the Let us now suppose that 54.. and let P be the property of being a positive integer. then (6) is true.gt. 1. N We . ^ = . then (c) is true.lt.001 or  $ (n) &..lt./&. &. viz. then (p (n) has the property for all values of n except a finite number of values of ft.1) (I/ft).0001./&. P being the property of being odd. at any rate for all values of n except a finite of such values.e.110 (b) LIMITS OF FUNCTIONS OF A [IV n. (n)=p n . ftth $(ft)=&. = 10000/ft.lt. that (ft) all values of n. .lt. 1 (ft) ( } /ft.lt. 3. ( ()&. For odd and even if n is even. (w)&. P being the property $ 001.lt.&.n&.. (ft) (ft) &. (n) &./&. &. prime number.. P being the property l0(ft)&. P being the property of where p n denotes the being a perfect square.gt. In either of these cases (a) (2) is true. VM P r AU..
Large is in fact a word which. We know now what is property kind that P meant by the n . . : Thus we may say that less all large primes are odd or that l/n . may be taken to be any number greater than n N the greatest of the exceptional numbers it is most natural to take it to be n N +l. such that $ values of n greater than .. or very large. . is than 001 for large values of n . and 400 runs is a 400 is not large score. And the reader must make by no more absolute meaning in mathematics than in the language of himself familiar with the use of the word large in statements of this kind. standing itself. l/n&. N why these N values should be the values &.gt. a number large in one connection is small in another 6 goals is a large score in a football match..53. for large values of assertion (n) has the It is with assertions of this we shall be concerned throughout this chapter. &. and 001 if n 1000. and what is large enough for one purpose : which may not be large enough for another. P definite number. what we mean is that we can determine some &. as the preceding examples first show.. ?? say.lt. the first 1000 values of n being the exceptions and 2. is to say. the exceptional values being 2. &. .gt.. 4. this is frequently the case in practice. in each of these cases the property is possessed That for all values of n from a definite value onwards. 2000. or equal to n (n) has the property for all This number n in the . 2000. It is a truism that in common life .. examples considered above.lt. But whether this is so or not we know that nN (n) has the property P if n Thus the nth prime is odd if n 2.gt. shall frequently express this by saying that $ (n) has the property for large. N.lt.. or all sufficiently large values of n. 54] POSITIVE INTEGRAL VARIABLE is 111 There of course no reason 1. We Thus when we say that (n) has the property (which will as a rule be a property expressed by some relation of inequality) for large values of n. 1 n &.gt. but a large income and so of course in mathematics large generally means large enough. .lt. 6. 1000 if {1 + ( l)}/w&. . &. has common is life. . though. n = 2 being the only exception to the statement. but 6 runs is not a large score in a cricket match.
if we like. and it is convenient to think of the variation of n as accomplished in time in the we shall say that n tends to as &. way of looking at the matter which it is natural to Suppose that n assumes successively the values 1.. venience. or n an abbreviation for symbol being usually employed The phrase tends to like the word successively infinity naturally suggests the idea of change in time. to assume these values at successive moments of time (e. There is a some what different adopt. 2. .112 55. 2147483647). It is convenient to have a short phrase to express this unending oo infinity. There is no number infinity &. (1) that it oo by itself means nothing. The word successively naturally suggests succession in time. .g. However large a number we may think of (e.. .gt. tends to oo oo means nothing at all except in the one phrase Later on we the meaning of which we have explained above.i equation as n is oo because as it stands absolutely meaningless equal to oo means nothing. itself manner described above. this number. and this last . J The reader cannot too strongly impress upon himself that when we say that n tends to oo we mean simply that n is supposed to assume a series of values which increase continually such an and without limit. 3. although phrases con taining (2) sometimes mean something. LIMITS OF FUNCTIONS OF A [lV The phrase n tends to infinity . every case in which a phrase containing the that in oo means something it will do so simply because we have symbol a meaning to this particular phrase by means previously attached of a special definition.g. and we may suppose n.. Then as the seconds pass n gets larger and larger and there is no limit to the extent of its increase. So far in fact the symbol . : n cannot be equal to oo . This however is a mere matter of con The variable n is a purely logical entity which has in nothing to do with time.gt. . a time will come when n has become larger than growth of n. shall learn how to attach a meaning to other phrases involving oo bub the reader will always have to bear in mind the symbol . at the beginnings of successive seconds).
the larger must be the least of values of n values which are sufficiently : when 8 has one value are inadequate when is is it has a smaller.lt. clear that if $ (n) oo .lt. It is obviously convenient to have some way of expressing the fact that any such statement as I/n is less than 01 for large of n is true. for example.lt.gt.(n) of a function of n as n tends to proceed. when we substitute for 01 any smaller number. The behaviour infinity. And so another way of for sufficiently putting the question what properties has values of n ? is how does $ (n) behave as n tends to oo ? large if and n tends to in the sense which &. We shall now &.lt.gt. last statement italicised that I/ft is is what really meant by the Similarly 8 statement (a). I/n&. . in the light of the remarks made in the preceding sections.gt. That this is true is obvious. to consider the meaning of some kinds of statements which are perpetually occurring in higher mathematics. For of n 8 if so that our sufficiently large values of n need 1/8. such as 0001 or 000001 or any other number we care to choose. in that it really stands for the whole class of assertions which j j we obtain by giving the smaller 8 is.f&.lt. Similarly it is true that I/n is less than 10000. And clearly we can do this by saying that however values small 8 may be (provided of course it is positive). one. Let us consider. small when n large. as being slightly the simpler. The H. : statements is small for large values of n. ensure that $(n) has the property P. have already considered the statement . has the property P for large values we have just then n will ultimately assume values large enough to explained. And of course and the larger the large sufficiently large 1/8. 0001 for large values of n in fact l/n 0001 if And instead of 01 or 0001 we might take 000001 or : &. to 8 special values such as 01. there is a good deal in them which will repay the reader s seem. we saw.gt.8 for sufficiently large values n&. or indeed any positive number we like. the two following 56. Let us take (a) first. in fact greater than 100.55. 56] POSITIVE INTEGRAL VARIABLE it is 113 Now of n. then . (b) 1 (I/n) is attention. (a) I/n Obvious as they may nearly equal to I for large values of n. We &. means that the inequality for large values of n I /n 01 is true for all values of n greater than some definite value. I/n is less than 01 This. 00000001. The assertion is however a complex only all be greater than 1/8. n &. I/n&.
if &.114 (b) really LIMITS OF FUNCTIONS OF A [lV !&. however that l/n&. The reader would growing smaller and smaller. cally in the form lim  = 0. are equivalent to one another and to the more formal statement for any positive number. The reader would on. 001 as soon as 72&. Similarly we say that statements l/n is small when n as is large .8 . another way in which it is common to state the facts expressed by the assertions (a) and (b).lt. however small. B . such as 0000001.gt. . Thus the : as n tends to oo l/n tends to tends to 1 as n tends to oo (l/n) .1000./&. 8 for is sufficiently large values (such as *01 or 0001) we attribute to 8 true is obvious from the fact that 1 whatever positive value That the statement (b) (n) &. = l/n. then l/n sufficiently large values of n if 8 is &. better of the argument.gt. of n then the statement is true ".lt. He might begin with 001. l/n tends to n tends to oo . which occurs in the last statement is of course We shall sometimes emphasize this fact by (8). This is suggested at once by 55. Instead of saying l/n is small for large values of n we There is say 1 to and these statements are be regarded as precisely equivalent to (a) and (b).gt.lt. introduce yet another way of expressing this shall say that the limit of l/n property of the function l/n.lt. but would try again with some smaller number.lt.lt.(ii) = l(l/n).~ a statement which we may express symboli as n tends to oo is shall We now We . 0000001 as soon as In this simple case it is evident that the reader would always have the n&. l/?i&. in the form n Q The reader should imagine himself confronted by an opponent who He would name a series of numbers questions the truth of the statement. The opponent would be bound to that reply admit this. then we can find a number n Q such for all values of n greater than or equal to n The number a function of writing ?i ?i 8. 10000000: and so reply that l/n&. or to the still more formal statement any positive number. if 8 is small.lt.. means ".lt. (?i) &.
in all these statements mean nothing whatever by themselves. as n . then n2 &. This statement equivalent more formal statements any positive number. 57] POSITIVE INTEGRAL VARIABLE 115 or simply lim (l/n) = 0. oo given. oo in this sense the use of the symbol suggests that it sometimes be convenient to write ?i 2 + oo for n 2 * oo and oo .j)(n) l(l/n)^l.lt. and we naturally say that oo as n tends to oo and write tends to &.56. &.(n) _W And will 2 ^ _ 00 . &. n] \ lim M \ W. and only acquire a meaning when they occur in certain special connectipns in virtue of the explanations which we have just + oo .gt.lt.gt. n] let is ri*. 57.gt. or lim ^oo or (liW. Finally consider the function is large. when n is large. Now let us consider a different example : Then to the n* is large when n is large . in order to secure greater But we must once more repeat that the symbols oo . however sufficiently large values of n if is . We shall also sometimes write as n * oo simply which may be read l/n tends to as n tends * In the same way we shall write l/n . * ri* In this case but negative. . A for we can find a number n (A) such of that n 2 &. And it is tends to oo natural in this case to say that n* tends to or ?i 2 tends to oo with n and to write .(rc) = ?i 2 . A large. . to oo . generally to use + oo instead of uniformity of notation. A for all values oo n greater than or equal to n Q (A) .f&.lt.
Ill. (n) is nearly equal to I when n is But although the meaning of this statement should be large.lt.gt. between c/&. (n) is said to tend to the limit however small be the positive number 8. . Thus if explicit the condition reduces to l/n&.n Qt which is satisfied if w =l+[l/8J*. however small be the positive number 8.d for n&. whichever positive. though the &. With this notation the definition : may be 8.gt.gt. equivalent to a whole class of statements of the than 8 Ilt type for sufficiently large values of n. we w (8).gt. greatest integer not greater than x. for $ (ri)*l. After the discussion which precedes the reader should be in a position to appreciate the general notion of a limit. : We I accordingly frame the following formal definition I.lt. positive or negative. &.gt. by is &. stands. f however small.gt. we are only considering real values.lt. from I by less than 8 for all values of n greater than or equal to I. and sometimes it is convenient.116 LIMITS OF FUNCTIONS OF A [iV 58.lt. DEFINITION as The function if. we can determine a number ?? (8) corresponding to 8. 8 when n to oo .gt. (n). &.(n) .lt.lt. in a few simple cases. . clear enough after the preceding explanations.x&. I Sometimes we may omit the brevity. n . \ (f&. to write *?&*&.f&. . as it to I a limit as n tends to oo if &. and (x) expression of n as a function of d.f&. it is not. n tends . The reader There * is one and only one case in which the same n will do for all values of i. precise enough to serve as a strict mathematical definition.gt. (8) so that (n) I I \ &.  (n) I . and agrees with the definition of the though at present stated modulus of (/&.gt.gt. Roughly we may say that (n) tends &. smaller 8 is the larger.gt./&. can find n that c/&.f&. (n) differs from I by less than 8 for sufficiently large values of n that is to say if.lt.lt. Here and henceforward we shall use [#] in the sense of Chap. is. Definition of a limit.e./&.lt. will find it instructive to work out.lt. It is usual to denote the difference positively.f&. nQ (8). in fact. given any positive number.gt. (f&. = (n) . and write lim n*&./&.lt. as the &. as a rule.gt. d. after a certain definite value n (8).gt.lt. II. (n)  I.gt. then we say (n) tends to the limit as &. will be this value n (8). It is equal to (n) (n) and I or to I taken &.gt. j more shortly as follows if. the = ljn then = 0. such that (n) differs tends to co (f&. as given in Chap. differs from I by less This statement has to be true for 8 = 01 or *0001 or any positive number and for any such value of 8 it has to be true for any value of n.
60.p . y= at distance 8 from Then Km (j&. &. illustrated geometrically as follows.lt. . iike the functions w or n which tend to a limit as n must now frame corresponding definitions for ?i 2 .5860] If. is The function $(n) is said to tend t o + oo when any number A.lt. assigned..N and all positive values of d. say equal to C. &.xV.lt. so that the inequality \&. and figure.lt. (?i) consists of a 3. DEFINITION II.. The graph of &. and the parallel lines y I 8. appreciating the point of The reader should by now find tend to positive or no difficulty in {positive infinity) ivith n.d &. &. 117 from a certain value JV of n onwards.gt.gt. we can determine n (A) so that (n) A when n ^ n Q (A). if.. it is &. 27. So much oo . negative infinity. I. in such shall find all points to the right of it. however large.lt. y=l8 Fig.gt. when once these lines have been drawn. for functions of tends to .gt. no matter how close = n0t as in the together. we can always draw a line x a way that the point of the graph on this line.lt.f)(n) &. they may be We this geometrical useful way when we come of looking at our definition particularly to deal with functions defined for all values of a real variable and not merely for positive integral values. (n) is constant. lie between them. if. then it is evident that so that $ (T&) is constant for all such values of n..JVe 2 functions which. then (7=0 for n^N. number of points corresponding to the values n= Draw the line y = it.lt. for l\ n&.lt.N and all (n)=l when n positive values of S. 2.gt.8  &. 59.f)(n)C\ evident that $ (n) And if is satisfied for n&.. The definition of a limit may be I. POSITIVE INTEGRAL VARIABLE &. (n) = I.
107. 11. The reader should be careful to observe the following points. and indeed we above. (1) finite We may obviously alter the values of $(ft) for any number of values of n.gt.gt. we cannot as a rule alter an injutite its number l/n of the values of oo . 237. 109. &. We may 61.lt.f&.gt. LIMITS OF FUNCTIONS OF A [iV &./&./&. form of statement is if we can make . would no longer be true that lim (n) = 0. less precise. 7. (n) * + On the other hand &. 11. So long as a finite number of values only were affected we could always choose the number ?? of the definition so as to be greater than the greatest In the examples of the values of n for which (n) was altered.lt. original function l/n./&.lt. (n) tends to + oo we (n) write * (/&. &. and not merely for some such values. in any way we . Similarly. (ri) A for sufficiently large values of n. for the function 7. 101.gt. 237 and equal to other values of n. altered the function it by changing its value to 1 whenever n is a multiple of 100. 237. 109. 101. n = n (A). Some points concerning the definitions.lt.lt.lt.lt.gt.gt. for instance. lim $ (n) = 1.gt. leave it to the reader to frame the corresponding definition for functions which tend to negative infinity. please. 2. This is (n) as large as we please by sufficiently increasing n open to the objection that it obscures a fundamental point./&. When &. without in . we could always take n &. equal to all = 0. in using this form of expression if harm we are clear But there is no what it means. (2) it is true that c/&.gt.gt. For this function. and to ft 2 otherwise. just as for the &. viz. l/n for functions from l/n by altering a finite number of For instance we may consider the function (n) which 3 for n = l.gt. however large A may be.lt. should be compelled to do so as soon as our imaginary opponent . 2. oo .118 that is to say if. &. thaT (n) must be greater than for all values of n such that A &./&. + oo . (n) without affecting fundamentally behaviour as n tends to If for example we &.gt./&./&. Another.lt. 107. the least affecting the behaviour of $ (n) as n tends to oo For We may deduce any as n tends to oo tends to example l/n number is of new its values. (n) which is equal to 3 if n &.gt.
lt./&. The (n) for.j&./&. all larger values It is obvious. \ this particular case.8 when is not true that &.f&. \ which are greater than (4) \&./&.(n) ?? all the S when n ^ n multiples of 100 are exceptions to this statement. This limit is itself the value of the function for an infinite number equal to is when n I of values of n.lt.gt.lt.lt.lt.gt. But now however large n Q may be there will be greater values of n for of which &.gt. for example. is of course 8 not merely when n = n but when n = n Q) of n. (n) is I. The limit if &. that.gt. : limit always greater than I. it &. Thus (n) 0.lt. I may itself be all (n). On not) be the value of the function sufficiently obvious in the case of the other hand the limit itself need not (and in general will for any value of n. (n) has been altered.lt. in (2) and (3) above.  l is some If however I) /n. for n and for &. is a multiple of 100 and to I/n otherwise.gt. (/&. \ 8. to instead of to 1. all multiples of 100. we can replace Thus the test whether I/n tends to the l\ by oo is simply whether ~L/n&.60. it : (n) is the function last can choose w so that \(f&.8 when n = ^ as n tends to If &. but $(ri) (n) times positive and sometimes negative. then given S we we have only to choose a sufficiently large value of n n=n which is not a multiple of 100. it is obvious that lim Again.gt. in the (5) (f) form  $ (n) &. i. then I is again 0. &.lt.lt.gt.f&. altered the value of the function. This is &.j&.gt. when n is a multiple of 100. (n) &.lt.gt. if considered.lt.lt.lt. viz. =( n &. in (n) (/&. The limit of this function as n tends to x is still obviously zero.gt. we should have obtained a function (n) which &. for example.gt.lt.f&. But.gt./&.lt. .gt. (3) absolutely essential that In applying the test of Definition I we should have I (n)  &. = (n) for one of the actual values of values of n. The reader cannot impress these facts too A limit is not a value of the function though it is it is something quite distinct from these values. In such a case we must I state the condition in the form 8.  &. = 1/n.lt.lt. but the function is never equal to zero any value of n. 61] POSITIVE INTEGRAL VARIABLE 119 56 had assigned a value of S as small as 3 (in the first example) or a value of A as great as 3 (in the second). limit is zero .(ri) . when n Q is thus chosen. defined by its relations .f&. if we had.e.(n)\&. strongly on his : mind.
then nk 1 for all values of n.gt. since sin \n7r is never than 1) the limit is numerically greater equal to the value which (n) assumes for all even values of n. &. (n)&.lt. The reader will find it instructive. If k 0. Euclid.gt. where k is a positive or negative integer or rational fraction. after a certain value of n. + (l/w).gt. This number is t since the remainder when evidently not divisible by it is divided by any of is It is therefore not divisible is by any prime save 1.11 . &=4. &.ll. sufficient illustration of this is given by can only tend to + oo or to it maintains a constant sign.. as n tends to oo : Consider the behaviour of the following functions 1. n is may be always numerically very large when without tending either to + oo or to .oo $ (n) = ( n l) n. (l)/w. If k is negative. of . Euclid s proof number of primes. 7.oo very large A (6) A function .fr(n) l/n. even in so simple a case as this. for all values of n (save n = I. then nk tends to k oo with n. and so on. but the c/&. k + = = = .. then w 4 10000 when 100000000 when ft 101. If there were only (n) would be defined only for a finite number first of values of n. 2.gt.. which It is contrary to our hypothesis.gt. let them be many 1 as follows.lt.. N}. l 1.& n&. &. as was is 1. Hence .2.7. infinitely there are only a finite N. however k when large. If k is positive.g. If e. any of 2. . (ri)=n . 2. (n) : for = &.lt. x Examples XXIII.gt. 11. We wish to choose n so that n We have in fact only to take for n any number greater than #A.. Hence lim n k I &.. 3)./&.gt. rc&.LIMITS OF FUNCTIONS OF A to [iV them and may possibly be equal to some of them./&.n . &.gt.3. n*&. A function if.gt. shown by If primes.0. and therefore itself prime. these numbers is N 1. &. values assumed for odd values of n are all different from the limit and from one another. . then lim n = 0. 5. 5. 3. For the functions * the limit is = 0. 1 + {(sin J?wr)/7i} (whose limits as w tends to oo are easily seen to be and 1.n moreover obvious that (n) * f oo . Consider the number + (1 .lt. &.fr(n)=p nt where p n is a finite number of primes then the nth prime number.lt. equal to all the values of &.5.gt.gt. (w) for (n) = (sin Jwr)/ra.lt. 2. to write down a formal proof that the conditions of our definitions are satisfied. : 1 + {(!) it is not equal to any value of &. Take for instance the case of Let A be any assigned number. There are however. 3.
lt. a continual (f&. as in the case considered in the last example above.gt. 6=plq. : a function oscillates denned in a when it does not do . as Then sin (npirlq) = ( l) ap n increases from to q 0. = l/{ft . then lim &. nor to + oo evident that when this is the case (ft) .gt. goes from Zq to 3q these values are repeated and so also 3^ to 4^ 1.gt. 1. where p and are positive integers.(ft) for all values of ft. is the quotient and b the remainder when n is divided &. . Let be the number of primes less than ft. 8 be rational. (n) does . (ft) to 0. then first (%)(ft) is &.gt.(n) A repetition of a cycle of values. Oscillating Functions.. i/r (ft) = 1000000/ft. (f) (ft) = (sin ft#7r)/ N 2 (a cos n6 + b sin 2 nff)ln t where a and 6 are any real numbers. 4. purely negative manner certain other things. Here again (ft) + oo .lt. as n tends to infinity. . nor to oc . ( }.lt. If B is integral then = &.lt.l) n ft .? to 1.61. oo .lt. where a is any positive number. nor to .lt.lt. and &. 5. &. Here and so on If f . When as ft ft increases from &.(ft) = (sinft#7r)/ft. (n) oscillates as n tends to function certainly oscillates if its values form. sin nBir 1. /ft. sin (jOTT/y).l/ft. (ft) * + o . real number. &.(ri)\ &. It is cannot tend to a limit. DEFINITION.. n{l(. the second to 7. e. (ft) Suppose. But of course Oscillation it is may oscillate without possessing this peculiarity.oo . and Next by let Let n = aq + b where a q.lt. where 6 is any and lim $ (ft) = 0. as n tends to oo we say that &. 62] 3.lt. sin (bptrjg).f&. &. .  since 8. oo . Here \&. not tend to a limit. &. 9.?  form a perpetual cyclic repetition of a finite series of different values. nor to + oo . $(ft) therefore lim = sinft#7r. being in fact larger until ft = 1000000.. then. These conclusions are in no way affected by the fact that at much larger than ^ (ft). 6.gt./&.lt. and so on. takes the values .f&. p even .lt. for example. POSITIVE INTEGRAL VARIABLE (ft) 121 &.^. $ (ft) = 0. When &..lt.l) n The first function tends + oo the third does not tend either to a limit or to f oo ( }.lt.. sin (2p7rlq\ sin {(y l^Tr/g).lt. It is discussed in the next set of examples. Thus the values of (ft) 1 2&.? . is The case in which 6 irrational is a little more difficult.1). (n) =0 : and if ^ (ri) = ft/1000000.gt.lt.$&.f&. 62.lt. (ft) = [aft].
0. + (l) n .lt.122 LIMITS OF FUNCTIONS OF A [IV is The simplest example which of an oscillatory function given by $(n) = is (!)".. (1000000//0 n %. (n) is numerically = (!) 100 + (1000/w) When n &. are further illustrated in the following examples. .. (n) oscillates as to oo . In this case the values of 0. &. 12. n {1 + ( . 8. (l) n n. I + (1/5). 4. l+(l/3). as well as those of 58 and 60.f&. Consider the behaviour as n tends to oo of the (_i).lt. If &. Examples XXIV. it is nearly equal to +1 or does not tend to a limit or to + oo or to (n) It oscillates: but the values do not recur.lt. 1000000 (!) + (Ijri).. (n) are . large every value co is . or to 100. 0. is large. DEFINITION. 2.gt.gt.lt. nor to + nor to And in this case we cannot any not limit rise. 3.gt... beyond which the numerical value of the terms does The distinction between these two examples suggests a n tends further definition. not tend to a assign limit. following functions: 1. The odd are all zero and the even terms tend to +00: &.lt. and and be less to observed that in this case every value of than or equal to 3/2. 5. l + (l/4). The numerically (n) every value is nearly equal to 100 = l). 4. equal to + 1 when n is even and to In this case the values recur cyclically. 1 when n But consider is odd. greatest value is 900 (for n ?l But now consider 3.lt./&. 1000000 + (l) ( 1000000%. These definitions. 4.. i. 2. K for all values of n. l + (l/2).lt.. 16. = (oo l) n. the values of which are . &. terras l) n ( 1000000 ?t). the values of which are 1 + 1. When n obviously therefore is &.(ri)\&..l) n).e.gt. for it does .. Similarly (w) oscillates. function oscillates.f&. (n) will be said to oscillate finitely or infinitely according as it is or is not such that all the values of $ (n) are number possible to assign a K numerically less than K.. This oo . 5 + 3(l) tt.1. 0. (n) oscillates infinitely. . 1./&. then &..lt.
than + /v/(A + l).gt. as the sum of their squares is unity. &. In the first place (/&. Hence &. (%) The case in which 6 is irrational is a little more difficult. Similarly ^) 07r = cos ndrr \ cos  Bit \ + sin nOrr 1 sin  6ir and so each of cos nQrr cos \ BTT. very small. 9. sinft07r. &. reader cos % BIT 5 . however small 8 may be.gt. numerically smaller than any assigned number.lt. cos ^drr  sin nQir sin Brr must be numerically cos (ft less than 8j sin \ 6ir .lt. if n is sufficiently Such an assertion is palpably absurd when made of a fixed number such as cos#7r. 8.lt. #TT #?r must be numerically less than fi/ sin sin ft#7r sin cos nOrr cos % ^?r is very small if n is large.gt. we mean that we can choose ft so that 5 &.is large. when (ft) &.lt. . so that the function progresses to infinity by a is ft&.(ft) oscillates 0. but the first is when n is large. which is not zero. unless 6 is an even integer. &. of course.lt.62] 5.l. We can without loss of generality suppose 0&.lt. should it not be the other factor cos$#ir which is very small ? The answer is to be found.gt.(n) sSll oscillates finitely.lt. It does not however oscillate 7. ft POSITIVE INTEGRAL VARIABLE 2 123 + ( n I) 2ft. Thus the hypothesis that \ . very ft much larger than the second 2 The second term oscillates infinitely. sinft#7r + n ft. Then. 18 and and so  l +d for all values we can choose ft so that sin ndrr lies between Hence of n greater than or equal to n Q .l tinual series of steps forwards and backwards.j&.lt.(ri) must oscillate finitely or tend to a limit.1. a cos 2 nQir + b sin 2 n6n. Similarly sin nQir must be very small. sin07r cos cos Hence + ^dn = cos ndir (ft + J) 6ir (n \&.lt.0&. according to our definition of the term.lt. in the meaning When we say of the phrase very^small as used in this connection. 9) that = 0. greater than any assigned value A if It should be observed that in this case $(2&+l) is con (2&). and this can only be the case if cos ndn is very small. he may ask.(?i)". and this can only be the case if cos nQn is very small Why. siiift^7r + l. unless 6 is an integer.f&.lt. (ft) is very smalf for large values of ft.gt. We shall consider whether the second alternative is really possible. siiift#7rh(lAO&. have already seen (Exs.gt.lt. Prove similarly that cos ndir oscillates finitely. We finitely when 6 is rational. In fact (ft) ft 2 .8. ( l) sin n6ir. The cos ndrr should is consider with particular care the argument very small. and therefore sin ndir oscillates as ft tends to oo. Let us suppose that lim sin ndir = I. xxm. so that I But it is impossible that cos ndrr and sin nBir can both be must be zero. must be numerically less than 8/ sin \Bu That is to say. . sin (ft+1) 07Tsinft#7r is numerically less than 28 for all such values of ft. sinft07r tends to a limit I is impossible.2ft and &. But it is not difficult to see that &.(ft)&. a cosft$7T + 6 sin ^$77. sin 2 ft07r. 2 2n=(ftl) l f Thus $ (ft) * always less oo &.
f&. 6=0. !!. example to the contrary.(ri) oscillates finitely. 485.gt. 11. If &. case in which 6 is irrational cannot be dealt with without the aid of considera tions of a 15.~\ we being any fixed integer. = 2. ! 0?r). 19. then lim = 0.0. then \^ (%) tends to + co or to co or oscillates infinitely.gt. If lim If (ft) (n) = 0.m 6ir} + b sin 2 (n 17. &. and  \ ^ (w) &. \//&.gt. or a&. [ v/?0.(^ + ao  and ^(n)&. The largest prime factor of n. as may be shown by arguments similar to though than those used in Exs.lt. If (n)*l as ft*. 7.*&.gt. If is rational but not integral. always numerically * ^V or if p See Bromwich s Infinite Series. then 12. then (ft)^O. &. ! cos (n 0?r).   (w) for all values of ?i. 18. A/A (11) \  when  then ^(^) tends to a limit or If (H) oscillates finitely. LIMITS OF FUNCTIONS OF A a + bn + (I} n sin ndn. 7*. 7i (ft) oscillates infinitely. to + co or . 6 Consider the special = 0.GO if both are negative. tends to a limit or oscillates and  &. or to or oscillates infinitely. &.lt. If (w) oscillates and.. co . cases in which a=0.D.co or oscillates so also does 8. If a and have opposite signs tional cases. xxni. I and f() when 6.lt. 6&.lt.124 10. n.gt. see that if (ft) tends The same is conclusions hold (except in the case of oscillation) less if p j varies with n but than a fixed positive integer varies with n in any way. In this case tends to foo if a and b b are both positive. 16. t h . then n B is certainly The integral for all values of n greater than or equal to q. p. (n) 2%e smallest prime factor of n. When is even. so long as it is always positive. Examples XXV.lt.co .lt.f)(n) (acos ft#7r + 6sin 2 ft$7r). Consider any excep sin 2 (ft 07r). n&. Thus (ft) oscillates infinitely.0. tends to + GO . (?i) When n is a prime.gt.[6ft]. ft an . a cos 2 (n n l) (&.lt.[Jri\.p (n)=Q.lt. .nQ. 5. then &. If 6 is integral. If & has a rational value pfq. Otherwise &. (?i) 3. ! ! more complex 14.lt. but to . 20.*]) ( B 1) [vH V^ . and values (ft) greater than n for which Is this true ? If not give an then ^ (ft) oscillates which ^ (ft) (ft). (ra)*0. n. the year n A. 13.0(?i) for all values of then 2.gt. then also (ft +&. ( [&&. Similarly (n+p). sin (n BIT). (w) *(). 9 and xxiv. p [This follows at once from the definition. 2 [iV n If n is integral. (w) finitely. or irrational.gt. The number of days in 1. Hence (n) *0.lt. where 6 is rational. much more ! difficult character. 4.). or a=0. then ^ (i)*0. (n) generally oscillates infinitely. % i^ \. however great be n Ql we can find values of n of n greater than nQ for (ft).*) %^w * . ^ (n) If * + oo .
least value of (n&. The object of mathematics But sometimes (as for example here) we mean by this is almost obvious something quite different from this. nQ such that + l)&.gt.gt.gt.gt. POSITIVE INTEGRAL VARIABLE least value of n Q for (ra&. But the inequality given is satisfied by = 1000001.00001.w ). according as [A] is odd or 12.io. i. Determine the (a) ^ . if a proof can be the baseangles in this sense does is to prove that certain premises imply certain conclusions. (l/^) + {(l)y?i 2}&. the truth of the propositions 2 +2=4 or of an isosceles triangle are equal it is .wo). for example. 10. w&.w ). and even if the theorem is true. And often.62. I. and often not even from the interest of the proof. 1000000 ).gt.999999 Determine the ).A % + (l) n (ra&.gt. 63. theorem One may mean it is difficult to doubt the truth such as commonsense instinctively accepts as . When a certain ambiguity in this phrase which the reader will do well to one says such and such a theorem is almost obvious one may the theorem is mean one of the it or other of two things. accepts. The argument which the reader will There notice. .] least value of (6) 2 or 2+ [A]. not because the proof is in any sense unnecessary.gt.lt.A (a) +2&. 63] 9. If $ (ri) and ^ (n) + b. and the fact that the conclusions may be as obvious as the premises never detracts from the necessity. tend to limits a. b.Ho). least value of (rc&. and this is the value of w required. then (n) + ^r (n) tends to the limit a is is This * almost obvious*. place when 2 000001 easy to see that the least value of % such that (n+l)/n is 1000002. sense are often found to it is the fact that also obvious is That a theorem is obvious most confident of the intuitive judgments of be mistaken. n2 + 27i&.1000 Determine the 2 w + (l)&. nQ for (6) which it is true that (ra&.lt.gt. when a statement is in this sense. Some general theorems with regard to limits.e.] it is . &. but should also suggest to him the general lines of a rigorous proof. The behaviour of the sum of two functions whose behaviour is known. (a) 11.gt. [(a) 72 =[ V/(A + 1)]: rc even.gt.0001. since the common found.lt.w 125 Determine the (a) which (6) it is true that w 2 + 2ra&. A. not prove that true.gt.gt. We mean a moment s reflection should not only convince the reader of the truth of what is stated. 1000000 (w&. THEOREM (f&.w w + (l) n &. obvious . but because it is a waste of time and space to state in detail what the reader can easily supply for himself. =1 + [A] (6) =l + [A] + {l + (l) [A] }.gt. In the first when and n^n Q [Let us take the latter case. one may well omit the proof. A being any positive number.gt. nQ for (6) which it is true that &. no reason for not proving it.
&... 001. equal to a + however.  \ .lt. *(*)& . Thus It follows that the desired condition will certainly be satisfied if n can be so chosen that \$(n)a\ + \1r(n)b &. Now by a proposition proved in Chap./&...(n) =a But this is certainly the case. j \ &..^III (more n indeed than we need here) the modulus of the sum of generally two numbers is less than or equal to the sum of their moduli.....gt.lt. b .J* (*i). ^ (n) + + oo or oscillates finitely.lt.. so that JS when n^n lt Similarly (n) a b Now take ?i we can find n 2 so that ty (n) ^8 when n^n 2 Then a to be the greater of the two numbers n i} n 2 (n) JS and ^]r (n) b ^B when n = n and therefore (2) is satisfied and . f (n)ab &. (w) + when n = .. we can choose %..lt. &.. and . 7 We Let S be any assigned positive number (e.lt.S .lt.  \ the theorem is proved.). then (n) + (n) ^ + GO .. difficulty in verifying &./&. (/&. .. lim (^ (n) + ^ (n}} = a+b. or to oo or oscillates finitely *&.g. If (n) tends to limit. . (2). (n) ^ fy + oo . &. The reader should have no 1. [iV at once form in his mind is roughly this when n is large. For since lim when Q so that $ (n) a B we can. Nothing prevents when n ^ our taking 8 = S. but ty (n) tends to &.gt.lt..lt..lt. + &. or infinitely.. n 2 so lim (ft) \// =a and that I !$()and then. 0000001..n&.lt. the following subsidiary results.gt. // &. and therefore their sum is nearly nearly equal It is well to state the argument quite formally.gt. (n) is to a and ty (n) to b.. if I &... &. require to show that a number n can be found such that ..gt.. (1).gt..lt..lt....lt. then (n) + ty (n) behaves like 2.gt. &..lt.S .. I &. Results subsidiary to a Theorem I. nn ?? x &.lt.126 LIMITS OF FUNCTIONS OF : A &.lt.. by the definition of a limit./&.  &./&..lt.gt.x&.f&. stated thus: since lim$(?i) The argument may be concisely = 6.lt.J* (nnji * n is not less than either HI or n 2 and therefore 64. find n 1} and this however small B may be.
* (i) &. (n) and ^r (n) oscillate infinitely. ( l) n. then &..lt. Hence (n)+^(ri) must either tend But if it tended to + oo then &. \//&. j additional examples of each case. (n) 4 ^ (n) may either  ter^a either to a limit or finitely or infinitely. 5. On must assume values numerically greater than any assignable number (e.}n.n 2 The reader should construct . consider (i) &. ^ (w) = . to co: hence infinitely. .. = {$ (?i) + ^ (n)} \^ (n) = n\ . oo or oscillate infinitely. then &.lt. 1 may obviously change into oo If &. = ( l) n n. number (e.(n)=?i.t&. 10/T.lt. and.g. +(n) = n\ (iv) 0( w ). (v) = n2 + * n l) n.oo In this statement we throughout. to +00 or would also tend to +00 .lt. + oo and may be permuted throughout.lt.lt. {l + (l) w + 1 } n.gt. if (%) + ^r(w) behaved in any of the (w) three last ways. or to For or oscillate finitely.gt. or to oo .p (n) 4. nor.gt. &. for instance.^5 ejOA* ^vj (TKSU.gt.lt. 4.lt. finitely. that oo (n) *which is not the case. . the other hand ^ (n).(n) ty(n) = (\} n n. say K. it would follow. oscillates 7.gt. (n) * oo and to *&. = n. ^r (n) infinitely. . If /r &./&.gt.()=*. n We thus obtain examples of all five possibilities.+ (l). GO (ii) (j)(n) = n.). that (%) the functions (l) + 1 w. Hence (f)(n)+^(n) must assume values numerically greater than any assignable since it oscillates infinitely. (ii) *(*) = *() = (!).oo and ty (n) oscillates infinitely. . (n) + ty (n) As examples (i) 4&.). As examples of the two cases which are possible. 6. 100^. Thus it (n)+ ^ (n} cannot tend to +00.lt.g. ^ (n)  oo . // (n) and ty (n) both oscillate must tend to a limit or oscillate finitely. 64] POSITIVE INTEGRAL VARIABLE 127 4. These ^ve^possibilities are illustrated in order by (ii) &.gt.63. (.lt. Here again the signs of &. in virtue of the preceding results.lt. from the previous results. ^f(n}= n. or oscillate either finitely or infinitely.lt. For (n) is in absolute value always less than a certain constant. or to + oo . {l + (l)". (%) (iii) ( (n) ^ (n) =n. 3.lt. then (j)(n) + ^ (n) may tend to a limit.f&. and then + i{r (n) oscillates infinitely. ^(n)n.M take = (m *() = ((n) oscillates +1 1)". + oo or o oo or may oscillate * *j ". 99#.. &..f&. then (n) + (n) may tend to oo + . . while \js(n) is in turn each of Suppose.(n) If (j) finitely. for similar reasons.t&.gt. 9^. .oo or oscillate infinitely. If both &. ^(n) = (l) n n2 .l) n + 1 (w+l). &. but cannot tend to a limit.
8 reader should supply a proof for the case in which at least one of a and b is which &. n * B. we may point out that the result of Theorem I may be immediately results 1 The 7 cover all the cases extended to the sum of three or more functions which tend to limits as 65. (n) then (j) ". we may choose n Q so that when n &. concerning products analogous to those stated now six different ways in which (n) may We must distinguish &. &. Then (n) If Km &./&./&.gt. fa (n).gt.gt. bfa (n) fa (n) fa (n). \  zero. Km fa = a and Km ty (n) = b. in 64 for sums.gt.lt. and so the theorem follows.ab} = 0.! $ (n) ty (n) = ab + a^ (n) + bfa (n) + fa (n) &. ^(n) = b + ^ (n\ (n) = and Km ^ (n) = 0.i have (*) .lt.()* is a strictly formal proof.gt. like Theorem I.w .128 LIMITS OF FUNCTIONS OF A [iV which are really distinct. The following !&. = + (n). is principal result the following. need hardly point out that this theorem.lt.lt.lt. The known. The behaviour is of the product of two functions whose behaviour set of We can now prove a similar theorems concerning the product of two functions.gt. THEOREM Let so that II. We t&.ab I = a I ^ (*) i + b( I ()! + ! & (*0 1 1 *i ( % ) I Assuming that neither a nor b is zero.n .gt. ab is Hence the numerical value of the difference (n) ty (n) certainly not greater than the sum of the numerical values of a fa (n).lt./&. Km (n) (n) = ab. Then is certainly less than 5 if 8 % a \b\. From this it follows that } Km $ O) ^ (n) . be immediately extended to the product of any number of may There is also a series of subsidiary theorems functions of n. { which proves the theorem. The HQ so that (n) ^(n) ab\&.v/r l (f&. Before passing on to consider the product of two functions. We behave as n tends to oo . It may (1) tend to a limit other than . co . That is to say we can choose when n&.lt.
l) ( .f) \ that lim subsidiary (n) = kci if lim $ (n) = a. to . ^(ri) = am$nir.f) (ri) *  according as k &&. for It is not necessary. (n) = a+ .t&. and 0(w) = cosi7i7r. of the difference or quotient of is known.lt.lt. a similar set of theorems for the difference of two given functions. III. to take account separately of (3a) and (36). as the results one case may be deduced from those for the other by a change of sign. POSITIVE INTEGRAL VARIABLE 129 (2) tend to zero. if &. when of course if = (n) then k(f)(n)++ao or or negative. To than we can state these subsidiary theorems at length would occupy.lt.1). V^(w) take (a] (c) = l(l) n . (4) oscillate finitely.(.gt. (ri) ty(ri) must tend a limit (which may be zero] or oscillate finitely. k&. select the two which follow as We more space examples.lt. (i) If or to (n) aoo + QO and ^ (ri) oscillates finitely. Let H.f&. then jrT&.lt.gt. then &.lt. ( ^ 0( n ) = i + (_i) A in which particular case of Theorem II which is important is that The theorem then asserts simply ty (n) is constant. (ri) . (n) a fa (n). is To this we may join the theorem that GO . positive values of n and lim for all k$ (n) = 0. $ (ri) = + (n) = . of course.6466] zero. 0. and 1 =. then so does (n).(ii) and ty(ri) oscillate finitely.n .f&.gt. (5) oscillate infinitely. THEOREM If lim r lim (n) 1 = a.(ri)*+ao k &. = (n) oscillates finitely or infinitely. which are obvious corollaries from what precedes. as a rule. then &.f&.f&. be Examples of these three possibilities may be obtained by taking n and \^ (ri) to be one of the three functions 2 + ( .lt. leaving the I verification of them to the reader.gt. In C. (36) tend to  oc . a is not zero. For examples.lt.lt. The behaviour two functions whose behaviour order to deal with the quotient we begin with the following theorem. (ii) to + oo ^ (ri) must tend or oscillate infinitely. There is. (3a) tend to + oo .gt.l) n 2. unless k = 0. afford. k&.l) w to (ii) If &. He will find it an instructive exercise to formulate some of the remaining theorems himself. And $ unless k 66.
% (n). ^ (n). 68. that Q .f&. For P is is a sum of a finite number of terms of the type where A a constant and p. b. . THEOREM zero. .: and if lim &. viz.. [&.. = R (a.e. Q b. &. .). c. c. positive integers. . (n) Then = a  \ J__l a  ..gt. subsidiary theorems THEOREM &. i. (ri). (n) lim \r = b. what is the behaviour Q ^_ ~ as n tends to * oc * ? We naturally suppose that neither a nor & is zero. b.130 so that lim fa (n) LIMITS OF FUNCTIONS OF A [iV = 0. (n). lim b. Theorem I.gt. then IV.gt... 67. q...gt..). by examples some of the corresponding to Theorems III and IV.}. This term. ir x (n).)=j=0. by Theorem II (or rather by its obvious extension to the product of any number of functions) tends to the limit fia p b q .lt. . .lt.() a  j + fa (n) \ and we can choose n so that this is smaller than any assigned number 8 when n = n From Theorems II and III we can at once deduce the principal it is plain. Similarly tends to by the similar extension of Q (a...(??).. since lim fa (n) = 0.. IfR{&. any function oftJteform (n)..!&./&.#&.} is any rational function o/ ^r (n).lt. . (n).j) = a. ^ (n). % (n). The preceding general theorem may be applied : to the following very important particular problem of the most general rational function of n.. .lt.).. . %(^). c.f&.lt. {&.. lim % (n) c.f&.. (n) a and lim fy (n) b. . (n). }/Q where P and Q denote polynomials in . and so P tends to the limit P (a. . . . P (n). and the result then follows from Theorem IV.gt. + (n).lt. . % (n). . c. .. viz. V.lt. X (n). .lt.} R {&. + (n).. . and b is not The reader and illustrate will again find it instructive to formulate. (n) .. theorem for quotients.gt. prove. If lim &. and then Q(a. ..
1 if p = q and n*?*. 8(n) 69. A special but particularlyjimportant class of functions of formed by those whose variation as n tends to oo is always in the same direction.6669] POSITIVE INTEGRAL VARIABLE 131 it in the In order to apply the theorem we transform form S (n) by writing in curly brackets is of the form JR (n)} where and therefore tends.lt. .lt.gt. 1.lt.l 2. q.f&. not necessary to consider the two kinds of always decreases. show that in all cases .lt. functions separately. negative). 92 . &. as n tends to oo to the limit 0(n)=l/w. 5).f&.if p q n p* = 1 and w ??*. R (0) = a /b Now m. S(ri) Functions of n which increase steadily with n is n.+ oo co (p&. . . for theorems proved for one kind can at once be extended to the other. Which (if any) of the functions 2 (w cos ^w TT +sin 2 ^nir}l{n (cos^^mr + n sin 2 WTT)} oo ? tend to a limit as n 3. (p&. + oo if p q. Hence. it is DEFINITION.gt. Denoting by (w) the general rational function of n considered above. The function .***.lt. by Theorem II.q. do/bo What is the behaviour of the functions n+l as n*~x&. . (n) mil be said to increase steadily n if $ (n + ^ $ (n) for all values of n. a /b positive).gt.gt./&. with The function 1) &. $ (n) always increases if (n)  &. &. S (n) .gt.gt.gt. [&. S (n) *Examples XXVI. lim 8 (n) = (p &. . that is to say those which always increase (or always Since decrease) as n increases.
(and so of course for greater values). (n) has the same value for several values of n Thus the function possible decrease.lt.9. is putting f in values of n.lt. &. does not. 3. there can be no doubt as to their behaviour as n tends to oo this . (i) (j&.1.gt. THEOREM. since 8 is number. (n) ^ f for some value of n &.gt. n.lt.lt.gt. are 1. the class R may or may &. while there are in general ^^alternatives as to the behaviour of a function. There is one exceedingly important theorem concerning functions of this class. or (n) &. 2. f for all The If it class L certainly exists. Thus &. LIMITS OF FUNCTIONS OF A [iV be observed that we do not exclude the case in which . 1..lt.132 It is to &. Then (n) On the other hand $ (n) a 8 for some (?i) ^ a.lt. then./&. L or R according as all &. however large.(n) as these functions are extremely special ones. is said to increase steadily with n. hand R exists. (n) tends to a steadily increases limit as n tends to oo (n) with . ^ That is to say..gt.lt.f C .lt. &. and so. &. $ value of n. then either or (ii) * (n) + oo . and as However. given any number A. and so for all sufficiently large values. there are two only for this special kind of function.gt.gt. // c/&. and so (n) A $ (n) * + If on the other oo . Let a be the section of the real numbers in the sense of number corresponding to the section. and let 8 be any positive a + 8 for all values of n.9. . &. all we exclude is whose values for n 0. &. for all sufficiently large values of n.5.gt.lt. We a simple corollary of Dedekind s Theorem divide the real numbers f into two classes L and R./&.lt. Our definition would indeed include even functions which remain constant from some value of n ^ ^onwards. thus = 1 steadily increases according to our definition. 4.j) (n) ^a V\V&..gt.. a 8 &. &. apparent incongruity in the definition is not a serious defect. not./&. the classes L and R form a 17. arbitrary..lt.5. This theorem A (17).
must be. and tends to a limit.lt. If (f&. be noticed that the limit then every value of &. (n) lim (f&. : (1/ri).gt. If this is so.lt. The great importance of these theorems lies in the fact 70. The reader should theorems and corollaries increases. (n) &. &. &. (n) increases steadily with n.lt. then If(f&.gt.lt. (n) is less than but the limit is equal to 3. we can use the (n  (/&. (f&. (n) =3 3. ^ K for than if e.gt. 8 n ) : as for example in the case of the limit can only be zero.f&. 2.g. then $ (n) tends to a limit and this limit is less or equal to It should &.gt. otherwise L has no largest member. 70] POSITIVE INTEGRAL VARIABLE 133 for all sufficiently large values of &. shall find this corollary exceedingly useful later on. 1. (n) increases steadily with n. it COR. all values of n.__ If one.lt. and &.lt. If K. Thus (n) can never be equal to a except in the case in which the values of $ (n) are ultimately all the same.f&. in a great many cases.lt. member of L . K ^ We COR.gt. COR. whether a given function of n does or does not tend to a limit as n * oc .lt.gt. write out for himself the corresponding for the case in which (n) decreases as n cf&. without requiring us to / be able to guess or otherwise infer beforehand we know what the test limit. a is the largest It should . (n) increases steadily ivith n. where it is obvious that to determine But suppose we have whether . (n)  I &. &. i.gt.lt. (n) is &. (n) for all values ofn. if there is what the limitjis. then it is will tend to a limit or to K or is not possible to find a number such that $ (n) v^c TWt Ut )%? (Wtv\ for all values of n.e.lt. oo + according as &.gt.f&. may be equal to K 3. that they give us (what we have so far been without) a means of deciding.69. (n) = 1/n. n .lt.gt. be observed that in general a for all values of n for if (n) equal to a for any value of n it must be equal to a for all greater values of n. (n) + a.
if both do. there is LIMITS OF FUNCTIONS OF A [iV if In this case : it is not obvious what the limit. and so Pn tends to a limiting position T. each of which is P Q2t PA. Similarly.lt. +8).1. to prove by means of = (. two equal parts. of a half of its predecessor. The points P. to decide whether it is and exists or not. and consider any interval of the type ( d.lt. which cannot be used. . For suppose that is a point of accumulation of S. . P 1? 2 . and each which contains infinitely many points of S. . Pn Qn will lie entirely inside this interval*. . impossible.g.f)(n) = Q.. at any rate directly. . . + $) of . Hence T coincides with T. 2 $ 2 which does so. of Alternative proof of Weierstrass s Theorem of 19. P the same point as P. select the lefthand half and infinitely many We select the l one which does. and if x = then (tt)=0. lim limit of as n tends to oo Let us apply the 69 to the particularly important case in which If x = 1 then $ (n) = 1. n equal to P$/2 tends to zero. progress steadily from left to right. lim $ (n) = 1. us.l) w it is clear ad absurdum that I cannot exist. we we denote the selected half by P 1 Q^ (Fig.(n} . 71.f&. Proceeding in this way we can or. contains infinitely many points of S. 28. * This will certainly be the case as soon as PQj2 n 6. if we We P define a sequence of intervals PQ.lt. Fig. one at least of them must contain points of S. 72. and Pn Qn being P .134 tends to a limit.. so that these special cases n oc need not detain &. which is obviously that I would have to be equal to 1 and also equal to a reductio &. TT P . is If PQi is the lefthand half. If e. if both do so. Of course the test can sometimes be used indirectly.gt.gt. divide P\Q\ into two halves. its sufficiently large. we select the lefthand half. one at least of them must select the half contain infinitely many points of S.. 28). one. The results n $ (n) = x &. will be I. The results 69 enable us to give an alternative proof of the important theorem 19. or. n Hence If ($&. a . and Pn and Qn is Then is T coordinate. both tend to T. proved in If we divide PQ into . Similarly Qn tends to a limiting position T is plainly less than But n Qn whatever the value ofn. PI&. involves I evident that the test above.
These examples are particularly important and several of them will be made They should therefore be studied very carefully. both positive and negative. x n is &. (n) &.lt. 1 alterna n 1.1.lt.lt. 1.lt. and x= y.f&. Thus lim x n = I. Finally if x n takes to + x and therefore x values.7072] First.gt. y &.p (/&. follows from what precedes that lim y n = and therefore lim x n = 0. 1. taking the values 1. for all values of ?i. and therefore l=xl: and as x and is are both greater than 1 this impossible. decreases as n increases if x &. ). The reader may give an alternative proof. Then.gt. and x 1.lt. Hence x n _^ _j_ QQ (#&. then x n must tend either to a limit (which must obviously be greater than 1) or to +00.gt. so that I must be zero.lt. POSITIVE INTEGRAL VARIABLE 135 (n). If x = 1 n it is obvious that x oscillates.gt. = x lim (n) &. . (n + = lim 1) x&. Examples XXVII*. x&.lt. lim &. finally to consider the case in which x then it x &. suppose x positive. To sum up &. by Exs.lt. 1.J&. (n) xl. is negative.f&. then *~+ oo . (x = ( 1 &. Example. (n). cf&. 1. 1).lt. (n) is positive (n) and (f)(n+I)&. Hence xn oscillates numerically greater &.K&.gt. + oo if and deduce that xn tends We If 1 have &. then y tends y.lt. (n) oscillates finitely (n) oscillates infinitely 1.gt. 1).lt. Example. as above I = xl.lt.gt.gt.lt. and so that &. showing by the binomial theorem that xn + n8 if 8 is positive and #=l + d.gt.lt. Prove as in the preceding example that (l/#) n tends to to 0.f&. 1).lt. If &. + oo =1 (n) = (n)  (x &. so that &.gt. . a decreasing function if x On the other hand oo must therefore tend to a limit or to Since x n is the second alternative may be ignored. than any assigned number. (x (x &. &. &. &. since (n + 1) = n if x 1. xxv.gt. 1).l xn * + oo . Suppose it tends to a Then Km (n + 1) = lim (n) = I. so that y tively. &. Hence . &. where K &. 7 but limit 1. * use of later in the text. \ . : (?i) = #n lim lim &.lt. 1.lt.gt.gt.lt. increases with If x &.gt.lt.gt.gt.lt./&. infinitely.lt./&. . and and positive say. x 1).
n&.lt.gt. l. [lV &.lt. tfn*l. for which .gt. factor n* only asserts itself 10. 1&. [This follows from Ex.f&. {&.?? result is true if the conditions above stated are satisfied .lt. integral values of L special cases k is immaterial except in the Since lim {(n + l)/n} k =I.K&. .gt.1.] Draw up a table to show how nkxn behaves as n *...lt.lt.gt.(w)0 when x=\ or 1.n .O. since ". as ?i^oc. then lim = 0. 5) and 0(w)*0 if r Next suppose x negative. 1.lt.lt.(n !)}/{&.lt.gt. x&.* + 1 &. of $(n}=ri x n where r is any positive integer. whether Tc be the limit of the ratio $ (rc + 1 )/$(%) depends only on positive or negative. it tends But if $n*l where to a limit which is greater than or equal to unity. as w*oo. a decreasing sequence.gt.lt.gt. (7? !)}/{&. (?i + l)&.(n} \im{&.f&. the factor xn The A and the behaviour of (n) is in general dominated by [The reader will observe that the value of when #=1 or . 5 follows from Ex. and.l . . 8. (nV) (n 2) .] Prove that t if x is positive then ?&.1 and to ^ (n) oscillates infinitely if #&.(n) = Q.gt. and. take = where I is numerically less than unity.  \ cf&.(n) when n&. only 3.K&. This result also true if the conditions are satisfied only when n^n 4.lt. lim&.gt.l (Ex.] Determine the behaviour.gt.lt.gt.lt. then lim &.(n + l)\&.gt.lt.lt.l.\ for all values of n.g.g. If is (j&. L] halfway between 1 and may. [For n+ \f(n + l)&.j)(n) = Q for all values of n.. then 7.(ri) positive and is &.gt. (ft) when + we [For we can determine nQ so that {$ Now apply Ex.l . as large as But if Z&. [The results are the same. Discuss .lt. &. (w) . In all other cases First suppose # positive. we can find values of n. Then &./f0(w). Hence Then $(ri) \=nr x n tends to +00 if #&.lt. 1. where we please.^&. 8).lt. 7..lt.f&. this is impossible.lt.gt.gt.1 (%)=7i *+oo #=1.Vn if (w + l)&.gt.gt. then certainly untrue for sufficiently large ?i&. is $x+\ astt*co [Suppose. since l n ln* + ao with n (Exs.K  &. [If #=0 then &.1 and (n)*0 if If \ . as AT n oo .lt. 6).}&. #&. e. If \&.l %&. A&.lt.A^~i [For from which the conclusion follows at once.1.lt. 2.gt. where 0&.gt.] (n)&.+ oo is positive and then + l}}/{(j)(n)} = 5. and 0&.j&.gt. + (n)} If&. &. if ^&.gt. . and $x&. . and &..gt. (?I)H. Thus #n decreases which is certainly satisfied if (see as n increases from 3 onwards.lt. when x is numerically equal to .136 LIMITS OF FUNCTIONS OF A 2 &. and all positive and negative 1. 73 for a proof). as it is always greater than unity.gt.oo for all real 9.gt.. except that &. (1).K\(j&.j)(n) 0.] .3 t &.&. Then #. Jx $x.n.\.w : A". The same when ?i&.f&.gt.1.gt.(?*)* + oo if o?&.gt.lt.1. 1.f)(n l&. values of #.] n~ Txn in the same way.lt.lt. n which is values of n. If lim 6.lt.ra + or \l + (l/n)} n &.lt.1.A".l or n x&.lt. then (Ex.lt. e..gt.gt. 4 as Ex.gt.lt. (ri)}&.0.1 $x&.] 11. Thus $x+l.
6).2. and the number ) of terms also increases with Hence . .lt.lt. which tends that un does the same (Ex. 1. 1 + ) . and will be considered later. 1. to zero as ?i*oo..lt. n. ..2. A more difficult / problem which 69 arises when = &.2 n2 ^ .+j) 1 1 1. Otherwise unless #8=0. 73] 12.gt. increases with n.gt. Un= tends to zero as w*oo [If . ! &. is positive and an increasing function of n. m is a positive integer. 137 n is large enough. n if Show that if  1&.2. viz.lt.72. . which is what is used here.. x &. 3. n /n\ then n + 1 /wn =A/(ft+l). I then .] 13.+ 00 [However large A may be.3 +. e is a number such that of 2 &.21. The (j9 + l)th term in this expression. ra(ml) ~~ (m n+l) ^* n __fm\ . 1.j&.7i Thus f ( 1 + 1\ ) cannot tend to 1 \ n/ +x n .. WB for n &. from the binomial theorem* that 1 i ".. so A $(n !).(ri) {1 + n l/n} . POSITIVE INTEGRAL VARIABLE . The other cases of the theorem belong to the theory of infinite series. is a elementary algebra. ( 1 + .gt. and so + n) (l\ where *  ) =Q ^ 3. m.gt. .] n (l\ can be solved by the help of It follows j . n For if un = A. ". \ ft/ and so tends to a limit or to + oo as n *~ oo But 1\ (&.. e The binomial theorem theorem for a positive integral exponent.
gt... If we put a=yM the inequality takes the form .138 74. obtain TLVu rar ( r (al)&.1 s lf &...gt. LIMITS OF FUNCTIONS OF A [lV Some algebraical lemmas. (ii) The s that r and the more inequalities (3) and (4) have been proved on the supposition are positive integers... and this we have proved already....1..gt.. (6)... (iii) In what follows is s to be understood that all the letters denote positive numbers... (i) number It will be convenient to prove at of elementary inequalities which will be useful to us It is evident that if a&.. l^^1 a^) .. where a....gt. are rational.... c..... But it is easy to see that they hold under general hypothesis that r and s are any positive rational numbers..r(aI) . the first of the inequalities (3).. 1 (1 /3) . so that ad&. we .gt.. (4).. obtain rfKfari (!).. we deduce 1 al&. that a and than 1..gt... lf s r Here 0&.lt.a \fcfl 1) to l and adding r(ar  each side.. we l see that r (al)&........ and for a./S^S/S 8 "...lt.... (7). when 5=1. a positive rational less than 1.... a if s is K^ol). (8)... then ol r 0.gt.. for example.. and r&.... from (5)..a......bc.l and r is a positive integer then Multiplying both sides of this inequality by a.. that r and 1. Combining (4) and (6). The same argument applies to the re maining inequalities and it can evidently be proved in a similar manner that . s=c/d.. we /3..l&. d are positive integers.....lt...(l/3) *1 ....s. . (5).. we obtain Similarly we can prove that r + l _r (2).Sa (al)./3&. Let r=a/b. It follows that if r and s are positive integers... 6. ..gt. and dividing by r (r+ 1). number it lj8&.. this stage a later on....lt.... Let us consider.a rar I&. Similarly.. In particular.. we have 0) . and s less than Writing 1//3 and I/a for r are greater in (4).
.lt.lt.lt.1..^". 139 Writing x\y for we obtain (xy} . 2. if k 5. where /(a)&. and Hence.. 9 to the case in which r or k The limit of n ($x s 1). & being / &.0. as n*oo.gt. And the same argument.lt.x*y*&.Q.gt. 75. 8..). Now = /v/al.gt.. 10) . [We may suppose ^l &.lt. s  () decreases steadily as n in Hence (n) tends to a limit Z as 74.lt.. l/a&.sy&..1) = for all values ).0..0.gt. if # = 1. if a&.. Thus &.. pp.gt.l)/#a. = l/n. ^u &. in virtue of Theorem III of &.l.lt.lt. we have lim 00 *U/a !)=/(&.lt.1) ". Next suppose /3&.0. xxvu..gt..gt.lt.lt.2 ( according as lies or &. Verify (9) for r=2.. ii. (10).. (9) if x&. as is certainly the case from a certain value of 66.. $.. if /3 (Exs. 2..lt.gt.1.gt. then w n(^l)^/( a (^. 4345.i (xy] 3.lt.i then creases..f&. .. (#/3 . applied to (5) and (7).1) = . we if &. Extend the results of Exs.^&.* &.. Show that (9) and (10) are also true that (9) also y&.0. where l&.&. put r =!/(!). = n (#a. that &&.lt.(w)W.gt.gt. ^&.l (l/a)&.lt.k &.gt.] 4. leads to *Examples XXVIII...gt.gt.. [See Chrystal s Algebra. Show holds for r&.. 75] POSITIVE INTEGRAL VARIABLE a.(n) If in the first inequality (3) of 74 we see that when * a&. w/al)^/(a).74.. &. 3. . Again in the first inequality (7) of we put s = ljn we obtain t Thus l&. always positive. ...x&.y&.gt.0.\ $ fc *Z fc. If or /&.0.gt. and let /3 = I/a then n xxvn. &. 1... (xy}&. of w.* Z* and  ? between & (^) fc1 and * (20*".. vol. It follows that the ratio of  &. The &.l..lt. .n (^fa ...l.lt. Hence...&. then rf^Y^&.lt. The proof is similar when 0.* any rational number. 1 result is still true when =0..] 7. If &.gt.. Also (%) is t and if.gt. are any rational numbers.. we have Finally.gt.lt. if and (10) for s=.lt. and that n onwards.
or call s the sum of the the sum of all the terms of the series.lt. . n. . In these circumstances we a convergent infinite series.w(w) is any function of If we add up the values of u(v) n. 2. now we suppose we have that sn tends to a limit s n when n tends lim 2 u v = s. more If shortly. + u (n). s n = u + u. n defined for v for all values of n. It is generally to alter our notation slightly 1 this equation in the . Prove that / (xy) =/(#) +/ (y). shall call the series if n=n (8).gt. .. we can choose n (S) so that the sum of the first n (8) terms. Later on according as able to identify this function with the Napierian logarithm of x x&. is that by adding more and more of the us together we get nearer and nearer to the limit s. +u n or. More precisely. This equation is usually written in one of the forms 00 the dots denoting the indefinite continuance of the series of u s. sn &. and we shall series. lies between s S + & ) or in symbols s S &.. of the above equations. viz. we shall be Example. Suppose that.. Infinite Series.lt. [Use the equations 76. This function f (x} possesses the properties and is positive or negative 1 or x 1. &.140 Thus we LIMITS OF FUNCTIONS OF A arrive at the result : [IV the limit lim?i(^l) defines a function of x for all positive values of x. if any small positive The meaning number & and s is chosen. we obtain another function of 8(n) = u(I)+u(Z) + and write 2 . to oo . most convenient form also defined for all values of n. + . or any of greater number of terms. = 1.lt. . expressed roughly.. sn =2u n v . s + 8.
gt. ft (n) &. is the = + + merely u n of the n terms tends to the limit s as of such infinite series n * oo and the consideration introduces no new ideas beyond those with .. + w TO+2 f If is is convergent and has the convergent and has the sum sum s.gt. or first . In that of Bromwich s Infinite Series. 2 +. and has the sum s.. may : be.f&. then (1) are shall General theorems concerning dealing with infinite series we infinite series.gt. 1 }u 2 + .. and it is sometimes convenient to say that oo of tends ) to the limit I. then are used differently by different writers.gt./&..lt. . occasion to use the following general If Wj + t&jH.. then so do the others.lt.lt.. by writing &. in this case say that the series u l +u + . oscillates finitely. we constantly have theorems.. Similarly convergent and has the sum then (2) wjn+i If w 1 +M . . is convergent and has the }^ 2 4. In fact the sum sn is merely a function $ (n). &..gt.gt.. or These phrases too may be applied to (j)(n) as the case &. as n If sn is (n) converges (instead + oo or sn * oo . (4) If 2 1 +u + 2 kut * + ku + . (3) any ^ series considered in (1) or (2) diverges or oscillates. . 77. expressed in a particular form.. um .(n I)} ......gt. 4&. = (n) (!) + ft (2)  (1)} + &./&. say. . to say that the series u l + u z converges to the sum s or simply converges to converges u2 + u another way of stating that the sum sn + and has s. series which oscillate infinitely being classed as divergent .. 2 oscillates finitely or infinitely*. is convergent and has the sum convergent and has the sum ks.lt.f&. we shall say that the series u oo . Any function (n) may be expressed in this form. Many foreign writers use divergent not convergent . When a + Ul f u + z .. is convergent. s u 1 u2 . such as we have been considering. + ..7577] POSITIVE INTEGRAL VARIABLE 141 Thus sum s. is sum a + s.& we may say that not tend to a limit or to infinitely : + we oo (n) diverges oo then it oscillates finitely or or to . divergent or diverges to + 00. if any function to (n) thus *~+ &. + oo If sn does .. . which the early part of this chapter should^already have made the reader familiar. The reader should be warned that the words divergent and oscillatory The use of the words here agrees with Hobson s Theory of Functions of a Eeal is Variable a series as meaning merely said to oscillate only if it . (/&..lt. is s. .
sn is an increasing function of n. + u2 u3 The reader the converse is will be able to supply the proof of this theorem. then so is any series formed by grouping the terms in brackets in any way to form new single terms. the series (u. then the series 2w n must either converge or diverge to + oo If it converges. + vj + (u 2 + v z ) sum of the first two series. The sum of the first four terms is The sum of the next four terms is is i+ + ^ + and so on. and vl +v + 2 . . . n : hence the series diverges to + oo . according to the definition and we can apply the results of that section to sn .and sn and sn_i have the same Hence lim u n s s = 0. . when of course its . = sn sn . Thus 11 + 11 + . sum of is zero). .i. (7) For un lt limit s. what different character. .LIMITS OF FUNCTIONS OF A [iV If the first series considered in (5) (4) diverges or oscillates.. The reader may be tempted true and that is if to think that the converse of the theorem is lim w n =0 then the series 2w n must be convergent.=. while or + + 0+ . 2 Ifui+u +. the sum of the next eight terms greater than &=4. Ms sum must be positive (unless all the terms are zero.gt. . Let the series be so that un =lfn. and the sums of the two series are the same. . ... Here again not true. The sum of the first terms is greater than and this increases beyond all limit with ..&.. then its sum is the All these theorems are almost obvious and may be proved at once from the definitions or to by applying the results of the sum * = Wj + w. Those which. For 69. then so does the second. . 0. . That this easily seen not the case is from an example. (6) ^ If Wj +u + 2 . + is convergent and are both convergent. + + 1^. . is + + (8) If Uj_ convergent. oscillates. converges to (9) If every term u n is positive (or zero). follow are of a some 6366 . . then lim un = 0. . is convergent. unless k 0. . .
. . and so s n 1 or 1. The commonest example is given by an ordinary recurring decimal.lt. of an infinite geometric series Recurring decimals. then sn sn &. 78] POSITIVE INTEGRAL VARIABLE 143 (10) // every term u n to is positive (or zero). . and oscillates infinitely 1. ifr&. sn oscillates infinitely. + 1 =n. then the first series is also to that t where K is a constant. + oo . and. not true perhaps hardly if the condition un is positive is not For example .. 2 ? tend to a limit if and only if rn does we see that the series 1 r) if +r+ + . sn will In the last case s n * + In the general case so..lt. to sum up. If r 2 Thus. the series diverges to = according as n is odd or . Examples XXIX.lt. = 1. Referring to the results of is convergent and has the 72 sum 1/(1 and only ifI&. and if u n ^ Kv ny of n.r&. . . v + v2 + If u + u 2 + and the second series is terms. . and that sufficient condition that the series it 2wn should be possible any number of terms the is less find a number than K.. l . If r : = then sn = n. .e. divergent. + . are two series of positive (or convergent.e. when + . 69. For if v and so ?^ +v + +u + 2 . We shall now con ~ the geometrical series. . even i. . is This also follows at once from necessary to point out that the that every theorem fulfilled. 78. . 1. for all values of n. + vn vn proves the theorem. 1 and to 0. s n oscillates finitely. 11+11+.lt. = t then v + v + + un ^ Kt which 2 . . is less than or equal l of the second.77. the series 1 + r + r converges to 1/(1 r) if 1 &. 1. 1.lt. * If r = 1. and l its 2 sum . if K K can be so found. this case except in the special case in which r sn = oo . &. for all values convergent. sn being alternately equal to . and ^ Ku nt then 2vn is sider geometrical series. l + 1 = 1.lt. then It is then the necessary should be convergent is such that the sum of sum of the series is not greater than K. then . + oo i. if %u n is divergent. . . . Conversely. oscillates finitely if r + oo ifr^l.. whose general term is u n = r n In The infinite l . obviously (11) zero) oscillates. diverges to r 1.
form t : trtrl as a pure recurring where Q is prime to 10. or in the 1) is divisible such that 10Wl and 10rl2 give the same remainder. and conversely. I?. Show that in general . The expression of irrational numbers as nonrecurring decimals. For the decimal to a definite number between 1 a2 34. 2.. 103 see that every terminating decimal can also be expressed as a mixed recurring decimal whose recurring part is composed entirely of 9 s. if k. i. 3. + 105 1000 how any //i /\ 10 2/ \_ ~ 2687 12375 The reader should consider where and of the general theorems of 77 have been used in this reduction. stands for the series .. figures. But this is not true of 10M r. ^ + _ ".. A decimal with m ljfZ\ no higher power of 5. This stands.. Decimals in general. (10 is divisible by q. ^ P P + w ^ w* + t^. &. the decimal 21713 ordinary rules of arithmetic. Show nonrecurring and n recurring decimal figures is m m equal to a proper fraction whose denominator is divisible by 2 or 5 but by 4. corresponds and 1.*". nominator prime to 10. = we !0 999 + + 10~2 Finally. we divide q different remainders.. by q. either. according tcrtne 1 J_ 10 2 ". n 9 s and m O s.e.. for _2_ ". Wl . Thus every proper fraction can be expressed as a recurring decimal.^y n decimal with.A greater is therefore expressible as the sum of an integer figures. The converses is of Exs. It is Let r=p[q. hence the decimal for r has exactly m nonrecurring p. for any value of less than m . 7 103 10 + 1 10 4 _3_ J^ _3_ + 1Q6 T i 6 T i 7 + _ ~ 217.n 1} 10 n . then 10m r has a de. and and If on the other hand q 2 5^ Q. For example. 25 we must add that of Ex. where %=w 1 w 2 Hence r may be expressed in the form Pj(lOn 1). all % 2 where n 2 . whether recurring or not.10%2 = 2 10". i.. and suppose first powers of 10 by q we can obtain at most therefore possible to find two numbers n^ and 4 are also true. = a m is the  To the results 6. \ &. 7.gt. and a pure recurring decimal..W2  Hence . of a and /3.1. Any decimal. 21 7 = 2 169.gt.1 144 LIMITS OF FUNCTIONS OF A [iV Consider. and so 10". If 3. we observe that of Exs. (N/ r v L 4 * ^ the denominator containing 3. that q prime to 10. r ( pr L that a pure recurring decimal is always equal to a proper fraction whose denominator does not contain 2 or 5 as a factor. for example.*".
2020020002000020.gt. to decimals in any scale of notation. m+3p. is &. Appendix  I. 4) the recur. and its sum is and 2 are in agreement.. For suppose that 1 a 2 3 . Prove that the series 1 + 2r+2r+ . ( y) by adding the two series writing^ 2 l+r + + .... with suitable modifications. It also follows that every decimal which does not 26) recur represents some irrational number between such number can be expressed as such a decimal. 8.b . . the intervals 0...... xxix may be extended.. All the results of Exs. is convergent. are all prime numbers .. is convergent if Verify that the results of Exs.gt.. Nor can it recur: for if it did we could determine and p so that m.. + i+ . The decimal 11101010001010. since the series includes m m m + mp.. Thus.gt.78] POSITIVE INTEGKAL VARIABLE n terms 9 145 of this Since all the digits a r are positive.lt. for a fuller discussion see * Bromwich. and it is certainly not greater than 5n or 1. and 1. r &.... in which the nth figure is 1 if n is prime. are all 9V). b r + 2 . 6). and that its sum (l+r)/(lr). Hence tends to a limit between and 1.* 1 &. 2. &.. r +r +... 1/10. the decimal 1414. anyFor it must lie in one of 9/10... 1*1(1 The . in increases by one at each 1 s or 2 s 9. (/ 1/10.. 77. cannot (Exs. and this is absurd. 3. and which the number of zeros between two stage..lt. (/3) by in the form 1 +2 (r + r*+ .). and zero. obtained by the ordinary process for the extraction of /2. and so on. otherwise.lt.). and so (Exs..is convergent if 1 its sum is l/(lr)lr. m+p. +r 1&. If it lies between r/10 and 3. [Since the number of primes is infinite the decimal does not terminate.lt. b^b z . &.. 1. are two decimals which agree as far as the figures a r _ 1? b r _ ly while a r r &.. (4)).. The series rm + rm + + . then the first figure is r.. H 10 . .lt. represents an irrational number. for example.. represent irrational numbers... 2/10. But decimal cannot recur. (a) by writing it in the form l + 2(l+r+r 2 + . Moreover no two decimals can correspond to the same number (except in the special case noticed in Ex..]* and Examples XXX. .2 . In each case mention which of the theorems of 77 are used in your proof.r) ( series r 77. ( 1 a".. 1 r 1. + 1)/10....b r br + l br + 2 ... By subdividing this interval into 10 parts we can determine the second figure. . the sum sn of the first series increases with n. 1 (2)). + 2p. Infinite \ Series... Conversely.b r + I&. 1. The decimals 1010010001000010.. It follows that the expression of a rational fraction as a recurring decimal is unique.. (unless 6 r + 1 .
and its convergent. LIMITS OF FUNCTIONS OF Prove that the arithmetic series A [iV is always divergent.lt.r 2 + ) 2 (? ?)+. 2x 2 + . If in addition the series i aQ + i + 2 + . when it is convergent. ".r 2 sum If we assume that 1 4 + .] Answer the same question for the series ~ r r r r 10.146 4. is sum is 1/(1 r) by means of 77.. and its sum then the series a + air+a 2 r 2 + is not greater than l/(lr). is 0.. [The its series is is 0.. Its sum is 1. Consider the convergence of the series and find their sums when they are convergent. Sum ? + 2 .lt.l.. when r&. and l/(lr).r^l.] . .. then the series sum is not greater than 13..2 1 2 3 + &. if . it diverges to +00 or  according to the sign of What is the sum of the series (1  r) ? + (r . when r 4.e. 1 n~l /2 .a&. The series +1 + n) &.lt. If 0&..0. except when r=l.lt. Sum the series r+ r&.. is convergent for 12.l. (1) and convergent then we can prove that its For if l + r+r2 + .lt. rz ^ + /. 7. It is also convergent when r=0. . when the series 2 sum rz is 0. 11.] 6. + 1~". .lt. S. i. if when the series is convergent [The series converges only its  l&. . Show a.l/(l+r)&. 2 .. [For 1/(1 .. [The series is convergent if l&.gt. is convergent. . unless both a and b are zero. if b is not b } while if zero.lt.] always con vergent. Its sum is 1 4r except when r=0. is air + a2r 2___ the lesser of a f ao _l_ i convergent for + a 2 + .lt..lt. that.2 or if and its sum is 1 4r.l.gt.lt. . . =s then (4).r&. the series diverges to +00 or oo to co according to the sign of 6=0 5. its sum 9.l 0^r&. s 1 0&..
diverges to 4. any series formed by grouping the terms in brackets in any way to form new single terms. the function which is equal to 1 /(I x) is + if .4 l + l7273 + + 1. = lim {^ (x) + W&. x &. POSITIVE INTEGRAL VARIABLE 147 The 1 series + lT2 + + 1. II are capable of a simple representation of this kind. 79] 14. 102 . 15. Many of the apparently arbitrary or unnatural functions considered in Ch..2.gt.. representation of functions of a continuous by means of limits. Any series.l/{ft(a+()}.oo .3.lt. . . In such cases the limit is of course a function of x. is itself convergent. is a function 1 &.*7&&.. which the function of n whose limit we are seeking involves. Thus in 75 we encountered the function in f(x) = lim n (\/x n*oo 1): and the sum of the geometrical series 1 + x + # 2 of x. 1 and undefined for all other values of x. . 1G. The real variable lim $ n 0).gt. In the preceding sections we have frequently been concerned with limits such as 79. *) and series tfi such as (0) +u 2 (as) + . oo besides n.475  are convergent.lt. viz. If is divergent then so is ! + w2 + 3 + .3. Now compare with Show that the series is convergent if and only if un tends to a limit as n 17. formed by taking a selection of the terms of a convergent series of positive terms. The general harmonic series where a and 6 are positive.gt. another variable x.2..[Fartvl/(0+*ft)&.. 18. as will appear from the following examples. + u n ()}. uz (a) + ..
gt.(o?) &.lt.1).p (#) &.] ^&. &. \ 1) . &.lt.lt. and 12.0.lt. arc tan (ra?). = sin (n!#7r) then = 0for all rational values of x (Ex. &.^&. and &.j&. 2 is lim (365 + (cos JMr n) 2 (cos ^N*Y + ( cos2 4 iu^". Here l.f) n does not appear at for all values of x.D.lt.gt. If ^Y &. xxiv. expression of 2. when $MW 18. [&. &. 1 or n x&.lt.gt.gt.lt.l.#&.] 14. = nx. nx Here &. 5 in that &.lt. LIMITS OF FUNCTIONS OF A 1. and is usually denoted (j) by sgn x. = x.] 15. &.~] : 11.lt.0Gr) = 1.0. 0(^)=l when a?&. when a?&.+or).lt.1). ^ n (a7) = *{(o*l)/( B (a. Here = lim n (x} x &. ^d &.} . when x= and l as well as when # = 1. = if 3. (a? l/("./(a7) + flr(^)}/(+l). [In the case ^(a?) = ^&. &./(a + l).(#) = (). ^&. n (^) l/(^+l). (#&.gt.j&. n (x) (#) differs from the &.gt.lt. (#&. when x= 1.lt.gt.] when x is an integer.] 0(a?) is not denned when x= 9. (#=1).lt.(!#).lt.(#) x.lt. Thus ((#) = when # = and n x = Inx.lt. [iV all in n (x) n (#). 17.(ff) 5. = when o?=l and (x) is not denned when x= l. [0(^) = 7). (*)= !.(#) = except when x is integral. Construct an example in which &.(#) is otherwise undefined (Ex.gt. If xxiv.lt.lt. (a?) of Ex. = x/n.gt.gt.(#) = !. 10.1 for both these values of x. first = (^ l n l)/(^ + l). and (a?) not denned for any other value of 6.)". $ n (#)*. and &.lt.&. #&.lt.lt. 16. 13.gt.lt.lt.o?).lt.gt. (a?) is undefined n (a.p # &. 1752 then the number of days in the year JV A. (o?= 0) .(x) it has the value 7.. &. &. is $ n (x} = x\ = 0.) +l)} /(A* + [Here &.1)J ^ WitfW+^^Bi (^=1). (J) when x Here &.k(*0 = a".#&.(l&. &.r) = 0.lt.lt. 4. n (x)*cc : only when is 4 =0has n (x) a limit (viz.(x) +w). &. (^)=^/(^l).lt.lt.$&.) = (cos 2 #71 n ) .gt.0). = x/n.(#) differs from the of Ex.^Cr) = ^(^=l).!).I) . (w^ n fl).lt. (#&. = 1. = #".. (^ 0) (. (!&.lt. (x)=f (x\ (\ x &.lt. for all values of x.l. [0(^) = 0. not denned for any other value of x.gt. n (x) = smnzir.(.148 Examples XXXI. 6 in that it has the value 8.(#) [The consideration of irrational values presents greater difficulties. Here If (x) = lim &. 0) as TI^OO. The second and third functions differ 1 the from the first in that they are defined both when #=1 and when x second has the value 1 and the third the value . 0(*)^(*X (I*I&. &. n !/(. This function is [^ (x) important in the Theory of Numbers.l). = {". &.lt.gt.gt.+ co (f&.f) n (x} .) = (2/Tr) (jb(^)= 1.gt. %)/( (f&.gt.lt.lt. (x=l and a = 1). 14).lt./&. l)/(^ n l/(s".lt.gt. the (x) n (x} &. a . l.(#) = !. &.^&.
Let S a aggregate of real numbers s. we say that S is bounded below.lt.lt. Any aggregate S which bound M.lt. Any aggregate S which is bounded below has a lower bound m. any number greater than K.lt. The number Af+fa would then belong to Z. but there is at least one member of S which is less than any number greater than m. Then every to one and one only of the classes L and R. or bounded below. and to R. If there is a number we say that S is bounded above. POSITIVE INTEGRAL VARIABLE 149 The bounds of a bounded aggregate. where is positive.lt. An infinite aggregate of The set consisting of the numbers does not numbers necessarily possess a least ill 2 3 for example. No member of S exceeds M . which we shall not exceeded by any member of S. by at least one It will be observed bounded below. that. &. and there is is a number M dividing the rj classes. (n) is bounded member. all This number M we call the upper bound of S. k m M &. member of L is less than some member of S. This number J/ than is exceeded M We divide the real numbers into two classes L and R. any number less greater than M\ and this is impossible.lt. When S is bounded. If there every s of s &. and bounded below. putting R according as it is or is not exceeded by members of S. while belongs to R. than Thus M belongs to Z. &. If S is bounded above. Each class exists number less into L or any belongs . and when K. K. any than any member of R. we say that * &. Thus the three conditions of Dedekind s Theorem ( K 17) are satisfied. The number first place. but every by at least one member of S. because it is On the other hand. or bounded. if In the there were such a member s of /S. defines a set S. and therefore exceeded M has the properties required.K such that for number k such that If S is both bounded above for every s. has l ". for example. and we may enunciate the following theorem.7981] 80. possesses among these numbers there is a least*. M the number whose existence we had For M cannot be exceeded by any member of S. we say simply that S is boulided. We less shall prove that call number M. and therefore less Finally. m k. to which we may apply all the arguments of 80. There will be an infinity of numbers which possess the property possessed by K . S is 81.lt. bounded above has an upper but any number less than is exceeded is M member of S. is it. when S is bounded above. M&. . K is be any system or s&. because it is less than s. for than any member of S belongs to L. &. &.gt. No member of S is less than m . no least member. Suppose that (ri) is a func The aggregate of all the values $ (n) tion of the positive integral variable n. is by at least one member of S.gt. In exactly the same way we can prove the corresponding theorem for an aggregate bounded below (but not necessarily above). The bounds of a bounded function. n 1 ". to prove. we could write *= Af+rj. . Suppose first that S is bounded above (but not necessarily below).
in case (2) that it is number.150 LIMITS OF FUNCTIONS OF A (71) [iV is If above. and both k and K exist then that bounded function.(%). Let us take any real number .lt. n. () &.lt.gt. &.lt. X . and consider now the relations of inequality 82. and in case (3) that it is an intermediate number.gt. (3) if 8 is values of n.gt.lt.lt. .(ft) for an infinity of values of n. There are three mutually exclusive possibilities : cf&. m such that (n) ^ m for all values any positive number then $ (n) This number m we call the lower bound of {ii) if 8 is &. &.&. if k exists. (ii) if 8 is any positive This number values of (i) &. &. since none of its members It is bounded all superior numbers. It is plain that no superior number can be less than m. which we shall denote by A.lt. then $ (n) A 8 for an infinity of values of n &. A and X are the upper and lower bounds if of the aggregate of intermediate numbers.gt.lt. any such exist . and has therefore a lower bound.lt. then there all values of n .gt. If K exists.lt. The limits of indetermination of a (n) is M ( \ ( m and the values assumed by which may hold between of n.X&. that . Suppose a bounded function. we shall say that is a superior number. In case (1) an inferior Let us consider the aggregate of below. m &. Mof bounded below. to say if there is a number k such that $ (n) then there a number of n .p ^ K for values of n. limits of indetermination A=lim &. or bounded.lt.&.lt. : These numbers have the following properties (1) (2) wi&.lt. ^ (j) (n) for all sufficiently large values of n . (n) for large values (1) (2) (3) &.K. and also &. X = lim$(7i). which we denote by X. We of $ call (n) as A and X respectively the upper and lower n tends to infinity and write .(p(n) for all sufficiently large values of n. M&.p M we call is number then $ (n) the upper bound is &. m + 8for (n}. or bounded below.gt. and no inferior number greater than M. are less than m.. such that say if there is a number (n) is a number such that (i) is $ (n) M ^ M for K bounded above. Similarly the aggregate of inferior numbers has an upper bound. that all to &. .&.lt. and (4) A + 8 for all (n) any positive numbej.(ft) for an infinity of values of n. and .8 for all sufficiently large values of (n) similarly X + 8 for an infinity of values of n &.gt. sufficiently large &. and and its upper and lower bounds. if at least one value of n.lt.A^J/".
1 and A = M\. all sufficiently large values of ft.. 4. (1) is an immediate consequence of the definitions and we can prove (2) as follows. 2& As A X ^ 0. If 6 is an integer then m = X = A = M= 0. so that X ^  A.lt.lt. then . so that Hence I d is inferior and l +d superior. Let (ri) 6&..lt. intermediate numbers as near as we please to either X or A.lt.lt. A nor X then affected by any (11). 5. then for every positive value of 8 (ft) &. this can only be true if is A = X. alteration in any Examples XXXII. =a for all values of n. If is rational but not integral a variety of cases arise. Neither of &. If $(w) = (l) M M tnen m=l. sin (2/i7r/^). sin (Zqpirlq). may be proved as follows.. m = X = A = M= a.COS (ir/2q\ \=^M=COs(ir/2q}. m = X= The reader may not both odd.81. where then 0..gt. Suppose. It is easily verified that the numerically greatest and so that (ri) are cos (77/2^) and cos(7r/2&. &.. Any intermediate number is less than any superior and greater than must be A then any inferior number. &. Conversely. ./)(70 = (l) n {l + (l/w)}. X = A = 0. &. M=. finite number of values 2. &. 1) such a way that. A = l. p assumes the cyclical sequence of values Then &. To prove (3) we observe that A + 5 is superior and Ad intermediate or The result is then an immediate consequence of the definitions and . if is .?&. 7. If A=X= . Suppose then that A X. and there is nothing to prove... 3.gt.gt. if $ (ri) * then the inequalities above written hold for all sufficiently large values of n. If m = X = A = and M= 1. &. that and q being positive. . and in this case the limit is I.(n) smQoTr/gO. the common value of X and A. If A=X = there can be at most one inter mediate number.lt.. and I.gt.lt. 6. and therefore A X &. . odd.Vprr/q}. Hence there are inferior. m=2.gt. 82] (5) is POSITIVE INTEGRAL VARIABLE &.lt. since it is plainly neither superior nor inferior. and 6=plq.g. then w = X=l and \ = M=l. (n} = 1 In. least values of .f&. But if X . . and prime to one another. = sin n6n.?). X=l.gt. If $(ft) = (l) n If&. 151 that the necessary and sufficient condition that (ti) should tend to a limit A=X.. M=l.lt. so that l...!.. sin{(2q. 1. discuss similarly the cases which arise when p and q are The of (ri) case in which 6 is irrational is more It difficult : in this case m = X= . viz. may ( also be in it may be shown that shown that the values are scattered all over the interval 1. Of these properties. I.lt. the proof of (4) Finally (5) is substantially the same. e. If (n) (f&. intermediate.
the condition is necessary. Hence &. ..8 and therefore we can (n) infinitely many such that &. Some Problems of Diophantine Approximation".lt... Acta Hathematica. infinitely many values of n such that A . for but a finite number of its values are less than if + 8 and greater than 18. and so a .. * The 83. which is greater than (1).. The necessary and to sufficient condition that a bounded function 8 is (n) should tend a limit is that..gt.f&. when any positive number to find a number n (8) such that given.. however &.....gt. 2 )&. For if &...f&.X may be. restricted Unbounded ..15*2 .. The general principle of convergence for a bounded function.lt. +8 and &.. So far we have ourselves to bounded functions for but the unbounded as In the for is the same bounded functions. the condition of Theorem 1 is satisfied..28.gt.. and the words a bounded function general principle of convergence 1.(n)*l then we when n &.gt. if &.gt. (i) when n t ^ HO an^ n2 = n osufficient. xxxvii. % ^n Q (8). &. it should for all values of % and n 2 such that ft 2 &.. This plainly contradicts the inequality 84. then there is a sequence 1} ?i 2 . and functions..gt. (A .lt. the condition is have only to show that it involves X = A. such that as ^~*".lt.lt.. we have and ^2 = ^0whenever HI Then than n Q &. find values of HI and 7i 2 .gt. In order to prove this we A then there are..n .. But if X &.(nj A . n . * A number of simple proofs of this result are given by Hardy and Littlewood. In the second place.(w)&.. may all be omitted from the enunciation of Theorem first place. and such that &. .X) if 8 is small enough. small 8 .gt.X .gt.lt. results are very similar when &. In the second place... THEOREM (j) 1. a condition usually referred to as the general principle of convergence cf&. so (j) (n} tends to a limit. each greater than any assigned number ?i .lt. In the can find n n so that first place... results of the preceding sections enable us to formulate a very important (n) should tend necessary and sufficient condition that a bounded function to a limit... LIMITS OF FUNCTIONS OF A interval. (n) is the fractional part of n6.. (n) is bounded and the last section applies also..lt.. Let us cnoose some particular value n^ greater (n^ 8 . The to a limit. [iV ?i any number of the $(%)~*".lt. (%) + 8 so the second part of the proof of when n 2 ^w ..lt. ". Hence X = A. (w 2 ) &.. vol. (n) tends to a limit I then it is certainly bounded ....
if it exists. up to the present. concerned ourselves only with real functions of n and series all of whose terms are real. But in elementary all with it. and = lim (n) &.. when u n is complex and equal to vn + iwn we . s respectively. VIII can be simplified by the use of the principle. + w na )&. converge &. must be it has not the limitations inevitable in theorems of such a special character . without requiring and us to be able to tell beforehand what the limit. and to obtain work it is generally possible to dispense we want from these special theorems. And it will be found applications are that. shall say that the series is convergent and has the sum I = r + is. it should be possible to find so that ^+1 + ^+2 + . a (n) are real functions of Then oo . practically no will only made of it in the chapters which follow.gt. if p (n) respectively to limits r l s as n * we and a.8385] The POSITIVE INTEGRAL VARIABLE 153 theoretical importance of the general principle of convergence can be overestimated. Like the theorems of 69. A few proofs given in Ch. Similarly.lt.(n) shall say that (n) converges to the limit = r + is. Suppose that (f&.* We remark that. viz : sufficient condition for the convergence of THEOREM of the 2. Limits of complex functions and series of complex In this chapter we have. if the series are convergent and have the sums * r. sufficient condition for the convergence given any positive number d. where p (n).gt. 8 for all values of n\ and n% such that 85.j&. . (n) is complex and equal to p (n) + iaand (n). in spite of the importance of the principle. The necessary and series u\ + u% + nQ I is that.lt. as those of 69.lt.. n. write /. it gives us a means of hardly deciding whether a function $(ri) tends to a limit or not. if we suppose that we obtain at once a necessary and an infinite series. There is however no difficulty in extending our ideas and definitions to the case in which the functions or the terms. terms of the series are complex.
up . and ^+v + 2 z . to the sum m. . is convergent and has the sum is of course the same as to say that the sum converges to the limit I as n &. In the case of real functions and series we also gave definitions of divergence and oscillation. When necessary to many possibilities make further distinctions of this kind. . . finite or But in the case infinite.f&.j&. is convergent then lim u n If MJ + u 2 + u 3 + is convergent. . .u 3 + . If lim = = I and lim ^ {(f) lim (4) (5) = m. the sum (3) I u (f&.gt. then ki^ + ku + 0. . = .lt.J converges to the z 2 sum sum I..gt.. oo . then = + m. . .... . then lim (n) = kl wxi&. .(n) l then lim (f)(n + p)l for any fixed value of p.lt.lt.. then (7) + v^ + (u + v ) + . . the (1) If lim = (f&. . l. and the sums of the two .. If lim &. 86. (n) + ty (n)} (n) l k&. converges to the (8) sum If Wj + u2 4. u1 . we shall make them by stating the way in which the real or imaginary parts behave when taken separately.gt.lt. .gt. . there are so it is hardly worth while. I and lim \fr (n)=m. u2 (n) .. (2) If . of complex functions and series.gt. where we have to consider the behaviour both of p (n) and of that this is a (n). then so is any series (9) formed by grouping the terms in brackets. If M! + uz + converges to the kl. The reader as will find no difficulty in proving such theorems following. then (6) If Wj +u + (M.(n)ifr(n)**lm... . . which are obvious extensions of theorems already proved for real functions and series. z^j 2 . . + u + . If lim &. .j) (n) (n) 2 I.. is convergent and has the sum then is + k+ +u + convergent and has the sum k+ and up+1 + up+2 + . is convergent and has + 2 I. converges to the I sum l+m.f&..gt. +c+ a+b+c+ a f b .154 LIMITS OF FUNCTIONS OF A [IV To say I that u l +u +u + 2 s . series are the same.
gt..lt..r .lt. &. and when w 2 ^o. ) The following theorems are of a somewhat different character.(n) should converge to oo ..(ri) tends to a limit s. it is necessary and sufficient that should converge to zero.gt.cro.j and no".. a p&.r (w)W. where n Q the greater of it is sufficient n&. pp = pp ...gt.lt. (p (ri) ^ (ri) ..p (ri)*l then p (ri)*r and a (n)*~s.gt...85. in order that necessary and sufficient that &..e..lt.(r + is) (/ f is = Im.ss + i (rs + r s). Hence when n%&. r (nO &.r).. us prove theorem (5).... 86] POSITIVE INTEGRAL VARIABLE let 155 As an example. I 8384 are still true when We have to show that the necessary and is sufficient condition that &.% in the same way it may ...* .lt.r both converge to zero then it is plain that follows from the fact that the numerical value of p or cannot be greater than v/(p 2 + 2 ) If p(ri) so.p &.gt.. and we can apply (10).oo... (10) zero as n In order *~ that &. and &.gt.H the first inequality holding when n 2 &.f&.. p (tt)W.lt.lt. To prove I that we have I only to observe that &.lt.lt.lt. (n) p (n) + ia. n^ ^w and the second when n 2 &.lt. be shown that o.ss + p . HI ^n .lt.lt.(?i) should tend to that ^W^(i)l&. P (2)p (%) 1^10 (* 2 )0 (l) Thus p (n} tends to a limit r. Let p(ri)*~r. &.. More it is generally.+ i (per + p (n) ^ (n) ..r(ri) The converse o".lt.. &.. ?i 2 &..rs + r s * rr .(n) should converge to a should converge to zero. Thus the condition S (1) is necessary. and so we can find numbers 7i and n Q depending on 8 and such that I P W P W is !&. For (p(ri)l converges to (12) zero.lt. (p (ri) ^ (ri) *. n^ ^n ". Then But and so that i.gt. $ (n) Theorems 1 and 2 of and u n are complex. (1) when If ".j&.* rr ..r(ri)*~s. &. does &.H I cr(na) .lt. &. (11) limit I....ni ~n ..
lt. Thus tends to zero if and only if r &. we have or 1 + r Cis +r 2 Cis 2 (9 = = (lr cos + r cos r sin &.r cos 0)/(l . from this special case the limit..r Cis 0) + ir sin 0)/(l . The n number.lt. .. &. if and only if r &.lt. + TT.lt. . 1.lt. = (1 . we see that these provided r results hold also for negative values of r numerically less than 1.lt.2r cos 1/(1 +r 2 ). ).2r cos + r = r sin 01(1 . + . by (1) of 86. . sn  then zn  = rn . we obtain 1 + r 2 cos 20 2 ? + sin 20 + + . Thus they hold when . if (a) I and therefore If z I = zl. can only be z zero. # + z* + . . Now so that  if = r (cos zn + i sin 0). = rn (cos nO + i sin nO). &. into If we change 1. Separating the real and imaginary parts. If z* * I then z n+l I. zn ~ ^ = zzn ^zl.156 87. . unless 1 f z\ if y when is the value of s n is n. except when 88.lt. . when z is complex. Apart if it exists. . it follows that the series \z &. But.2r cos 6 + r 2 2 ). which is = =1 then Km z n =l. sum w/ie/i convergent is 1/(1 #). 1 . &.1 r 1. . . Thus z r (cos 0+i sin 0) = r Cis 0. .lt. series 1 The geometric Since + z + z* + . . and it follows from (10) of 86 that lim z n = to a limit. &. 72. 0=1 In no other case does z n converge and z n * 1. convergent if and only if r and r 1. by (4) of 86. where r is positive. only possible or (b) z = 1. LIMITS OF FUNCTIONS OF A [IV limit of z as n x z being any complex Let us consider the important case in which zn (n) This problem has already been discussed for real values of z in . . 1.
.(ri) = 0. .l. p. then a multiple of 27r. 2. when r&. 69.lt.87. Prove also that (w) for all values of &.z\ n sin w0. . will be foiind in Chrystal s Algebra. IX by a different method.gt. if 1 and &. $ () oscillates infinitely. 1. Trip. 78.lt.()^lim^(?0. &. MISCELLANEOUS EXAMPLES ON CHAPTER 1. .9 is not a multiple of STT.1 } n + in + ( (rc) *)". that r&.gt.gt..lt. 1.) A proof that lim {^ (n) (n)} the limit e.j&.gt.&.lt.] 3. then both (w) tend to (ri) ^ ^ limits.lt. The function 1. We shall however . then 2. [The first result has already been proved 4. 2. r 2 &. when 1.gt. (Math.] \^ (ri) &.l if r =l and 6 and 6 is not a multiple of 2ir. 0. then + 2r cos 6 + 2r cos 20 + .r 2 )/(l . Show that this equivalent to the condition that z has a real part greater than \. steadily increases.lt. Prove directly that (f)(n) Prove further 1 and to 1 when r = I and 6 is a multiple of 2ir.gt.lt. 0.lt.f&.lt. $ (n) = r = z ml(l . vol.lt.lt.(rc) oscillates finitely. if then (jf) (?i + 1) &.lt. and is &.}] oo . 0. and lim&. Establish a similar series of results for 3. in 73.lt. if j s  &.(ri) of the preceding examples) that both tends to GO * .(ri) as n tends to (ri) steadily decreases. = (1 .lt. and ^ (n)&. 1903. 0. of all distinct pairs of positive Show that lim 2 ( n /?i ) = l/6. and that therefore each function ii. and and for all values of n.lt. 5. . (?0 and ^ (n + 1) ^ &.2r cos + r2 ). takes the values &. n by a formula which does not = J (1 + ( . [This is an intermediate corollary from Prove that. if r&..p(n) in terms of [&. 0.gt.. and deduce (by means and ^(n) tend to limits as n rc: The arithmetic mean of the products integers whose sum is n is denoted by Sn 5. *  tends to =0. 1.. 7i IV. (ri). 0. If if (ri) . if and only 4. prove this in Ch... The series 1 converges to the condition is sum 1 / (l  ^\ =1+z if \ z{(l +z)\&.gt. Which of the theorems of I 86 do you uso ? Prove that 1 if  1&. 1. Prove that m+l+ z m +z . then (w)^ + co . 88] POSITIVE INTEGRAL VARIABLE 157 to = r n cosn0 converges Examples XXXIII.j&. involve trigonometrical functions. Express &.
1901. If # #3&. x b&. . .. and then equal to Show that the graph of the function * Exs. . is equal to except when x an and then equal to 1. tends to as 7i* co .lt. 8 12 are taken from Bromwich s Infinite Series. then \\mxn =*JA.#&. (! f a 2 + If is . #4&. (Math. . distinguishing the cases in is increasing and continuous (see Ch.&.. #6&. .lt.1 values of w..gt. then the sequences +i an ^ ^25 ^45 ^6? are one an increasing and the other a decreasing #i) sequence.gt.. 12. x and A being positive.) unless ^ is an integer. and each sequence tends to the limit a. a^ifo + C^M)}* and so on.. [Prove 7. . + # n)M #3. Consider in particular the case in which this equation has only one root. The function /(#) sc.gt.a?2&. is denned by the equation Discuss on general graphical grounds the question as to .] If (ri) is a positive integer for if 0&.gt.gt. (5. #2 are positive and #n + 1 = (^n +^ni)j then the sequences #l5 #3 are one a decreasing and the other an increasing and and they have the common limit J (xi + 2#2 ). and tends to if #&. sequence. the positive root of the 10.1. first that n ". ..gt. line y = x from above to below and from If #j.. f IA = fe4^ all . . #5. then the sequence x lt #2l an increasing or decreasing sequence according as x\ is less than or greater than a. . 11. 2 equation # f x = k.) The function y= is lim integer. X y denned by the equation . If a n increases or decreases steadily as n increases. ^  .lt. the positive root of the equation x*=x + k t and in either case x*~a as ?ico. 13. an(^ ^ and #1 are positive.gt. QO with n. Draw a graph of the function = hm 14.xsm*irx is equal to 15.. . #n + i = */(+ # n ).* . which the curve y=f(x) crosses the below to above. then x of 4? 8.. and to +00 Discuss the behaviour for other values of x. #2 #3 .158 6. V) for all and a sequence #j. then the same is true of 9. values of #n + i=/(#n)whether xn tends to a root of the equation #=/(#). LIMITS OF FUNCTIONS OF A Prove that if [iV Xi = %{x + (Alx}}. = ^/(l+#n) and k and x\ are positive. The function x n*oo n&. m . . Trip.* +J.
ft are real and distinct.&. TO*.  lim [lim (cos (m I 7rx)} 18. so that a0 2 f2&. B. be the roots of az* + 2bz + c=0.1. 6. together with (as a x= 1. may be represented in the form is rational.lt.gt.lt. according as 5 ac or 6 2 If 6 2 &.POSITIVE INTEGRAL VARIABLE composed of parts of the graphs of two isolated points. .ac /3 then the roots a..s + c = a(. \z^a^ &.lt.r).(#) 159 rule) (a) and ^(x). (c) x = 1 16. 2 19. I/(z . If *&.then . a.gt.00 sgnxEx.a.. (6) x= .7?i{sin (w! irx)}. Is y denned when is &.). / andif . zft ac.~is equal to zero from a certain value of 2 irrational then sin 2 (m irx) is always positive. to 1 when x when x Prove that the function y which is equal to is irrational. It is easy to verify that Az+B a There are two (1) I _ (a ft) &. while if \z\ lies between a and \ft\ one fraction must be ex The reader should panded in ascending and one in descending powers of If \z\ is equal to a or /3 then no such write down the actual results.a)(z/3). powers of z. and so sgn (sin (m irx)} xxxi. and therefore onwards if nx)}. 14. and a and unequal. 2 cases. 3 We then shall suppose that A. 18).= 2a  (l a\ + a + 2 + a .a.] Prove that y may also be represented in the 1 form 2n ]. a) If \z is less than either a or and lj(zft) in ascending powers of z If \z\ is greater than either a or \ft\ we must expand in descending (Ex.then = a. 2 lim (2/7r) arc tan (nx\ sgn (sin x is is 2 [If x is rational then sin (m 7r. and y= where as in lim 2 s&. Let Expansion of (^0+5)/(a2 +2^ + c) in powers of z. we can expand expansion is possible. ! (m ! m : \ ! always equal to 1.gt. c are all real.lt.
lt. p cos &.6. 43).p we obtain a is similar expansion in descending powers.l \ z\ = no such expansion 20.lt. ) then (Math. If lim sn =l then lim ^ &.p). [We have 24.6/a.gt.gt.gt. Trip. Then we have to prove that (*iK 2 + . 2 r __ . is .lt.\ then ^^sin^ + 25.] Expand L](za) a  2 in powers of z.lt. + O/ w tends to . 19 in the case of the fraction 1/(1 +z 2 ). 1 p then each fraction may be expanded { in ascending The coefficient of z n will be found to be Ap sin ncp + B sin If s&. 2. Prove that if \z\&. If 2 1 &.160 (2) LIMITS OF FUNCTIONS OF A If b 2 &. so that cos =powers of z. 2 2 l). Ill a = pCis(p. If a/(a + bz + C2 2 = 1 +p& +p 2z2 +. ascending or descending according as or \z\&. and find the corresponding ex which holds when a&.1 =? + #. 1900. 22..p = % (a + /3) = .lt. and we can write /3=p Cis(&. pansion. \z\&. \ b \ then a? where p n ={(a) n /b n+2 } {(n + l)aBnbA}.lt. Show that if then [The sum to n terms 21. s&.) 27.}.lt.ac [iV then the roots are conjugate complex numbers (Ch.a. For what values of z do your 3 in ascend 26. [Let zero if t n does so. Show that if 62 = ac and az  \ &.. while if possible.gt.. 2 ing Expand (l+z)/(l+z \ (l + 2 )/(l + ^ ) and (l + ^2 )/(l+^) results hold ? powers of z. Verify the result of Ex. in descending powers of 23. where 2 p = a/3 = c/a.lt.
but the terms are all small.gt.j&.g..ft . When this is the case the proof is never very easy. &..l) n }. Hence each of the parts into which (t + t2+. (n}&. : small The point of the proof is this we have to prove that (ti + tz + . f p + 2. numbers ^. 27 oscillates as M*OO .] If &...gt.+8 n . if he desires to become expert in dealing with questions about should study the argument above with great care.n Q) if nQ is p/n * as n*. since . divide the ) .. if A is the greatest of the moduli of all the numbers t lt t2 . (n) l   1) rc*. But.n . The slightly different ways. + tn l )jn has been divided 28. Thus when n&. we might suppose p to be the integral part of Jn. and the theorem re duces to that proved in the last example..] 29. but their number is small compared with n.. so that their sum is small compared with n.( .] not true : for s n 30. If *= (*l (1 so that s n is equal to 1 or according as n is odd or even. ..gt. tn : Let be any positive number.!) = *. tn into two sets t lt t2 .POSITIVE INTEGRAL VARIABLE 161 .gt. then (n)/n^l. so that p+cc and p/n*0 e. (n. 11 .. oo . the t s being small when their suffixes are large. is [This example proves that the converse of Ex.gt. If c n . The number in the second group is not small compared with n.lt... [If is small (n when n ^ I as then is large. We . which proves the theorem. to split it into two parts which have to be proved to have the limit zero in limits.x large enough. we have and this also will be less than %8 when n&. It is very often necessary. We The terms in the first split up the terms in the bracket into two groups.+*)/n! as ?IQO..lt. then +*j+. (ri) = s +8 2 + . fp + 2 * are all numerically less than %8 be.. group are not all jsmall. reader. we can choose when and so ft&..lt. but more slowly than n.. in proving the limit of some given expression to be zero. and their number at any rate less than n.$&. sn denote the sums of the first n terms of the series then n. t 2 ..tp and tp + l Here we suppose that p is a function of n which tends to oo as %ao. f However small 8 may n Q so that p+ 1 . + tn )/n is when n is large.
lt. We shall confine our and we shall denote such a function by We suppose x to assume successively all (as).e.gt. We shall now return to functions of a continuous real variable. tends to infinity. selves entirely to onevalued functions*. 1 of of the last chapter. CONTINUOUS AND DISCONTINUOUS FUNCTIONS 89. &. values corresponding to points on our fundamental straight line A. theorems to which we are led are practically repetitions of those 58 we Thus corresponding to Def. there is a very close correspondence between functions of x and functions function of n may be regarded as a selection from of n. &. or to oo c . . and that x assumes all values as it tends . that the point P A which corresponds to x coincides in * to the right of its initial position. have * : Thus in 26) for the function + xAr *Jx stands in this chapter for the onevalued twovalued function whose values are +*Jx and . whereas n tended to x tends continuously to oo distinction by saying that We . . and not (as . I at the beginning of the last chapter. and write x to oo is between the tending of n this tending of x to oo to oo . cc The only difference discussed in the last chapter.lt.x/a?. Limits as x tends to oo . turn with every point of can express this GO by a series of jumps./&. starting from some definite point on the line and progressing In these circumstances we say that x always to the right. ^ i. the peculiarities which may characterise the behaviour of a Now we are concerned with the function $ (n) as n tends to oo and the definitions and same problem for a function (at) .CHAPTER V LIMITS OF FUNCTIONS OF A CONTINUOUS VARIABLE. As we explained Every f the values of a function of In the last chapter we discussed x.
using the words small in nearly large a sense similar to that in which they were used in Ch. IV. Similarly we define 3. If the is less satisfied. simply lim Similarly we have is : &. DEFINITION as x tends to GO if. i.f&. then  conditions of neither of the two preceding (x) is said to oscillate as x tends &.8 =x (8). 0(2).gt. t In the corresponding definition of values of n. we can choose $ when x ^ x (A).gt. &. a. matter is not quite so simple. The function x if./&./&.p(n)\&.gt. 0( 1) and K.lt.gt.lt. (x) * Finally we have : DEFINITION definitions are to oo 4&.gt. is assigned. (x) &. If is  &. Similar (n) will . 1.K for all values  K .lt.GO.f&. ( . there (x) no risk of ambiguity. and otherwise infinitely.gt. &./&. where is the greatest of 0(1).(x) .f&.lt.gt.lt.)*. when any number A. &. hoiuever a number x (&) such that (x) is said to tend to oo with large.x&. &. for all values of x equal to or greater than x (8). and not merely when n &.j&. for if 0() of n.*oo.89] LIMITS OF FUNCTIONS OF A CONTINUOUS VARIABLE 163 The function (x) is said to tend to the limit I when any positive number 8./&. however small. &.lt. (x) = /.f&.K  instead of (#). sometimes find a.lt.lt. if \$(x)l wlien x &.. we postulated that (n) n Q .lt. then \&.gt.lt. Thus we may say that (x) is small or nearly equal to I or large when x is large. * We shall oo . .lt. Here the many values of x less than XQ. is assigned.r? . &. a?*co or. when *1.lt. &.e.gt. in the last chapter we con sidered very carefully various less formal ways of expressing the * oo facts represented by the formulae (ft (n) * I. &.* + oo. (x) differs from I by less than 8. (x) = I.gt.lt.gt.lt.lt. been equivalent. as there are infinitely K for all 62. .lt..0(o.lt.. it convenient to write +&.lt.. (x) ^ oo cc *. When this is the case we may write lim &.gt.+ Go &.lt.gt. &. (x) than some constant K when x = x^ t then said to oscillate finitely. But then the two hypotheses would have when w&. .gt. The reader remember that modes of expression may of course be used in the present case. a number x (8) can be chosen such that. 112 . A We then write &. or DEFINITION 2.
lt. IV. y =$(*) and y 5. or x assertions x tends to meaning lim a:*co (x) =1. Consider in the same way the functions (sin#7r)/#. then y tends to oo as &. .\x\ oscillates and .gt. IV. 4).(#)*oo sponding property /It for $ (ri). analogous to the geometrical If (f&.lt. [x\ four functions correspond exactly to functions of n fully dis cussed in Ch. (x) x tends to as x tends to y tends to oo and the question of the behaviour of is the same as that of the behaviour of . $ &.(#) $ (x) cos XTT and $ (#) sin #?r oscillate finitely. y=F(x}J\ F (x} are some pair of [The Limits as x tends to &.164 Examples XXXIV. a cos 2 XTT + b sin 2 #71. The graphs of the last three were constructed in Ch. (x)**l t and oo I is not zero. The a wavy curve oscillating between the curves oo.gt. It is between (n) derived equal to zero whenever x is an integer. finitely. $(#).lt. 4. (# sin #?r) 2 . tan^Tr.g. If $ (#) or graph of either function is then they oscillate infinitely. 2. 1. x .j&. In fact.lt. : #sin#7r. x[x]. (x) = &. &.lt.lt. explanation of Ch.lt. xvi. and the reader will co . [V Consider the behaviour of the following functions t x\ a* first [a]. (x) ^ oo . (n) = sin mr = 0. Discuss the behaviour. your remarks by means of the graphs of the functions.lt. so that the function from it is always zero and so tends to the limit zero.gt. 1. and 1. \lr (y) as . The reader will have no of the difficulty in framing for himself definitions of the oo and oo .j&.) 2 Xir + F (x] sin 2 xrr t and is graph of y 90. The LIMITS OF FUNCTIONS 1. x and x ). see at once that [a?]* oo . but or ^ (#)that the converse involves the corre is by no means always 2. oo . or &. 59.(#)* &. as is obvious from the form of its graph.f) ( y) = ty (y).gt.[x]} * [#] + *J{x One simple remark may be oscillates inserted here. The same is true if (j) (x] = sin #77. oo is evident that &. Give a geometrical explanation of Def. 2 simple functions (e. then &. The function $(x ) = x[x] (f&. if x = y and oo oo .p (x) * oo . a curve oscillating between the curves y=j(x}. illustrating 3.gt. as x*~ oo of the function y =f (x) COS where /(a. II (Exs.
8 t &. (x) be such a function Then say. when any the statement expressed by lim (x) number 8. = I means that. [#]. is assigned. Let \\x. of functions proved in Ch.gt. IV are all true (with obvious verbal alterations which the reader will have no difficulty in supplying) for functions of the continuous Not only the enunciations but the proofs remain substantially variable x. of course. except : C } ( &.gt.gt. he shall find that the distinction is often important (cf. The theorem which follows in that section of x onwards.(#2)&. i. to oo . when x. as saying that we can choose y less = l/a? so that .gt.e.8993] OF A CONTINUOUS VARIABLE 165 91.lt. As x tends I. But this is the  same thing I \ &. ir (y) 8 for all positive values of than v ( or equal to y We are thus led to the following definitions : . of x that lim Limits as x tends to = I.(#) will 1 1 V\ * f The reader should consider whether or no the following functions increase steadily with x (or at any rate increase steadily from a certain value of x onwards): x^x. we can choose positive as we please from XQ so that  &. however small.lt. 93. # + 2sin#.lt. y tends to the limit 0.oo .&\ : for obvious verbal changes.gt. \x\+*]\x [#]}. 92. products./&.p (oc) I &. Steadily increasing or decreasing functions.fy To put the matter more precisely. and let y (x) &. this condition is only satisfied from a definite value #. The theorems concerning the sums. &. 108 109). Theorems corresponding to those of Ch. And by making y little sufficiently small I. and quotients 6367. We are for the moment concerned only with those values of y which correspond to large positive values of as. ^2 + 2sin^. [#] + sin#. #+sin#. whenever wenever be said to be steadily increasing in the stricter sense. All these functions tend to oo as #.lt.gt.X)&.lt. In many cases.lt. IV.(^i) whenever x^x^. 0. the same. If We tne possibility of equality being excluded. and i/r (y) tends to the limit Let us now dismiss x and consider ty (y) simply as a function of y. then i&.lt.$&.2 requires no alteration but that of n into x and the proof is the same.lt. ue. that is to say with small positive values of y. The definition which corresponds to that of 69 is as follows the function $ (x) will be said to increase steadily with x if 0(^ 2 )=^&. we can make ^r (y) has the property that (y) differ by as &. 8 for all values of x greater than or equal X to 7/0 a? . &.lt.
lt. and we write lim &.?n positive values.lt.lt. we can choose y (8) so that. however small. if &.lt. it is worth while to give a formal f when any positive number S. //. in definition A. when any positive number we can choose y (B) so that Q however small. we say /*a and we write  &. 0. /?&.gt.f&.lt. (y) differs from I by less than B. when any number A. A. is assigned. lim y +o &.lt.lt. lim &. is assigned.lt. for all values of y different from zero but numerically less than or equal to y (B).lt.lt. &.lt. then by positive we say that $ (y) tends values. We define in a similar I way the meaning by negative values of . l If lim V ++0 6(y) = and y lim + 0 lim = &.gt. y &.b(y) l. and write &./&. y (A). however large.lt. when tends to &.lt. [y is assigned. we write simply This case is_so important that definition. we say manner the statement that * co as .lt.lt.(?/) terccfo to oo as oo . to or &. (y) * co as ?/ y * + and also as y * * oo (y) as * 0. 2/0 (y) = I. &.166 LIMITS OF FUNCTIONS B. (?/) tends to (f)(y) the limit as y tends . then we say that (y) tends to the limit I as y tends to 0. y y (B). as y * + or y  0. and similar definitions in which ?/ when y ^ (5) We have in fact only to alter =y &. choose If. also. We define in a similar * 0. =l y y (B) to There is of course a corresponding analogue of definition B. to the limit I as y y++Q B. we can (A) so that when to &?/ &.lt. (y) = I So that &. If.
lt. \ } Thus lim a. i. then lim (f&. (x) = L By restricting ourselves to values of 0&.f&.j&.l&.gt.lt.\x x greater than a.gt. : 94.x a + e (8).lt.lt. by (#) replacing tends to write as a  ^ e (8) by a&.f&.(#) x as ^a it is .lt. to If y * then x * a and ^r (x) lim *&. x a \ ^ e (B).r the limit always determine e(8) so that when &.93.gt.lt. &. finitely as y * + 0.gt.a) = ^ (x). y * in a similar manner. can give similar definitions referring to the cases in which or &. y * 0. variable denoted The preceding definitions have been stated in terms of a by y what letter is used is of course immaterial. we say that (y) or infinitely as the case may be.j&.gt. and we are naturally led lt write i/r (x) = or simply lim ty (x) = I or ^r (x) * Z.(x) l is equivalent to the two assertions a lim $ x**a+Q (a?) = I.gt.lt.p right which we may (x) = I. oo . * I. 167 co or to (y) does not tend to a limit. and we define oscillation as &.lt. (a?) =L \\ a?*a0 We than a .gt.gt. be formally and directly defined as follows: if.lt. and to say that (a.jf&. Limits as x tends to and write y a. or to Finally. 94] OF A CONTINUOUS VARIABLE . * c/&.e. (x . In the same way we can define the meaning of lim cf&.) tends to The meaning of this equation may I as x tends to a. through values greater or less probably unnecessary to dwell further on these since they are exactly similar to those stated above in &. define when x approaches a from the lim &. &. and we may suppose x written instead of y throughout them.gt.lt. Suppose that &. lim &. we .lt.(y)**l as = xa.gt.gt.lt.* = cf&. we can &. (x) = I.gt. given S. (y) = &. if oscillates as y * + 0.(a?)*oo but definitions.
= ty(y} is a steadily increasing function of y. In particular.$)(x }&.&.lt. then will of equality being excluded. never greater than $(a). whenever be said to be steadily increasing in the 96.lt.(#) a &. and by putting y=za to functions of 0.(#".gt. () We can define (a .x".lt. IV. [We saw functions of in 91 that the theorems of Ch. when d is given. $(x] ^(x}+ll .e. the possibility &.) is a number will ^ 6 &.lt. /f M 95.lt.x + then &.lt. 1. Steadily increasing or decreasing functions.lt. hold also for x when them when * see to functions of 0*a. Limits of indetermination and the principle of convergence. W+ ^. Suppose first that x&.a . If there such that 0(# )^ whenever ae&. an interval . K &. .lt.168 the special case LIMITS OF FUNCTIONS [V when a as x*a 0. we can choose e (6) so that \ (o?2 ) $ j Examples XXXV.(%) and we can always discuss the by putting x a y and supposing that 7/^0.lt.a.b(x} &.e.a + e.x &. If as x*~ a. and prove that the necessary and sufficient condition that ^ (x}*l as x~a is that X = A = We can also &.gt.lt.lt. the lower and upper limits of indetermination of (x} as #. #*oc ?/. then we can define X and A. shall write lim &.lt. and when a e a + e) *.$(x". &.# .lt. 63 et seq. (x] should tend to necessary and sufficient condition that that. & ! If &. of the notion of a function bounded in For some further discussion 102. if $(#) is .f(a?) JJ * + = 0(a+0). and &.}. By putting x\\y we may extend when y*0. and 92 that It follows from We (a?) tends to a limit not greater than $ (a). &. and put y=\l(ax}. Then y^co as xa 0.lt.K X so that bounded in an interval including a (i.&.(#) be said to increase steadily in the neighbourhood of x=a.lt.lt.lt.Xf. i.lt.t&.4 sHfr(x}&.lt. or#oo. prove that the is a limit as x**a (#1) &.x".j&.0) in a similar manner and it is clear that It is obvious that similar considerations may be applied to decreasing functions.lt. All of the argument of 80 84 may be applied to functions of a con tinuous variable x which tends to a limit a.lt. . establish the analogue of the principle of convergence. behaviour of &. then (#) unless in the last case + ^ (a?) / = 0. if we can find e.
0. unless a=0. + K.9497] OF A CONTINUOUS VARIABLE 169 definition given above.. we then get a. H x. according have like or unlike signs. (x) when x has any value . . and if m negative. 6.a \ \ . is any polynomial. \x a &.0. R (a) It is very important that he should see exactly where ffiejis wrong) We shall therefore consider this point carefully before passing on to any further examples.gt.gt. is P (x) 7. When the result follows from Ex..lt. or x = a1 Of course may why P (a).n .lt.lt.gt..oo as #*. +kxm = a. 63 and write first result he need only take the proof of Theorem I of The reader should however throughout x for 2. Discuss the cases which arise when a=t=0 and more than one of the first coefficients in the as a and denominator vanish. lim x m = am ** is . m= then xm and xm *\. try to prove them directly from the formal in order to obtain a strict direct proof of the Thus. = 0. It follows at once that lim P (x}P (a)..am equalities (9) or (10) of 74. a/a.a R (x} = R (a). } 5.* oo or \ m odd or even. from &. its if R denotes any rational function and a not one of the roots of 8. If ^m^ + oo 4. 1 above. H is the greater is (cf.] If m is m a positive integer then x m *~Q as #*0. [This follows at once. is a negative integer then as x*~0. 6. .. 7. xxxvi.^)} = a/a. n. [If m&. the case is reversed if #* (). according as #m f oo is as #^+0. x=y + a and apply Ex. He ask not simply put a. For x m ~ l and mam ~ l of the absolute values of mx  \ that lim xm =am for all rational values of m. if m is any positive or negative put integer. If a negative we write x= y and a=  b. then the function tends to +00 or .gt. when a is positive. am The statement is lim x*Q =l (f&.lt. Then The reader will probably fail to see at first that any proof 97.lt. lim (a + bx + cx2 + lim \(a + /3=t=0.. 5. 8 above is necessary. denominator. arid a=t=0. bx+ .+ (). when a = w&. xxviri. 3. Show is and m m .(x) * a statement about the values of &. 4. Ex. If while #". 4# m )/(a + /5#+ . The /3 case in which both a and a vanish is considered in Ex.e for n &. where 4). except 4. 5. of such results as those of Exs. Ifa=0 . except when a=0 v^a the in negative.] lim x*&. a for oo and 0&.
&. then if x = (x) 1 Or &. and one consists of the axis of x.j&. 1).lt.lt.lt. let us consider y = x/x This function is is 24.y Q . So far as we know. than 1 .gt. from either side. with the point x = isolated point. for # = 0.x* ^ 1 &. When we make the state about the value of (x) ivhen x is nearly ment we_assert that. or it may have some For example.gt. to zero. while if x 1.lt. denotes as usual the greatest integer not greater = [!] = !.lt. But ^r (0) left out. (x) may value all not be defined at other than values of x I. equal not equal to 1 when x = Q: * it is in fact not defined at all for x = 0. the fact that if The equation (2) we move along the graph towards the expresses axis of y.lt. &. Now that consider the function ty (x) which differs from fy &. It is obvious that the equation (x) by &. &. The graph of this function exactly equal to zero. It to 1 for all values of x save x = 0. (x) only in = 1 when (x) x = 0. (2). when x is nearly equal to zero. A of of Q&.lt.x] x &. already discussed in Ch. For 0. 2 the function again. 1 . the point (0. tends to the limit zero. viz. the first 93 we make a statement about values of y such that these inequalities being inserted expressly in order to exclude the value w = 0. Then for. = 0. 1). &. II. &. consider the function defined for = 0. and so ^ (x) = [1 .(x) when x is actually equal to all 0. when x is nearly equal to zero.f&.lt. being always equal This fact is in no way affected isolated point (0. then the ordinate of the curve.lt.lt.lt. &. then &. One is N where or [1 .170 LIMITS OF FUNCTIONS [V It is not a statement distinct from but differing by little from zero *. about what happens We assert nothing whatever equal to I.y&. For when we say Thus in Def.lt. .x ] = 0.lt. by the position of the The reader may artificiality : but it is object to this example on the score of easy to write down simple formulae repre like this near # senting functions which behave precisely = 0. ty (x) is not only nearly but = 1.
(#). . [This follows at once from the inequalities x+l lim (x7  %& + 1)/(^ .0. of $ (x) for is x=Q.] result of Ex.lt.(. 1 .) that we mean (as (x) is defined for x we can calculate its value for (f&.] .lt./&. n.lt.gt.j&. we explained x = by putting x In this case we cannot. This of in fact.(#)*oo n . lim = 0. however small the difference Similarly may &.e. Show that the a 74.&B 2 + 2) = 1 . even. solely in that it is not defined for x expression. or &. /6 . &.m is Ifm&.gt.(#)* + 00 as #*().e. $ (x) as x tends to zero from being equal (0). None the less 0.lt.gt. [Observe that x 1 is a factor of both numerator and denominator. we obtain 0/0.lt. the ordinary case. (x) = {(x + is 2 I) !}/. Examples XXXVI.gt. but undefined when x = None the = 2.co as .] 5.lt. 2.0 /6 &. Thus if &. == I. x*a lim (xm ** am )/(# a) = mam ~ ! . if m is any integer (zero included).lt.*?). 3.+ () and &.97] that &. the value &. (x)  co according as /6 or /6 &. On limit of the other hand there is of course nothing to prevent the to &. lim (x) =a &.lt.lt. which is a meaningless in we put as = (x) The reader may object divide numerator and de But he must admit that when x = this is nominator by x Thus y = as /as is a function which differs from y = 1 impossible. (x) = 0. =x+ 2 so long as 0.j)(x)=x then = (0) and lim i.M is odd.f&..+ and &. II. from a practical point of view. or or as o.gt.lt. according as /6 &.  If m = n.lt. 2 remains true for all rational values of m. When in the actual expression of (x). view of what most frequently occurs in from the point applications. provided of (9) and (10) 4. 0 n and n .lt. is positive. x/x is equal to 1 so long as x differs from zero. (x) &.+ oo as #*. to zero.n and &. x is not less equal lim (x) &.a) = 2a. (#) * +x &. lim (x2 . Discuss the behaviour of as x tends [If m to by positive or negative values.v^0.gt.lt.lt. If or m &. OF A CONTINUOUS VARIABLE 171 in Ch.lt. lim (x/x) for = 1.lt.a2 )/(# .gt. be.
&. with x would be of either the This is were complete.x)}/x=l.lt. x2 x 3 . Orders of greatness. ..j&. therefore..lt. And as a matter of fact the integral orders of smallness defined above are so much more important in applications than any others that it is hardly necessary to attempt to make our definitions more precise. x3 than #2 and so on. These definitions have reference to the case in which There are of or course corresponding definitions relating to the cases in which Thus if (x) tends to a limit other than zero. when x tends to 0.. 19. . (x) is of the &th lim v/(l+^) = Hm v/(l. # 3 much smaller in other words .] = l. meet the case in which Similar definitions are at once suggested to is &. . For example $(#)=# If it rapidly than x and less rapidly than x2 . [Put l+x=y or lxy t and use Ex. . cf)(x) is of the &th order of smallness when x is large: while if &. since lim (2x + 3x2 + x7 )/x = 2.lt. x3 .. that &. lim ( v/(l +#) . Thus we might 7 5 say that # / was of the th order of smallness. and so on lim(o. We first say. xxxv. [Multiply numerator and denominator * In the examples which follow it is to be assumed that limits as x * are required. tends to a limit other than as x tends to 0. fourth. tf 2 . 8.172 6. x^a. diminishes with x. to 0. whose limit. (x) is of large (positively or negatively) when x is the &th order of greatness when x is small jc to a limit different from as x tends #*(). as x*cc then #*&. Thus 2x+3x2 +x7 (x}Jx of the order of smallness. we say . *7. It is convenient to have some scale by which to measure the rapidity with which a function.. must not be imagined that this scale of orders orders of smallness.. .J(l . = 0. 7 /6 tends to zero more obviously not the case. then every function (x) which tends to zero first or second or some higher order of smallness. We shall however see later on that such a scale of orders would still be altogether incomplete. lim (#3/# 2 )=0.. all also tend to 0. /#) x*Q 2 [y is When x : is small x 2 very much smaller.lt. It of smallness is in any way complete.*0 Another way of stating the matter is to say that. #*&. LIMITS OF FUNCTIONS Orders of smallness. The reader may not unnaturally think that our scale might be made complete by including in it fractional orders of smallness.(x) We shall say that if $ (x )lx~ k =x (f)(x) tends small.gt. .r) 8. 22) the contrary is explicitly stated. third.lt. Similarly we define the second. and it is natural to take the simple functions x.. .lt. then we say that order of greatness when x is nearly equal to a. is 0. that (x} is of the first order of smallness if is &. as x*a. unless (as in Exs. as the measures of our scale. as x tends to 0. but x2 tends to more rapidly than x.) (x fc (x) tends to a limit other than zero. a.lt.. still.
VII. [rut# = sm r r&.] .V(l . 10. . 7 =a . lim x cosec = a. .  x x ". when y is not denned. .! + cos lim 19.] ^^Wo.=1. lim x x cot x = k. OF A CONTINUOUS VARIABLE Consider the behaviour of ( N/(l 173 as x*~0. .. i im !H^ &. T v. . The justification of these assumptions must be . sm x x [It if may be deduced from the positive and less definitions of the trigonometrical ratios* that x is than \K then sin cos#&..^+0\ follows. * As x is an even function. the result :Ll2!f = i . . that the area of the sector is greater than that of the triangle inscribed in the sector. and y*l as #*().gt. I S this true if a = CH arc sin  16.^ Hence lim (l o. Draw a graph of the function Has it a limit as #0 ? [Here y=l except for # = 1.gt. x +i tan^ 7i * The on certain pro proofs of the inequalities which are used here depend of a circle which are usually taken as geometrically perties of the area of a sector iutuitive .97] 9. lim x x tailor = 1. and * / lim *+0 ^^=1. ^.lt.xm)}/xn m and n being positive integers. . lim .j 17.lt. 13. postponed to Ch. J. 12. J. 15. for example. arc tan cur lim x =0. +xm) . 18...._. 2 2 &.
. Continuous Thus he would functions call real variable. place it is evident that the property of the of which C is the graph may be analysed into some property possessed by the curve at each of its points. ai:d [V]wil. reader has no doubt some idea as to what curve. See Exs. Does the function tend to a limit as x tends to sin ( l/#) ? [No. Let us there on some particular value of a?. 29 continuous. l/2rr.] tends to the limit 21. say the value x = j . the curve C generally continuous but discontinuous for =f and Either of these curves function may be regarded as the graph of a $ (x). 1 /.lt. i. 7. when #=!/. (I/a?) sin (!/#).f&.174 20.] 22. of a is 98. The function is equal I/TT. as #0 [The oscillates finitely. # #m0.r*m + 0. the second infinitely. It is natural to call a function continuous if its Let us a continuous curve.. and otherwise discontinuous. to 1 except when = 0. 0/0. xv. #sin(l/#) behave the third 6. To be able to define continuity for all values of x we must first In the function y = &.gt. 8. ..(x) define continuity for fore fix any particular value of x.. None is defined when #=0. For these values the therefore not defined formula for y assumes the meaningless form for an infinity of values of x near # = 0. and y is Prove that if m is any integer then [x]*m and # [#]*! as [#]*() as . CONTINUOUS AND DISCONTINUOUS FUNCTIONS [v How ? do the functions sin(l/#). first 0.e. . The meant by a continuous as the curve C in Fig. take this as a provisional definition and try to distinguish more graph is precisely some of the properties which are involved in first it...
. x^if given   8.(f)  S I/".j&.lt. (f ). 98] CONTINUOUS AND DISCONTINUOUS FUNCTIONS 175 corresponding to the point characteristic properties of &.lt. we see that our y definition is equivalent to we can choose e (8) so that  (x) is continuous for &. including x = f in its interior. DEFINITION. = #.lt.gt.? = 6 (3) We (a. We function can now is &. The said to be continuous throughout a certain is continuous for all values of x in that be continuous everywhere if it is continuous Thus [x] is continuous in the interval interval of values of x if it interval. If where e is number less than J and .lt.lt. were not denned there would be a point &. It is said to x.gt.gt. Thirdly if x approaches the value % from either side then &. If (f) This is obviously (x) is defined for x %. b). 1 and x we recur to the definitions of a limit &./&. we Secondly can find an interval. (x) approaches the limit &.lt. and each of these limits is equal to (f)./&.gt. the graph. for (e.97.(#) define continuity throughout an interval. first place &.gt. What are the associated with this value of x ? (x) of P In the essential.lt. for all points of which cj) (as) is defined. &.lt.lt. &. &. satisfy our geometrical feeling of what a continuous curve should be.lt./&. have often to consider functions defined only in an interval In this case it is convenient to make a slight and obvious . common This picture of a curve is a generalisation from curves such as straight lines and circles. The which properties thus defined are far from exhausting those are possessed by the curve as pictured by the eye of sense. &. But they are particular the simplest and most fundamental properties and the graph of any function which has these properties would.e. (x) is defined for all values of x near x = %\ i.lt.j&. so far as drawing : it is practically possible.lt. missing from the curve. We therefore select these properties as embodying the mathematical notion of continuity. every value of 1 *).lt. any positive are continuous everywhere. We are thus led to the following x The function (x) is said to be continuous for % if & tends to a limit as x tends to % from either side.(#) &.
The definition of continuity given in the last section may be illustrated geometrically as follows./&. Any rational fraction is continuous except for values of x for vanishes. 7. [This follows from Exs.lt. C (Fig.(f) lines y = 8 ex (f ) (a) presses the fact that the point on the curve corresponding to x lies &.lt.lt. &. is cut off a vertical strip of the plane by such a way that all that part of the curve contained in the strip lies between the two horizontal lines in is we can always This value may shall evidently true of the curve have. now Some discuss the continuity of some special types of of the results which follow were (as we pointed II.gt. Thus our definition asserts that if we draw two such horizontal lines.lt. unless the denominator vanishes at the point. 6. 1.] Any polynomial is continuous for all values of x.176 CONTINUOUS AND DISCONTINUOUS FUNCTIONS change in our definition of continuity in so far as it concerns the We shall then say that (x) is con particular points a and b. &. 2.lt.f&. The sum or product of two functions continuous The quotient is also continuous at a point is continuous at that point.] which the denominator . and for (a x = b if (6 0) exists and is equal to (6). 99. two vertical which lines. 1. whatever We functions.B and y = (f ) + 8.lt. out at the time) tacitly assumed in Ch.  Fig. &. tinuous for x = a if + 0) exists and is equal to (a). xxxv.f&.lt. no matter how close   together. Then . Draw the two horizontal ./&.gt. &.f&. &.  &. Similarly x f ^ e expresses the fact that x lies in the interval ( e.gt. xxxv. &.gt. 30. Examples XXXVII. +e). [This follows at once from Ex.gt. 29).lt. between these two lines.
.* by positive or negative values.99] 3. For what : any) values of x are the following functions discon tinuous n. *J(x\x\\ . [. sec x. 13.r = 0. is so too.x)}. Discuss the continuity of J{(x a)(b. V {(x a)(b. 9. The standard rational function R ix] = P (x)jQ (x} is discontinuous for x=a. #=0. and &. continuous for x = Q in virtue of the remark made at is the end of 98. where a is any root of Q(jc ) = 0. sin (1 /A ) except when # = 0. tend to a limit as #0. 1/7(1 + sin. is continuous for all values of x. continuous for any particular value of nomial in $ (x\ such as a (0 (#)} +. is continuous for all values of x. then/{0 (x}} is continuous for #. 8.lt. / true of # */. The same even. yx].^J. #=. and #[#] are discontinuous (if for all integral values of x. CONTINUOUS AND DISCONTINUOUS FUNCTIONS njx is 177 continuous for &. . #=0.r).f&. is .a)l(b  . sin (1/.. and n are positive integers. h.oo In fact ljx+./ = ^. Z/{(x  )/(& . it It has no value for Ijx is not continuous for . Discuss the continuity of #". d . is odd.lt. v/(2 + cos #). defined when x 0. (x) is a continuous function of x which 12. 8).gt. xxxv.v. cot ? . It will be noticed that in the case of rational functions a discontinuity is always associated with (a) a failure of the definition for a particular value of x and (6) a tending of the function to + oo or . and sin (1 are continuous except for 1 zero The function which is equal to a. but is is all positive values of x It is not (Ex. 7.r}} sin x and cos are continuous for all values of x. [We have which 10. and when ^7=0.v 2 x /. The function x m n where n . x sin J{(x . 15. equal to 9 (#) is when # = . Such a particular kind of point of discontinuity is usually described as an infinity of the function.?f&sin 2 ^). where m for xQ. 6. nor does according as A I/a. \x\ 16. Thus (# 2 +l)/(^ 2 3# + 2) is discon tinuous for x=l. + oo or 5. and coscc v continuous or discontinuous n  If /(y) is continuous for is .C).] numerically less than the numerical value of For what values of x are tau#.. 17.sin #=2 sin %k cos (x + /i) (x + \h].v)}./. ^sin (l/^)..v) (l+sin#). to 14.oo as x approaches this value from either side. where m and n are any positive inteers of which n 4.x}}. If then any poly Discuss the continuity of 2 l/(acos . An infinity is the kind of discontinuity of most common occurrence in ordinary work.
ocas #.a2 ] (a?) 0) and $ (a+0) may be unequal. or (l/#) may differ from (0 0) = anda = 0. and (0) is (0 + 0) undefined.r) (x) (x) = \x\. and + 0).a from one side. or (x) is not defined at all on the (#).0) = .[ _ x \ for which (0 .378 CONTINUOUS AND DISCONTINUOUS FUNCTIONS [V 18. for which (00)=  1.  (a. (2) It majr be the case that only one (or neither) of (x) (a 0) and (a + (V or as exists.0) = . for instance. or 0(o + 0). or be undefined. for which (0 . Denote these limits.*0) x= a. exists.?]+# sin (I/A).1. supposing for example (a + 0) not to exist. but other side of x=a. if and a = 0. (f&.0) and 0(a+0). (0 + 0) = 1 = [. for the functions sin (I/a )* 1 the point 19. suggest a classification of different types of discontinuity. as in 95. Such is the case. but 0(a) if may not be defined. for continuity. . function which is defined only for rational or for irrational values cf x. Ex. and that (a . (#).oo as x** a + tends to a limit or to + oo or to so that 0. In this case (a) is may be The first case . (#) but that. by (a 0) and (a + 0) respectively.0) = (a) = (a + 0). Then.1. if (x) = \lx or = 1/x2 and a = 0. as the case not defined when x=a or has when x=a but a value different from may be. 7) that x = a is an infinity (x} And again we may add to these cases those in which (j?)*. = [1 . + 00 oo x approaches a from either side. it is necessary and sufficient = a. It is plain from 95 that a function which increases or decreases steadily in the neighbourhood of x = a can have at most a simple discontinuity for x a.gt. Some of the preceding examples Classification of discontinuities. (0+0)=0 (0) . but 0(0) = 1. $ (o?) = #sin Or for x = 0. And to these cases we may add those (a0) or (a which 0(#) is defined only on one side of (#) is either $(&. xvi^ 10) is discontinuous for all values of x. In such cases we say (cf. (a) (a 0) may be equal to (a+0).lt. Thus + 0) =0. (a (j3) equal to one or to neither. (. In any of these cases we say that has a simple discontinuity at in x = a.+QO of (x). The function which is equal to 1 when x is rational and to when x is is any So too irrational (Ch. = (0) the second by . but O) is not defined (0 (00) = $ (0 + 0) = 0. = and the third = 0. Suppose that (x) tends to a limit as x+a either by values less (1) than or by values greater than a. #=0 is C 1 /^) g i n (I/2 1  ) What the nature of the discontinuities at x Q of the functions 20. (3) Any tinuity nor an infinity is point of discontinuity which is not a point of simple discon Such is called a point of oscillatory discontinuity. illustrated by (# ) by = [#] . . Ex. Dis that (x) should be defined for x continuity may arise in a variety of ways. II.
lt.lt.lt. We conclude then that a continuous function cj&./&.. 2/0.&.lt. 5).gt. &.lt. must cut \ is the same thing as value of x between a. cannot be equal: (a? ).! : it is equal to I when 1&.then there is a value of x between x and x^for which In other words as x varies from XQ to &\.&.gt./&. We shall now prove that if the sense defined in ^8 then There is &. Let A and B be two points on the graph of $ (x) whose coordinates are # Draw any ) and x lt respectively.f) 77./&.(^) sense certainly declares that must cut \./&./&. = (f&. Here y is only defined when .lt. function y = lim xn (Ex.#. (x) is continuous it we consider this property of continuous curves loss of generality in supposing property as an intrinsic geometrical it is clear that there is no real \ to be parallel to the axis of x. discontinuous.gt.lt.lt.lt.(x) 77. &. a certain range of values of x.g. The points x=l and x= are points of simple discontinuity. y= &. for which For and so (x) is certainly less than 77 if it &. to say that the &.1 e. . 77 &.lt. 179 and to \ 1 J{( l +P 2 The function which is equal to as when x is l )i( + f)} when x is a rational fraction p/q (Ch. 100] 21..(#) to say that there a = &. (x} must possess the following property : if &.lt. If if the graph of &. be thought that the analysis of the idea of a con 98 is not the simplest or most natural j may perhaps tinuous curve given in possible.lt.lt. let us And let X be the graph is of where Then 77 (a? ) (x^. . In this case the ordinates of suppose.gt. IV.gt.(^ )&.lt. and (a^) &. XVL 11) is negative and for positive rational values of but ". It The fundamental property of a continuous function./&.gt. y mast assume l at least once every value between y Q and y l . Then common straight line X which passes between A and B. and 7/0 &. to the right of &. xxxi 22. &. &.f&.1 and to 1 when x=\.lt.?. y . for definiteness. k \ continuous for positive irrational values.gt. B &.?.(# &.r).lt.99..lt. Exs. does in fact possess this property. For what points are the functions considered in Ch.lt.] 100.lt. all discontinuous for Ex. &. = yi&.gt.gt.lt.HO = &. and what is the nature of their discontinuities ? [Consider.gt. COXT1NUOUS AND DISCONTINUOUS FUNCTIONS irrational II.^&.lt.lt.gt.gt. the &.lt..and x for which (x) 7. Another method of analysing our idea of continuity is the following. (x) is a continuous function of x in $(a )&. that line A &.lt.
so that f belongs to the upper class (f ) &. It is easy to see that the converse of the true./&. though this functions./&.gt. (a. true for continuous Thus such a function as the function (x) theorem just proved is not whose graph is represented .f&. 77 or f such that &.lt. i.lt.lt.lt.gt. 77. number and the theorem is established./&.gt.) continuous for X XQ) this condition is certainly satisfied if x is near enough to a? Similarly there is a certain range of values. of a section of the real numbers. as near to f as we please and such that the corre values of differ by more than jc. It should 97. and let (/&.gt. Then. and so for all values of f less than f ./&. rj k (f ) ^ x &. satisfy all the conditions imposed upon the classes L. &. It is not obvious.lt. (a.e. CONTINUOUS AND DISCONTINUOUS FUNCTIONS c &. Let us divide the values of L.) &. (f ) either = (/&. (x) &.(#) .lt.lt.180 (f&. be observed that we have proved more than is asserted explicitly in the theorem.gt.lt. all numbers f such that either () = 17 or there a? and f for which (a. And this again sponding contradicts the hypothesis that is continuous for a? = f (a?) &. Hence ( ) = of x for which (f) (x) is = rj. Then ?.f)(j.) = 77.lt.gt.)^rj. a value of a? between Then it is evident that these two classes 17.lt.". we have proved in fact that f is the least value &. or indeed generally true.gt.lt. which contradicts the condition of continuity for x = f &. is .gt. &. .TJ. to the left of a?. 77. all the other values of is a*. &. that there a least among the values of x for which a is function assumes a given value. Let f be the to the section. (f f &. (f ) = 77 say. Next suppose greater than f between f and find a . if f is any number we can find a number In either case we can ". between a* and xl into two classes R as follows (1) : in the class L we put all a? when x = f and (2) for all values of values f of # such that between # and f .lt. . ) &.gt. for which &.lt.lt. (a.lt. R and so constitute number corresponding : First suppose &.gt.) a? &. 77 in the class R we put &. [V But since (a? ) is numerically less than 77 (# ).
say the first.r) is discontinuous". prove 101. as x varies from . i.gt. i. It is clear that. of the function &. 31. which is contrary to our Thus assumption.lt. that if (#) takes every value between (# ) and (^) then it must be a continuous t.gt.&. however. C but not the points ^4. if &. is shown in Fig. and capable of precise statement in mathematical terms. (#) assumes once and once only every value between (0) = and 0(1) = 1 . The function through values greater than will be continuous for x= (x) tends to the if if and only either of these equations is untrue.lt. then the two notions of continuity are really equivalent.100102] by CONTINUOUS AND DISCONTINUOUS FUNCTIONS 181 assumes at least once every value between (# ) and Indeed it is not even true that $ (#) must jet (#) is discontinuous.(#) be any value of x between X and Q tends to the limit 0(0) . be defined as follows from #=0 to #=1. As^ as Similarly (1 through values x*~j. 31 obviously ( x i) let (#) = !#. than . ( 98 For less let 95). (V) and if x=l let (#) = !. 32. steadily increases or decreases. but 0f.(x) But In this and the following paragraphs we shall state and some general theorems concerning continuous functions. B. for^=0 and Fig. limit + 0). . the curves which usually occur in elementary mathematics are composed of a finite number of pieces along which y always varies in the same direction. it includes the points 0. (x) must be continuous.lt. 32. as x varies from to 1. It is show that if ?/ = (#) always varies easy^to in the same direction. The net result of this and the last section is consequently to show that our commqrLsense notion of what we mean by continuity is substantially accurate.lt.e. As a matter of fact. If 57 = let = if 1 4&. be continuous when it assumes each value once and once Thus let (x) only.e.r to tfj. Fig. then it is evident that never assumes any value which lies between (0) and (). function in the sense of *i. The graph 0&.t&.o? 0. Fig.
M : M tf&. referring so that &.9 we please. . b). . is continuous for a? = f. and that Then we can determine a positive number e &. and then &. (a?) is continuous for x = f. &.gt.lt.(a?) such that &. &. 102.lt.182 CONTINUOUS AND DISCONTINUOUS FUNCTIONS [V THEOREM!. Let which we shall only assume at (a?) about defined for every value of x in an interval (a.gt. i The values assumed by (x) for values of a. f+ e). or assumes. If (f) is not zero. (x) is continuous throughout (a.i*(f)&. then it is (a.gt. 4.lt. theorem THEOREM 2. for all values of x in In this case (a?) (x} is bounded above. If&.gt.lt. question.lt. it would be positive for all values of x sufficiently near to f which &./&. b). contradicts the hypotheses of the theorem.gt.lt..lt.lt. positive. . as we aggregate S to which we can apply the arguments them in 81 to the aggregate of values of a function of n. then This is an obvious corollary of Theorem 1. to negative values of (a?). The range it is us consider a function present that of values of a continuous function. There is plainly a corresponding (a?) is positive.o. &.gt.gt. for values of x as near to f as we please./&. in (.lt. and &. Suppose that &. b) form an of 80. (a. variable x.() in the fundamental inequality of p. b). applied to functions of a continuous THEOREM bounded in 1.gt. we say that value of $ (x) exceeds M. (/&.() is positive. 175. . for example.lt.gt.gt. must be positive or negative and if it were. //* ^&.gt. .lt. it . we can choose that throughout (f e. For./&./&.(x). both positive and negative values.f&.e). but any no possesses an upper bound is exceeded by at least one value of number less than bounded as lower bound Similarly we define bounded below (f&. applied If there is a number K such that $ (x) ^ K. taking = e so &.) vanishes for values of x as near to f 0. (f ) ts positive throughout the interval (f e.
lt. divide the points f of the interval (a.lt. f) throughout which &.lt.lt./&. Similarly it may be shown that there is a point at which (x) m.f&.j&. is not continuous. (a.f ()&. But ft + 77 belongs to .lt. * If = 6 we must replace this interval by (8rj. 183 We (a. R.gt.lt. given any positive number & however small. f).gt. venient to postpone these for a moment f. (x) is continuous for x = /3.72 and so bounded in (a. $ (x) 8 or 1/{M . for M &. &. we can.lt. M which For. it is in and therefore bounded in the whole interval (a.gt.j&. Now ^fy\^ ^U^+cT It follows from is what precedes that L certainly what .lt. determine between &. continuous for x = a. /3 ?. *09)*&.lt.gt.102] CONTINUOUS AND DISCONTINUOUS FUNCTIONS can certainly determine an interval &. by Theorem 1. There must therefore be one point at which the denominator vanishes: at this point (f&. $ + 77)./&.(x)} we can &. then (x) assumes the values and m at least once each in the interval. and in J? if this (j&.lt.lt.lt. If (x) is continuous throughout (a. putting f in L if (f) is bounded in (a.gt. j8). given any positive number S. extending to (/&.gt. b).(#) bounded in (/3 is bounded ?. f).lt. But M&.p (x) lies 8. and therefore.lt./&.lt. f See 104.&. and let ft be the number corresponding to the Since section whose lower and upper classes are L and R. in view of the great importance of the theorem.gt.(x)= M.): belongs to L. And so (x) is bounded in the &. however be con throughout and p + rj by /3. The proof may just given is somewhat subtle and indirect.lt.gt.gt.we propose to prove is that R does not.gt.gt. and it be well. the right from a.. determine an interval (j3 ij.gt. and so I/{M is &.#) +& 77 Thus $ (x) Therefore is &. in which For since (x) is (#) is bounded. however small B may be. exists: (/&. and and m are its upper and lower bounds./&.(x) is a continuous function. /3 4. &. . 1).j&. l/S. we can. whole interval (a. ft + 77). This contradiction (V) is rco shows that R does not exist. and obviously (#) is bounded in }his interval. (f&. Hence 1/{M  (x)} not bounded.(#)! continuous at any point at which its denominator does not vanish (Ex.gt. &.lt.gt.77). It will to indicate alternative lines of~~proof. &. Now /3 (a.&.lt. find a value of x &.gt.gt. xxxvu. Suppose that R does exist./&.gt. M is THEOREM 2. &. jS the argument which fo lows. b) into two classes L. b)./&. (a) 8 an interval and (a) + &. j3 + rj) * throughout which not the case.
gt. and = 0.#) upper nor a lower bound in any interval. (3) (a. b\ . Let p be the greater of (a. 1) its is discontinuous for is all integral 1. and m (a.gt. and (a.].gt. and the upper bound 41. x\. and 0(o7)=0 when # = 0. +1).lt. b) of the properties of the functions M &.p &. b) to M (a. +1). b). b) = M(a. b). Let x. c) greater of M (a./&.gt. c) or in (c. it is &.( [v = If $(#) = !/# except when # = 0. and let B be any positive number. the lesser of m and m (c.lt. the interval ( 1. . ^ Then (x) throughout (a./&. b) for M. when #=0 or equal to never assumes a value equal to 5.lt. Let and m (x) be any function bounded throughout (a.lt. and (x) = q when x is a rational Then (x) has the lower bound 0. then has &. but no upper bound.(#) Let &. The oscillation of a function in an interval. Examples XXXVIII. b). c) and (c.(x) = Q when x = 0.f&. are as follows. as e. (2) (a.b)m(a. m. The simplest (a.lt. b). then $ (#) has neither an But if (.v) fraction pjq.gt. &. c) b). its never equal to upper bound.fi (x) = when x is irrational.lt.184 CONTINUOUS AND DISCONTINUOUS FUNCTIONS 1. &. (a. Let &. and each of these values +8} the lower bound is is assumed by $ (x) an infinity of times. b). b) is an increasing.gt.lt. an increasing function of b.f&.? 103.gt. b). b). In any interval ( &. in the interval (1.(x) 2. (1) If a = c ^ b then and M(c.lt. Then 8. then 0. interval (a. pendence of M and m on a (a. We shall now use the notation M M m (a. and therefore somewhere in (a.f&. &. /t somewhere in (a. b).gt.lt. and neither an upper nor a lower bound in any interval which includes #=0 in its interior. but Thus &. first M m (a. and its upper and lower bounds. b) b)^0 (a. b) and B (#) The M M &. (#) = # . values of It is In the interval lower bound and its upper bound 1.r = 0. c) and (c. (x} = sin (I/a:) 57 (x) is 1 discontinuous for except when # = 0. (a. in any = ( l) p when x=p/q. b). b}. two theorems are almost immediate consequences of our definitions. m (a.j&. 6).lt.lt. &. &. and we shall write b). c) + (c.g. b) is equal to the (a. but no upper bound. This number lower bounds of &. except when # = 0.lt. the difference between the upper and we shall call the oscillation of (x) in (a. and therefore throughout (a. 4. If $ (x} = \lx has the lower bound 3.lt. 0(a. when . (x) in (a.lt./&.lt. b).[. b). This function (0. and in order to exhibit explicitly the de b.gt. /JL &. &. b) a decreasing.lt.j)(./&.
m^ . be proved by the method of repeated bisection (a?) in an interval PQ. c) and in (c. The most straight 104.2 numbers correspond to the same one of the intervals (a. (3) If so = a then J/(a. We can therefore construct a section of the numbers &.m. a + M(b b}=M. by /3. = M .n^ . Sets of intervals on a line. ) putting the number corresponding to the section. values as near as (x) is continuous. arid (2) is /I/a is an obvious Suppose now that and and M (c. ) M exceeds M. then it is possible to find a half P Q l in which the upper bound of 71. and it is (x) is M. Let be any number of and never the interval (a. 2 2 . forward proof of Theorem 2 of 102 is as follows. M . 6?.lt. and case the argument may be completed as before. 6) or not. Alternative proofs of Theorem 2 of 102. And if 3 = 6 then 3/(a.p (x) at this point is M. Pv Qi^ P^Qz. particularly. similarly. and PQ is divided into two equal parts. In /3 in L or in It according as M(a.lt. The proposition concerning is m may be proved corollary. ) = M. as in of 71. (c.lt. we construct a (x) is also M. (1) of 103.lt. These intervals.. 6). (c) is not Hence 2 ^ m 2 whether these greater than AF2 nor less than m. we havo for all positive values of /. and that m^ is the greater and 2 the less of the less and the greater of 2 M M (a. Jf.m ^ M + M . ( . Proceeding as in in each of which the upper bound sequence of intervals PQ. since c belongs to both intervals. (c.. . and so #0317. 105. converge to a point T. b) 185 Hence M = //. &. We shall now proceed to prove cerning the oscillation of a function which are of a somewhat abstract character but of very great importance. x 2 2 . as we shall see later. used in If M is the upper bound of } easily proved that the value of &.lt. The HeineBorel some theorems con Theorem. In either /3 T]. and so. ^^J/". b). These theorems depend upon a general theorem concerning intervals on a line.102105] CONTINUOUS AND DISCONTINUOUS FUNCTIONS (a.m = J/ + M. also The theorem may 71. The function increases steadily with (a. and (a. in the theory of integration. c) c) m m (a. 6) c) + 1 1 . Hence (#) assumes. b. J/(#. since must be equal to M. Thus (1) proved. . b).r as near as we please to &. Then.lt. . Let be /3 &.) &.lt. for values of we please to J/. If a M or &. (a. b) But and (3) follows.
1). A pair of intervals such as (0.gt. ) and with 1 we associate (1f. overlap. at least one interval (ii) of I . overlap with one another. The HeineBorel Theorem. ^) and (^. interval We make no restriction as to the nature of these intervals. end point. Suppose that we are given an and b). of intervals I each of whose members is Suppose further that I possesses the following a. /3). and any number of number. point. they may be finite or infinite in may or may not overlap*. (iii) We take the rational points p\q of the interval (0. we We an infinity of rational points. 1). interval (a.set properties (i) : every point of (a. of is a the lefthand and b the righthand end Then the set it is I which form a possible to choose a finite number of intervals from set of intervals possessing the properties (i) and * (ii). 1). at least one interval of I.overlapping intervals which just cover up the (ii) line. 1) is thus formed define a finite set of divided into n equal parts then the n intervals non. except that with we associate (0. every point interval + e). We regard as 0/1 and 1 as 1/1 : in these two cases obtain thus an since there are reject the part of the interval which lies outside (0. b). and in general we reject any part of any interval which projects outside the interval (0. We thus define an infinite set of intervals. The word overlap is used in its obvious sense: two intervals overlap if they (^. other than a and b. 1).f If the interval (0. It is worth while in passing to give a few examples which we shall have occasion to return later. and associate with the where e is a positive number less than 1. (i) sets of intervals to /&. 1). they them may bo included in others. b). which plainly overlap with one another. infinite set of intervals.186 CONTINUOUS AND DISCONTINUOUS FUNCTIONS [V line. 1) may be said to abut. common which . and associate with o the interval where e is positive and less than 1. and it is obvious that many of them We take ( e. of the interval (0. other than p/q in the interval y associated with plq. t That is to say in and not at an end of. ) and are not end points of either. 1) have points in Thus {0. Suppose that we are given a set of intervals in a straight that is to say an aggregate each of whose members is an (a. included in (a. lies inside ]".
say (a./. a3 . . a n +i) associated with a n in our construction of the sequence a lt a 2 a 3 . . . If a n never coincides with lies to b.105] CONTINUOUS AND DISCONTINUOUS FUNCTIONS 187 We know that a is interval of /. Similarly a2 lies inside an interval a3 ) of /. to Z.. say (f f some term a of some sequence.. It is plain that this argument may be repeated of steps in b. b Fig. The L certainly exists. Let us suppose now that the process just indicated. right . Either (ii) sequence or with regard to any point f between f lies to the left of some point a n of some (i) We divide the points f into two it does not. for we have obtained a finite set nothing of intervals.. and f belongs to L... (a/. so that we obtain all Then we can prove . But then we can take (f f as the interval (a.or (ii) is true. is performed in all possible ways. unless after a finite number a n coincides with number of steps then further to prove. and the classes form a section of the real numbers between a arid. and possessing the If a n does coincide with b after a finite is there properties required. . a aj ax a2 a2 $ o3  Q I. We shall now prove that R does not exist. 33.) . but this limiting position may. starting a.. say (a/. and all points to the left of There are therefore points of L to the lie to the left of a n+1 It is of f and this contradicts the definition of R. selected from the intervals of /. lie anywhere in (a.). The point f lies inside an interval which corresponds a number f If R . ". R exists then L lies entirely to the left of R. so that every point f belongs to L. . f". must (since each the right of predecessor) tend to a limiting position. . There are two a and classes class 6.2 that there must be at least one such sequence which arrives at b from . b. b). a ) x belong to L. the lefthand end point of at le ast one We know also that a^ lies inside at least 2 ). therefore impossible that R should exist. a a ).. after a finite number of steps. . 3 possible sequences of the type a lt a.. since all points of the interval (a. then the points its a l5 2. and so lies to the left of of /. possibilities L and R according as to whether (i) . so far as we can tell. one interval of /. definitely. ".
the supposition that every point of (0. .lt.. (2*. is The reader may be tempted elaborate. if small enough. points of (0. say (&. 3/w. The sum of any finite nuBrber of intervals of / can therefore not be . 1) lay inside an interval of / (with the obvious reservation as to the end points).. 1)... (ii) do not lie inside any interval of I satisfied. If every point of (0. and so we obtain a sequence in which the term after the nth coincides with b.. 1) lies inside an interval of / leads to a contradiction. for which &. We shall now apply the HeineBorel Theorem to the a proof of two important theorems concerning the oscillation of continuous function. 186 in the light of this theorem. . that there lie in any interval of /. 2e. and 3 ql intervals. b). V ^r greater than 2e times that of the scries e which e i  ~L p +P+33+43+ 113 to will be shown to be convergent in Ch. . 1) which do not by using the theorem. such that an But then we may take the interval (a n a n+l ) corre . Thus the theorem is proved. of total length 2e(^l)/^ associated with any other value of . The sot of intervals 2 f ). 1  possesses the properties re In this case we can provo. It is instructive to consider the examples of p.. Now b is the righthand end point of an interval of /. (l2 f e. associated with the points quired. sponding to a n to be (6 l5 6). 4e). Now there are two intervals^ total length 2f.?=!. 108. .. 3e.. if e is small enough. f &. 2/n. 3*).gt. follows at once from the fact that the sum of all these intervals is less than 1. not in any interval of /. (iii) e. then we could find a finite number of intervals than of possessing the same property and having therefore a total length greater 1. think that this proof is needlessly and that the existence of points of the interval. VIII.. . / \ q. and can only be proved rigorously by some such use of the HeineBorel Theorem as is made in the text. arc.b l . (i) Here the conditions of the theorem are not . Here the conditions of the theorem are (0. and ^ belongs to L.. and therefore a finite set of intervals having the properties required. . Hence it follows that.188 CONTINUOUS AND DISCONTINUOUS FUNCTIONS [v Thus every point f belongs to L. (e. But the theorem to which he would be appealing is (when the set of intervals is infinite) far from obvious.. satisfied the points 1/n. Hence there is a member a n of a sequence a l} a 2 a s .
. and so an the same property. f + e) such in this interval is less than 8.lt. less than an assigned positive number B. any particular value of e.105. tli If c/&. and which possesses the same property as / intervals The points as in Fig.lt. (x} is continuous throughout the interval  en we can (#!. having call the ^intervals of a. Thus Theorem I is proved. then the oscillation of. &. b) into ajfcnite number of subintervals which the oscillation of (x) is &) ?n ea c ^ f &. the set / of intervals formed by taking all the It is plain that this set satisfies b).gt. included in an interval of / of &. 106] CONTINUOUS AND DISCONTINUOUS FUNCTIONS I. . the conditions of the HeineBorel to the interval is Theorem . Let f be any number between a and 6. b). determine a set / which is formed by a finite of /. b) is divided in any manner rj into subintervals of length less than 77. Given any positive number B.lt. and in each of which the oscillation (a) is less than B. and a and b are end* points of at least one such interval. divide (a. then value. 189 THEOREM (a. we can determine an interval (f e. if the interval (a. we can find a such that. But which compose the their end divide set 1 will in general overlap obviously up a b (a. Since $ (#) is con tinuous for x=%.lt. tfj). %. and we can define in a similar ^intervals of 6. It is that the oscillation of (a?) &. 6) into a finite set of ineach of which is tervals I". II.gt. (a.** . These we manner the Consider now Sintervals of all points of (a. If f =a then we can determine an interval a f e). tf^X (#n&. We can therefore of intervals number itself. every point interior interior to at least one interval of /.gt. indeed obvious that there are an infinity of such intervals corre sponding to every f and every B. What values of e We shall call the intervals thus associated with f the ^intervals of (a.^(x) THEOREM number in each of them will be less than B.gt. 34./&. infinity of such intervals.j&. for if the condition is satisfied for it is satisfied a fortiori for any smaller are admissible will naturally depend upon f we have at present no reason for supposing that a value of e admissible for one value of f will be admissible for another.
in one of the parts of length less than ?. Continuous functions of several variables. know what is meant by a continuous function of and we accordingly give the following definition. sequel. t and is therefore This theorem is of fundamental importance in the theory of It is impossible.).. It is a straightforward generalisation of the last form of the de finition of 98. in the . Their application to such functions. then any j in each of and construct. cf&. of The notions continuity and discontinuity may be extended to functions of several independent variables (Ch.gt.190 CONTINUOUS AND DISCONTINUOUS FUNCTIONS Take B l &. when ^ x % ^ e (B) and ^ y TJ ^ e (B). b) into parts each of length less than rj. greatest oscillation of $ (x) in a subinterval j less than 2S 1? and therefore than B.gt. [V 5. however.lt. and in particular at the point &. intervals (x) is less than B lt be the length of the least of these subintervals j./&. how ever small. y) at less (f.lt. as in Theorem which the oscillation of I. without the use of definite integrals (Ch. we can choose e (B) so that . If now we divide (a.lt. whose sides are parallel to the axes of coordinates and of length 2e (B).gt. 107. y) is defined at points of the square in question. II.gt.f&.lt. (f&. ??) by any point inside than B. The function (x. require to two variables.* it or on its boundary all This definition of course presupposes that (x. the oscillation of (a?). this or some similar theorem.f&. Hence. 31 et seg. . It would be impossible for us to discuss these questions in any detail here but we shall. VII). cannot exceed twice the &. y) of the two variables x and y is said to be continuous for x = f y = TJ if] given any positive number B. * The reader should draw a figure to illustrate the definition.gt. that is to say if we can draw a square. (x. whose centre is the point (. to prove that a function continuous throughout an interval necessarily possesses an integral over that interval. a finite set of sub Let 77 such part must lie entirely within at most two successive subintervals j.lt. in virtue of (3) of 103. raises questions much more complicated and difficult than those which we have considered in this chapter. vj). &. and which is such \  \ that the value of differs from &.
gt. i.. for example. II we were content to a function y of x. carefully that to assert the continuity of y} with respect to the two variables x and y is to assert much more than its continuity with respect to each variable considered separately. (x. Thus.f) The reader should observe (x. the products.. y} = ^ T .. are continuous except for pairs of values of x and y connected by all values of the variables . met with 108. that (or a:).(#.r.gt. In the same way it may be shown that tends to the limit zero as x*0. and (x. ^ which may have any value between and 1.f&. and it . y) is continuous for #=f.. y} is continuous with respect to x and y then it is certainly ?/ continuous with respect to x (or y} when any fixed value is assigned to But the converse is by no means true. We have already. y=r] if when x*~%...lt.. taKeTh is for granted. But &. II.(. but if x and y tend to zero along the straight y = ax.... ^ y) is not a continuous function is x andy x=0... x... CONTINUOUS AND DISCONTINUOUS FUNCTIONS : &.... in Oh. y = Q.... Suppose. plain that if &.. polynomial in two variables is continuous for A and the ordinary functions of x and y which occur in everyday analysis are generally continuous.... and in general the quotients of continuous functions of two variables are them selves continuous... (i). y} a continuous function of and for its value when x in particular continuous for x is zero. ?/)=0 (j) when is either x ory is zero. (x. zero or not. the idea of an implicit function. Km $ 1 (x.. but it in any manner./&.. explained and can only be explained by the help of inequalities like those which occur in our original statement. (x. or several such functions.106108] (f.gt. y~^y y) v} This statement is apparently simpler. Then . *&.. it is But . It is &.lt.gt. = $ of (Xj y} is a continuous function of for ?/.e. 101 Another method of stating the definition is this 77)..lt. //) = r ^ &.f&. (x.gt. then $ (x. if x arid y are connected by the relation i jl jgfy*o . special relations. Its value line when = 0. We are now in a position to consider whether the assumption we made then was justified. It is easy to prove that the sums... then y is an implicit function of far from obvious that such an equation as this does rejdlyjlefine In Ch. y has any fixed value. \) when if neither x nor y is zero. y = zero . contains phrases the precise meaning of which has not yet been (/&.lt.lt. Implicit functions.
$ (. It and so (. in virtue of A and negative at (iii) and 100. Suppose that x+a. y] is positive at Q and therefore. or near (a. y} being con take Q and R Q Q tinuous at Q and at R. remains to prove that This is most simply effected (x) is continuous. it follows from (iii) that f(x.(i) f(x. (#) is y= ^&. If we as in the figure. when (x M* ) N^Vu* substituted in the t satisjies it identically for all values of x in the neighbour hood of a. (#). We Suppose that it is possible 107. the line. (x} which. a. and similarly A=6. vanishes once and only once at a point P on The same construction gives us a unique point at which on each ordinate between RQ and R Q It is obvious.(/.y) is.7. in the stricter sense of then (1) there is a unique function t$H . we have / = 0.} the values of the sequence. f(x. RQ parallel OY and f(x.192 CONTINUOUS AND DISCONTINUOUS FUNCTIONS shall find the following terminology useful. moreover. It is obvious that of (a. This being so. y) = 0.". y} is positive at Q and negative at R. b). points such as It P gives us the graph of the required function y= (f&.^^&. 6). Fig. 6). points of . id y in the neighbourhood of (ii) (iii) /(a. equation f (x.lt. [V We to surround a point (a. square being replaced by an interval on a THEOREM. with a square throughout which shall call such a square a neighbourhood and say that the condition in question is satisfied in the neighbour hood of (a. 6). is a continuous function of x a. a through is and /(#.?/) a con Hence X = 6. If . . by using the idea of the limits of indetermination of (x) as x*~a ( 96). as in a certain condition is satisfied. meaning by this simply that it is possible to find soiw square throughout which the condition is satisfied. The aggregate of /&. and since tinuous function of A. .gt. (a.* we can find a sequence of values of x such that {#. a steadily increasing function of y. Thus for x=a. &)=0. and let X and A be the limits of indetermination of 0 (x) as x*~a. X) and (a.lt. similar language may be used when we are dealing with a single variable.r) is continuous (#) is tends to the limit b as x**a t evident that we can show in . we can draw so that lines is QQ and to 111? parallel to OX. A) lie on QR. that the same construction can be carried out to the left of RQ.y} (a. 6). for all values of x in the neighbourhood of 95. In the figure the square represents a neighbourhood of the conditions (i) and (iii) are through and P the point (a. and f(x. 6). (2) continuous for all values of x in the neighbourhood of a. y} is positive at all QQ and negative at all points of RR . Moreover. 6) out which satisfied. It is evident that the points (a.and y. 35.gt.r)*X when (#. In par ticular /(#.
6 example that y*=x. Moreover when x. viz. let us consider the equation (1). Suppose in particular that form F(y] . in question is in this case *. and each of which is defined only for positive values of x. sign opposite to that of ?//. of x # ^ +^ . the Hence condition (iii) (with decreasing for 5 has. follow if The same conclusion would The function the equation were Vi y*xyy* = 0. y.lt. in which the sign of the changed.3 . does not satisfy the condition of vanishing with x. for y is not a steadily increasing function of y in any = interval which includes y it decreases when y is negative and increases when y is positive. supposed that the hypotheses of the theorem were satisfied *m that is to say throughout a certain square (a. a = 0. And in this case the conclusion of the theorem does not notjhave been : y =x defines two functions of x. We then obtain the following theorem. then a unique continuous function y = which is equal to b when x = a &. is 193 continuous for any value of x in the neighbourhood of to It is clear that the truth of the theorem would not be affected 3 if we were It is change increasing to decreasing in condition (iii). pjiejaoint in the proof which the reader should be careful to ob is positive.lt. 6) e. / 1. = # ^ + * ^e^y^fj + a\ that is The conclusion holds e in the neighbourhood e! &.**} yf{i+*V(i +*+*% where the square root square root is There serve. As an example. increasing ) is satisfied. and I 109. We the neighbourhood of * . we had supposed that y^ = x then the conditions of the theorem would 2 satisfied. and F(b} = a.108109] exactly the CONTINUOUS AND DISCONTINUOUS FUNCTIONS same way that $(#) a.fr(x) (or decreasing}. The inverse function x=%y. There of the hypotheses. for 2 t The reader should consider the more general equations H.(. is The second root. and satisfies the equation F(y} = x identically in the neighbourhood of x=a. in the in the stricter sense of neighbourhood of y = b. to say throughout a certain interval is nothing to show that the ^ of the conclusion is the indeed this is generally untrue.. /(a?. y) is of the If F(y] there is is a function of y. It follows that there is satisfies the equation (1) one and only one continuous function y which identically and vanishes with x. evident that the conditions (i) and (ii) are satisfied. y ^Jx and y= *Jx both of which vanish when #=0. 6=0. Inverse Functions. . hold. defined is The function thus Suppose If for called the inverse function of F(y}. continuous and steadily increasing 95. taking a = 0.x. so that the equation has sometimes two solutions and sometimes none. is Then all the conditions of the theorem are satisfied. = 0. and y are sufficiently small.
&. . 10).lt. and would we could find a value of y such that y than The not be the upper bound of the class considered. any axn n~l . &my = x has just one solution which vanishes with #. So far particular we have considered only what happens in the neighbourhood of a value of x. b) and it is plain that the finite set of functions fa (#).. given by the other values of arc sin x (cf. already considered in Ex. where F(y] is continuous and if x A to B as x increases from a to 6. increasing (or decreasing) throughout an interval (a.y &. where c x is of the first order of smallness If P(x} = + . Show that. in virtue of the HeineBorel Theorem. the value of arc sin x which vanishes with x.. Hence F(rj) = = has therefore a unique solution # = = an(^ sa J equation F(y} increases steadily and continuously with . plainly 108. Show that in general where a = a/A.gt. : = F(y]. then as x increases 2. We thus obtain the theorem increases steadily and It is worth while to show how this theorem can be obtained directly with &.lt.lt. .194 in the CONTINUOUS AND DISCONTINUOUS FUNCTIONS same way.. &.gt. which do not satisfy this condition. If F(rj) were less and F(y}&. xv. . ft = (bA aB)/A 2 . B. out the help of the more difficult theorem of and consider the class of values of y such that (i) a &.+ = axn (1 + t x \ when x is large. b). Another interesting example is [V given by the equation y*yx = 0. then ax* + bx 1 + .. viz. Similarly the equation 7.lt. Given any point and a unique and con From the set 1 of intervals i we can.lt. Ex. 77 &. and e z is of the first order of smallness when c is large. corresponding to the subset of intervals i thus selected.lt. and plainly F(rfi&. which proves the theorem. . define together a unique inverse function $ (x) continuous throughout (a. V. +bx where X is P(x) has ultimately the sign of a and so has P(x+\}P(x\ constant. pick out a finite subset covering up the whole interval (a. if neither a nor b n is zero.lt. Let us suppose now that F(y] is positive and steadily 6). and a is not zero. Indicate any exceptional cases. then there strictly from is a unique inverse function y = $(x} which is continuous and increases steadily and strictly from a to b as x increases from A to B. xiv. 3.%. &. of (a. Suppose that A 6 and (ii) F(y] This class has an upper bound rj. There are of course an infinity of solutions. b\ we can determine an interval i including tinuous inverse function fa (x} defined throughout i.lt.()&.. 5 77 MISCELLANEOUS EXAMPLES ON CHAPTER 1. .+.rj T) .
Tangents are drawn to a circular arc at its middle point and its A is the area of the triangle formed by the chord of the arc and the two tangents at the extremities. touching that as in T. and 0(*) = and discontinuous for is irrational.cos x} &.#&.. e x is of the first order of smallness 7.lt.has the limit c) ft and prove that lim x y(ax* + 2bx + c) . Prove that is (a?) =# sin (sin a radius a?) .6 2 )/2a.tan #)*0 as x*~%n.ax . the function may be represented by the formula". order of smallness Show X* lim *Jx{J(x+a}.*&./#) (1 +ex where ).lt. and TN is drawn NAjAP^l P moves per up to A. when x = \. to \x and to 1 when and 1 once and once only as x increases x=Q. 16.Jx} = %a.lt.15. [Use the formula 6. 17. it pendicular to OA. Show that A/A ^4 as the length of the arc tends to zero.(#) = 1/0 when continuous for all irrational when x a?=p/q.lt. small . 14.) + (y/x*) (l+ x \ when x is large. three tangents.A when  J&.CONTINUOUS AND DISCONTINUOUS FUNCTIONS 4. produced beyond the circle. x=$.0. oo . Evaluate lim x 9. J(a/&.gt.cos (1 . and &. then all rational values of x. ft} (ac . lx lim (cos 2n + 1 is irx). to from assumes every value between to 1.lt. small .lt. = l. Show that V(#+a)=V# + is large.sin 2 x &. 195 Express (ax 2 in the form cx is a+ of the that first + bx + c)/(Ax2 + Bx+C) (ft fa.lt. of the sixth order of smallness x&.ax .= .QO 8. when x zero as x*. 32 may be repre x+ [x] [!#]. but is discontinuous for when #=0. Show to the circle._l: the function (2) x. Show that the function 0(#) which to \ when . as a?^0? For what values of a does {a + sin (!/#)}/# tend to (1) [To GO if a&. find the limit of (#)/#* as is x+Q.lt.lt. and #=1.1. if oscillates ISaJSL] If . (#) = 1 . Show also that equal to f. when x 12. of the fourth order of smallness when x 11. and A the area of that formed by the extremities .r ()&. 13. Prove that Prove that is (sec x . is 10. and find the limit of 6 (rf/x as From a point P on 0^ a tangent PT is drawn of a circle.gt. Find values of a and ft such that N/(cu?2 f 2bx + . Show that the function whose graph sented by either of the formulae 1 is drawn in Fig.^ (x) is &. where 5.lt.l. to co if &. 132 .
g.gt.A) limits tend to limits when A0 by positive values only. &.(x)}/h h tends to zero.*)}/*. {&. NX RPQ PT is v/r. as in Fig. If these limits exist but are not equal. which is not perpendicular to the axis of x.lt./&.) RQ of P .f&. (x . Thus the assumption that the curve which {&. not parallel to OF.MP)/MN = RQ/PR &.gt. and that the two (f&. (x} has an angle at the particular point considered.lt. to the limit is unaffected by the sign of the graph of (x) has a tangent at P.j&. (x + h) . for the particular value of x corresponding tends to a limit when to P.198 DERIVATIVES AND INTEGRALS P two along the curve from either side. This of course implies that both of {0 (x + A) $ (.lt. let us suppose that the curve has (like the circle or a tangent at every point of its length.gt. and so r lim  * tan (1). or at any rate every ellipse) Now .A)  (*)}/( .gt.gt.$ (x)}/h .(NQ .gt. and that the convergence h. 37. We have now to distinguish cases. The reader should be careful to note that in all these equa tions all lengths are regarded as affected with the proper sign.lt.lt. &. the property that (x + h) is &./&.lt. is negative in the figure when Q lies to the left so that (e. that in which ty is not equal to JTT.gt. a general case and an exceptional one.f&.r. {&. so that In this case tends to the limit and RQ/PR^tanRPQ tends to the limit tan Now . implies that (x) has.gt. then the curve y = are equal.f&. ON The general case is M Fig. 36.
gt. and which may be calculated or derived from the original function We shall &. 37).e. every function a differential coefficient for every value of x for which it is continuous.lt. in we proceed to consider the which = TT. &.(# + h) = cf&. is generally known as differentiation./&.gt. . and is true. &. For it is evident that (x + h) (x)}/h cannot tend to a limit unless that &. tan ^ : tan ^r is a function of x. Hm &. firmly established for historical reasons is see Before above.lt. 111] DERIVATIVES AND INTEGRALS 199 portion of its length which corresponds to a certain range of Further let us suppose this tangent never per variation of x. and it is this which is the property denoted by continuity.f&. and the operation of calculating This : (x) (x) terminology 115.lt. &. (1) The existence of a derived function (x) for all values of x in the interval a ^x b implies continuous at every point of this interval.$&. (x) is {&.lt.lt. call this function the derivative or derived function of &. we i/r shall illustrate special case mentioned our definition by some general remarks and particular illustrations. i.* The answer sufficient No: it is to is obviously consider the ^  \ curve formed meeting to * by two straight lines form an angle (Fig.f) (x) .f&. Then an equation such as (1) holds for all values of x which fall To each such value of x corresponds a value of inside this range. (x) is the differ ential coefficient of &. which is defined for all values of x in the range of values under consideration.gt. the axis of x in the case of a circle this would of pendicular to : course restrict us to considering an arc less than a semicircle. (x).gt. 111.gt.f&.110. Some general remarks. It is natural to ask whether the converse (2) whether every continuous curve has a definite tangent at every point. Fj g 37 We in which the curve leave out of account the exceptional case (which we have is supposed to have a tangent perpendicular to still to examine) OX: apart from this possibility the two forms of the question stated above are equivalent.(x)./&.lt.lt.lt. (x).lt.lt.lt. and we shall denote it by for the derived function of &. Another name from &.gt.gt.
xxxvu. &. the graph (Fig. On the other hand the function defined by the equations function is Draw is is continuous for #=0 (Exs.lt.0 This is of course a case in which a curve might reasonably be said to have difficult. ~ \ ^ It has even been shown that a function of x may be continuous and yet have no derivative for any value of . pp.lt. But lim {$ i &. according and &. by limit exists. and the graph of this function a continuous curve. 14. (x) has or has not The geometry of as this limit does or does not exist. a function denned. or Hobson s Theory of Functions The notion of a derivative or differential coefficient was suggested to us by geometrical considerations. (0)}/A or lim sin (I/A) and no such . definition.gt.f&. the idea of a curves is Another important application moving in a straight line = point on the line is s &. but the proof of this is much more difficult. The notion of velocity is in fact merely a special case of that of the derivative of a function.(#) (h) &. The derivative &.lt. 15). But the following example. for any particular value of x.lt. The reader who is Bromwich s Infinite Series. without any reference to any kind of geometrical representation of (#). 6205. 14. The not denned for #=0.lt. and so is discontinuous for #=0. although a little more shows conclusively that there are cases in which a continuous curve cannot be said to have either one direction or several directions at one of its two directions at a point.lt. in (t)./&. For $ (0) would be.lt.200 DERIVATIVES AND INTEGRALS will see at [VI The reader once that in this case {&.gt.r. p. Variable.gt.p has no derivative for # = 0. interested in the question may be referred to 4901. (x)}/h has the limit tan/3 when h*~0 by positive values and the limit tan a when h by negative values. of a Real (3) pp. is in dynamics. points. Suppose that a particle is such a way that at time t its distance from a fixed Then the velocity of the particle at time t is by definition the limit of as A*0.lt.p a derivative. But there is (x) of nothing geometrical in the notion itself. . 53) of the function x sin (l/#). (x + h) &. by the equation $ (x) may be &. 1 of mathematics in which merely one of many departments derivative finds an application.
(x)=x &. 13). and so necessarily discontinuous for x 0. then (x) = cos # . then $ = . J). Meanwhile the reader will find it instructive to determine (x) when m has some special fractional &. and the equation point of contact) of the normal (the perpendicular to the tangent at the (y is . Examples XXXIX. have assumed that the tangent this special case it is obvious that the tangent and normal are X=XQ We and y=y 6.g. Interpret If (f&.lt. by means (x) x.^o = o. &. Prove this (i) from the formal de finition 3. which are not defined for #= 0.gt. then = (#) .gt.sin [For example. 112] DERIVATIVES AND INTEGRALS 1. xxxvi./.gt.yo) &.(x) = ax + b by . at Write down the equations of the tangent and normal at any point of Show that if # = 2a/wi. If $ (x} (#) is a constant then (#) = 0. 201 &.lt. respectively. Its equation is therefore 5. the parabola # 8 =4ay.0.] The Equations of the tangent and normal to a curve is the line through (# to the curve at the point (# which ) ) tangent y y makes with OJTan angle ^. value 4. this result geometrically. . (e.Ill. If sin #. when k&. then $ (#) = [For (#) = lim The reader should observe that this method cannot be applied to is a rational fraction. We shall show later on ( 118) that the result #&.] &.t&. $ (#) = sin (#+A)  #. is cos a?. as we have no means of expressing (# + A)P/ as a finite series of powers of h.lt.lt. y = a/w 2 then the tangent .r 112. 2. cannot have derivatives for # = 0. we have {&. .lt.gt. In not parallel to the axis of y.lt. Or again the function [#].lt. has no derivative for any = such value of #. We it have seen that if (#) is not continuous for a value of x then cannot possibly have a derivative for that value of x. where tan \^ = (# ).(x). since lim cos (# + ^A) = cos # (the cosine being a continuous function) and lim {(sin A)/A} = 1 (Ex. . # /o is . m where m is a positive integer. if of some special device. and If (f) (ii) geometrical considerations. which is discon tinuous for every integral value of x.lt. where p/y of this example holds for all rational values of m. (#o) is + x . and if (#) = cos x. Thus such functions as \\x or sin (I/a?). (#)}/A = (2 sin JA cos (x + JA)}/A f the limit of which. y=&.
and consider the special part of the curve corresponding to fa (x). .f&. DERIVATIVES AND INTEGRALS [VI its Since [#] is constant between every two integral values of #. (1). of which the most typical are shown in In cases (c) and (d) the function is two valued on one side Fig. all discontinuities of$(x) are But the converse is not true. 111. of P and not defined on the other. when we are dealing with the very simplest and most obvious kinds of functions.lt. the upper case.202 Example.lt. has the value zero. as we may easily see &. which we may represent by [#] . which occur on one side of P or the other. whenever it exists. We saw also in Ex. tive of [#]. as was pointed out in &. has exactly the same properties. In such cases we may consider the two sets of values of (x).f&. N /T.gt.gt./&. such as polynomials or rational or trigonometrical functions. as in such cases as those con for certain special values of x. 7 that the types of discontinuity which occur most commonly. as defining distinct functions fa(x) and fa(x). also discontinuities of (x). if we return to the geometrical point of view of 110 which the graph of Y.gt.lt.lt. 38. Thus the deriva derivative. xxxvu. is a function equal to zero for all values of x save integral values and undefined for It integral values. (x) * oo . hitherto left aside. 38. is interesting to note that the function 1 sm trx . This case may be subdivided (x) has a tangent parallel to into a number of cases. in &. \ & ^ In fact. there no derivative these cases. Fig. are associated with a relation of the type (j) (x) * + 00 or &. In all is sidered above.
/&. 113] DERIVATIVES AND INTEGRALS will easily 203 The reader convince himself that in (a) as h 90. xxxv. (x) has a derivative For A faWlim ^ ". out the theorems which follow (1) Some // (/&.gt. (&.gt. (x) =f(x)F(x). (d) y* = x.112.lt. then &.gt.(&.gt. and in (6) while in (c) and in (d) (a. ~ r * . (a? + A) &. *". though of course in (c) only positive and in (d) only negative values of h can be considered. (x) has a derivative f /() (2) derivative //". ^ v \ J ^ f ".lt.lt.! + A) . ajact which by itself would preclude the existence of a derivative. ^ ". 2 (#)}//* ^+ oo .gt. 2 (c) 7/ = #. (x) =/(#) + F (x).lt. \^ \ If (f&. from the general theorems stated in Ex. f = x. (x) = kf (x). then &.&. the special value of x under consideration being x = 0.& O)}//* 0) ./&. +&. then &. We can obtain examples of these four cases by considering the functions defined by the equations (b) ^5 113. y* = x. {.j&. general rules for differentiation.gt.lt.lt.gt./&. Through we assume that the functions and F(x) have derivatives / (X) and F (x) for the values of f(x) x considered. (x) has a We leave it (3) as an exercise to the reader to deduce these results 1.gt. where k is a constant.
Our tion.V^M vv t(\\ 5WJ (5) $ If W= / \ i * lim T n f( . where ^r (y) is continuous and last steadily increasing (or decreasing).lt. (x) has a derivative ^~. and &. (x\ where is the inverse function ofty.gt.gt. in = (/&. then (x) has a derivative . and k/h*f (a. Then we may write y 95. Another interesting special case that in which f(x) = ax + b the theorem then shows that the = derivative of F (ax f b) is aF (ax + b). ~ ^V V x v /%/ (6) 7/ ^&. And =r lim ^ A.gt./(*} is In this theorem we of course suppose that f(x) not equal to zero for the particular value of x under consideration. Me?i (x) has a derivative 7 * H For let Then k+Q as ^0.gt. } .lt. (x).204 DERIVATIVES AND INTEGRALS [VI (4) If $ (x) = f r . theorem requires a few words of preliminary explana Suppose that x = ^r (y). is the inverse (/&. (x) = then &.gt. (7) x = ^r If y = (y).gt.gt. ^ . ( 109) of ty. where &.(#)}.). in the stricter sense of a certain interval of values of y. so that ty (y) has a derivative ty (y) which is not equal to zero. ~ / Then . as we see on is taking F(x) kx or F(x) = I/a?./&. /*e/i (/&.gt./&. (ic) =F {/".lt. : r ^^ x lim j This theorem includes (2) and (4) as special cases. f (x} F (x} This follows at once from (3) 1 f (x) F (x) \ I ** V st\ [F(x)Y and (4).f&.gt. 7*as (#) has a derivative ( . (/&. function &.lt. ^&.
lt. x of the independent variable x . We = have already explained that what differential coefficient. but these depend upon the general notion of a notion which we have en a function of a complex variable countered at present only in a few particular cases. which is many purposes the same thing as .(#) (5) complex. D = (/&.$ (#)}/A is the difference of the values x+ h. The notation of the differential calculus. and the reason was as follows.gt.lt. be called the increments of x and y respectively. The reader will have no difficulty in (x) &.gt. we call a derivative is often called a different &. Not only a . + h} = y + k.lt. 115. So far we have = (#) is a purely real function of x.lt. (x] of the dependent variable y. The denomi nator h of the fraction (0 (sc + h) . and The last function may now be expressed in terms of x by means &. These (#+A). &. know 114. so that This theorem enables us to differentiate any function if we the derivative of the inverse function.lt. i (x)}jh  and making h*~Q. Derivatives of complex functions. &.lt.gt.113115] DERIVATIVES AND INTEGRALS (x 205 For if (/&.lt. and it is for convenient to denote the limit of the fraction. at first sight so peculiar Of course a notation would not have been adopted without some reason.f&. then k * as h^0.gt. and denoted by 8x and Then the fraction is &//&i % ./&. then we define the derivative of y complex as being + ify (x).gt.t&.$&. complex functions.fi(x) {&. (x)}.lt. is the reciprocal of ty of the relation y = $ (x)./&.f&. seeing that Theorems (1) is &. Theorems above retain their validity when (6) and (7) have also analogues for . (x) .gt. different name but a notation is the derivative of the function y often denoted by one or other of the expressions is often used (x) Of these the last is the most usual and convenient the reader must however be careful to remember that dyjdx does not mean a certain number dy divided by another number dx it means : : \ the result of a certain operation x or d/dx applied to y the operation being that of forming the quotient (x + h) {&.gt.lt. similarly the numerator is the difference of the corresponding values differences may dy.gt. &. If y is a supposed that y function (x) + i\jr (x).
or using the notation preceding sections of this chapter. ^x (6) if y is a function of x.yn then SIn particular. (#). and if y=xn y then 3. of 113 may of course at once be translated into : They may be stated as follows (2) ify = tyi. these limits being simply equal to zero.gt..lt. and standing by themselves they would mean nothing in particular dy and (purely not mean lim dx do dy and lim &?. dx and dy The theorems this notation. (x).. then dz dz dy dx dy dx Examples XL. if dy dyldx=nx ~\ n as then dyldx = nz (dzjdx) y=z was proved otherwise in Ex. n~1 . VII. we shall show how it is possible to define the symbols in such a way that they have an independent meaning and that the derivative dyldx is actually their quotient. n . : the form x y.lt. and z a function of y. ( ) v y ien y\y*&. D &. but so long as it puzzles him he will be wise to avoid_it by writing the differential coefficient in (#) symbolical.gt. xxxix. If y=y\ y^s then and if y= y^ . DERIVATIVES AND INTEGRALS hy dy\dx. as we have done in the In Ch. however. [VI But this notation must for the present be regarded as The dy and dx which occur in it cannot be separated. &. The reader will have to become familiar with this notation. .f&.206 &.lt. 1.
t&.) =n of The binomial form $ (x) is often written symbolically as and then &.. (x . even if numbers a are equal if but of course tl^en some of the terms on the righthand reader will easily verify that (f) side are repeated./&. (x  an \ where the a s are real or complex numbers.. 116.gt.(#* a. known x n ~*+ as the binomial form. +a ft .gt. (x) = a. Then the notation implying that factors. y 1 dx y2 dx n . + . . (x) =a ^m l (x .lt. Standard forms.115. (a.. standard form of a It is sometimes more convenient to use is for the polynomial of degree n in x what viz. .lt.a )m 2 * . 71 ". We shall now systematically the forms of the derivatives investigate more of a few of the simplest types of functions. Polynomials...lt.lt. a lt (#) ./&. In this case &. .x^ 2 1 +. rt^j^.. then &.y n then y dx T l^ = i^l + I^2. DERIVATIVES AND INTEGRALS l 207 If y=y y2 . (a?) = n (a ..lt.) m . then (/p) = naox". The (x) a (x . + an _j. we form and add them several of the all together.. 2 . A. 1 I) . . .= 7y dx z dx ..xn + a.a^ (x  a 2 ) m .. If &. 1 ". . can always be expressed as the product of n factors in the (a?) form =a &..)". 1 possible products of n This form of the result holds all ./&..gt.aj (a?  cr2 ) .lt. 116] 2. . (x  a.+ (nl) aixn~ + 1 .. + yn dx i^ In particular.gt.gt... if y=2 ^ then 1 dy n dz ^.a l Y lt (x 1 . &... . We shall see later that &..
by mer.4bd + 3c3 g% . If 4&. 8 we deduce that will be equal.ad 2  2 eb". (47) a)". the result of Ex.gt. is of the form a(#a) 2 then must be divisible by # . (x .(%) is divisible generally.lt. a^+6=y. 3 + 6c. Conversely. i.7) a then //2 #4/#32 =0. where P (x) is a polynomial. equation. powers of h.r2 + 3 c# + d = has a double root G f2 + 4 ff = 0. P". equation reduces to where (?/ . This The reader may verify that if a.r 4 8 are the roots of + 4fo. ft 7.6) + l/(a.) equal roots only if 8. *= xl and we cannot solve the first J and l ffs z ffi/ff3 H t /ff2*=afixFind all H . and so on. ff^/Hf^Q. then the equation whose roots are and two similar expressions formed by permuting 3 4&. (#) is divisible &.. ft y.208 Examples XLI. 3 J [Put must have a root 9. then (x) is divisible a)". with their degrees of multiplicity.? 2 0#3 =0. . .e.lt. Show that if $ (#) is a polynomial then in (x) is the coefficient of h in the expansion of 2.b/a. P(#)=0.] 5. [Vi &. then x . cubic It is clear that if two of a.lt. Show that if ax* + 3 6.*".lt. if &. Using g&.ns of the elementary algebraical operations. the roots.a. by x . This value of x must .lt. . 8 are equal then two of the roots of this . Trip. a) m then (#) is divisible by x a and m~l by (xa) : . is . with their degrees of multiplicity.9 a. The equation !/(#a = b = c.? . 3.lt. when the in common with H=aob\ G = a 2 d3abe + 2b s 3 f 3ffy + Gf=0. is 1 .r 2 + dx + e = 0. /3. divisible by (x 2 a) and $ (#) and if $ (a?) by (# (#) are 6o/i divisible is divisible . ( where = ae. so that a= .. 1905.. a) + l/(a. Thus if P(*) = (ffl)3(0sa.. 7 cyclically. DERIVATIVES AND INTEGRALS 1.lt. a. H 2 the highest common the freWe P . 1 of Show how to determine as completely as possible the multiple roots 4.9^3 = ce + 26ccZ . Verify that the condition thus arrived at is 7. + A) &.c3 . then the roots of H H lff l 2 3 2 =^ are the double roots of P=0. the roots of //2 //4 /#3 2 = But it may not be possible to complete the solution of roots. a# a + 2&# + c=0. If 2 (oA* + 6) satisfy ax t + Zbx+c has a double root. if .c) = can have a pair of (Math.a and (x) by (#by &. (#) is by (x divisible by (x 2 a) then . P [If HI is the highest common factor of P and and so //3 that of Jf2 an d factor of HI and . on.a. of 6.27#3 2 =0.
if But convenient. signs.. 117] 10. in differentiating a rational the method of partial fractions. will be given later. It will be the simplest possible if Q (x) and Q (x) have no common factor.lt. e. Hence F(x\ and so vanishes for some value of which have opposite x between a and 117. repeated respectively A m and n times. any polynomial. u .lt. it follows at once from 1 13. u. i. (5) that R and rational function. applying not only to polynomials but to other classes of functions. vol. Chrystal s Algebra. are two successive roots. i.lt. B. factor.g. as in 116. employ We shall suppose that Q(x). is expressed in the form function. Now F (a) = m (a . pp. If where P and Q are polynomials. to It is very often Then it is proved in treatises on Algebra* that R (x) can be expressed in the form * See. The following is an algebraical proof valid for polynomials only./3.0) (a) &. &. general proof of this theorem.116. so that where B (x] is a polynomial which has the same sign. 151 et seq. We suppose that a. however.gt. say the positive si&.e. for a&.lt. DERIVATIVES AND INTEGRALS s 209 is Rolle Theorem for polynomials. Then (^ say. (x) = ^~~ ~ L^ ^ ) the derivative of any which we obtain it. Rational Functions. may or may not be the simplest possible.lt.#&. and (#). if Q (x) has no this formula enables us to write down The form in repeated Q(x) has a repeated factor then the expression which we obtain for R (x) will be capable of further reduction. F (0) = n (/3  a) 6 (0).m.(x) then between any pair of roots of =Q // (x} lies a root of (x} = 0.j&.
3. a. simple Hence no Algebraical Functions. Examples XLII.210 DERIVATIVES AND INTEGRALS is [VI where H. dx \ Prove that d atf + Zbx+c \ _ (ax + b} (Bx + C)(bx+c) (Ax+H) . 1 1 is mxm ~ l y The method explained in this section is particularly useful when we have to differentiate a rational function several times (see Exs. if a is complex.(^o)*~". The results of the pre of 113. obtain the derivative of any explicit algebraical m The most important such function is x where m is a rational 118. of the polynomial and it follows at once from Theorem (4) derivative 114. .e. XLV). in the form Incidentally we have proved of m for all integral values that the derivative of x positive or negative. enable us to ceding sections. together with Theorem (6) function whatsoever. If Q has its a factor (#a) m then the denominator of is divisible R (when R is reduced to of lowest terms)  by (x t^ a) w + 1 but by no higher power have a simple factor xa. Prove that ~~  dx \l +x*J 2.A^f In no case can the denominator of R rational function (such as I/a?) whose denominator contains factor can be the derivative of another rational function. i. _ 1. We have seen already ( 117) that the derivative of this . from its extension indicated in where a is We : that the derivative of the rational function last written is _ pA (x. __ (1 +tf )* 8 . as the sum of a polynomial and of terms of the type know already how to find the a root of Q (a) = 0. any x 4. of 113.of)^ _ ". or.(x) the sum of a number a polynomial. C. ^fiV PfV\ / f* f \ ^. 1 pA (^a)* m We are now able to write down the derivative of the general rational function R (x). number.
118] DERIVATIVES AND INTEGRALS 211 or negative and rational values . function is we shall of m..lt.Saxy = we find Differentiating with respect to x ". Suppose q positive and . VII. may also be deduced m we have (x) = x . xxxvi.= mx ~\ S~x _ xm more general formula T. a ( x + y * j y ~ a (y ^ J dx \ + x dy i dx and T so dy dx ^1. Find the derivatives of 2.(ax + holds also for all m b) = ma (ax + b) 1 rational values of m. Prove that a 3. where and that y 1 m ~l all let z =x p l v. Examples XLIII. h**Q n tin = It is clear that the lim $+x m ^. But there is no practical in the actual calculation of the difficulty : derivative of such a function illustrated sufficiently by an example. as a corollary from Ex. Then dx This result 3. so that x = sfl and y = q are integers and gP.117.ry+&y 2 + 2#o. 4 Find the (i) differential coefficient of y when (ii) o^2 + 2/i. . the equation + y". + 2/j/ + c=0. x* y* ay ~ ax . ^ 142 . if &. 3 the method to be adopted will be Suppose that y is given by 0. For. mx when m is an integer positive now prove that this result is true for = xm = xP *. The differentiation of implicit algebraical functions involves certain theoretical difficulties to which we shall return in Ch.
&. * In these is a rational number and a. 2 2 .212 119.gt. a.* 1. (4) and (5) of 113. sin. .. Dx arc sin x = + 1/V(1 ~ # DZ arc tan = 1/(1 + x Dx arc sec x = + 1/{#V(# ~ 2 )&. : ordinary inverse trigonometrical should verify the following formulae The reader 2 ). The more general formulae D x arc sin (as/a) = + l/\/(a2 #2 ). 113. Find cosm #. x) 2 positive or negative. Dx cot x = Dx sec x = tan x sec x. xxxix.. /3. DERIVATIVES AND INTEGRALS D.. b examples values that the functions which involve them are real. by means of Theorem (6) of differentiate composite functions involving symbols both of alge braical and trigonometrical functionality. have such .lt. . x\ x sm 2 x) x arc sin x + v/( 1 x2 ). cos# m the derivatives of . . and so to write down the derivative of any such function as occurs in the following examples. the reader will Dx tan x = sec x. Dx arc cos x = + 1/V(1 Dx arc cot # = 1/(1 + x) ^e } 2 !)}&. [VI Transcendental Functions. 113. . We have already proved (Ex. Dx cos x = By means of Theorems easily verify that 2 sin x. sin (cos r . m . a? 2 ). which are also easily derived from Dx arc tan (x/a) = a/(# Theorem (7) of 2 4 a 2 ). A arc cosec x + 2 l/{#\/(# 1)}. since sign a VU according as a is . ( 1 f ^ ) arc tan Jx t Jx.*?. derivatives Theorem (7) we can determine the functions.gt. are also In the first of them the ambiguous of considerable importance. cos (sin cos x sin x\ x 62 . Dx cosec x = And by means of the of cosec 2 x.V(a cos + r. Examples XLIV.. sin^m.O /a )} = 2 2 2 V&. In the case of the inverse sine and cosecant the ambiguous sign is cosine and secant the the same as that of cos (arc sin x\ in the case of the inverse same as that of sin (arc cos x).gt. cot x cosec x. is the same as that ofa cos {arc sin (x/a)}. 4) that Dx sin x = cos x. we are enabled to Finally..
#2 ). Find the derivatives of arc sin V(l . Each of the functions arc cos r:T: fa cos x I ) .f&.) 7 Show that 8. ^ o^ .lt. Show that the values of x for which the tangents to the curves y=x cos are roots of cot x=x.27 y=(sin #)/# are parallel to the axis of x respectively. more complicated cases. Differentiate x cos a?. / (0 (#)} (^). 13. a?.#2 )}. 1904. arc sin (2^ ^(1 . 07. Trip.ffl} has the derivative 6 Prove that rf arc cos VVcos //cos30\) 3 *. The The derivative of y = (tan x + sec x} m is my sec derivative of y = cos x + 1 sin ^ is ly.119] DERIVATIVES AND INTEGRALS 213 2. a?.* arc tan //a6 ^ T U / (s+6 IV f ( ) tan has the derivative l/(a + b cos 07). (sin#)/. and arc tano7+arccot# for all positive values of x.gt. ? arc tan ( How 4. If u and are functions of x. . tan #=. Prove that the derivative of F[f {$ result to still v (x}}] is F [f {&./ VV // ^cos 6 cos 3^/ (Math. 2_ ) arc cos then 10. 14. Show that each of the functions Sarcsin J(^\ Cretan J(^\ arc sin VK )(*. If X= a + 6 cos 57+ c sin y= . and 2x f 2 _. 9. \\ax ^ do you explain the simplicity of the results Differentiate 1 ax+b 1 ax+b 5. (a?)}] and extend the 11.r. 12. then Dx arc tan (ujv) = (vDx u . 3. Verify by differentiation that arc siri x f arc cos x is constant for all values of x between and 1.uDxv)/(u + v 2 2 ).
where values required are the values of a for which [The =.f&. The normal is xyQ .gt. ellipse (#/a) 19. Ex. define the nth derivative or &. and the tangent is 1 which may be reduced to the form XXQ + yyQ = a which of course passes through the origin. y = ax 16. Find the equations of the tangent and normal at any point of the 2 2 + (#/&) 2 = l and the hyperbola (#/a) a .r) . a&.v=ax.214 15.lt.cos when (2)). DERIVATIVES AND INTEGRALS [VI is positive.lt. The equations of the tangent and normal are to the curve x= y = ^r (t\ at the point whose parameter is t t 120.lt. y ) of the circle j^+y ^a 2 2 .lt. (l/#) (cf. touches y=sin If (#) #.gt.lt. 17. where a 1 a finite number of if and if has no real roots except #=0 &. But it is which follow.(y/6) = l. from $ (x) just as we formed (#) from &. #4=0. It is easy to see (cf. function coefficient is called (a?). then $ 0?) = ZK sin (l/. . 18. xvn.gt. [Here y = /v/(a ^ 2 dy\dx= x\^(d x\ i ). (0) = 0. Find the equations of the tangent and normal at the point (# .lt. .y?Q = 0. and &.jT) that the equation sin . when x =4= 0. the second derivative or second differential of The second derivative of y=$(x) may also be written in any of the forms d*y In exactly the same way we nth differential coefficient of y = of the forms may (a?).] = #2 sin (l/#) &.^ (^) Repeated differentiation. // &. ^nd ^ ^ ) is discontinuous for # = 111. which may be written in any d*t/ y./&. and (0)=0. We may form a new function This (^). roots which increases as a diminishes.] . Prove that the values of a for which the number of the equation tan is a positive root of roots changes are the values of cos . d \n that it is easy to write down a only in a few cases formula for the nth differential coefficient of a given general Some of these cases will be found in the examples function.1.
To prove the theorem we observe that .gt.( m is a positive ) The formula dx enables us to write in the standard down the nth derivative of any rational function expressed form as a sum of partial fractions. so that u n for example. and n&.119. This result enables us to write down the nth derivative of any polynomial. . and we can form the u and v. which gives the rule derivatives of 1 where suffixes indicate dift erentiations.(ranfl) # . 1/(1 4.. 3. It is obvious that by repeating this process we arrive at a formula of the type for r=l. In these two examples m may have any rational value.. Ti1. 2. then w (#) = 0. If y is a product ue...m. and show that It will if this is so then a n + i. . then we can form the nth derivative of y by of Leibniz Theorem. If $(x}=x m then ". (u y) 2 and so on. If (f) (as) = (ax + b) m then If integer.r v r is This establishes the theorem. &.. n means first Leibniz Theorem. 2. then follow by the principle of mathematical induction that in question. 1. denotes the nth derivative of u. 2. 0W(^) = w (ml).. Prove that the nth derivative of #2 ) is 5. 120] DERIVATIVES AND INTEGRALS 215 Examples XL V. Ui r = n\ ( } for all values of n and / When we of form (^^)n + 1 by differentiating (uv) n it is clear that the coefficient un + l .lt. .r( f r } ?* = 1. n.
] 11.gt.l) n n (a.0). And if y = A cos mx + B sin wi# f Pn (a?). sin (x Prove that If Dxn cos x = cos (a? 4. 2 + a2 )( n + sin {(w + l) arc tan (a/a?)}.l)7i !/2i} p[Cis (O+ 1) = (. then a? 9. is a polynomial of degree n.ZBx + C\ then [First obtain the equation when ?^ = . DERIVATIVES AND INTEGRALS The nth derivative of [VI xm /(a?) is m ! m (m ! (*)! .m y = 0. Dxn sin x 2 + 1? 8.] ( If Un denotes the nth derivative of Lx + M )/(^ 2 .Cis {.(n + ^ )} = {( . {(cos a:)/a.216 6. n times by Leibnitz Theorem.\nrr}. and is the and a/p.+ ^ sin mx then Dx2 y 4. then differentiate 7i times by Leibnitz Theorem. Pn and Qn are polynomials in ^ of degree n and w Establish the formulae _i =_ dy z = dx*f \dx) dy* dy~ /\dx) __\dx* dx . The nth derivatives of a/(a?+x^ an(i xi(a? + x 2 ). ZV* 2 {a/( . Prove that A. where Pn (x} If a. where 13.9} I 1 ) /2 Similarly 2 ) ~ (n + 1)/2 cos {(n+l) arc tan (a/a:)}.} = {Pn cos (a: + ^TT) + Qn sin (a? 1 respectively. then 6&. 2 1 1 &.". terminates. Since \ a 2 4.1 terms or until 7. 2 D^y + D^ y + y = then [Differentiate 10. y = ^4 cos ma.lt.x 2 2 \a? ai x + aij * we have y^lt_l \i n+1 2 ((a?ai) 2 2 and a similar formula If p = \/(a? for x {a?/(a + a: )}. numerically smallest angle whose cosine and sine are x + ai*=p Cis 6 and xai=p Cis (. and so n * D 4 a?/p a 2 ).*(* T it the series being continued for n 4. 12.
then 16. respect to #. I then z ?/9 Z2 Z3 (Math.lt. (x) for all values of x in question.gt.(x). &.c/* 2 x + by +f}\ &. DERIVATIVES AND INTEGRALS 217 If 3/2=1 andyr =(l/r!) 1 I f.gt. 1905. acute angle with the axis of x.(abc + 2/#A .(#)&.gt.f&. generally known a^Rolle s Theorem. //* (# ) &. THEOREM B.lt.f) &. sign of for &.2 .lt.gt. ) for all values of x greater than XQ but sufficiently near to x .lt.gt. If then and 121. 15.lt.lt. tt ) dashes denoting differentiations with z u y". c/&./&. then there must be at b least one value of x which between a and that and for which (x) is equal to There are two possibilities : the first is &. {(f) (x (x )}/h converges to a positive limit (ar ) as This can only be the case if + h) $ (a? ) and h have (a? the same sign for sufficiently small values of h and this is precisely Q &. assumption of course involves the continuity of The meaning &.a/2 .gt. The reader should formulate for himself the corresponding theorem for the case in which 0.lt. &. &.(# of x than XQ but sufficiently near to XQ and &. (a? ) all values &.lt.lt. (^ ) . 121] 14./&. Some general theorems concerning In all tions.lt.lt.^&.f&.lt.# )/(/&./&.gt.gt. (a?). An immediate deduction from Theorem A is the following important theorem.gt. z.120. of intuitive.) . Trip. that follows &. /ie?i &. *. Of course from a geometrical point &.gt. THEOREM less A.lt. IfW(y. of the &. (x) &. the inequality (x) expressing the fact that the tangent to the curve y (x) makes a positive = &./&. &. t what the theorem view the result is states.lt.gt.". In view of _the great importance of this theorem it may be well to repeat that its truth depends on the assumption of the existence of the derivative &.gt.lt./&. = (!/*!) 1 ! Dx z. (x) is derived func a function of x This ^ which has a derivative (x) for all values of x in question.gt.lt. Ihe".lt.f). we suppose that (j&.lt. dyldx =(ax + hy +g)l(hx + by +/ ) . h^Q.gt. . If (a) = and lies (6) = 0.gt. For + h) c/&.
be greater than &. then there must be 0.x l (tfj &. (a^) when # is (#) is. a. then &.gt. Nor can (# ) be greater than (a?2 ). a value of x between a and b such that (f&.lt.gt.lt./&.lt. (x) k fy (x) and apply Theorem B ^ (as).gt. say to so that &.gt. cannot be equal to (#2 ) for. so that there would certainly be values of (x) greater than (). +e a &. &. that there is a value Let #! XL and &. by Theorem A. by Theorem 2 of 102.lt. a c/&. &.(#) &. b).lt. and such that $ (f ) is at least as great as the value of (x) at any other point in &.lt./&. &.gt.lt. greater than than a^ and sufficiently near to xlf It follows that there is greater = (^) and so.gt. In the first We have to show that $ (x^) (a?2 ). then there must be values of for which is Let us suppose.lt. [VI (x) is zero throughout the whole interval (a.lt. example.lt. &. For. b. not equal to a or b.lt. The conclusion of Cor.lt. there (a^) place would.lt. (x) is (x) for all values of x in a certain interval.lt.lt.f&.gt. provided $ (x) is COR.p If on the other hand x (x) is not always equal to zero.gt. throughout (a that remains + e. or is not positive.lt.lt. &. COR.lt.lt. &. since (X) (x) is positive.lt. COR. would. + e&. &.lt. (f) for values of x greater than f but sufficiently near to f.lt./&.lt. and that corresponding to an end of the interval.A for by Theorem B.gt. continuous even for these exceptional values of x. that (x) is sometimes positive.gt. in the stricter sense of 95.lt. &. Then. &. . . .f&. there is a value % of x.lt.gt. so that 2. (%2 ).lt. &. If an increasing function of x.gt. also equal to zero throughout the interval./&. Similarly we can show that () cannot be negative. (b) = k. &. #2 . x2 &. &./&. c^&. throughout that interval./&.lt. the interval. positive then And &.lt. by Theorem A.lt. b) for which $ (x) does not exist.(#) For if (f) must be equal to zero.gt.218 DERIVATIVES AND INTEGRALS In this case &. It is plainly sufficient to consider the case in which there is one exceptional value of x only./&. 2 still holds if the interval 3. &. If (f&. &. be a value of x between x and #2 for which = 0./&. which $ (x) = 0. &. &.gt. &. considered includes a finite number of exceptional values of x (a. (x) We to have only to put 2. &. we can is choose a b). it were &. x &.& If a (x) &.j&.lt. &. (a) &. if this were so. for positive or negative.gt. and x2 be two values of x in the interval in question. a value xs of # between ^ and X* such that (x8 ) of x between xl and x. by Theorem B.gt. and by Cor.lt.lt.gt. All to prove that . 1. &.lt.lt.
lt. Now &.^&. &. and (f&. COR. (V) when f and we define a minimum in a similar manner.lt. &.lt. &.gt.gt. b). throughout the interval (a.// or * A function which is continuous but has no derivative may have maxima and minima. (a) ^ 0. . as &. (/&.lt. (X) decreases steadily.&. %+ e) of values of e x x such that (f ) and when f a? f + e. and illustrate it by an actual example. &.gt. Thus in the figure the points A correspond to maxima.* minimum value of at x=% is that (f) $&.gt. The reader should compare the second of these corollaries very carefully If. 122] (a) sense.lt. We ( Maxima and Minima We shall say that the value $() assumed by (x) when x = is ^maximum if (f is greater than 122. A necessary condition for a maximum = 0. 4.gt.*a+0 (X) &. &. .lt. of x=%. the function whose graph is there shown. 39.121.lt.(x) &. DERIVATIVES AND INTEGRALS (ayj). (x) in the immediate neighbourhood &./&.lt. ) any other value assumed by &.gt. &. i.lt. we assume only that fr) is positive at a single point #=# then we can prove that (#2) when x\ and #2 are sufficiently near to # and x XQ x z For $ (xfi (# ) and (a? 2 ) (^o).lt.gt.gt./&. x^ 219 &.lt.lt. in a moment &.lt. &. the points B to minima of &.lt. But this does not prove that there is any interval including with Theorem A.lt.e. XQ throughout which (x) is a steadily increasing function. &.lt.lt.lt.. B^ to a minimum no way inconsistent with the function is greater at B than at A 3 in l fact that the value of the THEOREM C. for the assumption that x l and x2 lie on opposite sides of XQ is essential to our conclusion. then cj&. (#) &.lt. by Theorem A. (X). as in Theorem A.gt. &.^&. and in the stricter a.() &./&.lt. if we can find an interval (f e.gt. &.lt.lt. 6).gt. f &. B.lt. decreases towards and so &. (&) is positive throughout the interval (a. &. = &. &./&.gt.lt. //". &.gt./&. (a + 0) = Km ce./&. It is to be observed that the fact that is A s corresponds to a maximum and .lt. Fig. .lt. (^i)&. &. We are of course assuming the existence of the derivative.lt.gt.&.lt.gt. shall return to this point.
lt. when x is less than f but sufficiently near to f. then 123. This result may &. f ) throughout which this ensures that cf&. obvious from the form of the graph of a? (Fig.f&. as f gives a minimum. &. p. Let us assume (x) has a second derivative follow from the existence of (x).(?) has a &. Thus positive when x is greater than f = gives a maximum.lt.".". decreases as x increases and obviously at () shall be a maximum.f&.gt.f&.lt.(?)&.gt. (f) = 0. 45).gt. &.".lt. then x = f gives of and if the sign of (x) (x) changes in the : &.gt. THEOREM D. (x) increases it with x.lt. for There or is another way of stating the conditions is a maximum that (f&.gt.. j~ But &. of these two inequalities are reversed there will certainly be a For then we can (by Cor. (x) : this of course does not &. and an : f 4. minimum.lt.220 DERIVATIVES AND INTEGRALS This follows at once from Theorem A. there will certainly be a maximum at x= if &. that Then.".lt. Thus Theorem B fails in the case of the function . = 3#2 which vanishes when x = 0.gt.lt.lt.0.lt.lt.lt. &. minimum which often useful.lt.lt. &. In what has preceded (apart from the last paragraph) we have 124.gt. interval (f interval (f .f) (x) follows from that of &. minimum if (f) &. opposite sense./&.gt.(x) &.() &. any more than the existence of But in such cases as we are is likely to meet with at present the condition generally satisfied. $". 3 of 121) determine an e.gt.". by Theorem A. (x) changes a maximum x = % from positive to negative. a &. (x) &.e) throughout which &.g. for all values for all values of x less than but near of x greater than but near to to %. then (?)=0 and = %. x Suppose.gt.lt. of x in the interval assumed simply that (#) has a derivative for all values If this condition is not fulfilled the theorems cease to under consideration. But if y = x 3 then (x) does not give either a maximum or a minimum of a?.gt. [VI is That the condition not sufficient is Thus x= evident from a glance at the point G in the figure. &.lt.lt. also be stated thus. as is . (x).$&. &. $ (x) is e.lt.lt.lt. a maximum if maximum or minimum at x 0.f&. : and (x) and if the signs &. 10.f&.lt. &. 0. and negative but sufficiently near to f. If the sign of &. be true. // + 0.
for which it is (#)&. Consider the function (#) defined.122124] DERIVATIVES AND INTEGRALS 221 where the square root is to be taken positive. The graph of this function is shown in Fig. not difficult to see that this answer is wrong. between the limits of indetermination a.lt. bare_existence of the derivative however. (0)j as it does not exist. 40. Can a function (#) have a derivative for all values of # which is not itself continuous ? In other words can a curve have a tangent at every point. will probably incline to the answer No.l.lt. conditions of = when #4=0. What is practically the same example enables us 121. we can find values of #. and never figure. however. and throughout which therefore impossible to find any interval. and 0(0)=0. But if #4=0 then (#) = 2# sin (l/#) cos (I/a?) + a. It is. and that \ subtlejbut is that (a?) itself is a con 1 tinuous function. is all that we have as sumed. as near to as we like. which oscillates As al&. function of #. as is evident from the 1 if # is positive.a&. 2#sin (I/a?) tends to as #^0. cannot be equal to zero. and yet the direction of the tangent not vary continuously ? The reader.lt. including a?=0. And there is one_jtssumption in particular that we have not made. but at P. where 0&. Then Thus the (0) Theorem A of 121 are satisfied. so that the test for a maximum The (#).0. still This raises a rather O pjg 49 a very interesting point. and no tangent to the graph And in this case #=0 obviously gives a maximum of 0(a?).0.1) = 0(1) = 0: but (#). also to illustrate the point referred to at the end of (a?) Let ( = #2 sin 1 /#) f a#. when # 4= 0.lt. is equal to 1 if # is negative and to vanishes. There is no derivative for #=0. if he considers.0. while (0) = li Thus for (#) exists for all values of #. so that of indetermination (#) oscillates between the same limits. = 2# sin (!/#) cos (l/#) . what the question means and tries to answer it in the light of common sense.1 and a + 1 as #*0. Here (. But (a?) is discontinuous for #=0. and cos (1/0) oscillates between the limits 1 and 1. If #4=0 then (#) Then 0(a?) is continuous for all values of #. fails.gt. (a?) is a steadily increasing . by the equation and suppose that 0(0) =0. a&.
Verify the theorem when UCGSX.lt. 3.1 2# + 7.gt. then this interval can include at either of the equation s/(. 1?* c) discontinuity. &.lt. 5). 3^4 + 8^3 . and so equation 5.r)=0. xxxvii. And na] 6.~ (x) is continuous for # = 0.a) m  * (.f) (Ex.gt. and conversely.r. 3 and from cos^.2? {(m + n}x. p are positive integers b &. ^7 . 6.a) m (x . [The first function vanishes for &. Examples XL VI. v = sin.] ^ ^ . Ex.lt. unless a = b = c. then the [If v does not vanish between two roots of w function ufv is continuous throughout the interval (a. xvii. XLVII. 4. or c. generally.r. . contradicts our hypothesis. when x&.b} n and a &.+ 0. that (x}^a when and $ (0) = c. K ^ (r Deduce from Ex. For a proof see where m.r&./ (#)=0. cf) (#) = (. prove that if and .gt. If f(x) /".0. Ex.lt. that this sin ##&. x*. x=a .24.increases as x increases from TT to \TT.() ^&. say a and /3.l. continuous and have the same sign at every most one root of throughout a certain interval of values of u v are continuous and uv u v never vanishes Show that between any two roots of u = at any point of the interval. (3} and vanishes at its Hence (u/v) =(u v uv )lv 2 must vanish between a and /3.] 2. e. then C2m and SZm +i and (x} are are positive or negative according as m is odd or even. which extremities. in which case 125. = 0.2.G^2 .lt.0.gt.lt. (x) = (x.g. w.lt.gt. 7.sin a? is an increasing function throughout any interval and that tan. from this that that sin^^ + i^3 And. The functions v and their derivatives x. [VI Ch. 4&. n b) (x n. if 4).when x^&.gt. &. 4j (a?) = (37 . Show that tan 57 to x = \ TT. (#)sT".222 It is. 18) a &.l + ^. Show that the polynomials %xP + 3^2 .0.i 2 &. DERIVATIVES AND INTEGRALS however. from #=3r and deduce that there is one and only one root of the (cf. lies one of v = 0.? +19 are positive 3. in each of these intervals x also increases from X=^TT to ^=TT.0. tan#=# on.&) and 1 xb. impossible that simple 0. point of an interval (a.gt. For what values of a is ax sin^7 a steadily increasing or decreasing function of xl Show that of values of x. V $ (%} should have what was called in &.ml) vanishes for x=(mb+na)l(m\n\ which lies between a and second case we have to verify that the quadratic equation In the (m + n +p) x* {m (b + c) f n (c + a) +p (a + b}} x + mbc + nca +pab = has roots between a and b and between b and c.v a) m (x Verify Theorem B when p .
that (ax + b)/(cx + d) has no maxima or minima. and y*~a/A as x*~ + cc or x^. 2. #=0. We where \ = bAafi. ~b\a&. In each s case sketch the form of the graph of the function. 3^4^ + 1.oo It is easy to verify that the curve cuts the line y = a/A in one and only one point. (Ay a). which [Consider the last function. or according as greater root of (1) gives a vice versa. Thus the maximum if b}a&.] 13. whatever d may have. It is easy to see that x 3 gives a 3. The maximum and minimum l values themselves are the values of X is for which axP + 2bx + c\(Ax2 + 2Bx+ C) a perfect square. X 3x. and that it lies above this . 6. since y is continuous for all values of #. where m and n are any positive integers. c. c.gt. DERIVATIVES AND INTEGRALS Determine the maxima and minima (if 223 2 any) of (x I) (# + 2). as (#) is negative &. on both sides of #=0.c) 3 distinguishing the different forms of the graph which correspond to different hypotheses as to the relative magnitudes of a. 4^3 18^2 + 27.B/A.15^ + 3. while x=0 gives neither. for example. line for large positive values of x.lt. Sketch the graph of the function. and below or it for large negative values.] 9.c 3 3^2 36^+10. values 12. 6. and this is impossible.lt.6) 2 (x . Discuss the maxima and minima y of the function when the denominator has complex roots.a) (x . we obtain an equation Writing further of the form for \x + p and rj for j) (*)}. &.BjA. Here (x) vanishes for x = and #=3. For if not the derivative would always have the same sign. Discuss similarly the function (x . 15. The derivative vanishes if This equation must have real roots.r7.lt. = 5x2 (x2 9). If Ax2 + 2Bx + C=0 have has real roots then it is convenient to proceed as /^ follows. (A/X) p. [This is the condition that y=\ should touch the curve. .] In general the maxima and minima of R (x} = (x)/Q (x) are among 14.gt. a 6 . the values of X obtained by expressing the condition that P(x) \Q(x) = Q P should have a pair of equal roots. = cAaC. Show . 11. maximum and x=3 a minimum. a minimum algebraically in the contrary case. [We may suppose a and A positive. Discuss the maxima and minima of the function (x .B\A b{a&. m considering the different cases which occur and n are odd or even. Draw a graph of the function. according as 10.124] 8.a) m (x n b} . .
The preceding if a/ A = b/B. 42. The maximum is  l/(v/P . 41 a.J(pq) a minimum. corresponds to a minimum of rj This transformation from origin. i. 77) amounts only to a shifting of the keeping the axes parallel to themselves. 42 corresponds to the former case. and otherwise assumes all values except  in an interval of length 4 v/(ay * /3 y  ). y] to (. is The graph shown in Fig. 41 if c. 41 a. . 41 c. Show y lies between a and that (x a) (x /3)/(#y) assumes all real values as x varies.] 16.gt. and (if X minimum of y. &. considered as a function of #.On drawing a graph becomes clear that this value gives a maximum or. considered as a function of . it fj. 77 the graph of is as shown in Fig.r 1 f #. Fig. q are positive the general form of the graph easy to see that is as shown in and it is =\J(pq} gives a q the maximum and = &.\/2) 2 minimum &. 41 6. by purely be found in Chrystal s Algebra. .2 ). and dy\dx = gives the single value x = (. 416 Fig. which the latter is the greater. When p and Fig.e. Thus there are two roots of the derivative if p and q have the In the latter case the form of sign. minimum according as positive or negative. and similarly if for a maximum. say.* In the particular case in which p function is and its graph is of the form shown in Fig. The derivative of 77 with respect to vanishes or if same g*=pq. n Fig. [A full discussion of the general function algebraical methods. will 2 y = (ax + %bx + c}l(Ax? + ZBx+ C\ i. a change of scale along each 0) a reversal in direction of the axis of abscissae and so a axis. 4647.gt. if those included /3.224 DERIVATIVES AND INTEGRALS [VI (#. and vice versa. vol pp. But discussion fails in this case A=0.lt. none if they have opposite signs. we have Fig.
Trip. where a and b are positive. Show Show that tan 3# cot 2# cannot that. A cos 2 x + 2H cos x sin x + B sin 2 #. 25.b). or minima.*? + 5) 2 (3. 1898. .) (a. sum in A line is drawn through a fixed point (a. Find the maxima and minima of a 2 cos 2 x + b 2 sin 2 x. 24. The maximum value Discuss the a? of (#  2 l) /(# + 1) 3 of is ^. 1909.124] 17. The maximum and minimum of (x + a) (x + b)l(xa) 6). maxima or minima) 2 and 3 when x = l and 3. and draw a graph of the function in (Math. H.v2  14. 27. Determine the function of the form (ax2 + 2bx + c)/(Ax2 +2. maxima and minima (tfl)/(aa + 3a? 2 + 3).) [If the last function be denoted by P(x}jQ(x\ it will be found that 22. (# I) (3^ 2 . 15 . of the lengths of the hypothenuse and another side of a rightangled triangle is given. 1910. } axes OX.) The least value of a 2 sec 2 x + b 2 cosec 2 x lie (a + 6) 2 . if &. Verify the result 23. then the area of the triangle is a maximum when the angle between those sides is 60.lt. a). 4 sin a sin 6/sin 2 (a .2#. 1908.e. DERIVATIVES AND INTEGRALS Show that 225 can assume any real value this case. (Math.r .c &. arY(#l) (A 3)3. Trip.Bx+C) which has turning values (i. 1. 6) to meet the 29.*= 1 19.1). 1909. that the function has an infinity of minima equal to and of maxima equal to 26. Show that the minimum values of PQ OP+OQ. P and Q. 4ab. are respectively (a 2/ 3 + 6 2 3 ) 3 2 (V + v/^) 2 and / / . Show Show that sin (#+a)/sm (# + &) has no maxima Draw a graph of the function. are 20.lt. (#a*A. Trip.) 18. Find the maxima and minima of a cos x + b sin x. (Math. OY OQ and OP.37)/(. TVip. by expressing the function in the form A cos (a.) 28. if the between $ and f. 21. Trip. and has the value 25 when # = 0. is (Math. respectively.
) is (& we give a strict proof of this theorem. Eliminating 6 between these two equations we find 32. b). If $ (x) has a derivative for all values of x in the then there is a b. 43) that if the has a tangent at all points of curve &. where x and y are positive and = 3K2 is then a + b(dy/dx) = Q. interval (a.lt. of ax + by. The relation between x [If ax + by is a maximum and y gives (2^+y) + (#+2y) (dy!dx} = Q. \ &. The Mean Value Theorem. simply (see Fig. where x and v are positive and 33. value % of x between a and such that * (a)Before (ft . $ are acute angles connected by the relation a seed 4. DERIVATIVES AND INTEGRALS [VI A least value of 31.&sec$ = 0. Find the lengths and directions of the axes of the conic of x 9 with the axis [The length r of the semidiameter which makes an angle is given by 2 cos sin + b sin 0. c are positive. The greatest value ./ Fig. We Theorem of the Mean . If and where a. Equate the two values of dy/dx. APB its length then there . which perhaps the most important theorem in the Differential Calculus. is a minimum when then a cos + 6 cos = c.226 30. it will be well to point out its obvious This is geometrical meaning. Show that the PQ is equal to the sum of the semiaxes of the ellipse. The greatest value of xm y n . 43. tangent to an ellipse meets the axes in P and Q.] 34. The condition fora maximum or minimum value of r is tan20=2A/( 6). 6. can proceed now to the proof of another general theorem of extreme importance. THEOREM. a theorem commonly known as The Mean Value Theorem or The 125.lt.
where the tangent is parallel to AB. Of course a + 0(b a) lying between merely another way of writing some number f between a and b If we put b = a + h we obtain where is &.lt. Similarly b = c. [In the latter case i.lt. Show that is the difference between the ordinates of a point on the curve and the corresponding point on the chord. where (6 + a6 + = 2 then lies between a and 6.] Establish the theorem stated at the end of 124 by means of the Mean Value Theorem.l&.(&) ( (a) + (b  a) $ [a + .]&. 125] DERIVATIVES AND INTEGRALS 227 must be a point.lt.a)}.gt. &. It follows from Theorem B for which its derivative vanishes.b. It should &. {&.lt. Examples XLVII. a small positive value of  such that () is nearly equal to c. 2.124.lt. It is easy to give a strict analytical proof.gt. 152 . we have 2 3 (6 )/(6 a) =3 2 . which is inconsistent with lim $ (#) = unless a = c. (a) + h$ (a + is 6h). is a number and 1.lt.(a + h) = &. But this derivative is which proves the theorem. by the theorem.gt.a) the tangent of the makes with OX. 3. it has not been assumed in this proof that (as) is It is often convenient to express the Mean Value Theorem (b in the form = &. and (6) angle which AB makes with OX. Verify the theorem when $ (#) =xi and when to prove that 2 ) &.] is *^+o ./&. (0) = e. such as P./&. (x) 3 = x3 .e.lt. [Since &.lt.lt. most often quoted. .gt. that if a&. Consider the function of which vanishes when x = a and 121 that there is a value a? = 6.&.$ (a)}/(b .gt.lt. For () is the tangent of the angle which the tangent at P . which is the form in which the theorem 1. be observed that continuous. we can find a small positive value of x such that $(0)}/# nearly equal to c\ and therefore./&.
DERIVATIVES AND INTEGRALS [VI A least value of 31.)*(). Show that the PQ is equal to the sum of the semiaxes of the ellipse. 6. Eliminating 6 between these two equations we (l/^} of (6 (!/*)} = A*. If 6 a. c and $ are acute angles connected by the relation asec0 + 6sec0 = c. then there is a value % of x between a and b. APE its length then there . We can proceed now to 125. The relation between x [If ax + by is a maximum and y gives (2# +y) + (a? + 2#) (cfy/ote) = 0. The greatest value of ax + by. Equate the two values of dy/dx. simply (see Fig. a theorem commonly known as Theorem of the Mean .] . 43) that if the has a tangent at all points of curve O Fig.)? (ft Before we give a strict proof of this theorem. The Mean Value Theorem. the proof of another general theorem of extreme importance. 32. b).] 34. where x and y are positive and then a + b(dy/dx) = Q. THEOREM. where are positive.&.lt. 33. The Mean Value Theorem or The a derivative for interval (a. 43. The is greatest value xm y n where x and v are positive and = k. which is perhaps the most important theorem in the Differential Calculus.lt. it will 0(0] be well to point out its obvious This is geometrical meaning. If (x) has all values of x in the such that *&. tangent to an ellipse meets the axes in P and Q. Find the lengths and directions of the axes of the conic [The length r of the semidiameter which makes an angle of with the axis x is given by 1/72 = a cos 2 0f2A cos & sin d + b sin2 Q.226 30. then a cos + b cos is a minimum when = 0. or The condition for a maximum {a minimum value of r find is tan2#=2/i/( &).
merely another way of writing some number f between a and b If we put b = a + h we obtain ( a + h) = (f&. [Since &.lt.] &./&. we can find a small positive value of x such that $(0)}/ is nearly equal to c. where 3. (a) + (b  a) &. (a is + 0/i). 125] DERIVATIVES AND INTEGRALS 227 must be a point. (0) = c. . unless a = c.lt. 3 =3 2 a 3 )/(6 [In the latter case we have to prove that (& ) 2 2 2 i.a)}.^+o 152 . where 6 is is a number lying between and Of course a + 0(ba) ./&. a small such that positive value of () is nearly equal to c. Establish the theorem stated at the end of 124 by means of the Mean Value Theorem. Show that is the difference between the ordinates of a point on the curve and the corresponding point on the chord.e.gt. most often quoted.(&) &. it has not been assumed in this proof that (x) is It is often convenient to express the Mean Value Theorem in the form = &.gt.lt. (#) = ^3 . Similarly 6 = c.lt. and therefore. 2. Consider the function which vanishes when x = a and x = b. where the tangent is parallel to AB.lt. For (f) is the tangent of the angle which the tangent at P makes with OX. which is the form in which the theorem 1.gt. {a + 6 (b .lt.& .lt.r) i$(#) a. by the theorem. such as P. 1. (#) =xz and when &./&. (a) + h(f&.l with lim (. which is inconsistent = a.lt.lt. It is easy to give a strict analytical proof. o &. It follows from Theorem B of 121 that there is a value f for which its derivative vanishes. But this derivative is o a It should &.gt.gt. {&. and $ (a)}/(b a) the tangent of the angle (b) which AB makes with OX. which proves the theorem. be observed that continuous.lt. Examples XLVII.124. Verify the theorem when &. that if (& + a& + )= then lies between a and 6.
lt.gt. throughout a (f&. : &. tf (x). where is a number lying between x and x+ h. DERIVATIVES AND INTEGRALS Use the Mean Value Theorem is [VI 113. by the Mean Value Theorem. We have in this chapter seen &.lt. if a and b are any two values of &. i. to prove Theorem (6) of assuming that the derivatives which occur are continuous.gt. f(x + Ii) =f (x) + hf ().j&. where tive of & is a number lying between / (x) and / (x) + hf is Hence the deriva F{f(x}} since +x and 1 /(#) as A*0. which is of great importance in what follows (a) if is constant a certain interval of values of x. (b then (b)  = (b (a) a) &. then the functions out that interval by a constant. throughout that interval. first place we want to know whether such a This question must be carefully (x) actually from the question as to whether (supposing that distinguished there is such a function) we can find any simple formula to function as exists.228 4.lt. &./&. It is natural to all those of the commonest occurrence. [The derivative of F{f(x)} by definition _ h But. how we can find the derivative of a given function (x) in a variety of cases. express (2) it.a)} = 0. {a &. that of determining a function whose derivative is a given function. parts. then (x) throughout 126.lt. Suppose that = ty (x).gt.gt. 127.gt. + (x) . And ().lt.] result The Mean Value Theorem furnishes us with a proof of a = 0. Then we wish to (x) is the given function. (1) In the &.e.lt. We want one such function know whether it is possible that more than should exist. (x} and ty (x) differ through Integration.lt.f&. we want to know whether our to . &.f&. including consider the converse question. as in the interval.lt. An immediate corollary is that if certain interval. For. A little reflection determine a function such that (x) shows us that this question may really be analysed into three ^ .
which is nowhere continuous. xxxvn.gt. or greater than 0. unless a = b = c. A . this is one of the cases in which higher mathematics has proved commonsense to be mistaken. &. in this case x = 0. or con tinuous curves which never have tangents.^20. The proof of this (x) (x) such that always a function follows. 229 we want know whether there is and any simple relation all between the different solutions which will enable us to express of them in terms of any particular one. (x) x m where m may have a derivative for x. how special values of x. has obviously no derivative for any value of x. a further question : which is at present beyond us. is a positive integer. all values of as. Whether there is are continuous functions which never have derivatives. : &. XLVII. (3) If there is a solution. for The example of the function {/a?. or sin It may generally. or c according as x is less than. like tyx or tana. is DERIVATIVES AND INTEGRALS one which admits of a unique solution or not to . however/we shall always suppose ^(x) continuous. In what will be given in Ch.lt. shows that a continuous function may not have a derivative some special value of x. Of course during this chapter we have confined ourselves to functions which are continuous except for : some ever. the function considered in it may never have one Ex.125127] problem if not. equal to./&. But at any rate it is clear enough that the question has (x) a derivative (x) ? is one which has to be answered differently &. but not always have one. we want to it. / . function &. And then the answer is Yes if ty(x) is continuous then there is = ty (x). or sec#. And we may $ too. as we have already stated in 111. . then the answer No is (Ex.lt. is This a case in which the given function discontinuous. VII. 3). Or again for example.lt. know how to find an actual expression for It will throw light on the nature of these three distinct ques three corresponding questions tions if we compare them with the which (1) like arise with regard to the differentiation of functions. question tive ? is there a function (x) of which expect that the converse is the deriva i/r (x) will have different answers equal to We is have already seen : that there are cases in which the answer No is thus if i/r (x) is the function which is a. in different circumstances. Commonsense says No but.lt. 6.
We shall use the notation It is hardly necessary to point out that \. Thus r \x m dx= J m+l xm ^ f . (. be regarded purely as a symbol of operation the I and the dx no more mean anything when taken by them d and dx of the other operative symbol djdx.gt.f&.lt. the answer is almost equally simple. In the case (2) of differentiation we have a direct definition of the derivative which makes it clear from the beginning that there cannot In the case of the converse problem possibly be more than one.lt.f arbitrary constant of integration. is &.gt. &.dx like djdx must. The most general integral is obtained by adding to the former a constant C.lt.?. (#).j&.f&.(l). at : present at any rate. The results of the earlier part of this chapter enable us to write down at once the integrals of some of the commonest functions.230 DERIVATIVES AND INTEGRALS [VI The second question presents no difficulties... is much more difficult. The operation of forming ty (x)from (x) we call integration. known as the t&. solution of the problem then another. then we call (x) an integral or integral function of &.. = I cos xdx = sin x f } I sin xdx cos#. the form $ (x) + C. selves than do the 128.). If ty (x) is the derivative of ty &.(x) +C The practical problem of actually finding (a?) is a fairly one in the case of any function defined by some finite com simple The converse problem bination of the ordinary functional symbols. for any value of all possible solutions are comprised in the constant C.f&. The practical problem of integration. J J These formulae must be understood as meaning that the function on the righthand side is one integral of that under course the sign of integration.gt. and that This follows at once from 126. DEFINITIONS.lt. ..lt.lt. (3) &.gt. It is that if (x) is one &.gt. The nature of the difficulties will appear more clearly later on.
!_! when x is negative. The proof of this is too detailed and tedious to be inserted in this book. 4) that I/a cannot be the derivative of any polynomial or rational as fraction. where the ambiguous sign these formulae hold for is to be chosen so that +x is positive : all real values of x other than x = 0. the function on the righthand side of this it is defined so far only for : is Next suppose x negative. Then Also defined by what precedes.. that In this case the formula becomes meaningless. functional symbols corresponding to them. where the properties of F(x) are investigated systematically. is only to be expected. F(x) independent of any of the classes of functions which we have considered yet. . that is to say incapable of expression by means of any finite combination of the will . and we shall call equation the logarithmic function positive values of x. and so log ( x) d^ so that. IX. .127. The formulae (2) and (3) may be united in the formulae log 101 .. This function F (x) is certainly not a polynomial or rational function and it can be proved that it is not an algebraical function. For the present we shall be content to assume its existence. *) (3). It can indeed be proved that is an essentially new function. since we have seen already (Ex. . 128] DERIVATIVES AND INTEGRALS is 231 There in which m= however one case of exception to the first formula. Then we shall write (2). XLII. That there really is a function F(x) such that D x F(x) \\x be proved in the next chapter. unfortunately but some further discussion of the subject will be found in Ch. 1. Suppose first that x is positive. x is positive.(4).
atfc p + a^(c + . All the general theorems of 113 may of Thus we have.gt. 129.. f^L =arc tana?. These theorems enable us to write down at once the integral of any function of the form S A v fv (x\ the sum of a finite number In of constant multiples of functions whose integrals are known..= + arc sin** y(l#r) . (3) the five most fundamental standard forms of the Integral Calculus.x + a.x) last of the if formulae that log. particular we can n write down the integral n+l of any polynomial: n thus r I (a. It is not really necessary.. .232 DERIVATIVES AND INTEGRALS of the properties of log [VI will The most fundamental Ch. . and to x &. log xy = log x + log y. to begin with.. first of which the second an obvious deduction from the and third. assumed. is = . + a n ) dx = . + a n x.. . (1). that the arbitrary constants are the formula (1) asserts that the sum of adjusted properly. .. the formulae . and in either case to 2 log x equal to 21og# if Either of the . (6).log a?.. . to assume the truth of any of these formulae but they sometimes enable us to write our formulae in a more compact form than would otherwise be possible.xn 1 + . J 1 +x* J f^. 0. l (2).. \ formulae (4) is therefore equivalent to the formula dx = are The five formulae (1) viz.. x which be proved in IX are expressed log by the equations log (I/a?) 1=0. Polynomials. Thus of f(x) and any integral of F (x) is an integral of Here it is any integral /(*) + ?(*). * See 119 for the rule for determining the ambiguous sign.r2 is x &. .. .lt... of course. for the purposes of this chapter.. To them should be added two more. It follows from the 2 log ( . course also be stated as theorems in integration.
lt. so that (x) is an integral of ty (x). b . In particular. all the terms in There remain the terms for hypothesis.O)P./&. we obtain f J ~ 1 a^^ * ( Thus. that all the coefficients in a = R (x) are real. viz. I J x a.a) p We can at once write down the integrals of the polynomial and of all the other terms except those for which p = l. We R (x) can therefore write down the integrals of for which is p complex. and in particular. and write if we take f(x) = ax for f b. In order to deal with these we shall introduce a restrictive which p = 1 and a = 1 and a is real. &. vol. After integrating polynomials natural to turn our attention next to rational functions.gt. = log I x a I. Chrystal s Algebra. since .gt.128130] 130. jugate a = 7 &. A pl I (x  a)P~i complex ( 117). which p = 1 present rather more It follows immediately from Theorem 113 that (6) of The terms for difficulty. for example.lt. Then if 7 + Si is a root of . the expression of to of multiplicity m. where a and &. pp. . viz. so does p j(x A A p /(asa) p occurs in where A p is conjugate . 1519. Q if (x) 0. so is its con and R (x). DERIVATIVES AND INTEGRALS 233 a number of terms of the form Aj(x . Let us suppose (x) to be any rational function expressed in the standard form of 117. f J dx ax + b j = + a log \ax I. Ap a partial fraction . as the sum of a polynomial II (x) and it is Rational R A whether a be real or dx .(&) F (x) b are real. are explained at length in treatises on Algebra*. i. and ty (x) for F aa (x)./&. Functions. if a is real. by means of which the * This follows from the nature of the algebraical processes partial fractions can be found. for example. and which See.
R (x) in partial fractions. suppose F {f(x)} to be . in virtue of the equations have Ix (6) of 128 and (4) above. B. and we are thus enabled to write down the integral of any real rational function. we obtain }.234 Thus. a number of rational functions of the type of A pIt only 1 1 .lt.". . (x  X) 2 + dx = log . c B in terms of being where A= ac 62 If in (3) we D = aB bA. so will a is term (X /**)/(# 7 + ^0 i and the sum of these two terms This fraction is in reality the most general fraction of the form Ax + B where A. b* &. b. and a number of inverse tangents. ac. a. ^. The integral of any such function is composed of the sum a polynomial. and log \f(x) . the two forms.X) + /i 2 2 And.a)*(x if 1 a number of logarithmic functions. remains to add that a is complex then the rational function just written always occurs in conjunction with another in and a are replaced by the complex numbers conjugate to which A them. we obtain (5). These two formulae enable us to integrate the sum of the two terms which we have been considering in the expression of R (#). The reader will easily verify the equivalence of 7. if all the factors of its denominator can be deter mined. we = 2o arc tan f /#X\ j . of if DERIVATIVES AND INTEGRALS a term (X f /u)/( [VI x . i and if we further suppose that f(x) = (x {(x X) 2 + 2 /i . and that the sum of the two functions is a real rational function. the formulae which express X.7 Si) occurs in the expression ".
ft . may be deduced from the [The form of the fraction corresponding to a facts that as x^aJ 4.  1) 8 (#*)* 8 (x i) __ _ a 8&. Prove that D (where X=ax2 +2bx + c) if A&. 1.gt. 0. a(ax+b) 3. In the particular case in which ac = b 2 the integral . A D p. and P(x) of lower degree than Q (x\ then the summation applying to all the roots a of Q (#) = 0. A and D having the same meanings as on 2. Calculate dx ^ I [The expression 4 (^72 in partial fractions is I) 2 (a.a + 2 . . If all the roots of Q (x] \ are real and a is is being simple roots.130] DERIVATIVES AND INTEGRALS 235 Examples XL VIII.gt.7\ H .0. 234. where and the summation applies to 5.Clog x.. all r roots /3 of Q (x} =Q other than a. and .fi) 8 (x+i) and the integral is 6. is a log I ax + b I. Integrate x (xa)(xb}(xcY .lt. is if A &. and is P (x] of lower degree than  a double root. Show that if the roots of Q(x}=Q are all real and distinct. the other roots Q (x\ then the integral A/(xa) + A \og\ x.
R : such a function as 2 px 4. XIV. y) dec.236 7. and is fairly Dr Bromwich s tract Elementary Integrals (Bowes and Bowes. See the author ". It is not part of the purpose of this book to go into practical problems of integration in detail. DERIVATIVES AND INTEGRALS Prove the formulae : [VI 131. i. We have to consider the problem of integrating y. q + V(a# + 2foe 4. 1911).. second ed. y) is any rational since (Ex. function (x. than directly as The integration of functions of a single variable". solved algebraically. 6) the form is only apparent.lt. This does not (Cambridge often happen in practice. Bertrand s Calcul Integral.*J(aa? f 2bx + c) px + q is far more conveniently regarded as a rational function of x and itself * the simple algebraical function *J(ax* an algebraical function of x.c) . where y is an algebraical It is however convenient to consider an apparently function of x. 1915). In simple cases (as in Ex. . The 130 gives us a general method by which we can find the analysis of integral of any real rational function R(x\ provided we can solve the equation = 0.. generality of this The choice itself an algebraical function of x. more general integral. I R (x. partial fractions naturally If the equation $(#)=0 cannot be 132. No. s tract Tracts in Mathematics. then the method of fails and recourse must be had to other methods*. pp. y) is convenience of this form is in fact dictated simply by motives of R (x. times prohibitive. The reader who desires fuller information may be referred to Goursat s Cours d? Analyse. 5 above) the application of the method &. viz. We naturally pass on next to the question of the integration of algebraical functions. + 2 ^+ c). Algebraical Functions. vol. where The greater function of x and y.2(#) In more complicated cases the labour involved is some simple. and other devices have to be used. Note on the practical integration of rational functions. 2. second edition. 246 et seq.
R.$&. y R 2 (t). and is of is Its application to the problem immediately under consideration obvious.lt. where reduced by the substitution x=tf to the integral of 2tR(t 2 ). where f (t) is any f auction of a new variable (t).(t)} RJ(t)dt t the latter integral.gt. if it is a rational function. 134. which it may be convenient to choose multiply by . con be would carry us beyond our present range to enter upon any general discussion as to when it is and when it is not possible to find an auxiliary variable t connected with x and y in the manner It indicated above. say x = Ri(t). in other words that the graph of y. This method of i. This equation supplies us with a method for determining the of integral of ^r (x) in a large number of cases in which the form the integral follows: put t is not directly obvious. for example. to the integral of a rational function of t. stated as a rule as x=f(t). Let us suppose by an equation of the form ax2 + 2hxy + bif + Zgx + 2fy +c= . It follows DERIVATIVES AND INTEGRALS 237 Integration by substitution and rationalisation.(x) then {/}/ ^= {/( )} (1).130134] 133. extremely R R by rationalisation. Thus denotes a rational function. considered as a function of x . to of ty {f(t)}f (t) express the be found that the function of t which we are led by the application of this rule is one whose This is always so. y) and dx = JR (Rtf). t. that x and y are connected Integrals connected with conies. integral can easily be calculated. is the integral of (\/x). from equation (3) of 130 that if \^(x)dx=&. integration is called integration wide application. then JR (x. f and determine (if possible) the integral It will often result in terms of x. It may be . We shall consider only a few simple and inter esting special cases.e. being that of a rational function of calculated by the methods o/ 130. and it is very often possible to choose the relation between x and t so that this shall be the case. If we can find a variable t such that x and y are both rational functions oft.
it assumes the form expressed let a conic. 2 ) .^ x) = 0. ~dl~ /x_i_A\". Suppose in particular that \~r/ 2 A J Y (CIX p 2tOX J~ G ) It will be found that. . and y 7) = Y.". if we put y+XfJa=t. If the relation between x and y is in terms of and F. 2 2 where F=h + bri+f. application to an important special case. a&.gt. (G^_Ft}_ ~ 2 * Hence the process can be carried out. X aX + 2hXY + 6F + 2GX + 2FY = 0. be found that expressed as rational functions of t. so that ( dx = 7 y jhx + by+f 2 ^ r ] a+2ht + bP dt ^r . ". It will then G = a% + hrj + g. 2* ". When follows h2 &. x f = X. of rationalisation described in the last section The reader should hx verify that + by +/= . f fly.".2 and so fdx _ f dt (1). n where ^ . ab it is is The conic in some ways advantageous to proceed as a hyperbola whose asymptotes are parallel to the lines f Zhxy + by* = 0.238 is DERIVATIVES AND INTEGRALS [VI Suppose that (. we obtain \". and it is clear that as rational functions of x and y can be calculated fr6m these equations We shall illustrate this process by an t.gt.". b (y = we obtain /JLX ax 1 or say If we put y t.lt.i (a + 2/rf + fo &. rj) is any point on the conic. . be so expressed. Oj. ~~ ". In this equation put and F can both be and therefore x and y can X 2 (g + ffi) ".gt. The integral y 2 = ax2 + 2bx + c. .0.fix) (y . 135. the actual formulae being _ &. Y=tX. ~i // *v ".
gt. we have f (\x + p) dx ) V In the _ ~ \ V(f . &. The It is integral I r. 0. c= a 2 we . 119). or a=l. these formulae should be associated the third formula With dx (3). J \/(ax be integrated sections.gt. (\b/a). dx If a last integral positive we put x \Ja a may be positive or negative. + (&/V0 = when we obtain t&.K  A for a and It thus appears that in any case the calculation of the integral may be made to depend on that of the integral considered in 135. in all cases by most convenient This integral can xr^: dx. when we obtain x JA put dt _1_ f ^a)J ^. c=a 2 . : The formula 136.134136] If in particular DERIVATIVES AND INTEGRALS a=l. X. which a &. is If a is negative we write b 2 )/a. equations whose truth may be verified immediately by differentiation.(b/^A) = t. N a f / ". &. 26uHc) means of the results of the preceding r + to proceed as follows. it will hardly the reader (3) appears very different from the formulae (2) be in a position to appreciate the connection between them until he has read Ch.lt. Since \x + p = (X/a) (ax + b) + //. and that this integral of the three forms dt may be reduced dt dt to one or other . 6 = 0. X6\ f "..lt. corresponds to a case of the general integral of this section in which if a then the integral is arc sin (xj a ) In (3) it is supposed that a . where tc = (ac . 2 9 . \ In practice we should evaluate the general integral by reducing (as in the next section) to one or other of these standard forms. (cf. 6 239 obtain = 0.
f (x) F can be ty (x) where Then (*) dx = xx". DERIVATIVES AND INTEGRALS The integral find [VI I(\x + n}*j(ax +2bx+c)dx.240 137.0 then a2  tf 2 ) + a2 arc sin Or/a).gt. It follows at once from tion of a product proved in 138. . /(#) we obtain a lydx = (ax + l}y. that (x) integrated.gt. is the second derivative of a known function (x) = x^r (#). &. by parts.( and the last integral may be reduced to one or other of the three forms In order to obtain these integrals it is convenient to introduce at this point another general theorem in integration. is Integration Theorem (3) of 113 that jf (x) F(x) It dx =f(x) F (x)  jf(x) F (x) dx.X 0) dx = X X () X Taking We can illustrate it the working of this method of integration by applying to the integrals of the last section.[ .lt. example. .gt. % (x). 2 In exactly the same way we /". 1 Prove that a &. may happen that the function which we wish to integrate is (x) expressible in the form for f (x)F(x\ and that (j&.f&. 0) dx = XX 0) . if Examples XLIX. and we have seen already 135) how to determine the last integral. The theorem of integration by another way of stating the rule for the differentia merely parts 113. so that (fo( (ax+fyy  + acb 2 [dx   . Suppose.
. bx \ c Ay J ~f b&. where m is any rational number. p+q 1 is an (ii) . that if a &. The integral (i) I (1 if +x^x^ dx. by the substitution ax+b=y 10. (ii) by the substitution by writing (#+a) .) is Show n . .(x) 3 /2 to the integral of a rational function. 138] DERIVATIVES AND INTEGRALS . Integrate . and (iii) if where s is the denominator of q put 1 +#=*. in three ways. 3.gt. viz. where p and p is q are rational. (Math. by multiplying numerator and denominator by J(x + a) . 6. and 1. [In case (i) put x=u s ./{( a) (b . or 7. by means of the substitution 2x + a + b = %(ab} (t 2 + (l/t) 2}. Calculate J ^ where by the methods of the preceding sections. Prove that ". where s is the denominator of p and in case (iii) put where s is the denominator of p.137. m (x+a) =t and t 4.x]l(x = t 2. that 130 and 138) fdx _ax + b J y 3 (in ~ fxdx ". (ii) by the substitution . 241 by means of the 2.x}} and J{(b . Show. y(ax + b)}dv reduced. verify that the results agree with those obtained 135 arid Ex.] an integer.a. . Calculate the integrals / J vv x~) U( a x )dx 2 2 J substitution in #=asin0. in case +x=xt8 if q is an integer. J (x + afi* + (x _ a) / F". 1899. and verify that the results agree. viz. 9. Trip. ^ 16 . (ii) integer.x}l(x a)}. by in three ways. (x}F(x}dx=f (x] F(x}f( X ] F (x) + dx and generally ffn(x)F(*)d*fW(s)F(x)/**W^ 11. b then 8. by the notation of means of the substitutions ax + b=l/t and #=!/M. (iii) by integration by parts. can be found in three cases. (b a} and (iii) by the substitution #=acos 2 + &siri 2 0. (i) 5.a for x and verify that the results agree. Prove.gt. that / R{x.J(x + 6). (i) \x(x+a) m dx. Find a substitution which will reduce I . Calculate viz. to the integral of a rational function.
The general integral I R (a. arises is that of the integration of a function of the form is a rational function of x. 39).] 2 2 ) ( ). 17. . If (^2 +3/ 2 ) 2 =2c2 (^/)then 9 2.. or. 16. Ex. integrals of three different types (i) . G into partial fractions. [In practice it is often most con venient to calculate a particular integral of this kind by a formula of reduction 13. the same thing. J(ax+b\ *J(cx\d}} dx can be reduced to that of a rational function by the substitution 4*= 14.JX) The only new problem which where A.242 12. Misc. We suppose that R is a real function. y(xy}*=x. 2?. are rational functions of x. When we may In the first place there may be integrals of the type /: . The most general integral. y = l/{t (1 .1)}. It may therefore be reduced to the form The P Q are poly (A + BJX)(CD.*)}. function. y} dx. what is G And the integral (2) ~ dx can always be evaluated by splitting up do this.] I The integral R {#. (6/a){* + (l/0} (rf/c) 2 ( (I/OP Reduce I R (x. DERIVATIVES AND INTEGRALS The integral / [VI to xm (axn \b} l dx can be reduced the preceding n integral by the substitution ax =bt. arise. where y 2 (x y} = #2 to the integral of a rational .. where y z =axz 134. and subject of integration is of the form P/Q.y\ when we obtain x = a*t (t* + a?)/(t* + a4 y = a?t (t 2 . where Fv A / .] (1 (a) Reduce the integral 2 same way when ).. where X=y =ax + Zbx + z i c.. where nomials in x and *]X.a 2 )/(* 4 + a 4 ). y) dx. of the type considered in the special conic y2 =a and associated with (1). * = * s log xy J 139. .r = (b) +# =a 2 Or 2 3/ 2 [In case (a) put xy t: in case (6) put 2 +y 2 = t (x . [Putting y=tx we obtain in x= lj{t the 2 . 15. (cf. QIJX.
The cases in which ra=0 or m = l have been In order to calculate the integrals corresponding to 136. on carrying out the substitution. a. &. y are constants whose values may be easily calculated. &. fJ so that fji and v are the roots of the equation This equation has certainly real roots.gt.lt.2+ a J/ t^* s . +& (6) of these integrals is rationalised t by the substitution which gives dt f j (at* + /3) v/(y* 2 + 5) ~ = f J /3 du + (ad  y) V? 162 . that the integral assumes the form (5) R The second tdt l i ( &. In this case (cf.. when we m (ii) In the second place there may ( be integrals of the type / where p is real.+ v) + c = 0.+ & . and v fulfilling our requirements. We shall confine ourselves to the simplest case. that in which all such roots are simple roots. for it is the same equation as and it is therefore certainly possible to find equation (1) of Ex. . 12 real values of p. we obtain a relation between three successive integrals of the type (3).138. we can calculate in turn its values for all other values of m. 130) a pair of con jugate complex roots of G gives rise to an integral of the type Lx+M In order to evaluate this integral we put x (5). where p and v are so chosen that a^^b( . If we make the t substitution then this integral is reduced to an integral in of the type (iii) Finally. /3. 139] DERIVATIVES AND INTEGRALS 243 where m is disposed of in larger values of a positive integer. XLVI. m we observe that where that. It will be found. ^TZ xp = \\t (3). It is clear integrate this equation.gt. there may be integrals corresponding to complex roots of the denominator of O. As we know the values of the integral for m = and 1.
but then the For further information be reduced by the substitution ax + b = t. ^ where 2 ?/ J (% &o) y = a^ + 26^7 + c.e. 12 and the author s tract quoted on p. is rationalised 7. pp.244 DERIVATIVES AND INTEGRALS [VI = 1/w in the first of the integrals (6). 1911).] elegant but straightforward than that explained in 6. Show that I .lt. Trip. by the substitution #=(l+y (#+l)dk 2 )/(3 y 2 ). may be &. Show that the integral dx JxJ(&. Prove that C I ^ si w &.lt. (Math. and may therefore be calculated in the manner just explained. + c)l(xp) that dx + c) where A = p 2 + 2fy less dy ( 2 J V(ty /*) is + c. dx Evaluate f dx f dx +\y /(*+l)V(l+2**^) 2. Bromwich. Examples L. Greenhill: see his A Chapter in the Integral Calculus. p.) Calculate * The method integral may of integration explained here fails if alA = bjB. viz..2 or other of the forms T to log 2 according as a^ + 26^ + c is positive and equal to y 2 or negative and equal 2 5. j(xp] *J{(xp}(xq}} 3.gt. seq given by Sir G. 331 etseq. i. et . 236. Grundzuge der Differ entialundintegralrechnung. it is transformed into Finally. i. 2 ) = u. concerning the integration of algebraical functions see Stolz. vol.lt. [This method of reduction 139. = _A_ qp O / //y \/(\xpj ch ^n /&.\ ) If a^ 2 + cA 2 = . if we put t an integral of the second type. .lt. Elementary An alternative method of reduction has been Integrals (Bowes and Bowes. 2  Show by means of the substitution y = J(ax * \ 2J}x &. 1911. pp.v then J (hx + g] J (ax^ + c) r \fv [_ agx J* expressed in one 4. = acb 2 . by putting u/^(y+du 1.
The equation satisfied + 3 + 2=0.139141] 8. This reduces the integral to by p and v is is appropriate substitution tdt The first of these integrals may the second by putting I/J(9t 2 . . ...)#} and the integrals of these terms can be written down at once. 1911. .22^ + 19) (J/^A. Trip. n. y} dx.)#[. ri. !_.] 9. where (p.] 140. be rationalised by putting /^(9^4) = M and Calculate . where y 2 = ax 2 + Zbx + c.. 134. . We few classes of transcendental functions whose integrals can always be found. DERIVATIVES AND INTEGRALS 245 Calculate #2 + 12# + 8) v/(5^ + 2# [Apply the method of 2 * 7) 139. b.. are positive integers and a.. m.(%t+ l)/(t + 1). any real For such a term can be expressed as the of terms of the types sum of a finite number a cos {(pa + qb + . i/=l. and the x = . . q) is is ration alised t= (xp)l(y + q\ y*=aaP + 2bx + c.. sin m a# cos n bx sm n bx. /3sin {(pa + qb + . Show that the integral by the substitution I # (#. Transcendental Functions. so that /x=2. can always integrate any function which is the sum of a finite number of terms such as 141. . We A cosm ax where m. Owing to the immense variety of the different classes of transcendental functions. shall consider in order a Polynomials in cosines and sines of multiples of x.) 10. [The substitution t=(xp)\(yq} conic integral is of course also rationalised : any point on the by the cf.. the theory of their integration is a good deal less systematic than that of the integration of rational or algebraical functions. numbers whatever._ x + 5) x /(7# 2 ..4) = v.
Integrate by any method cos ax cos bx. on making the substitution cosx=t. ^here . 2.5 sin 3# + 3 sin 5. cos ax sin for. by a process similar to that of the preceding paragraph. x cos 4#. I It we can always calculate . sin 3 ^ .cos x.17 sin 7#) dx  5 ~ ~\~$ cos x + 4 s cos ^ x Bff cos 5 ^ + 112 cos ^ x The integral may of course be obtained in different forms by different methods.V sin 3x\ we obtain (7 sin x . For generalise integrals. I x n cos xdx= = xn sin x xn cos x n \ xn xn ~ l sin x dx. I x n sin x dx the +n ~ l \ cos x dx. cos . 39). cos bx.r. . cos 2 . to I 6 (4Z .1 ) dt = f cos 7 x . . . for example. Multiplying these two expressions and replacing sin by 2 (sin 5. and clearly integrals can be calculated completely by a repetition of this process follows that if whenever n I is a positive integer. cos ax. \ xn sin x dx and associated The method of integration by parts enables us to the preceding results.r cos 2#.f cos 5 ^ + 1 cos 3 x . / For example sin 3 x cos 2 %xdx= I 4 (4 cos x 4 cos 2 x 4.17. cos 4 x. P (x. [VI In this case we use the x = (3 sin x . sin ax. 2 3 Integrate sin . Ex.] 142. cos 6 ^7 sin 7 . sin ax sin bx.246 Examples LI.r cos 2# cos 3. we can calculate dx.1) (1 cos 2 x] sin xdx^ which reduces.37. differ It may be verified that this expression and that obtained above only by a constant.) P is any polynomial.8t* + 5Z 2 . sin bx. cos 2 Zx= % (1 + cos 4#). [In cases of this kind it is sometimes convenient to use a formula of reduction (Misc. formulae sin 3 DERIVATIVES AND INTEGRALS 1. The integrals I x n cos x dx.sin 3#).r. cos3 %x sin 2 3. xn cos ax dx and x n sin axdx n is a positive integer and so.
J#.lt. 1 ^ 2 cos 2 a.$ sin #&. fche substitution 143. according as a2 2 2 reduces to a constant multiple of that of sec \x or cosec \x. I tan x sec xdx= sec #. 6 2 . 2t doc 2 so that the substitution reduces the integral to that of a rational function of t.gt. I 1.? n snl where Rational Functions of cos # and sin x. 3 cos A . a. /sec2 . Integrate x sin x. cfa= P cos*. then as x varies from to n one value of y also varies from 111 Show b 2 } sin also that x ~ v/( y a. Prove that lx n cosxdx=Pn cos # + &. a?. and its value written down.] j is log  tan (\ir+%x )  .] 3.lt. / cosec x dx = log tan \x  J. Deduce the forms of the integral when af b may at once be &. a third form is \ 2. = . #3 sin3 2. is negative. Show that if y is defined in terms of x by means 2 of the equation =a (a + b cos x}(ab cos y) where a is 62 .b cosy dx _ sin* ~ a + 6cos x dy sin?/ a.b cosy . I cot # cosec xdx= cosec . \iax\.r. as the results follow at once from of 130.141143] Examples LII.xdx= log cos x\. I cosec 2 a? efo.cot x. . x sin 2 x cos 4 . icoto7c?.v=logsin A* I. 119 and equation (5) use a substitution. DERIVATIVES AND INTEGRALS 1. 4. Prove that sec  sec xdx = log x + tan x \ . may be expressed in one or other of the forms arc tan t j If a2 =6 2 then the integral 6 2 or a 2 b2 where *=tan \x.gt. positive and to TT.4. The integral x and sin x may be calculated by of any For tan \x t. Find polynomials P and Q such that a?) I{(3*1) cos# + (l 2#) sin 3. &. 247 X* sin 2 x sin 2 2#. where a + b is positive.#c?*=tan*. Examples LIII. rational function of cos l 2 .gt. but there is no need to [These integrals are included in the general form. Show that the integral of 1 /(a + 6 cos*)..r. 2 a &. [Another form of the first integral  log (1 + sin #)/( I sin x) .
and log a.1 57).lt. loglaf ficosx + ysmx a +A /". when we obtain 2 xdx f dt a + 2b tan x + c tan 2 a. inverse to/ x is the function (y\ where For on making the substitution y =f(x) we obtain \&.0\ ^ cos arc ^V^O &.y sin b cos x + c sin x in the 6. /af2i /sec Integrals involving arc sin a?. form J(b*+c 2 ) cos a). . Then the integral is ux+v 7.^ / 2 ).gt. [Express # )/( a f /3 cos # 4.248 and deduce that if DERIVATIVES AND INTEGRALS &. 7 O ^(a* 62 ) 7. JB=^(a simply by putting C=b siii ^?). A=^(a + c). = .] sin . integrals of arc sin y and arc tan y Integrals of the form I P (x. ~ \ I ff Show 5. X". that this result agrees with that of Ex. It is easy to see that if we then we can always find that of can find the integral of y =/(V) &. x+c [Determine v so that a + b cos #+ c sin x= Af /* (a + /3 cos ^ + y sin #) + j/ ( /3 sin x+y cos . where but the integral may be calculated more tan#=. \ log x\dx = x (log 1).lt.lt. r \ arc sin xdxx arc sin x xdx xdx x arc tan a? x log f \ scdx ~jj J J Vv*r I ^ = x arc sin x + V(l . r I arc tan ^^ = y J 1 + = a? arc tan a? a. log x} dx. dx J a + /3 cos A.r) Integrate (a + 6 cos A.lt. arc sin x) dx. integrate (a? Show how to 1 /(a + b cos x + c sin x). arc tan a?. The of the inverse sine and tangent and of the logarithm can integrals Thus easily be calculated by integration by parts. The reader should evaluate the in this way.r).. [\i x &.l&. [The subject of inte form l/(A+J5cos 2#f(7sin 2.** /a cos x + b\ I I \a + 6 cos #/ 3. I P (x.+ y sin a? 2 : J Integrate l/(acos #f 26 cos^?sm^ + c the : gration may be expressed in c].lt. p. 144..(y)dy=\ xf (x) dx = xf(x)  \f(x) dx. J log (1 + a? 2 ). ir then 1 // dx a f o cos # .
44. which can be found by the method of In the case of the second form we have to calculate a number I of integrals of the type xm (log x) n dx. 44) is the graph of a continuous curve y = (x) which lies wholly above the axis of x. ONPP and has received the most careful mathematical analysis later on we shall return to it and explain precisely what is meant by : . for I a polynomial. and h being being &. N* The reader is of course familiar with the idea of an area . the point (x. Suppose that P Q PP (Fig.gt. . Making the substitution x = sin y. Areas of plane curves.lt.y + &). we obtain 142.__ and it is shall always arrive finally at the evident that by repeating this process often enough we complete value of the integral. y) and P the point (x + h. P O NI N Fig. We have to calculate a number of integrals of the type xm (arc sin x) n dx. P either positive or negative (positive in the figure). can always be calculated. and This idea we in particular with that of an area such as It is indeed one which needs shall at present take for granted.143145] DERIVATIVES AND INTEGRALS is 249 where first P form. Take the example. Integrating by parts we obtain I X (log X) _. 145./&. I n m y sin y cosydy. One of the most important applications of the processes of integration which have been explained in the preceding sections is to the calculation of areas of plane curves.
Moreover = lim h**Q {#.lt. an arbitrary constant.gt.250 ascribing an DERIVATIVES AND INTEGRALS area to .lt.&. we should of course deter = 0.J&. the area not however necessarily true in general. which we suppose Then the area required is (x). (#). This involves 3&.$&. Also &./&.gt. &. (h) and \ (h) ^0 as h 0. continuous function. ) &.gt. = (NN RP) + (PRF) = (x) + (PRP ).lt. [VI present we shall For the such a region of space as ONPP assume that any such region has associated simply with area.gt.gt. f from P to P. Thus We area are thus able to formulate a rule for determining the ONPP. where x^ is the abscissa of P^ If the and the area would be (x) would be negative. it a definite positive number (ONPP ) which we call its and that these areas possess the obvious properties indicated by common and so on.gt. because it is not neces 44 a) that the arc sarily the case (see for example Fig. e. that (PRP ) + (NN RP) = (NN P P).lt. &. . so chosen that (0) = 0.gt.lt. \(h)+0 as h*. ( ONPP \ all this for .lt. (x). sense.\ &. NPP.lt. Moreover. Thus we have is always less than \h\\ (h). As the figure is drawn.lt. x + h) ( &. granted it is it obvious that the area a function of x we denote For (x) by &. the absolute value of * (x). This is is less than hk. (x) is a continuous function. area is the integral of the ordinate.gt. Taking is (N.&.I&. () + M (h)} = and the the ordinate of the curve is the derivative of the area.0.. where \pQi)\&.lt.E&. the integral of $ (x). = (NN P P) hcf&. mine the constant so that * (a?i) curve lay below the axis of x.g. From this it follows at once that (x) is continuous.lt. Calculate &. If it were the area iV^PP^ &.gt.gt.lt. &. But the area PRP should rise or fall steadily PRP PP where \ (h) is the greatest distance of since (as) is a any point of the arc PP from PR.J&. which was wanted.
which we may denote by S(x). as did the corresponding as PP sumption about tinuous.gt.gt. Calculate the area of the segment cut off from the = xi\^a by the ordiuate x=%. is in reality a more difficult one even than that of an area.S (P)} = 0. Indeed it is not possible to proceed further.lt.lt. and the length of the arc which it. Hence lim If also (PP /h) = lim V{1 that we assume lim{PP }/PP = l. 146] 146. Answer the same questions for the curve a ?/ 2 &. does not suffice for our purposes. + &)&. where {PP is the arc whose chord is PP . Calculate the areas and lengths of the 145 146. Now and where ( &= lies &. continuously. between x and x + h.e. but less so in such a case as is shown in the smaller figure. In fact the assumption that (Fig.gt.] } dx. We that lim {8 cannot even prove that S (a) is con This looks obvious (F) . without a careful analysis of precisely what is meant by the length however easy of a curve. with any degree of rigour./&. (X&. so that that the curve has a length leads to the equation &.145. i. The notion of a curve. areas. showing that the length of the arc is 3. 1.lt. we obtain the result S and so (x) = lim {S (.*?) 8 = J V{1 + 2 W&. DERIVATIVES AND INTEGRALS 251 of the length Lengths of plane curves.lt. It is to see what the formula must be. =^3 . 44) has a definite length. parabola y Examples LIV. (so) = hfi (f). other than a straight line. Let us suppose that the curve has a tangent whose direction varies Then the assumption (#) is continuous. bounds 2. enough in the larger figure. by means of the formulae of circles . {8 (x + h)S (x)}/h = \PP\lh = (PP /h) } x ( (PP J/PP ).
&. As t varies from is [Putting towards the loop described once.e.gt. &.*!)*.// to 1} then the area of the region t varies from t bounded by the corresponding portion of the curve. . #=asin2. Also . the method. 6 that the coordinates of any point 7. y=asint is (Math. Thus the area of the loop is fa 2 . [Here &. where if x steadily increases as &. y=to we oo Find the area of the loop of the curve x*+y*=Zaxy.] 6.lt. result of Ex. the axis of ar. axis of Find the area bounded by the curve y sin x and the segment of the x from #=0 to #=2?r. y=3a 2 /(l + ).cos #. parallel on the curve can be expressed as in Ex.lt.252 4. as t varies from t to t it P moves in the same direction round the curve and returns after a single circuit to its original position. Suppose that met by any in P terms of t. Apply the given by 8. initial Show final and that the area of the loop is equal to the difference of the values of any one of the integrals dx .lt. . and that. 1908.(#)=. Suppose that the coordinates of any point on a curve are expressed as functions of a parameter t by equations of the type x=$(t\ y=^r(t\ Prove that and being functions of t with continuous derivatives. when every part i. dx dy\. this difference being of course taken positively. DERIVATIVES AND INTEGRALS Show 2 2 that the area of the ellipse (#2 /a ) + (//6 ) = [VI 1 is irab. and the two ordinates corresponding to t and l5 is. The explanation of this is of course that between X=TT and x=Zir the curve lies below the axis 5. is twice this. 9. 7 to determine the areas of the curves u &. counted positive. and so the corresponding part of the area The area from .cos x for #=0 and x=1ir is zero. [ t .~V^ which tends to 10. G is a closed curve formed of a single loop and not And suppose to either axis in more than two points.dfy\.) .lt. and the difference between the values of . whole area required. is 4. apart from sign. of #. 3 obtain #=3a/(l+ 3 ). Trip. &. .r=0 to x=ir is is is counted negative in applying cos 7r + cosO = 2 and the .gt. A (ti)A ( ). f 9a 2 * 2 3 2 as ^oc .] Find the area of the loop of the curve Prove that the area of a loop of the curve 11.
r=a + bcos0..l&.? #&. Denoting ax+b. ax + 2bx + c.. F(0 2 }F(0i).] +e cos 6} . e being the eccentricity.lt. is F(t^) by x=acost y=6sin. 1908. a&. . and the two branches which touch at the point by the radius vector = a 2 ( 2 l) 3 2 is r=a. Show that the calculation of the area of the region bounded by an arc of the curve and two radii vectores depends upon thai of the integral 1 1 ^ 1 2 .lt. and show that the area bounded 14.lt. continuity of and tf=2.4^ u3 + 3w 2 2 are independent 2 . #=l.. ^&. . when to x when and to 3xx2 when ^&.1.#&. 1900.lt. (Math.) a. (Math.gt.2. / . . 4) (1 by the help of the substitution e 2 .. where ". Trip.b .valued function of 0. And the length of the corresponding $ (02 ) $ &.b.gt.gt.A/H3Wthe area and perimeter of the circle Hence determine (ii) (i) r=2asin0. where F(6} = ^lr arc of the curve is 2 d0. .) . when 1&. where /(0) is a one.0. DERIVATIVES AND INTEGRALS The arc t 253 of the ellipse given points t=ti and = t2 . show that of x.2. 2. the area between the parabola r=Zsec 2 ^0 and its latus rectum. MISCELLANEOUS EXAMPLES ON CHAPTER 1. by u u lt u 2 and U Q u 4 . VI. Trip. and p the perpendicular from the origin on to the tangent. . (0^. and the areas of the ellipses 1 /r2 = a cos 2 + 2 A cos sin + 6 sin 2 [In the last case (cf. between the F(ti\ where ) 2 2 F(t] = aL/(le sin *) dt.lt. and ^ /r=l + ecos0.lt. we are led to the integral I which may be calculated (1 Ex. LIII. 0=1. a = b. and the length of the corresponding arc of the parabola (iii) the area of the limagon . Discuss the /(#) and the existence and continuity of / (#) for #=0.ecos = 1 0) Trace the curve 20 =(//) + (r/a).] Show that the area bounded by the curve Polar coordinates.. and the radii 0=0 i5 = 2 is r=f(6\ 13.146] 12. A function /(#) is defined as being equal to 1 +x to 2. [This integral cannot however be evaluated in terms of such functions as are at present at our disposal. and (iv) distinguishing the cases in which a&. A curve is given by an equation p = f(r\ r being the radius vector 15.
x &. /3. 1900. Trip. maxima and minima /(#) = 0.lt.cos x\ and Show that in the first IT. and suppose that da/dc=0.sin x . and whether they correspond to maxima or minima.lt.^ *\ #(!#) 4 V^ V when 22. case the condition for that tan a a should be a multiple of : Show that by choice of the ratio X real p. of the graph of the function (ax* + bx + c)j(a x2 + b x + c Exs. &. (cos a .) any maximum or minimum 2 2 5(^s)^ + 4A =0.256 18. Show that if (a.(a . XLVI. is x .(a x 2 + b x + c )=Q and having a nitude.cos x\ and a an angle between a double root 20. sb = sc = %a. and draw the graph of the function. 2 2 2 2 = 1 can greatest acute angle at which the ellipse (# /a ) f (y /6 ) 6 2 )/26}. Trip. Trip. is a lower limit magnitude of their difference. given that 24. Draw the graph of the function 77 COt TTX x x\ _ 23.lt.tan a (1 . and the real roots of x . Sketch the general form of the graph of y. then a root of / (#) = will fall in the fourth interval from /3 on each side.lt. how the paper must be folded to folded over so that one corner just reaches the make the length of the The be cut by a concentric 26. 1907. (Math. /(#) being either of the functions of /(#). y in ascending order of magni tude.] r 21.a) (sb} (sc) A determine Differentiate with respect to c. we can make the difference of roots of \(ax 2 + bx+c)+p. Discuss In a triangle the area A and the semiperimeter s are fixed. = 2 a and b [The equations a + 6 + c=2s. circle is arc tan {(a 2 (Math. from which we deduce that . y) are each divided into six equal subintervals. 25. unless the roots of the two quadratics are all real that in the excepted case the roots are always real. of c. DERIVATIVES AND INTEGRALS The [VI roots of a cubic /(#) = are a. A sheet of paper is Show opposite side.sin x . Show that of one of the sides is a root of the equation the reality of the roots of this equation.sin a) . but there for the any mag and interlace and .) 19.tan ^a TT.) } : [Consider the form cf. 12 et seq. 1. Prove that TT &. /3) and (/3. s(s. &. as functions It will be found that b = c. crease a maximum. What will falls at a point of division? Investigate the be the nature of the cubic in the two cases when a root of/ (#) = (Math. 1895.
r) = l t 2 The curve has a loop branching (2.gt. G is a. Trip. . A are the areas of the two maximum &. and making t tend to limiting values f.f&. Trip. Expressing 1 or 1. is f(a} 0(a) (6) ^( ) =0. 27A 2 and one in the contrary This equation has three real roots if s 4 In an equilateral triangle (the triangle of minimum perimeter for a 4 2 thus it is impossible that s 4 &. 1899. Of the two positive roots one corresponds to a 27.27A 2 Hence the given area) s = 27A . proaches the position [If (2.{ sin a (x) (x.t 2 ){t. two roots are positive and the third negative.lt. x 6. x sin da x^a T. and.lt. then lim / (x)} lirn x*a / f sec a T g sec a.7~ we take 3) as a new . which can be described with their vertices at the origin and their base angles on the cardioid r=a (1 +cos 0). (Math.] The area of the greatest equilateral triangle which can be drawn with its sides passing through three given points A. and the function given becomes ( we put 77 = tg.6# + 3) approaches as the point (x.DERIVATIVES AND INTEGRALS &. . then 256AA = 25a4 (Math. the equation of the curve becomes 2 2 2 2 + 4 ^r})/(r} 2 + 6rj 6). Show is that if &. since their sum is positive and their product negative. 1903. then there &. B. . 30.gt. maximum and one to a minimum. equation in a has three real roots.) 31 . 6. we obtain the .a} 3 r cos a 5 . 257 case.( n) = Qn (#)/( 1 + x?} n + \ (x} where Q n (x) a polynomial of degree Show also that (i) (ii) (iii) (v) all the roots of Qn =0 (a?) are real and separated If /(#). c being the sides and If A.lt.lt.gt. at the origin. If 4J? =0. A the area of A DC. ^ (6) 17 n. we obtain g = (1 . 1 and t=l.J5. y) on the curve x 2y 4x2 4xy+y 2 +I6x 2y7 = ap 29.) Find the limiting values which (x 2 4y f 8)/(y 2 . which corresponds to the two values t= the given function in terms of t. = 1 /( 1+ #2 (x} n. ) then &. 3). Trip. 1907. (Math.) origin.) isosceles triangles 28. &.] If f(x)=. a value of lying 32. have derivatives when a between a and b and such that (x\ \lr by those of Q n _ l = 0. 1896. (Math. Trip.gt.
. & VfUHf WWt I . :  .  Z .. .gt. A* \ft ..P N^&.".gt.&. j(JMP ^&.. U+a : t ". .Vi*5* *Ks&.# . . \4# &. /. . ". &. &.^JF.lt. * . ".lt. i// //U 44U4ili ttUtf .&.gt. flW h4+ ? _ **p*BW* ".gt.gt.lt..
.
sin 2 * ) r^ .arc tana. J (2 sm^a?)(2 + smo7 .(x} + (x}*~ a as #oo. W.// .gt.lt. (#)* then &. . Hog (a f/3. &.x&. for all sufficiently large values of #. then = a\fy(x\ so that (#) + \J/ (#)*().r) J (a + ..(#) (# (.  : &. If &.gt. .lt.] 35.  f . [cQ&xBmxdx rj4 ^7 + sm A/". Similarly we may dispose of the case in which ty (x) is ultimately negative. a large value corresponding to a maximum or minimum of ^r(ar).0. $ (a?). XLIX. 2 2 2 nog(a +/3 ^ J (TT^F^ J ^~ ^ ) .lt.gt. If fy(x}*~l is impossible (Ex. &. &.gt. 13. x /arc ^ ^ . V/(^+^)1 \**+// &. ". A fortiori are the *// other values of ^ (x} small when x is large.lt. and alternative lines of proof.gt. then increases and must tend to a limit I or to &.oo (#).gt. \^ (#).gt.oc as #oo. The simple proof sketched above was suggested by Prof.lt. zero. .] Calculate the integrals dx f I 77^  rfia.QO then (#) *~ then i// (#). 32 the formula /(&)/() 34.!&. and this \//&. (#) . /". /". If (x}*a as x*ao .lt. I (arc sin J /". /* tan # x arc sin x .. #. 4 J cos . see a paper by the author entitled ".gt. If . say positive. 35) unless =0. 38.".p(x}*~a and If i// (#). .258 DERIVATIVES AND INTEGRALS [VI [Consider the function formed by replacing the constituents of the third row by /(a?). then these are the maxima and minima of v//&. For generalisations of this theorem. Show how to reduce fsL J : J + IV (^] the integral of a rational function. &.fe to use Ex. where # &. then cannot tend to any limit other than 36.lt.j&.gt.lt.a. (x)/g +&. Hobson. then ^ (#) + \// (#) is small and *// (#)=0. ($} [Let &.j&. (#) is of constant \/r sign.] 37. in volume 43 of the Quarterly Journal of Mathematics.r) ) = (# # ) ().f&. This theorem reduces to the Mean Value Theorem = x and ( 125) when $(x} 33. surpass all limit.Generalisations of a limit theorem of Mr Mercer. Deduce from Ex.arc J^ sin xdx. If &. ^ (#) steadily . \mx nj inx+n= l/t and [Put . / COS6C Xj(8fSG j **i / aro sec x dx.lt.lt. E. If \j/ which contradicts our hypothesis. so that (^) is small. (#)[Use the formula ( then &. /#f sina? Jl+cos^ I ^ .lt.lt. If (57) changes sign for values of x which If x has (x}.
Show also. that &. (i) Show that [Put #p = dt . 259 Formulae of reduction.\Wk . 172 . 2 ^^ =A: then we obtain I dt 1 t 2 dt and the result follows on integrating is by parts.DERIVATIVES AND INTEGRALS 39.n cos /3. . 3/ 4 /231 61 (TT /iHM/7/r (v) If / = ^n cos /3. integral for every value of n in * turn.r + w/B _ l .gt. &Jn = . I most useful rn in terms when n of f I a positive integer. It is A formula such as this is called a formula of reduction.a.gt.3 + 1 . ! . y ! t and so evaluate the &. We can then express / (x*+px + q} dx f^rr . (ii) Show that if 1^ q = \x* (1 +x} dx then and obtain a similar formula connecting lptQ with /p _i means of the substitution x= y/(l +y).] . v .r cfo? I and / = I . n sin 3^ nJn .24a6^ + 2l6 2 s ) A".r 11 sin ^ rfa7 then /3/ n = o. by (iii) Show that if X= a + bx then \x*XWdx= (32a 2 .
?W&.". 1) (7n (viii) If n = / cosm ^ sin". that the most general function values of x. 1897. (/) where A.260 If DERIVATIVES AND INTEGRALS [VI (vi) In = lcos n xdx and Jn = /sin* a? etc then n If J=  (vii) In = (tMFxdx then (n7 m&.) (xiii) If Imtn is 40. n (Math.lt.p(x). x dx then = [We have (m + 1 ) /m cosm 1 ~ ! a? sin n + ! * f (m . _ (1)". 1 896. p.) (x) If ImHaPooa&Pxdx xm ~ l cosec w ~ l x (m sin x + (n 2) x cos x] . [Multiplying by 20 and . =. &. may such that + 2 = for be expressed in either of the forms A cos ax + B sin ax.] /sinw (ix) 7m n = . are constants. ^ sin 71^7 dx with /m _ 2 then .gt. 1898.gt. i rm + 1 SQ?S* ( nCnDGog^)". n. 41.lt. B.1) + 7m _ 2 . (Math. \ sin n x ^ (cos l x} dx which leads to the Connect first reduction formula. ) (xi ) If / = / (a + b cos #) ~n dx then (xii) If 7n = I 2 (a cos x + 2A cos # sin x f 6 sin 2 x) ~ n dx then U =^. Trip. (J/a^A. n.". Trip. &. all Show ). If n a positive integer then the value of \xm (log#) w dx 1 _ ^Gog^)*.! ..
Determine the most general functions y and z such that y + = 0. %?. should be divisible F F by 50. and that the area of either is a 2 cos /3 sin (/3 /3). The area . See the author s tract quoted on . y.] and z o&. (Math. 6.2 respect to x. (Math.DERIVATIVES AND INTEGRALS integrating 261 we obtain 2 &. c. 236. Show that the length of a quadrant of the curve (#/)* + (y/6)$ = 1 is (a? + ab + b 2 )l(a + b). 46. Trip. is Showthat racosa+flsing + J (I e (Math. Trip. Trip. show that the locus of the middle point of the chord consists of two loops. by .) 48. where b is a constant. and a. 1904. the integral when this condition is satisfied. 1910. is a positive acute angle. 1903. Trip. Trip.* Show that the necessary and sufficient condition that where / and are polynomials of which the latter has no repeated factor. is zero. should be a rational function of x. 1902. +a2 2 = a 2 62 . ) 47. 1 909. ax= I . 43. fF". then dx where PT. (Math. * (Math. 1911. 2 1) = is the equation of a conic. The (Math. projection of a chord of a circle of radius a on a diameter is of constant length 2a cos j8 ..) p.gt. g.gt. 1910. PT 7 of the conic on the are constants. sinasin$ lcos 2 asin 2 y ( ^ I sinacos0 cos 2 asin 2 (Math. at a distance b from the centre. whereto is a constant and dashes denote differentiation with a&.3/=0. Trip.lt. is that f F. Show that the Prove that if (a._ of the curve given . PT P /3 are the perpendiculars from a point tangents at the ends of the chord lx + my + n=Q.) where a 44. Show t hat if will be a rational function of # and only if one or other of ACB 2 and 265 49.) is inside A point A a circle of radius a. Trip. A locus of the foot of the perpendicular drawn from to a tangent to the circle encloses an area TT (a 2 + ^6 2).) cos off if y^ if a rational function of cos x and sin x and only ae+y=0 and determine .) 45. _ . from which we deduce that 42. is TT (1 + sin a) 2/sin a. /.
a and b 121) for some value of x between Hence there is a value f of x.%&. 125) we proved that if f(x) has (a. b) In the preceding a derivative f (x) throughout the interval then where (a.b. or that. b). This we proved by considering the function which vanishes when # = a and when # = b.lt. Let us now suppose that / (x) has also a second derivative an assumption which of course involves (a. 1. /". then where &. and its and this must vanish ( (exclusive of a arid b).lt. first derivative / (#}. Ol &. if f(x) has a derivative throughout h). and consider the of the the continuity function This function also vanishes when x derivative is a and when x = 6 . chapter ( Higher Mean Value Theorems.lt.CHAPTER VII ADDITIONAL THEOREMS IN THE DIFFERENTIAL AND INTEGRAL CALCULUS 147. a + a&.lt. between .) throughout (a.
+ ^ . hn f J \ &. fr .lt.lt.lt. ". the Mean Value The analogy suggested by (1) and (2) at once leads us to formulate the following theorem : Taylor f(x) is or the General Mean Value Theorem. ) \i ii ~~ /(". .. where 2 1.lt.lt. &. } I n I dn &..&."... The proof proceeds on precisely the same lines as were adopted before in the special cases in which n = 1 and n = 2. // a function of x which has derivatives of the first n orders s 6).147] ADDITIONAL THEOREMS IN THE CALCULUS b..lt. ().& If we put b = a + h we obtain the equation which is the standard form of what may be called Theorem of the second order. () + n .x)f (x} (n *^ 1) ! f". 1. then ( f(b) =/() + (6 .".a)f (a) + ~ .^V". consider the function We where Fn (a) =f(b) f(x)(b. for which &.lt.j. I ) I +h f(a + h) =f(a) + hf (a) + f &..* (nl) where &.(&. hn 1 ". throughout the interval (a. and ifb = a then (a) \l?f". 263 a and a+ 2 (6  and therefore capable of representation in the form a). b for which This leads at once to the desired result. (a) + ~^/".().lt.lt. ~ +\ where a&. }/". _(bxT^ J This function vanishes for x = a and x = b] its derivative is and there must be some value of x between a and the derivative vanishes.
&. all real values of is even if #&. by A=2a. 5.  + ( l)ii + xn 1 1 ^+A r a? a* 5 a . If #=.] 3. if the interval (#. The formula / (x + A) =/ (. By applying 1 the theorem to /(a?) = l/#.lt.. leads to the algebraical identity ) + /". of degree r.r) + hf (x + 6fi) is not true if / (x) = 1 /# and 2 &.0&.lt.+ x and 2.. can be put in the form is (x + h) &.0&. mn positive.lt.= . and the theorem ) &. +  (l)".264 ADDITIONAL THEOREMS IN THE CALCULUS [VII In view of the great importance of this theorem we shall give at the end of this chapter another proof. &. Obtain the formula sin (x + A) = sin x+ h cos x  A2 sin x A3 3 .gt. ^.lt. it not satisfied.(x + h) n + as evidently the case.. we can ) verify the result n + 1 or that xn + l .] 6. ^ + A) does not inclvide ^=0..lt. is satisfied theorem not may = ^ iV v/3 [This example shows that the result of the #1 hold even if the conditions under which it was proved are  satisfied.. obtain the result [Since x+h~x~ x* x?~ A 1 A .lt. and similar formulae involving 4. the formula still holds if Show holds for m . the formula and all positive integral values of n and that. not essentially distinct from that given above. and n a positive integer not greater than then also that. /(#) = ff 1/3 then the equation . z cos ^f .] .x xn (x\h] 1 .0 and A/ (a?+^A)= A/(# + 0iA) evident that the conditions for the truth of the Mean Value Theorem arc + A. 1. and supposing positive. (a) + . Suppose that f(x) is a polynomial Then /(n (#) is identically zero when n r. Examples LV.jA). by showing that n &. (.. if m is a positive integer.# is [For/(#+A)/(#)&.lt. Show that wi.+ 5 8 1 2 _ 1 _ h_ + A2 _ ". but different in form and depending on the method of integration by parts. ! the corresponding formula for cos 2n + 1 powers of A extending up to A . #&.gt.lt.# + A or # + A #.^ &(x+hY xn f M W (1) A r\ .lt.
such that )/ (x) enough..JA .+.. we find . a is small.r 2 ly=0. If the root is a simple root. where Show that (1 a) TT is a better [The method of Exs. . so that /( + /*)* 0. and then. show that ^(1 of the fourth order.a + a2 ) does not depend on /(#) = being an and/ are continuous. still. if h is regarded as of the first order of /() is of the first order of smallness. where y is approximately..gt. Show that the limit when Mean Value Theorem is h+0 is of the number l/(?i + l).8. in spite of the roughness of the first. taking = 3/2 as the first = 17/12= 1417. TT.. 10. and Hence But if we apply the original Mean Value Theorem to the function /() taking ^ n A in place of A. the actual root being + h. is in general of the third order.!&. Newton s method of approximation to the roots of equations. has a root nearly equal to a better approximation. 9. n which occurs in provided that /( w + 1 )(#) is [For/(^4A) equal to each of /()+. and the error in taking 7 {/(I)// ()} as the root is of the second order.{^/( 2 +y}} way the equation .. which is quite a good approximation. 710 as/ algebraical equation. that the error in taking the root to be 2 /3 (/// ) 1 (/ /". + y) = 1 + J y . \ K can.147] ADDITIONAL THEOREMS IN THE CALCULUS 265 7. If now we repeat the process.. where 6 n + l as well as Bn lies /+. taking + A = 577/408 = l414215.+ between /w W 1.] The equation sin#=o#..(*+). the argument of every function.gt.// ).. this process to the equation . Let  be an approximation to a root of an algebraical equation/ (a?) = 0.1/12. we obtain is correct to 5 places of decimals... and (1 . +A [We find A= #2=2. when h all is small K for the values of  are considering. find a positive constant we &. the error being Show is where 11. Apply approximation. Then so that A . It follows that in general a better approximation than x= is x which we smallness.. so long 12. which =17/12. By considering in this small. the general continuous.
266
where 6 also
ADDITIONAL THEOEEMS IN THE CALCULUS
lies
[VII
between
and
1.
Hence
from which the result follows, since
to the
1
+^
/("
(x
+ 6B n h) and f( n +
(a?)
1)
(x
+ d n + ik)
tend
same
limit /( n + 1 )(
as h*0.]
3.
Prove that {f(x + 2A)
is
 2/(a? + A) +/(#)}/A 2
(2) of
*/"
as A*0, provided
that/" (a?)
continuous.
that, if
[Use equation
n
)
147.]
14.
Show
the/(
(a?) is
continuous for # = 0, then
where or = /( r )(0)/r! and
15.
ea
() as
#0.
Show
that
if
where
ex
and
T/^
tend to zero as#0, then
=6
,
a1 = 61
,
...,
an = bn
.
[Making
&*0.
is
#.0 we
see that a
=6
.
Now
divide
by x and afterwards make
We
thus obtain
^=6^
and
this process
a + a l x+a&lt;1 x*+ ... (a n It follows that if f(x} necessary. r first n derivatives of f(x} are continuous, then ar =f( )(0)/r !.]
(&gt;
=
may
be repeated as often as
+
+ fx )xn
,
and the
148.
of
Taylor
s Series.
Suppose that f(x)
are continuous
is
a function
all
whose
7j,
differential
coefficients
in
if
an interval
h
is
(a
0.47?)
cally less
surrounding the point x than 77, we have
. . .
= a.
Then,
numeri
f(a + h) =f(a) + hf (o) +
where
&lt;
+
n.
1 "
() +
"/""
(a
+*
A).
(^yj/"
n
&lt;
1, for all
values of
Or, if
we have
f (a + h) Sn = R n
suppose, in
.
.
Now let us R n ^0 as ji^oo
/(a +
A)
addition, that
we can prove
that
Then
Urn S. X
=
=/(a) + A/ ()
+ J/" () + ..
"
When
This expansion of /(a+/i) is known as Taylor s Series. a = the formula reduces to
* It is in fact See E. H. Fowler, "The sufficient to suppose that/(w (0) exists. elementary differential geometry of plane curves" (Cambridge Tracts in Mathe
&gt;
matics, No. 20, p. 104).
147, 148]
ADDITIONAL THEOREMS IN THE CALCULUS
as
267
which
as
is
known
s
Lagrange
Maclaurin s Series. The function R n is known form of the remainder.
The reader should be careful to guard himself against supposing that the continuity of all the derivatives of /(#) is a sufficient condition for the validity direct discussion of the behaviour of R n is always of Taylor s series.
A
essential.
Examples
Hence
LVL
all

1.
Let /(#) = sin
x.
!,
x.
Then
&lt;

all
the derivatives of f(cc}
are continuous for
in this case
values of
\
Rn
hn\n
1 for all values of x and n. Also /n (^c) which tends to zero as 71300 (Ex. xxvu. 12)
whatever value h
may
have.
It follows that
!
3
!
for all values of
x and
A.
In particular
smA = A
,
A3
+
A*
...,
for all values of A.
Similarly
we can prove that
cos# +
A3
.
cos
(x+ A) = cos ^7
2.
Asm ^
A2
sin.r+... }
cos A = l
A2
j
+ ~r~\~
any rational
f
A1
The Binomial
Series.
Let f(x} = (\+x} m where
, )
m
is
n number, positive or negative. Then f( (x} and Maclaurin s Series takes the form
= m (m  l)...(m  n f 1) (1
x}
~n
\
When
2
+....
is a positive integer the series terminates, and we obtain the formula for the Binomial Theorem with a positive integral exponent. ordinary In the general case
m
and
in order to
show that Maclaurin s Series
really represents
is not a positive integer, we must any range of values of x when This is so in fact if n *~Q for every value of x in that range.
m
m (l+#) for show that
!&lt;#&lt;!,
R
and may be proved, when
for
^x
1 if
&lt;
1,
by means
of the expression given above
n ^0 as
x&lt;l
Rn
,
since (1
+ 6 nx} m ~ n
&lt;
n
&gt;
m, and
since
(j z
l
n^ oo
(l
(Ex.
n #)
1
xxvu. 13).
"~
But a
n&gt;m;
difficulty arises if
1
&lt;.r&lt;0,
+ 6n
and
+
B
&gt;
1 if
6n l,we cannot be assured that knowing only that m~n quite large. quite small and (\ + B nx]
&lt;
&lt;
+ 6 nx
is
not
In fact, in order to prove the Binomial Theorem by means of Taylor s Theorem, we need some different form for Rn such as will be given later
,
(
162).
268
149.
ADDITIONAL THEOREMS IN THE CALCULUS
Applications of Taylor
Taylor
s
[VII
s
Theorem.
A.
Maxima
and minima.
Theorem may be applied
to give greater
theoretical completeness to the tests of Ch. VI, 122 123, It though the results are not of much practical importance. will be remembered that, that has derivatives of (x) assuming
&lt;/&gt;
the
stated the following as being sufficient conditions for a maximum or minimum of % for a (x) at x
first
two
orders,
we
/7
&lt;&gt;
:
maximum,
&lt;^
= 0, ^ (f)
(f)&lt;0;
for a minimum, $
&lt;"
= 0, (f)
&lt;
//
:
^&gt;
(
)&gt;0.
It is evident that these tests
fail if
(f) as well as
()
is zero.
Let us suppose that the
&lt;/&gt;,
first
n derivatives
...,
&lt;p&gt;(#)
$
(x\
are continuous,
and that
all
save the last vanish
when x = f Then,
.
for sufficiently small values of h,
In order that there should be a
maximum
or a
minimum
this
expression must be of constant sign for all sufficiently small values of h, positive or negative. This evidently requires that n should be even. And if n is even there will be a maximum or a
minimum
minimum
derivative,
according as
(n
&lt;/&gt;
&gt;
(f) is negative or positive.
Thus we obtain the
and
test:
if there
is
to
be
a
maximum
or
the first derivative
which does not vanish must be an even
is negative,
there will be
a maximum if it
a minimum
if it
is positive.
1.
Examples LVII.
positive integer,
2.
Verify the result
when $
(x)
= (x  a) m m
,
being a
and
=a.
&gt;)
for
maxima and minima
Test the function (x  a) m (x  n where m and n are positive integers, Draw graphs of the at the points x=a, x=b.
,
different possible
forms of the curve
a,
y=(x
a)
m
(xb}
,
n.
3.
Test the functions sin#
X*
sin^^7 +
X^
a?
sin
x3
x5
XX+Q
or
^
,
...,
cos#
1,
cos.rl+
B.
,
cos#l+
X*
1
^7,
...
for
maxima
minima
at ^=0.
calculation of certain limits. Suppose derivatives / (x) (x) are two functions of x whose that/(#) and and $ (x) are continuous for x = t; and that /( ) and (f ) are both equal to zero. Then the function
150.
&lt;f&gt;
&lt;
The
149, 150]
is
ADDITIONAL THEOKEMS IN THE CALCULUS
269
not defined when x
=
f
.
But
of course
it
may
te),
well tend to a
limit as a?*f.
Now
where
x^ lies
/(*)=/(*) /(f) = (*f)/
;
between f and x and similarly where #2 also lies between f and #. Thus
&lt;f&gt;
(x)
= (a;
f)
&lt;
(* 2 ),
We
must now distinguish four
(1)
cases.
If neither/
() nor
&lt;/&gt;
()
is zero,
then
(2)
!/() = 0,0
If/ (f)
=}=
()*&lt;),
then
(3)
0,
&lt;
(?)
=0, then /(?)/
/(#)/&lt;/&gt;
(x)
(a?)
becomes numerically
tends to
oo
very large as x*%\ but whether or is sometimes large and positive
negative, in which
(4)
or
oo
,
and sometimes large and
we cannot
/
say,
without further information as to the way
(#)*0 as x^%.
(^)
(f)=0, then we can as yet say nothing about the behaviour of /(#)/( (sc) as a?*0.
If/
= 0,
&lt;
But
and
(f&gt;
in either of the last
two cases
it
(x)
have continuous second derivatives.
"
may happen that /(#) And then
/&lt;*)
*
W
= /(*) /(?) T f)/ (?)  K* (f  * (f )  (^  f ) f (?) = * (ar) J ( f)
a; 3
where again xl and
lie
between f and
a?
;
so that
We can now distinguish a variety of cases similar to those In particular, if neither second derivative considered above.
vanishes for
x=%, we have
argument can be repeated indefinitely, and we obtain the following theorem: suppose thatf(x) and and their derivatives, so far as may be wanted, are continuous for # = {. Suppose further that f (p] (x) and (x) are the first
&lt;f&gt;(x)
&lt;f&gt;W
It is obvious that this
derivatives
off(x) and
&lt;f&gt;
(x)
which do not vanish when x
=
%.
Then
(2)
ifp&gt;q,
270
(3)
ADDITIONAL THEOREMS IN THE CALCULUS
if
p&lt;q,
[vil
* oo
p&lt;q
and q
p
is
even,
either
/(#)/&lt;(#)
+00
(3)
or
(f);
,
the sign being the
same as
that
off (p)
(?)/&lt;
(4)
if
f(x)l$(x)
fw
(f)/&lt;p&gt;
(f),
and is odd, either or 0*0 * + as x^^\0,the sign being the same as that of wfoYe t/ ar^f0 fo wgrw HUM* 6e reversed.
qp
/(/&lt;/&gt;
00
This theorem
equations
is
in
fact
an immediate corollary from the
Examples LVIII.
1.
Find the limit of
as #*!.
[Here the functions and their
0"(1)=2.]
first
derivatives vanish for
and
/ (I) =(+!),
(tan
37
#=1,
2.
Find the limits as
#0
of (tan
 #)/(#  sin #),
^^  % tan
.r)/(w sin
x sin wo;).
1/y.]
3.
Find the limit of x {V (x2 + a 2 )  #} as x* oc
Prove that
.
[Put x =
4.
lim (a?  W ) cosec^Tr = (
* **
^^
*
,
Urn
J
x^n^n
Jcosec ^TT
\

1 (X ft)*]
/
(
~ 1)n
= (~ 1
6
^
xir.
n being any integer
5.
;
and evaluate the corresponding
limits involving cot
Find the limits as x*~Q of
G.
(sin
x arc sin #  x*)\x*^
C.
(tan
^7
arc tan x 
x2
)/afi&gt;
1,
as z*~0.
151.
The contact
of plane curves.
they
Two
curves are
said to intersect (or cut) at a point if the point lies on each of them. have the same They are said to touch at the point if at the point.
tangent
Let us suppose now that
/(a?),
&lt;
(a?)
are two functions which
and let us possess derivatives of all orders continuous for # = consider the curves y =/(#), y = $ (#). In general /(f) and (f ) will not be equal. In this case the abscissa x = f does not corre
&lt;
spond to a point of intersection of the curves.
If however
150, 151]
ADDITIONAL THEOREMS IN THE CALCULUS
271
=&lt;()
f () =
the curves intersect in the point x=%, Let us suppose this to be the case. Then
&lt;&gt;
(f),
in order that the curves should not only cut but touch at this point it is obviously
/y=0(s)
J
/?=/(*)
necessary and sufficient that the first derivatives (x) should also have (x),
f
&lt;/&gt;
the same value
when x =
of the
.
The contact
case
curves in
this
may be regarded from a different point of view. In the figure the two
curves are drawn touching at P, and
is
QR
equal to
f&gt;
(f
+ h) /(f + A), or, since *(?)=/(), * (?)=/(& to
and
1.
where
lies
between
lim
Hence

when h^O.
In other words, when the curves touch at the point whose abscissa is f, the difference of their ordinates at the point whose abscissa is % + h is at least of the second order of smallness
is small.
will easily verify
r
&lt;f&gt;
when h
that lim (QRIJi) = () / () when the curves cut and do not touch, so that QR is then of the first order of smallness only.
The reader
It is evident that the degree of smallness of
QR may
be taken
as a kind of
measure of the
closeness of the contact of the curves.
if
It is at once suggested
that
the
first
,
n
1
derivatives of
f
and
&lt;f&gt;
have equal values
?ith
order of smallness;
g then QR will be of the and the reader will have no difficulty
when x =
and that
~{&lt;P
in proving that this is so
Hm^ =
We
...,
/&lt;
(?)/"()}.
are therefore led to frame the following definition
:
Contact of the nih order.
() =
but
&lt;&lt;&gt;(),
&lt;j)
/
(n+1)
If /() = +1) (), ()4
:
(/&gt;
(f ),
/
(f)
the
=
&lt;/&gt;
(f),
&lt;P
then
curves
y =f(x\ y
(x) will be said to have contact of the nth order at the point whose abscissa is %.
The preceding
?ith
discussion
makes the notion of contact of the
fails
order dependent on the choice of axes, and
entirely
272
ADDITIONAL THEOREMS IN THE CALCULUS
to the curves is parallel to the axis of y.
[VII
when the tangent
We
can
deal with this case by taking y as the independent and x as the dependent variable. It is better, however, to consider x and y as
functions of a parameter
t.
An excellent account
tract referred to
vol.
ii,
be found in
Mr Fowler s
on
of the theory will p. 266, or in de la
et seq.
Valle e Poussin s Cours d Analyse,
pp.
396
Examples LIX. 1. Let $ (x} = ax+ b, so that (x) is a straight line. The conditions for contact at the point for which x=% are /() = + 6, If we determine a and 6 so as to satisfy these / () = equations we find
y=&lt;
a =f ()
point
or
&=/() 1/
is
()&gt;
and the equation of the tangent
x=
to y=f(x] at the
y f () = (#)/
2.
().
Of.
Ex. xxxix.
is
5.
simple contact with the curve In order that the tangent should have completely determines the line. contact of the second order with the curve we must have () = (), i.e.
&lt;"
The
fact that the line
to have
/"
/"()
point at which the tangent to a curve has contact of the second order is called a point of inflexion.
3.
= 0. A
2#/(l
4.
f
#2 ),
Find the points of inflexion on the graphs of the functions sin x, a cos 2 x f 6 sin 2 #, tan #, arc tan #.
So 4
 6/r 3 + 1,
Show
that the conic
point of inflexion.
[Here ax + hy +g + (hx + by +/) y 1 =
aa?+2hxy+fy*+2g+2fy+c==Q
and
cannot have a
suffixes
denoting differentiations.
Thus
at a point of inflexion
or or
a
(6  A
2 )
{oa
2
f
2^^ + 6y 2 + 2^ + 2/?/} fa/ 2 of*
But
this is inconsistent with the equation of the conic unless
 2fgh + bff 2 = c (ab  A 2
*
)
or abc +
2fghaf bg ch
2 z
2
=0
/
,
and
this is the condition that the conic
should degenerate into two straight
5.
lines.]
l
The curve y = (ax 2 \ 2bx + c )/(ax2 + 2 [3x + y) has one
the roots of
cu7
2
inflexion according as
+ 2/3^ + y=0
are
(cf.
real
or three points of or complex.
[The equation of the curve can, by a change of origin reduced to the form
Ex. XLVI.
15),
be
where p, q are real or conjugate. The condition for a point of inflexion will be found to be 3 3pq+pq (p + = 0, which has one or three real roots
&lt;?)
according as (pq
(
p
q}\
is
positive or negative,
i.e.
according as
p and
q are
real or conjugate.]
151]
6.
ADDITIONAL THEOREMS IN THE CALCULUS
Discuss in particular the curves y =
(l
273
#)/(!
+#2
),
y = (1
 # 2 )/(l +#2 ),
Show that when the curve of Ex. 5 has three points of inflexion, they 7. on a straight line. [The equation 3 3pqg+pq(p + q) = can be put in 2 = 0, so that the points of inflexion the form ( p)(q)(+p + g) + (p lie on the line g + + = or A
lie
&lt;?)
A()*r+
8.
Show
that
the curves
y=xsmx, y = (s\\\x)lx
have each
infinitely
many
9.
points of inflexion.
Contact of a circle with a curve.
circle, viz.
Curvature*.
The general
equation of a
contains three arbitrary constants. Let us attempt to determine them so that the circle has contact of as high an order as possible with the curve y=f(x) at the point (, ^), where 17 =/(). We write m 2 for
,
rj
/
(),/"
().
Differentiating the equation of the circle twice
we obtain
.............................. (2), .............................. (3).
&)yi
=
%2=o
when
If the circle touches the curve then the equations (1)
and
#=,
r) Z
.
y=
is
the contact
when y 2 =
This gives (^a)/ 7l = (^&) = r/ v/(l+77l 2 ). If of the second order then the equation (3) must also be satisfied
rj,
.
y 1 = r)l
(2)
are satisfied
Thus 6=77 + {(1 f^ 2 )/^}
;
and hence we
find
^2
7/2
r; 2
which has contact of the second order with the curve at the point (, 77) is called the circle of curvature, and its radius the radius of curvature. The measure of curvature (or simply the curvature) is the reciprocal of the 2 3/2 radius thus the measure of curvature is/" ()/{! + or [/ ()] }
circle
:
The
,
10.
reciprocal of the radius curvature is constant.
11.
Verify that the curvature of a circle is constant and equal to the and show that the circle is the only curve whose
;
Find the centre and radius of curvature at any point of the conies
12.
In an ellipse the radius of curvature at
P
is
CD3 /ab,
where
CD
is
the semidiameter conjugate to CP.
*
A much fuller discussion of the theory of curvature
on
p. 272.
will
be found in
Mr Fowler
18
s
tract referred to
H.
274
ADDITIONAL THEOREMS IN THE CALCULUS
[VII
13. Show that in general a conic can be drawn to have contact of the fourth order with the curve y=/(#) at a given point P.
[Take the general equation of a conic,
viz.
ax* + Ihxy + by* + tyx + 2fy + c = 0,
and
differentiate four times with respect to x.
Using
suffixes
to
denote
differentiation
we obtain
If the conic has contact of the fourth order, then these five equations
must
thus
be
satisfied
by writing
,
77,
71,
172, 173
&gt;
*U fo11
#
#&gt;
3/i,
#2, ya, #4
We have
:
just enough equations to determine the ratios
14.
a
:
b
:
c
:
f g
:
h.]
An
infinity of conies
with the curve at P.
can be drawn having contact of the third order Show that their centres all lie on a straight line.
[Take the tangent and normal as axes. Then the equation of the conic is 2 2 2y=a# + 2A#y+&3/ and when x is small one value of y may be (Ch. V, Misc. Ex. 22) in the form expressed
of the form
,
2 s y = Jotf + (iA + e^ x
,
where
fx
~Q with
x.
But
this expression
must be the same as
where
*Q with x, and so a=f" (0), /*=/ (0)/3/" (0), in virtue But the centre lies on the line ax+hy*=Q.] of Ex. LV. 15.
"
fx
of the result
15.
ellipse (#/a) +(y/6)
16.
Determine a parabola which has contact of the third order with the 2 2 =l at the extremity of the major axis.
The
locus of the centres of conies which have contact of the third
order with the ellipse (#/a) 2 + (y/6) 2 = l at the point (a cos a, b sin a) is the diameter x\(a cos a) =yl(b sin a). [For the ellipse itself is one such conic.]
152.
So
far
we have been concerned
Differentiation of functions of several variables. exclusively with functions of a
to prevent us applying the single variable as, but there is nothing to functions of several variables x, y, .... notion of differentiation
Suppose then that /(a?, y) x and y, and that the limits
is
a function of two* real variables
* The new of several variables points which arise when we consider functions The generalisations are illustrated sufficiently when there are two variables only. of our theorems for three or more variables are in general of an obvious character.
If x and y were connected by a relation the state of affairs would be very different. say y = (x) and then at all. to denote them by and dy or or that these The reader must not suppose. new process as ordinary differentiation. 1. that is to say that f(x.// OTfx . In this case our definition of// would fail as we could entirely. and then f (x y). notations imply any essential novelty of idea: with respect to x is exactly the same partial differentiation simply//. is essentially a function of two independent variables. is a function of the single variable t./&. Of course we may also really a function of the single variable x. then 2. If y depends on x.151. y=rsih0. y is a function of x. But then f(x. Prove that if x=rcos0. Account for the fact that *l and considering a function y of one variable x it followed from the definitions that dy/dx and This is no longer the cfo/rfy were reciprocals. A . we were *1/(). &. we may regard x and y as functions of a third variable t. It is usual to call these derivatives the partial differential coefficients of/. not change x into x + h without at the same time changing y.lt. is ?)=/{*.fy . Examples LX. II. 152] ADDITIONAL THEOREMS IN THE CALCULUS 275 exist for all values of x and y in question.y) possesses a derivative dfjdx or D x f(x. so that r= 0arctan(y/x). represent it as a function of the single variable Or. /(a. the only novelty lying in the presence in / of a second variable y independent of x. however. as we defined it in Ch.y) with respect to x and a derivative df/dy or Dy f(x. which is of the form (t\ t /{&.y) would not really be a function of two variables function of two variables. as is often most convenient.lt.gt. In what precedes we have supposed x and y to be two real variables entirely independent of one another. y) with respect to y. [When 182 .(*)} y.
] Prove that if z =f (ax + by) = a (3z. 0}. a theorem concerning the differentiation of a function of one variable./. increment MMi = 8x. the increment of r is P P^fir. Y=xy. x. x. . . .Any other choice would give us values of Ax.. .r=lim(Sr/6\tf). when X+Y=x. known generally as the Theorem of the Total Differential Coefficient. which is of very great importance and section re depends on the notions explained in the preceding This theorem gives us a rule functions of two variables. say. 5. = lim (PP PA) = sec 0. is so that u= . specified. 3JT/3. Y+Z=xy.e ADDITIONAL THEOREMS IN THE CALCULUS when we [VII Let P (Fig. =A . then 3w/3#=0. 7. A/ proportional to those used here. and z being the independent variables. increment of x is MMi = A#. P along 9r/9. but lim (Ar/Atf) lim (Sr/Ar) = cos 0. and the point (#. P 2 . when and find 3^7/9 JT. But if we regard u as a function of the variables #. 46) are dealing with functions of two variables... x and y being now regarded as and 0. Y and find 3^/37. and Now Aa?=&z?* : but Indeed it is easy to see from the figure that A PPi) = cos 0. Of course the fact that &x = 8x is due merely to the particular value of Ar that we have chosen (viz. upon the same hypothesis as then b (dz/dx) variation of P. If to PI. x+y = T]. by Ar. and x+y + z=. if on the other hand we want to calculate functions of r r 3#/3r..gt. in terms of X. y. This brings OPl we take OP = OP..%). where PP =Ar: 2 This brings P to the corresponding say . z Find 3JT/3^. express ^ [There is of course to functions of no difficulty in extending the ideas of the last section any number of variables. say by an y} or (r. so that lim (6Y/6X) = lim (P / 2/ 2 The fact is of course that 3#/3r and to the dr/dx are not formed 3. we must increase keeping 6 constant. Express x as Find dXjdx. then 3w/3# = l. say.. functions of X.. To find 3r/3# we must increase #. while keeping y constant. 4. garding for differentiating with respect to * t.276 case \&. y. 153. PP 2 ). But the reader must be careful to of a function of impress on his mind that the notion of the partial derivative several variables is only determinate when all the independent variables are Thus if u=x+y + z. Z=xyz.] There Differentiation of a function of two functions. Ar=t=6Y.
so is the circle where ^7 and y are to be put equal to carrying out the differentiations. + ^2 = l. where ^ and 7?^0.lt. y} xP+y (0 = 4tl(l + 1^. come as in question. ^ + T)) / {0 (&. y) f (0 +// (. where we are to put = ^(Q. say F(t). Hence y). by the Mean Value Theorem. y) * a? (0. Suppose (0 = (1O/(1 2 )J ^(0 = 2*/(1 + * 2 ). (1  * 2 )/(l + Z 2 ) and 2i/(l + t ) 2 after . are functions of t with continuous differential of the (t). Then (x. is And now let us suppose that the variation of and y restricted in that (x.lt. in the first instance.lt. The problem is to determine F t x+ Suppose that. dfjdf_dx dt dfdy dy dt dx dt Examples LXI.lt. 153] ADDITIONAL THEOREMS IN THE CALCULUS 277 Let us suppose.152. y) is a function of the two variables x and y. differentiations with respect to expressed in the form y = ^(t) after carrying out the a? and This result may also be y. y)} 5". F (t) = A/{* (0.gt. (*). y) lies on a curve where $ and coefficients ^ t. V (0 = 2 (1 1.n) /(&. that the locus of 4&. that /(a?.*. (. .gt. f ^0 and f/T^ yr+ (t: also /. As r^0. and that /. ^ (0}] lim \f(x + &y+&. / are continuous functions of both variables ( 107) for all of their values which . (* + T). + r.] TJ But. ^ (0) =// (^. x and y change to [/ (&. y) reduces to a function (t). and each (0&. (t\ ty single variable Then /(a?. when t changes to and y + ?. Then by definition t.. lie between and 1.
= (a?/ . e.g. y) = 0. (6) x = a cos t. 1 . Verify the theorem in the same way when (a) x=t y=l t r \ f(x. if y= Now suppose that z =f(x. adequately shown by in which case the dx dy dx though open to objection. which is obviously correct.#). y) = x*+f. = x2 + y z Suppose. by the same letter /in df\dx and this notation is also ^ 9//&. The distinction between the two functions denoting the first by dffdx and the second by theorem takes the form . y = a sin t. It was this case which led to the introduction of the notation 8//8o?. y} = x+y. in terms of h. y) is a function of the two inde pendent variables x and y. [VII We Then fx = 2#. For it would seem natural to use the notation df\dx for either of the functions D*f{x.lt. . y) . y\ in one of which y is put equal to ^(#) before and in the other after differentiation.k.lt. (#)} and f(xt y\ whose forms as functions of x are quite different from one another. + (x}}=Dx f(x. . or as it may be written. that /(a?. y=^(t] is /(a?.278 ADDITIONAL THEOREMS IN THE CALCULUS can easily verify this formula in particular cases. Many depended upon the Mean Value Theorem. k or 8x. then 5. is 9//8a?. y} + (*)./ are functions of (9 t. If a? and . then/ = (^a/+yy)/r. One of the most important cases is that in which t is x itself. .( of the results of the last chapter + )&.. in that it is a little misleading to denote the functions /{#. and that a? and y receive increments h. Sy respectively: and let us attempt to express the corresponding increment of z t viz. If the result of eliminating t between x=$ (*).$ (t)+$yy (t)Q. but Dx f(x. ^ (#)} and Dx f(x. Suppose for example that y=lx and /(#.#) = Z) x 1 = 0. //= 2y. 8//8y.ytf)/r* and 6 are the polar coordinates of 9 dashes denoting differentiations with respect to 154. 4.lt. O) = hf (#). since F(t)*=l. y) = 1. y) = x. (as + 6h). and r (#. 3. and it is easily verified that F1 (t)=Zx&. expressed by the equation &. m 2. y) + Dy f(x.+y. k and the derivatives of z with respect to x and y. Then D*/ (#. where y is to be replaced by ^ (#) after differentiation. The Mean Value Theorem for functions of f two variables. f(x. We then obtain Dx f{x.lt.
153. We must say the larger of Sx and Sy because one of them might be small in comparison with the other. &. be observed that if any equation of the form dz= approximately true in this sense. that the difference between the two sides of the equation small in comparison with the larger of Sx and %*. (x.. y+k) f(x.y + 0k) =/* (x + 0h. + r) htk . this can only be the case * \=f f . where e^k and ij tend to zero as h and k tend to the theorem may be written in the form h&.\fa or \\fx if &. if f any assigned positive number. T.  f. which is the formula desire d. where e..e. finally & =f(x+h.y + kt).lt. Hence. x Similarly fi=fv Or with \5x\ + \dy\ or .gt. as may be  we please.. less than &. 1. The result embodied in (1) may be expressed by saying that the equation is approximately true. is i.y + kt) + kfy (as + ht..... * ?/ all tend to zero as 8x and by tend to zero . fx fy (x + 6h. y. But.lt.lt. y + 0k).. y + 0k) =fy y) y) + M . Hence ^ = (/* where e + ) * + (/!. we might indeed have Sx It should = or Sy = 0...y + kt) (t) = hfx (x + ht. 8z \fa /% = f fa + rf 8y i. 154] ADDITIONAL THEOREMS IN THE CALCULUS ht.k zero. F Hence = D f(x + ht. we must have \=fx /*=/ have fx &V ~fy ty=efa + r) Sy. .r.153. + *7)fy . and.. and so (\fx ) &* + (/*/ where p and p tend to can choose &.s For we & . we so that for all values of I &r and 8y numerically ..lt. y)=hfx (x + 0h...lt. by t Then &. (1). y+ 6k) + kfy (x + 0h. 279 Le t / (as + where y + kt) = F(t).. Taking dy=*Q we obtain as small as (X fx ) fa\&. Since fx fy t f are supposed to be continuous functions of x and we have (*.r ) ty = p is dx+p toj zero.lt. and rj are small when Sx and Sy are small.
... venience in this.. we have .. These technical advantages are however so great. that the use of the notation inevitable... especially when we come to deal mate equation with functions of several variables...280 155. we obtain dx = Sx . dz. The advantages of (2).. y... it is In the applications of the Calculus.. If for a moment to a function y = / (x) of a single f (x) is continuous then 8y=.. If we divide both sides of (5) by dx we obtain where dy/dx denotes not..... thus acquires a double meaning but there is no incon dy/dx ..{/() + e}& where e^0 .. pass from the exact equation (5) to the approxi can.. when &r*0.......... (3). (5)... but the quotient of the differentials dy.. as heretofore.... since (6) is true whichever meaning we choose. but in terms of what are called their differentials dx.. Sy. which gives the the differential notation are in reality of a purely technical character.... the differential coefficient The symbol of y.. (4)... [VII especially in geometry........ We have up to the present attributed itself.. and its truth does not /depend on any assumption as On the other hand it is precisely the fact that we to the continuity of/ (a?)........ usually most convenient to work with equations expressed not.. (1)... under certain conditions..... just as we might say part of ctx + bx* when x is small.. no meaning of any kind to the symbol dy standing by now agree to define dy by the equation If We dy=f(x)Sx ... so that dy=f (x)dx ....... is almost When / (#) is continuous. like equation (1) of 154.. (2) approximately true.............. z.. as Sx*Q: in other words the equation fy=/&* is . Let us return variable x......... It is exact and (5) has two apparent advantages over (2). in terms of the increments Sx....... dy. This is sometimes expressed by saying that dy is the principal that ax is the principal part of 8y when dx is small.... . The equation former its importance... ADDITIONAL THEOREMS IN THE CALCULUS Differentials.. Sz of the functions x...... not merely approximate. we choose for y the particular function x. dx.......... ..
.. y.. = By . so that z is also a function of t alone.. Prove that dA and da db state the corresponding approximate equation connecting the increments of the axes and the area.............. V=f dz = j dz ~dt . dy ..Sy y in turn.. and then \=fx /*=/ This follows at once from the last paragraph of 154.gt. &... the exact equation corresponding to the approximate Here again it is to be observed that the 154.. .lt... we obtain dx=$x.... dt... 1. d T dii t ..... Examples LXII... which the same in form as Thus formula which expresses dz in terms or not. dx +fy the dy. The area of an ellipse is given A = irab. where b are the semiaxes. (8).. a function z of two independent variables x and differential dz by the equation We define the Putting z = oo and z dtf.. (7)........ ... equation (1) of former is of importance only for reasons of practical convenience which in working and because the deduced from it..... y\ x and y being not independent but functions of a single variable t. so that dz is f^dx +fy dy . of dx and dy is the same whether the variables x and y are independent This remark is of great importance in applications...... It should also be observed that if z is a function of the two independent variables x and f .... It is obvious that the theorems and definitions of the last three sections are capable of immediate extension to functions of any number by of variables.. latter can in certain circumstances be One property of the equation (9) deserves special remark. (9). then _ df dx 3/ dy di~~faditydi dz Multiplying this equation by dt and observing that dx = .. dx j . We saw in 153 that if zf(x... Sx+f..**&..155] ADDITIONAL THEOREMS IN THE CALCULUS pass 281 with We now to the corresponding definitions connected y. we obtain is dz =fx (9)...
dZ/dYin terms of dzfix. the area of a triangle ABC. F. /_. B. do 7 cos . dv 7 ". as a function of and (iii) a. Z be defined by the equations Z may be expressed as a function of and Y. C. where 3. dA = db be { do . dz/dy. dv du . c vary so that R remains db 1 constant. da  = cos A JD da . and let JT. _ dz du dz dv du dz dv_ dx~ du dx dv dx dy~~ du dy dv dy [We have dz = ^ dz du du dz f ^ dv. constant. . so that a may be regarded as a function of b and da db Prove that B cos A cos da do _ ~ cos C cos A [This follows from the equations 7 77 da = ^rdb da db da i ~. then da cos do 1 A cos B cos ~= C 0.do.5 + (7 are then If 2 is a function of dz w and v. 6. or Then X . Express dZjdX. Substitute for du and dv in the first equation dz .rt6 + cos . b. sides of a triangle vary in such a way that the area remains c. dA A n =2 da a cot A dA + rrH A asmB asmC dA=R (cos Ada+ cos Bdb + cos Cdc\ 1 cdB r bdC = ~&. which are functions of x and dz __dz y. A y b. . dx dy .. c. (ii) ADDITIONAL THEOREMS IN THE CALCULUS Express A.Then dz pdx qdy = 0. R is The the radius of the circumcircle.^ dy. If a. and compare the result with the equation dz . dv .282 2. dv = ^ dx \. cos [Use the formulae a = 2/2 sin constant. 6. cos and so . 9o cos ^ sco = A 77 (7 . du = . c. [Let these differential coefificients be denoted by PI Q an^ Pi g..du.. ~ Cdc=0. du 7 dx + ?r. Let z be a function of x and y.4aa + cos/&. and the facts that R and 4 4. ox oy ". 4.] 5. 4. and establish the formulae (i) [VII a.lt.gt.
155, 156]
ADDITIONAL THEOREMS IN THE CALCULUS
this equation with
283
Comparing
dZPdX QdY=0 we see that

_
c3

C3
7.
If
then
Q=aB X+b Y+c3 Z.
3
(Math. Trip. 1899.)
Differentiation of implicit functions. Suppose that /(#,#) and fy (x, y} are continuous in the neighbourhood of the point (a, and that
8.
its
6),
derivative
f(a,b)=0,
/6
(a, 6) 4=0.
Then we can find a neighbourhood of (a, 6) throughout which / (#, ?/) has always the same sign. Let us suppose, for example, that / (x, y) is positive near (a, b). Then /(#, y) is, for any value of x sufficiently near to a, and for
values of y sufficiently near to 6, an increasing function of y in the stricter 108. that there is a unique sense of 95. It follows, by the theorem of continuous function y which is equal to b when x=a and which satisfies the
equation / (x, y) =
for all values of
x
sufficiently near to a.
is
Let us now suppose that /(#,.y) possesses a derivative ft fay) which If /(.#, 3/) = 0, a=a + h, y=b + k, we have also continuous near (a, 6).
where
and ^ tend to zero with h and
k.
Thus
or
dx
).
The equation
.
of the tangent to the curve /(#,
2/)
= 0,
at tho point
vn
is
156.
Definite Integrals
145,
x,
that, in Ch. VI,
and Areas. we assumed that,
is
It will
if
be remembered
is
f(x)
a continuous
function of
and
PQ
the
y=f(x), then the region PpqQ shown in Fig. 47
graph
of
has associated with
it
a definite
number which we
It is clear
call its area.
that, if
we denote
Op and Oq by a and #, and allow x to vary, this area is a function of x, which we denote
\&gt;yF(x).
O
p
Fig. 47.
284
ADDITIONAL THEOHEMS IN THE CALCULUS
[VII
and we showed how
in 145 that F (x) =/(#), be used in the calculation might of the areas of particular curves. But we have still to justify the fundamental assumption that there is such a number as the
Making
this assumption,
we proved
this result
area F(x).
We
know indeed what
it is
is
meant by the area of a
rectangle,
measured by the product of its sides. Also the properties of triangles, parallelograms, and polygons proved by Euclid enable us to attach a definite meaning to the areas of such figures. But nothing which we know so far provides us with a direct definition of the area of a figure bounded by curved lines. We shall now show how to give a definition of F(x) which will
enable us to prove
its
and that
existence.*
Let us suppose f(so) continuous throughout the interval (a, b), and let us divide up the interval into a number of subintervals n, where by means of the points of division # #i
,
%z&gt;
&gt;
a=#
&lt;
x
l
&lt;
.
.
.
&lt;
ocn
^
&lt;
xn =
b.
the interval (#, oc v+l ), and by Further, let us denote by and let us write lower bound ( 102) of f(x) in
,
m
v
the
s = w
S
+m
Sx l
+
. . .
+ m n B n = 2m, 8,,
say.
It is evident that, if
M
is
sM(b
a).
The aggregate
the upper bound of f(x) in (a, b), then of values of s is therefore, in the
bound 80, bounded above, and possesses an upper language of which we will denote by j. No value of s exceeds j, but there are
values of s which exceed any
number
less
thanj.
,
In the same way,
define the
if
M
v is
the upper bound of /(a?) in
we can
sum
the lower bound of f(x) in (a, b), then bounded aggregate of values of S is therefore
is
It is evident that, if
m
S = m (b  a). The
below, and possesses a lower bound which we will denote by ^7. No value of S is less than J, but there are values of 8 less than any
number
*
greater than J. The argument which follows
vol.
i,
is
d Analyse (second edition),
pp. 171 et seq.
modelled on that given in Goursat but Goursat s treatment ;
s
is
Gours
much
more
general.
156]
ADDITIONAL THEOREMS IN THE CALCULUS
make
clear
285
s
It will help to
the significance of the
sums
and S
if
we observe
that, in the simple case
in which /(a?) increases steadily from x= a to x=b, m v is /(#) arid In this case s v is f(x v + l }.
M
the total area of the rectangles shaded in Fig. 48, and S is the area bounded by a thick line. In
is
s and S will still be areas, composed of rectangles, respectively included in and including the curvi linear region whose area we are
general
trying to define.
We shall now show that no sum such as s can exceed any sum such as S. Let s S be the sums
}
Fig. 48.
corresponding to one
mode
of
subdivision,
to
and
s
s
,
S
those
s
corresponding to another.
S.
We
have
show that
S and
a third mode of subdivision by taking as dividing points all points which are such for either s, S or s, S Let s, S be the sums to this third mode of subdivision. corresponding Then it is easy to see that
.
We can form
.(1).
For example,
occurs in s
is
that at least one interval 8 V which divided into a number of smaller intervals
s differs
from
s in
so that a
term
m
,
v
v
of s
is
replaced in s by a
sum
m Vtl w, j2 ... are the lower bounds of f(x) in 8V)1) 8 V 2 .... But evidently m Vtl ^ m v) m v ^ ^m v ,...,so that the sum just written is not less than m S v Hence s ^ s; and the other inequalities (1) can be established in the same But, since s way. S, it follows
where
,
,
.
that
S MI S 3 ff,
which
is
what we wanted
j
to prove.
J.
It also follows that
For we can
find
an
s as
as
as we please*, and so please and an S as near to j would involve the existence of an s and an S for which s S.
we
J
near to j
&gt;J
&gt;
*
The
s
and the S do not
in general correspond to the
same mode
of subdivision.
286
ADDITIONAL THEOREMS IN THE CALCULUS
So
far
[VII
We
shall
limit
J
we have made no use of the fact that f(x) is continuous. now show that j = J, and that the sums s, S tend to the when the points of division x v are multiplied indefinitely
way that
shall
all
in such a
cisely,
to
the intervals 8 V tend to zero.
More pre
possible
we
show
that, given
any positive
number
e, it is
find B so that
whenever 8 V
&lt;
$ for all values of
v.
There
is,
by Theorem II of
106, a
v
&lt;
number
a),
8 such that
M
whenever every 8 V
is less
v
m
e/(b
than
S.
Hence
But
and
all
Ss = (SJ) + (Jj) + (j  s)
= J and
;
the three terms on the righthand side are positive, and therefore all less than e. As J j is a constant, it must be zero.
Hence j
0^js&lt;e,
0^8
J&lt;,
as
was
to
be proved.
We
define the area of
S, that is to
definition.
PpqQ as being the common limit of s and say J. It is easy to give a more general form to this Consider the sum
o
=
2/A
&lt;r
/ denotes the value of f(x) at any point in $. Then plainly lies between s and S, and so tends to the limit J when the
intervals
where
tend to zero.
a.
We may
therefore define the area as
the limit of
157.
is
The
definite integral.
Let us now suppose that/(#)
__
a continuous function, so that the region bounded by the curve the ordinates x = a and x = 6, and the axis of a, has a y =/(#), definite area. 145, that if F(x) is an proved in Ch. VI,
We
integral function
of /(#),
i.e.
if
F
As
of
it is
it
(x)=f(x\
is
F(x)=\f(x)dx,
F(a).
then the area in question
F(b)
F (x),
not always practicable actually to determine the form is convenient to have a formula which represents the
contains no explicit reference to
area
PpqQ and
F (x). We
shall
write
156, 157]
ADDITIONAL THEOREMS IN THE CALCULUS
287
The expression on the righthand side of this equation may then be regarded as being defined in either of two ways. We may regard it as simply an abbreviation for F(b) F(a), where F(x) is some integral function of /(a?), whether an actual formula
expressing it is known or not or we may regard 156. the area PpqQ, as directly defined in
;
it
as the value of
rb
The number
is
f(x] dx
(/&lt;*) J a
called a definite integral;
is
a and
b are called its
lower and
subject of integration or upper limits; f(x] integrand; and the interval (a, b) the range of integration. The definite integral depends on a and b and the form of the function /(#) only, and is not a function of as. On the other hand
called the
the integral function
F(x)=!f(x)dx
is
sometimes called the indefinite integral o
The
distinction between the definite
and the
indefinite integral is merely
is
/b f(x}dx=F(b}F(a)
a
a
function of
6, and may be regarded as a particular integral function of /(&). On the other hand the indefinite integral F (x) can always be expressed by means of a definite integral, since
But when we are considering indefinite integrals or integral functions are usually thinking of a relation between two functions, in virtue of which one is the derivative of the other. And when we are considering a definite
!
we
I
*
are not as a rule concerned with any possible variation of the integral limits. Usually the limits are constants such as and 1 ; and
we
is
not a function at
all,
but a merejiumber.
/
It should be observed that the integral
Ja
f(t)dt
t
having a differential
coefficient/^),
is
a fortiori a continuous function of #.
Since 1/x is continuous for all positive values of #, the investigations of the preceding paragraphs supply us with a proof of the actual existence of the function log #, which we agreed to assume provisionally in 128.
288
158.
ADDITIONAL THEOREMS IN THE CALCULUS
[VII
Area of a sector of a
circle.
The
circular functions.
The theory
of the trigonometrical functions cos x, sin x, etc., as usually presented in textbooks of elementary trigonometry, rests
on an unproved assumption.
An
by two straight
lines
OA, OP;
angle is the configuration formed there is no particular difficulty in
definition into translating this geometrical terms. The assumption comes at the next stage,
purely analytical
when
it is
assumed
is
that angles are capable of numerical measurement, that that there is a real number x associated
to say
a real
with the configuration, just as there is number associated with the region
of Fig. 47.
PpqQ
mitted, cos# and sin
further
This point once ad a? may be defined
in the ordinary way, and there is no difficulty of principle in the elaboration of the theory. The whole
digiciilty lies in the question,
Fig. 49.
A
what is the x which occurs in cos x and sin x ? To answer this question, we must define the measure of an angle, and we are now in a position to do so. The most natural definition would be this: suppose that AP is an arc of a circle whose centre is and whose radius is = OP = 1. Then x, the measure of the angle, is unity, so that OA
of the arc
the length
AP.
.
This
is,
in
substance, the definition
adopted in the textbooks, in the accounts which they give of the It has however, for our present pur theory of circular measure
pose, a fatal defect; for even of a circle, possesses a length.
we have not proved that the arc of a curve,
The notion
of the length of a curve is capable of precise mathematical analysis just as much as that of an area; but the analysis, although of the same general
character as that of the preceding sections, is decidedly more difficult, and it is impossible that we should give any general treatment of the subject here.
We
must therefore found our
definition on the notion not of
define the measure of the angle length but of area. twice the area of the sector of the unit circle.
We
AOP as
AOP
\
Suppose, in particular, that
where
m
&gt;
by
&lt;
(m).
and that QP is y = moc, The area is a function of m, which we may denote 0. If we write p for (1 + m*)~f, P is the point m^), and
A
is
y=
(&gt;,
]58]
AND INTEGRAL CALCULUS
289
we have
&lt;/&gt;
(m)
~ = \mtf + [V(l x*) J
v
dec.
Differentiating with respect to m,
we
find

\^
s
Thus the
arc tan
analytical equivalent of our definition would be to define
m by
the equation
dt
.
l+
2
and the whole theory of the circular functions could be worked out worked out from a similar definition
from this starting point, just as the theory of the logarithm is in Ch. IX. See Appendix III.
integral.
Examples LXIII. Calculation of the Show that 1.
J a
/
definite
from the indefinite
xn dx=
nT l +
,
and
in particular that
/
xn dx=
.
f
2.
I
b
cosmxax=
dx
, _,
smmbsinma tan tun
mb
ma
,
J a
_
,
m
/si
Ja
C
l
f
b
.
_cosmacosmb
dx
3

[There is an apparent difficulty here owing to the fact that arc tan x is a valued function. The difficulty may be avoided by observing that, in the equation
many
dt
x must denote an angle lying between \ir and TT. For the integral when # = and increases steadily and continuously as x increases. Thus the same is true of arc tan x, which therefore tends to \ir as x * oo In the same way we can show that arc tan x *  \K as x * x Similarly,
arc tan
vanishes
.
.
in the equation
dt
o
v^it ;
less
where
Thus,
1&lt;^&lt;1,
arc sin
x denotes an angle dx
if
a and b are both numerically
b
lying between than unity, we have
arc
\K
and
f
I
~?h
2\
= arc sm o
sm
a.]
4.
,J=L, O ^/ 3
H.
r
J
Q
19
290
5.
I
ADDITIONAL THEOREMS IN THE DIFFERENTIAL
s
[VII
Jo
value of the integral
(
2
&gt;J(lx
a + x2
is
,
= tr2sma
which
if

7r&lt;a&lt;7r,
except
when
a
= 0, when
the
is the limit of
a
cosec a as a0.
6.
Jo
7
)dx=%ir,
\
J
J^o
,
* d I , J a 6cos^ integral see Exs. LIII. 3, 4.
+
=
V(&)
If

f
T9
if
a
b
&gt;
I
b
I
.
[For the form of the indefinite
a
\&lt;\
\
then the subject of integration has an
the value of the integral
infinity between negative and
and
b
\
TT.
What
is
when a
is
a&gt;\
?]
= ^r , if a and 6 are positive.  5 lf) o 2ab a 2 cos 2 x + b 2 sin 2 x J value of the integral when a and b have opposite signs, or
X
8
.
=
What
is
the
are
when both
negative
9.
?
Fourier s integrals.
Prove that
cos
o
771.37
if
m
and n are positive integers then
sin
nx dx
is
always equal to
zero,
and
/cos
o
mx cos nx dx.
m=n, when
each
I
sin
mx sin nx dx
?r.
Jo
is
are equal to zero unless
equal to
10.
Prove that
I "cos
mx cos nx dx and
is
I
"sin
mx sin nx dx
TT
;
are each equal
Jo
to zero except
/"
J
o
when m=n, when each
equal to
/"""
and that
I
cos
mx sin nxdx=^
271

J
n*m*
7 ,
I
cos
mx sin nx dx = 0,
JQ
according as
n
m is odd or even.
159.
tion as the limit of a sum.
definite integral
Calculation of the definite integral from its defini In a few cases we can evaluate a
by direct calculation, starting from the definitions 156 and 157. As a rule it is much simpler to use the of indefinite integral, but the reader will find it instructive to work
through a few examples.
/5
a,
xdx by
x2
,
dividing
(a, 6)
into
n equal
parts by the points of division limit as n * oo of
a=x x^
() ,
...,
xn =b, and
calculating the
158160]
[This
AND INTEGRAL CALCULUS
is
sum
which tends to the limit
reasoning.]
2
(6
 a2 )
as
rc
* QO
.
Verify the result by graphical
/b
3.
x^dx
in the
same way. by dividing
where
rn into
rb
Calculate
/
xdx, where
2
,
0&lt;a&lt;
&gt;,
(a, 6)
rc
J a
parts by
the points of division a, ar, ar
...
~ arn l , arn,
b/a.
Apply the same
#m dx.
^ /I
rb
/b cos mx dx
i
and
/
sin
mx dx by
.
the method of Ex.
1.
J a
5.
Prove that n 2
5*
$ *
*!
as n*~co
[This follows from the fact that
rc
,
^
which tends to the limit
of the integral.]
6.
I
o 1
+

9
as
n *
oo
,
in virtue of the direct definition
J
^"
Prove that

n r=0
\
*
z
&gt;J(n
 r2 )
+ %n.
[The limit
is
/
^(1
o
 ^2
) efo?.]
J
160.
definite
General properties of the definite integral.
integral
The
possesses
the important properties
expressed
by the following equations.*
J a
/to*.
This follows at once from the definition of the integral by means of the It should be integral function F(x\ since F(b}F(a)= ~{F(d}F(b}}. observed that in the direct definition it was presupposed that the upper
limit
is
greater
than
the
I
lower
f(x}
;
thus this method of definition does
a&lt;b.
not apply to the integral
as fundamental
dx when
If
we adopt
this definition
(1)
J &*
we must extend
its
it
to such cases
by regarding the equation
as a definition of
*
righthand
side.
All functions
mentioned in these equations are of course continuous, as the
definite integral
has been denned for continuous functions only.
192
b}.lt. 156 cannot be negative. (".lt. then J f(x) dx a ^ 0. ^ a The reader will find it an instructive exercise to write out formal proofs of these properties.i following theorems are also important.t The v*n^V&.lt. .&. rb Uw. Then the integral is f (x] equal to rj (b a). It only to observe that the sum s of be shown later (Misc. Theorem of now to be only another call (8) way of stating the Mean Value for We may the First Mean Value Theorem Integrals. *  .If and Kf(x). * * i * (6) ///(^) = when a^x^b. + * ()} a ^=f V&.a). 41) that the value of the integral cannot be this may also be deduced from the zero unless /(#) is always equal to zero second corollary of 121. (4) * [ f J {/&.lt. This follows at once if we apply (6) to /(a?) (8) where % lies between a and (7). But.292 ADDITIONAL THEOREMS IN THE DIFFERENTIAL [VU (2) r &..&=&. then (x) dx K(b. This follows from lies For we can take H to be the least is and K the greatest value of f(x] in ?7 between for is value of If and K. since which /() = . we can write the result of (8) in the form so that (8) appears 125./(*) a /&.gt. H (a.gt. there must be a F(x] the integral function. b. ( 100). Ex. in each case giving a proof starting from (a) the definition by means of the integral function and (/3) the direct definition. . will : We have (7) If H ^ f(x) K when a^x^b. where continuous.lt.lt. ^ + ^6 ^ r/o) &/0) dx = kf f(x) dx.t H&.
gt. . [The truth of these equations will appear geometrically the graphs of the functions under the sign of integration are sketched. I J a \Kf(x}} The reader should formulate for himself the corresponding holds when (x) is always negative. and H and K are defined as in b &.160] (9) AND INTEGRAL CALCULUS 293 for inte (7).gt. The Generalised Mean Value Theorem If (f&.lt. 3. definite integral. I J J a #0(. (x) dx =/() ( Ja (/&. 2 ) dx=Q . (6) to This follows at once by applying Theorem the integrals (*) dx. .(x)dx.. defined as in (8).j&. that F (x) a continuous function of x. that direct definition of the fa (i) / 2 &. as convenient to follows as It is a was pointed out in 157. it is This has been proved already in 145.3) x} + .p J I (A. (x) dx. (x) is positive.gt.lt. ( J a *{/(*) H}^ (x} dx. j o sn. &.0 (sin x] dx : JQ rK Jo Jo m being an intuitive. Examples LXV.] 2. sin nx cot xdx is equal to or Lo TT according as n is odd or even. (ii) Jo rrmr (cos x} dx = / &. Prove that / . by means of the and equations (1) (5) above. but restate the result here as a formal theorem. if integer.lt. Show.1) x} last term being 1 or 2 cos xJ\ Prove that I +2 cos {(71 .(x)d9*\ J a f(x)$(x)dx^K\a J * and where % is J fa f(x) &. has a derivative equal to f(x).lt. 5 sin x dx is equal to tr or to according as n is odd or [Use the formula (sin 7w?)/(sin x) = 2 cos {(n .gt.lt.p result which (10) The Fundamental Theorem of The function the Integral Cal culus . grals. then H{a J j&.. corollary. Jo (sin x} dx = \ \ ". 1.) dx 2 [a / ra (x * 2 ) dx.
7r. (x} =a + ai cos n. J a 9(*) ax. I tan". LXIII.sin 2 sin 2 &..lt.(*)&. if [Use Ex.^&. Prove that * &. 913 are taken from Prof. I .lt.3a?) if and hence that M ^(4573&.lt. Jo 9*.**&.] 13. x dx. then 7ra . Prove that (3#+8)/16 1/^(43^ + ^ 3 )&..rc then the value of each of the last two integrals 9. ^o [The first  then V(l 5&. ^ [Use Ex.  and a the corresponding  ^ o. &.) * a=$ = 14.294? 4. a and b are positive then ^ ^ ___ ^dk.lt.lt.J = .J I a /b /(. + . 8 and Ex.sin 2 a sin 2 x) lies J(l . x + a 2 cos 2# +. . and & a positive integer not greater than then I cos#&.lt. 12. (^) gbr I ^ 3/ p.gt. tann + x x dx &. inequality follows . are positive acute angles then f J If 1 . [VII (#) = + i cos x + 6j sin x + 2 cos 2# . &.^2n the second from the fact that )&. Prove that P^^^SOS.lt. 11.r) &.gt.gt. . by 2 [Put^=l+w: then re 3 2 4w and 2 place 2 + 3w + % by + 2 + 3w 2 . LXIII.lt. is ^ If &. /dip Jo Prove that *".5M. + b n sin w#.** / &. [ sin n + J .& a /& 8. ADDITIONAL THEOREMS IN THE DIFFERENTIAL If is &.lt..lt. &. BID** &.\ f sin n ^^. ". + a n cos ?*o.W(4. then the integral I between 523 and 541.is the sum formed from the function /(#) then . f) &. Jo If n&.gt. . If (f&.gt.lt. 10. and k is a positive integer not greater than I". v\*".lt.&.lt. [Use Ex. n then the value of the Prove that LXIII.lt. .lt.f)(x)dz= f Jo yf Jo last integral is zero.". sum considered at the end [If o. Jo from the fact that [^ 1. ^^.gt.lt. If a and (f&.lt.=7rajt.(#)cfo. 6.lt. I Prove that \ J a f /(*) dx \ I/O) dx.r 156. .] zero.". 1 above.lt.+ cos w#.] /".r dx &. 5.lt. 10.  Jo. of &. * Exs. & &. Gibson s Elementary Treatise on the Calculus.
xm f( m + V (x)dx=F(b}F(a\ where rb More generally.lt. cf&.gt. Prove that ( J a 2. b 1. (b) d. Prove that r\ I ri arc Jo smxdx=^7T 1.1/^^ 3. Examples LXVI. Some instances of this will be found in the following definite integral examples. we then know 133) that if F(t) is Hence. AND INTEGRAL CALCULUS parts 295 substitution.lt. particular effected by the use of more advanced methods than are at present at our disposal. 161] 161. and sometimes even when the integral function cannot be found. we have (&)} I J c f(t) dt=F(d)F(c) = is F{&.lt. (x] dx={bf (6) /(&)} .160. (a)  J a /V (*) &. 1 #aro Jo . xf".f&. which the formula for the transformation of a definite integral by substitution. It Integration by 138 that follows from b and by Ja ( f(x) # (x) dx =/(&) * (b) /(a) is &. (a)} /*/{* (*)} J a f (x) dx. F {&. I J a **/^*)*^. That the value of a may sometimes be found without a knowledge of the integral function is only to be expected.lt. for integration by parts and for transformation enable us to evaluate a definite integral without the labour often The formulae of actually finding the integral function of the subject of integra tion. for the fact that we cannot determine the general form of a function F(x) in no way precludes the possibility that we may be two of its able to determine the difference F(b) F(a) between But as a rule this can only be values. (*) d*. ).{af (a) /(a)}.gt. This formula known as the formula for integration of a the integral function of definite integral by ( parts. Again. d if $ (a) = c.
*) ^+/(6) J [ ) dx.(nl) 3.] if f l (x}= \ Jo then f.6. o TT [Integrate by parts and use Ex. then there is a number J a f V(*) (*) to=/(a) f J a % (. Deduce that un equal to i 2.7..gt.4..(* Ja 160 generalised r i) Mean Value Theorem of : i.r then un + wn _ 2 7o evaluate the integral for n [Put tan 8.W according as 11.(nl) 2. JJ parts.(x}dx= ( Ja f(x}* (x}dx=f(b}*(b}. 4.t&.(x} = ^y. If /(#) is a function of # has a differential coefficient of constant sign for all values of x from between a and b such that to x=b.] if [Integrate repeatedly 6. by Prove by integration by parts that integers.296 4. Hence all positive integral values of n. x 5.gt.which x=a The Second Mean Value Theorem. Prove that if un \ smn #c?# by then u n parts.r (sec 2 # last and integrate by lies parts.n I Jo xm (\ .5.6.gt. LXIII.] ..] {(n l)/ft) w n2 [Write 7 o sin 71 ". um n 7n&. ^^.e.lt. X [Let ( * Then b J a Ja by the f f(x)&. n where m and n are positive then (m+n + 1) w = ?iw m&. = l/(w 1). 1) #=tan n ~ 2 . 1 x sin a? for sin n # and is integrate 10. J a which is equivalent to the result given.w _ 1 and deduce that m 7. ! n ! Prove that if un = I tann a? o?. x} dx. ADDITIONAL THEOREMS IN THE DIFFERENTIAL Prove that if [VII a and b are positive then x cosxsmxdx I. 8.] 1)} Deduce from the example that u n between l/{2(w and /Jn9. 7i n is odd or even. "..
X&.lt. \ \(ab cos y} n dy. ?r) into stitution.^ X &. ever is and note that the integral of sin x over any what numerically less than 14.l/(# Prove similarly that iff (a) and/ (6) /(a) have the same sign. y. Establish the results of Ex.6 2 ( n ~ ) i) Jo &. ^^ dx 2. then where will be X lies between a and 6. obtain the first equation to obtain the second divide the range (0. [Apply the interval first formula of Ex. a).a. AND INTEGRAL CALCULUS 297 (of) of constant sign. In (iii) divide the range into m equal parts and use the substitutions ^=7r+y.lt. case is that in which 0^f(a] Prove that!/ 12. LXV. . 0).. is a positive integer and a b .gt. \ Prove that / ^  n ^ dx=\ir 19.. .. [Put #=77 Jo 17. Bonnet s form of the Second Mean Value Theorem.0. Prove that J a 1 6.of. then y is f where JT lies between a and 6. J Prove that / cosm x sinm x dx = 2 ~ m I cos x dx. and evaluate the integral when 2. $ (sin .gt.gt. It found that the integral can be expressed in the form The important 13. \ F(x}dx={ a F(a + bx] dx. [Use the function *()=/ (#) d#.x)} dx=lir(b. when n 9i=l. Show by means of the transformation #=acos 2 + 6 sin 2 6 that J{(x a)(b. In (ii) use the substitution x=\ir y to (0.r) cfo. 3. [In (i) 1 by means of the rule for sub divide the range of integration into the two parts ( . : two equal parts and use the substitution x=%ir+y. Jo X(fr(smx)dx=^Tr Prove that / Jo 18. Show by means of the substitution (a + bcosx] (a ~n bcosy) = a ~l 2 b2 that Jo l(a + b cos x) dx = (a 2 . #=27r+y.] if &. . Jo i .161] 12.] 1 5. and so is the value of * (a?) for a value of x such as The important case is that in which 0^/(6) X &. and put x= y in the first.. 20. If/ and/ (6) and /()/(&) have the same sign. [For /(&)* (6) + {/()/(&)} *() = / where /* lies between ^ () and * (6).
+* Bn = r^fvJ^l *)". 1) &. l f Jo (i _ t) n~ f l (n) (a + th) dt = I (l o n = (1 where &.(6 . (2).. ADDITIONAL THEOREMS IN THE DIFFERENTIAL If [VII m and n are positive integers then b Ja [Put [ (xa} m (bxYdx=(ba} m + 6. If hand we take p = 1 we obtain Cauchy s form.r) ^ W now we a\th.lt. &.gt.&. + ( ^TT)1". write a +A and transform the integral by putting 7ni x we h) obtain (a) /(a + where =/() + A/ \n + . if p is any Theorem (9) of by 1 positive integer not greater than n.a) y. 1.298 21. . and use Ex.: (x) __ (*).Then and so F. .lt..lt. .. ?i ** * (l0)ni/w(afflA)a 147 can also be modified so as to obtain these The method used ". Fn (a) = If 1 J? . Taylor s Proof of Taylor s Theorem by Integration by shall now give the alternative form of the proof of Theorem to which we alluded in 147.1 ! / (n ". we have.lt. Now. (&.x)f (*). /t^i (*) =( \~^f J ~~ \". viz.] + i m!n! (m+n+I)\* x= a + (b . We Let f(x) be a function whose and let first n derivatives are continuous.lt.. 6) Pfw (a + Oh) ^ o (1  6 &.gt./ ) ij o (a + th)dt .gt. a &. 162.. (6)  ".. Parts. 160.in alternative forms of the remainder.. Fn (*) =/(&) /(*) . I Fn for (x) dx = a 6.. *^ ! ^ f (6 a *)/&. Hence If we take p = on the other we obtain Lagrange s form of J?n ( 148).
The truth of the Binomial Theorem is thus established for all rational values of m and all It will be remembered that the difficulty in values of x between 1 and 1. 13. AND INTEGRAL CALCULUS 299 f (x} = (\\x) If Application of Cauchy s form to the Binomial Series. by the equations rb J a rb J a rb J a and it is evident that the properties of such integrals may be deduced from those of the real integrals already considered.1 (1 0)/(l K 7* ". rb and so * I J a fdx = lim  %f & = lim S/r 5. being in fact less than (l + l^l) 1 1 if wi&.lt. so long as 1&. difficulty from the 156. whether ~ for or negative. So far we have always supposed that the a definite integral is function f(x) ty (x) limits rb J a + ify subject of integration in define the integral of a complex (x) of the real variable x.#&. arose in connection with negative 164. then Cauchy s form of the remainder is ) ^^Trlfe^ Now x is all (l\x\) + 0#) is less than unity. using Lagrange s form.lt.1.lt. m where m is not a positive integer.l and than m. 14. $v + i Km S ^r = . LXV. v v  The corresponding inequality for a real integral was proved in Ex.lt. and (\ + 6x} m l is less than a constant positive values of n. values of x.ity v ) $v lim /". This inequality definitions of may be deduced without 156 and 157.(!)*. If of &. v $v 2 (&./&.lt. in Ex.161164] 163.gt.t&.lt. and &. between the We a and 6.. real. is one of these properties that It is expressed by the inequality we shall make use of .gt./&. xxvn.gt. 2.. ^ are the values then we have rb &. Integrals of complex functions of a real variable. if m&. and so R n *~0. There later on.lt. has the same meaning as in and ^ at a point of B V} and iijr v} fdx = J a J a I rb rb I dx + i t I Ja tydx = lim 2 lim &.gt.l Hence say But p n *0 as w^oo by Ex.$. LVI.lt. . 4. .
a &.lt. Verify the terms given of the following Taylor (1) (2) s Series : (3) (4) 2. where a &. first [Here three orders then (b 4. 4. value of 6 which occurs in (0)4=0. b. [Apply to the function ) / (*) /() J (*  {/ () +/ 3 (*)) (E". ex ^0 as #0. Show that under the same conditions 6. LV.a) C/(6) 147. and 6 n is the after n terms of Taylor s Series.gt. W + 4/ i* ( a+6 ~ 1 6 ) 5 / Show that under the same conditions () + i (6 ){/ () +/ (b)} ~ ft ( 6  2 ) (/ ( 6) /".S r. (a).lt. and Lagrange s form of the where 3.300 while ADDITIONAL THEOREMS IN THE DIFFERENTIAL [VII J a f /^ = lim2/.] ~ /(a) ~ 4 (& ".lt.] Verify the last result when/ (a?) = !/(!+ #). [Follow the method of Ex. Show that if f(x) and its first y&. The result now follows at once from the inequality It is evident that the formulae (1) and (2) of 162 remain true when f is a complex function &.a) 3 /". Show that if /(#) has derivatives of the / (6) =/ (a) +1 (6 a) {/ (a) +/ (6)} & . ( a + 7^ ( 6 ~ .ity MISCELLANEOUS EXAMPLES ON CHAPTER 1. then remainder + i) n + 2 derivatives are continuous. 12. Show that if /(#) has derivatives of the first five orders then 2H Sa ( / (6) =/ () +*(*) [/ () +/ 7.gt. arguments similar to those of 5.f&. VII.
. and #=c. lies n (x)}lh when n = 2. 301 Establish the formulae /() where lies /() 6. and so on. ^ x .b) A (a) h (a) h (6) A (6) A (c) which vanishes when x=a. will now complete the proof without difficulty. Apply this result to (*!)! and deduce Taylor s Theorem. Its first derivative. x=b. /3 between a and /(&) and /() A (a) /(&.lt.( &. then where  {A h n (t&. then A (x jn \}&.lt. We thus obtain the formula A (a) A (b} A (c) A (a) A (6) A".0.1 vanish when # = 0.0 ! / \ I i / A lie (6) A(c) A  (a) c.gt.F(x}^B (x /n n n !) when &. . n between x and x + nh.gt. of &. Deduce that if (x} is continuous then ^ ( ^. [This result has been stated already (a?) as A ) (f&. in Ex. 1 1. A (3) A". [To prove (ii) consider the function /() g(a} g(b] /() g( A ( a)(x.) &. first Afc 1 ^ () a?)}. x^ A. n Deduce from Ex. by Theorem B of 121. and A B when (x} &. .) where j8 and y between the least and greatest of a. 10. F(x) is a function which has continuous derivatives of the first n which the first n. In particular prove that . 10 that xn ~ m & hn xm ^ m (m .lt. c.W 1)".AND INTEGRAL CALCULUS 8. LV.lt. 7) If A$(*) 4 (*)$(+ A).lt.. 6. must vanish for two distinct values of x lying between the least and greatest of a. 13. x ^ A.lt.1 ) (m n + l)h as m being any rational number and n any positive integer. and its second derivative must therefore vanish for a value y of x satisfying the same condition.lt. and has derivatives of the n orders.] .gt. F(". j.(y) The reader 9.] If orders.(&. 6.
. [Cf.] .$&. Misc. Deduce that the conic of closest contact at the point (. assume the continuity of . of the curve y=/(#) where T=(yn }ni 17. y). at the point (#..) then u is un when x. . so that y= where ex (x) =X + (2a 2 2 # 2 + 3 3 X +( 4 f f ^ &.. are all increased in the ratio A 1.. and that $ 2 = 0. . for y that value of x which vanishes with y f 3) 2  \ and prove as # . 1899.. y). 77) of the centre of curvature of the curve x=f(t\ y = F(t). = nu. (0) = 1.lt.lt.gt. are homogeneous and * In this n 1. t. z/x. Establish the formula 2 x = ^ (y ) =y where that ey y as y^0.. 1907. are given by . ADDITIONAL THEOREMS IN THE DIFFERENTIAL Suppose that y =&. are given by and the radius of curvature of the curve is dashes denoting differentiations with respect to 14. Trip. &. a y = a*x 3 2 +a 2 bxy+(acb*)y rj) 2 . y) will have contact Prove of the third order with the curve if (l+yi 2 )y3 = 3y 1 y 22 at that point.. The coordinates (. 3a(+#) + 2#2 =0.. Trip. as #*0. z.. x*.du du du dy This result 18. . y oz air". homogeneous and of degree n then du/dx.t (J/aA. Ex.z. VI.* If u=x f(y/x.(#) is [VII a function of x with continuous derivatives (0) 3 of at least the first four orders. (x^1 . . The coordinates (.gt.y. is known is If u Theorem on homogeneous functions. 10 (iv). du/dy.) n altered. 17) of the centre of curvature of the curve 27a?/ 2 =4. Homogeneous functions. at the point (#. 3 .) Prove that the circle of curvature at a point (x. is to all and the following examples the reader the derivatives which occur. . Ch.302 12. 77=4y + (9ay)/.. . also that the circle is the only curve which possesses this property at every and that the only points on a conic which possess the property point . The conic is of closest contact with the curve y = ax2 + bx3 + ex* + is + kxn . at the origin. as Euler s of degree . are the extremities of the axes.+ 9 ^r + z ^ +. n : In these circumstances u is called a homogeneous function of degree n in the Prove that if u is homogeneous and of degree n then variables #. y.. (Math.0. 15. save for a factor X . 1 6.. 13.
we 90 du du dy 90 cu du 9. y.AND INTEGRAL CALCULUS 19.y) Similar results hold for functions of any number of variables. v)=0 obtain (1). Dependent and independent functions. all values of x. Two functions u and v are said to be dependent or independent according It is usual to call as they are or are not connected by such a relation as (1).Z 90 dv 90 dv dv dx dv 9y and. w of three variables #.af 2 . v. . v. y. z= = I. v} 9&.lt. /the Jacobian or functional determinant of u and v with respect to x and y.u v*= ( 3 )&. y) = is where Ff F^. 303 n n Let/(#. eliminating the derivatives of 0.gzx + 2.f&. It can be proved that the condition is also sufficient for this we must refer to Goursat s Cours tf Analyse.gt. . i. Show that the equation of the tangent at the point (. and z if and only if i l Show l product of abc + 2fgk . z.ck2 = 0. = vl . and let F(x. This condition is therefore necessary for the existence of a relation such as (1).gt. rj) of the curve/(#.bg* . z when x=g.(u. and to write J= 3(u. [Write down the condition that px+qy+rz and p x+q y + r z should be connected with the given function by a functional relation. # +y #=0). vv = u * v . z are or are not connected by a relation (u. Suppose that u and v are functions of x and y connected by an identical relation &.g.kxy can be expressed as a two linear functions of x. Thus three functions u. 20. . y) = be an equation in x and y (e. vol. 125 et seq.gt. xn \yn xz n l = Qi).g. . w)=0 according as uy J= does or does not vanish for 21. F$ denote the values of t Fx Fy F .y + 6y 2 and Ax2 + 2Hxy + By l 1 are independent unless a\A 22. Differentiating (1) with respect to x and y. that ax2 + by +cz 2 + 2fyz + 2. li\H= b/B. Show that ax2 + 2/o..] .y=r). where ux uy vx vy are the derivatives of u and v with respect to x and y. pp. z}=0 be the form it assumes when made homogeneous by the intro ~ duction of a third variable z in place of unity (e. y. Jacobians or functional determinants. y.
B / arid so that the ratios a 30.C. Trip... 9 &. .] a relation in (ii) virtue of is Suppose that three variables x. that A A B B C . 1900. . Prove that if / (a?) = * * . v such that vanishes identically and conversely.. show must be/(#)+/(y) =f(xy). Show that if a functional relation exists between then / must be a constant.. of the determinant that /3y /3 y and it follows from the vanishing To ... p. y. 1909. C are three functions of x such A". relation then /to x) is independent of x. . then we can find constants X.. with derivatives a?y xz . (Math..) A.) are the equations of three homogeneous as in Ex. that u and v if Let/ (#) be a function of x whose derivative Show that if u =f (x} + f (y ). 19.. [ J o then vC 1 f*) 26. v=0. x\ and /to y) are connected by a functional circles. But aA+$B + yC=0.gt. Prove in a similar manner that of f(x) is 1/(1 +#2 ). : : y are constant. [The condition for a functional relation will be found to be / (*)/ 27. then f(x] must satisfy the equation 25. (y)/ () {/(y) /(*)} {/(*) /(*)} (/(^) /(y)&. If them orthogonally.gt. (Math.. and Prove that a function of y and 2. is when x\. result to any number of variables.?) to y) Extend the 24. vanishes identically. [The converse is almost obvious. and /(0)=0.gt. By putting y = l.304 23. 2). [VII u and v are functions of then and ?. then are connected ux vy \\x and which vanishes u y v x = 0. = l If /(y. z are connected by which (i) z is a function of x and y. . with derivatives zx %. 2) all represents the circle which cuts 29. y.B". . If M=0. w = rendered 8 (a?. /(z.. which are themselves functions of x and _ 9 to y) s (&. ADDITIONAL THEOREMS IN THE DIFFERENTIAL If y. B. x . B". Trip. v = xy. then the equation 28. an d hence that this relation the derivative by a functional relation.. C. C C". prove the direct theorem let a BC = 0.. = Then a = BC".
&.gt. 0. Show that where y* denotes the derivative of y. when expressed as a function of z and u.b 2 &. u are connected by two relations in virtue of which any two can be expressed as functions of the others. (Math. In connection with Exs. 35. xxxv of the Messenger of Mathematics.x/(a 2 . If a &. 1913.lthen f ^ ^2^=7r{a. X. and TT TT. In the latter case the inverse tangent lies between and [It will be found that the substitution t= X ^? reduces the y+yo integral to the c?r /a *TV\ 1 77=5 a ~^ form 2 [ Jo ov=j7r. z. * Ifa&.1)}.) C. Trip. and A.gt.) 37.lt.0 when ^ ^^^^. 38. Four variables x. C.) [The value of the integral (2n is \ir if . 33 and 40 see a paper by Dr Bromwich in vol. when #=0. Trip. z.* Evaluate the integral ma ? / J _! 1 . Find A. TI being any integer Zmr and &. and  IT.lt. 1897.gt./(^) = v/ (^ 2 + 26^lc). 20 .gt. with respect to 32. t which can be true only if zx xy +zy = Q zx xz =\I\ 31.(2n + l) TT. / r^aJ i2 J (J/a^. 2ft7r. 0.gt. 305 dx = xy dy + xz dz result of substituting for dx (zx in the first equation is dz = xv +zy )dy + zx xe dz.a&. x0) then _ _dx__ J Xo ax* + 2bx +c o p *J(ac the inverse tangent lying between 34. ^n&. H.1) TT 35.. TT.gt.lt. . and x l &.] if if a is a multiple of If a^ 2 + 26^7+c&. JB. and according as is positive or negative. B. ac . y. ^ 2xcosa + x  For what values of a is the integral a discontinuous function of a (Math.lt. a &. /* so that the first six derivatives of ~a+x vanish 33.AND INTEGRAL CALCULUS [We have The dz = zx dx + zv %. 1904. D. A so that the first four derivatives of vanish when i: #=0 .
gt. then [ J a . .gt. \2 rb &._ a )( 3. a)/(2w but that if m=n then the J a value of the integral 45. which possesses the property that fft I Qn (x] 6 (x) dx = if 6 (x) is any polynomial of degree less than n. ADDITIONAL THEOREMS IN THE DIFFERENTIAL Ifj0&. b &. If Q n (x) is a polynomial of degree n. and then the value find an interval (d.l. = ^.] 42. pB (a. KPn is [We Qn of degree n.) 40. any polynomial of degree less than n. and I f(x)dx=0.] If 156 and 157. 10). If a sin &. then we could.). Prove that I Pm (x) Pn (x) dx=0ifm* n. and the inequality (Ch. and observe that 0( (A) if 6(x] is m+l 1 ) ffi 44. 0&.2&.] times.t&. 0.306 38. 43. Schwarz s inequality for integrals.)=0 for all values of # between a and ft.gt. where m is the degree of 6 (#). o&.lt.lt. in virtue of the continuity of /(#). Prove that rb ab a &. Prove that / if a &.gt.1 : then . Qn (x) is a constant multiple of can choose * so that Pn (x}.gt. which possesses the property that /.lt.)} f then P tt (*) is a poly IP ) \ nomial of degree n.gt.)_L^^{( a .~ a n (a. then y a y(a. Ex.lt. J [Use the definitions of Misc. [Integrate by parts m+ = O.+dx) / J a tfdx] a y*dx. + ? sn* where 39.gt.^k of the integral would be greater than .gt. J the inverse tangent lying between 41. say. and rb lff(x) is continuous and never negative. [If/ (#) were equal to a positive number ^ when 0. I. 1904. is (/3 + 1). is the positive acute angle whose cosine (l+pq}/(p + q). x/(& + c2 then 2 d6 o n TT.a / . then J a. ^f ) throughout which f(x}&. l [VII then o P ~ ** dx *   2 is /&.1. Trip.b cos 2 ^^ = ) * ft 2 {a c _ x/(a 2 _ 62)} (Math.
Trip. } 3 (V(4 + ^)^&.a) 5 where is the maximum of 0( 4 )(#).] / (#) = in the interval (a. is known as Simpson s Rule. 4) (x} is any cubic polynomial then 0( (x} = 0. [In the last integral the subject of integration #=0 but if we assign to it. and a parabola with its axis parallel to OF and passing through the three points on the curve y=0(#) whose abscissae are a.. Show that the error (x) than ^M(ba) 3 . dx is less is the a) and that the error in taking (b 5. Now apply Ex. x=b. 4 and Show (a) that the error in taking (6) . b. where M . Approximate Values of definite integrals. \ (a + 6).lt.lt. and Simpson s Rule . Calculate the integrals 1 1 two places of decimals.. and this curve is a parabola. /3 being the roots of the equation # 2 #+ ^= (Math. b. we can draw through them an infinity of curves of the type y=a+/3x+yx 2 + 8x3 and all such curves give the same area.a) 3 [Write (a + b}} is less than ^M (b {^  0". 0.AND INTEGRAL CALCULUS and so { J * 307 (On  K Pn 2 ) dx = 0. 6.) TT Apply Simpson s Rule fin to the 7833. This rule.) 50. Show 89 &. given three points whose abscissae are a. i as the value is less (b [0 + + 40 {$ (a + 6)}] than ^fe^M (b ..(#)  .a} /b a. 41. 47. which gives a very good approximation. (x} in Exs.] 49..9.. calculation of If from the formula integral into two.] to is it not defined when : 202 . becomes continuous throughout the range of integration. b) maximum of . (Math. is we divide the and and apply Simpson s Rule to the two separately. The correct value is 7853981.J 46. when #=0. If (x} is a polynomial of the fifth degree. . a and 48. Trip. For one curve 5 = 0.y = 0.] M Show that the approximation assigned by Simpson s Rule is the area bounded by the lines x=a. 1903. [Use Ex. It should be observed that if is exact... 1909.. then (a?) dx = Jg (50 (a) + 80 ( J) + 50 03)}.. the value 1. J (a + b). That is to say. integrals /I j^2from to  [The result to 1. It amounts to taking onethird of the first approximation given above and twothirds of the second. we obtain that 7853916.
LXVII. when dealing with the series 1 . that an infinite series illustrated IN Ch.. We shall therefore adopt this as our standard But we shall not adhere to it systematically. is clearly more convenient.gt.. 166. for the infinite series u Q \u l {ii 2 + . + 4 + $ + . . e. IV we explained what was meant by saying is convergent.* The. and that the series begins series begins with M O . (as in Ch. simplest series which commonly occur in analysis are convergent.. (as here). 4.: for these the latter notation notation. theory of the con vergence of series is comparatively simple when all the terms of shall consider such series the series considered are positive f. and we shall suppose that u : the first term whenever this course is more convenient.g. . or simply 2u n . to suppose that t* n =l/ with Wj than to suppose that w n =l/(n + l) and that the This remark applies.CHAPTER VIII THE CONVERGENCE OF INFINITE SERIES AND INFINITE INTEGRALS 165. divergent... It is more convenient.. shall be concerned with series of the type a + a 1 x + a 2 x 2 + . is for example.. . and prove a number of theorems which enable us to determine when the and other series closely connected with it. and our definitions by a few simple examples.. We * It is of course a matter of indifference whether we denote our series by Later in this chapter we Ui + U2+. Series of Positive Terms. to Ex.. mainly derived from the geometrical series In this chapter we shall pursue the subject in a more systematic manner.. t Here and in what follows * positive is to be regarded as including zero. IV) or by U Q + U I +. We shall often use the notation and write ^u o n&. . or oscillatory.
oo then 2v n is convergent (divergent). to oo . all values If %u n vn ^ u n for of n. The comparison theorem. . * will The last part of this difficulty in theorem was not actually stated in 77. then 2vn is convergent and *5Lv n ^K^u n And if %un is divergent. + un &. The one important fact which we know at present. finite number only of negative or complex terms. . First applications of these tests. particularly useful case of this theorem is If %u n is other than zero as convergent (divergent) and u n /v n tends to a limit n *. The necessary and w sufficient condition that %un should be convergent is that there should be a number K such that + MI + . and and %v n ^ 2u n More generally. . but also because the discussion of the convergence of a series containing negative or complex terms can often be made to depend upon a similar discussion of a series of positive terms only. and cannot B. .lt. if v n ^Ku n where is a constant.v n is divergent* by means of one of these Moreover. Thus. vn ~ Ku n .165168] first. v ITSiin . but the reader have no supplying the proof. When we series we may are discussing the convergence or divergence of a disregard any finite number of terms. for all values of n greater than a definite value ?? But of course the con clusion that does not necessarily hold in this case. A series of positive terms must be convergent or diverge oscillate.e. A D. then 2&. THE CONVERGENCE OF INFINITE SERIES. when a series contains a remainder. as regards the convergence of any 168. . %v n is K convergent. i. 77. C. K is for all values of n. it is sufficient to know that the test is satisfied for sufficiently large values of n. in inferring the convergence or divergence of 2? n tests. and convergent. we may omit them and apply the theorems which follow to the 167 It will be well to recall the following fundamental theorems established in A. .lt. ETC. then . 309 not only because they are the easiest to deal with.
in that Cauchy s can be applied whenever d Alembert s can. that if v n+l /v n ^ r = 1 for all values of n. and the result follows by comparison with the convergent This test it is series 2r n . %v n is l n divergent if v n ^ 1 for an infinity of values of This hardly requires proof. whereas in Theorem 2a is We shall use r in this chapter to denote a number which always positive or zero. than Cauchy g. con vergence of a series of positive terms 2. when n = n then . We at once find The is convergent if v n ^ Krn where . series is The Zv n = r r wnere r &.310 THE CONVERGENCE OF INFINITE SERIES is [VIII if special class of series. r &. convergent if v n +i/ v all sufficiently large values of n. We shall see later that less general. C. for v nlln = 1 involves v n = l. is known as d Alembert s test. where r &. this condition is Hence we obtain what may be written known as Cauchy . It is true. not true (see Ex.lt./br all sufficiently large values of n. . all sufficiently large The series *v n ^ r.* un %v n It therefore try to apply Theorem 1. But it is or all sufficiently large values. in the form v nl n s test for the ^ r./ or To prove this we n observe that V n V n i if vn+ i/vn = ". viz.lt. is. theoretically. then %v n is divergent. &. for There 2a. The n. 1. and sometimes when the latter cannot. When K 1. that is Hr 11 is convergent natural to r &. taking series rn .gt. 1 and divergent if r^l. as the general than the test given by reader will easily prove for himself. Moreover the test for divergence which test for convergence is much less corresponds to d Alembert s test Theorem 2a. viz. viz. l&. a corresponding test series for divergence. LXVII. ~ Vno+i ^ v ^ni 7 n~2 ~^~ ^?i n ^o = ro r &.lt.lt. 3.lt. ln is convergent if v^ values of n. 1. but for some purposes it is more convenient to use a different test of convergence. 9) that this is so if only for v n+l /v n * ^r^l an infinity of values of ?i. The two theorems 2 and 2a are of very wide application.gt.
4. LXXXVII. Consider series ffiS+. while always less than 1. proof of this as an exercise to the reader. Since lim n lln = 1 (Ex.oo . 11).0).gt. be proved in Ch. It is almost obvious that if either function tend to a limit greater than unity. Ifv n lln or v n+ Jvn tends to a limit less than unity as n then the series 2v n is convergent. [The * behaves like The case in which r=l. If the coefficient of r n is denoted n 167.!.] It will That the converse is not true and v = 2when n is even. Cauchy s test leads at once to the same conclusions.oo . In the simplest cases which occur in analysis it often happens that vn+l /vn or v n lln tends to a limit as n *.gt. it assumes for an infinity of values of n values approaching indefinitely near to 1. n n may be seen by supposing that v n = l when n is odd . 86) that if v n+l /v ** I then v l /n * I. But it is clear that there is a wide margin of cases in which some more subtle tests will be needed. &. +A". And the tests which involve v n +i/v n fail even when that ratio oscillates so as to be sometimes than and sometimes greater than 1. +f ".l requires further consideration. 0+i/tV. by P(ri). [Here v n + : 1. the series behaves like 2?i*r . where k a positive rational l lvn ={(n is + I)/n} k r^r.a&. positive. by D 3. }r n. so that d Alembert s test shows at once I and divergent if /&. [We may supposed then P(n)/nk *~A and. k&. xxvn.0. Consider the series of 2(Ank + Bn k ~1 + .lt. 2?i k.. The test fails if convergent if r r= 1 but the series is then obviously divergent.gt. number. When v n lln behaves in way Theorem 2a is sufficient to prove the divergence of the series. because vn is a complicated function v n+1 /v n is often much less complicated and so easier to work with.168] AND INFINITE INTEGRALS 311 None when our test had only to be satisfied for such an infinity of values.* When this limit is less than 1.] ". less this Examples LXVII. IX (Ex. Apply Cauchy s and d Alembert s is tests (as specialised in 4 above) to the series 2wV. and they way that. (^&.] that the series 2.l n r . then %vn is divergent. Thus *. it is evident that the conditions of Theorems 2 or 3 above are satisfied.lt. tends to 1 these tests generally fail fail We leave the formal v n l!n or But when also when v nlln or v n+l /v n oscillates in such a completely. the less d Alembert s test is very useful in practice..
applied to them. n . Show p convergent.. Ex. For example the series 1 and divergent if r convergent if r 1 &. Sum the series l + 2r+2r 4 +.lt.. It has already been shown in 77. Misc... convergent 7.. and to 2 places when r= *9..lt.lt. . r r . .. . + r +r* + 9 + 8. If r= 9.&. . [For 10. mn + &. is convergent. .gt. is [Since lim {*(+!). then the first 8 terms give the sum and the error is less than 9 2^/(l?*i7)&. . i&.3/10 . 25 ) &. 5.. 2 (1/w 2 ) 13.un + (Iln 2 ) and 12. 4.. .lt. IV. [If *=*!. If is 2w 2 re is 2un /n. a positive integer... and l+r+ r + p + .] Show that l4 + 4^. 003. if l&.m then the series 2rmn 1 is ? con .l+l4.lt. = 2r 25/(l .r 11 5*458.. but that d Alembert s test fails.gt. may be applied to this series.312 4.and it is obviously divergent if that the series .gt.1..lt. and the error is 2r*5 to 24 places of decimals when r= l first 5 terms give the + 2^ 6 + 25 .lt.. vergent is and divergent if r 1. THE CONVERGENCE OF INFINITE SERIES [VIII We have seen (Ch. then the sum 1*2002000020000002. Show that the series ~l an is l + l fin + .] fail when 2w~ where p is an integer not less than 2.&&. Show l/H [For lim un = lim (un + .lt. and &. Theseries l+r+^j + ^ + r.. &.. If 2w n is convergent then so are convergent then so is 2^^rl2 and 2 [For 2u n /n&.lt.Jc + \ and divergent if l&. (7) that the series is divergent if /? = l. 17) that the series l are convergent.] If &.lc +K + \. r are convergent for all positive values of 11. Show that Cauchy s test then the series a + & + a 2 + 6 2 4a 3 .lt.. convergent. 6. and . If if mn is r &.(n+p !)}/* =1.!.2r + 2r36 + 2r47 + ..gt./wn ) = that Cauchy s and d Alembert s tests both 1. this follows from the convergence of the series considered in Ex.
] (8) and (6) of Prove by a reductio ad absurdum that 2 (I In) is divergent. +p+p+". But we can show in exactly the same way that s ^ t. say. 170. : of u a. and therefore the us. sum of any number of terms selected from the v s is not greater than s. [If the 13. and iu particular to Cauchy. is not greater than s. An Multiplication of Series of Positive Terms. we shall prove an important general theorem concerning Dirichlet s terms is the series of positive terms.**4 77. Let s be the sum of the series Then the sum of any number of terms. . and its sum t is not greater than s. since every term of the corresponding term of the second. It * This theorem seems to have first been stated explicitly by Diriclilet in 1837. + x 2 .] first series is less than the 169. . . Theorem. and vice versa. was no doubt known to earlier writers. immediate corollary from Dirichlet s Theorem is the following theorem if u + w x + 2 + and v + v : + v2 + are two conver : . Thus s = t. .168170] [To prove the AND INFINITE INTEGRALS first result 313 we note that i byTheorems 14.* The sum of a series of positive same in whatever order the terms are taken. selected from the But every v is a u. and form any other series out of the same terms. Before proceeding further in the investigation of tests of convergence and divergence. . then the second series is convergent and has the same sum as the first. . u we have a convergent series of + w + u. by the argument used in Ex. by taking them in any new order. series were convergent we should have. if This theorem asserts that positive terms. . or which is obviously absurd. The proof is extremely simple. Hence %v n is convergent. . Of course no terms must be omitted every u must come some where among the v s.
.2 v u 2 v^ and u. Thus the theorem is proved. of these groups of terms are respectively equal to ..o 2 Ui) (V + Vj). We begin with the single term U O VQ for which (1) for which then we take the two terms u^. begin with the single term U Q V O for which both then we take the terms U^VQ. by Dirichlet s Theorem. When of the the sum of the series is .. rectangles The sum of the series formed in the second manner is st. and tends in this to 5^ as n  oo .2 Vt. We thus obtain the series of the theorem.. m+ u^ m . . two.. comprises all the terms in the first. second. third. then the series (U 2 V is +U 1 V1 + M Va + convergent and has all the sum st. Arrange the possible products of pairs u m vn in the form of a doubly infinite array UV We can rearrange these terms in the form of a simply infinite series in a variety of ways. (2) We suffixes are zero. . and therefore. . formed groups manner the sum first one.314 series THE CONVERGENCE OF INFINITE SERIES of positive terms. it con verges to the sum st. n = 0.2 v u v lt u v 2 for which m + n = 2 and so on.U V + Vi + V ) . . u^. and s [VIII and t are their respective sums. The sums VQ . . . . But the first series is (when the brackets are removed) a rearrange ment of the second. + n = 1 then the three terms u.gt. u v. u l v l) u v l which () involve a suffix 1 but no higher suffix then the terms u. r . three. so on. Among these are the following. indicated in the diagram above.2 which involve a suffix 2 but no higher suffix .Oo + (U + U! + U ) sum of the first n + 1 groups is (U + Mj) (V z (&. U and the + Vi) .
r 2 r 3 . For example. . (2/3) 2&. w n = 2uivm . except in the trivial () 1 ..lt. xxvu. .36 .* If either of the series w + % + .lt. . The tests is much We viz.l. 2 12 = 10. } n It follows that the sequence convergent. p such that m + n+p = k. WQ + WI + . r n 2 *. t. in which u n+l /u n tends 168 generally fail when this limit and we had to fall back upon a special device. converges to the sum st. 311 313 suffice to show that there are simple and interesting types of series of positive terms which cannot be dealt with by the general tests of 168. diminishes more rapidly than the sequence This seems at first paradoxical if r is not much less is than unity.!..lt. n. w 2 vo + w i yi the series Vo + (% l o + Vi) + ( v +v + is divergent. 5 Thus of the two sequences the second seems at far 95 . is zero. . 2n k rn . + . s. LXVII. necessary in order to ensure convergence...lt. to a limit as n ^ oo the tests of is 1.lt.l/5.. If the series ^ + w. what may have...(l/10) (2/3) oo&. Verify that if r &. the geometric series. r.(l/5)3&. more rapidly than is derived from comparison with it are therefore naturally very crude.. The fact is that the geometric of far 168 were obtained. sight to go far enough into the sequences we shall very much the smaller... 3 * } I k^ ..S. converge to sums then the series 2A A where \ k = 2u m vn iv p the summation being extended to all sets of values of m.. rst. and in Ex. series. instead of . . If ". . 171. if we consider the simplest type of series. . . 1 we proved more than this. . LXVII. 1 then 2.lt... that the series ..(i/io)i60 ) while * 1000In Exs.. and more delicate tests are often wanted. t +#! + . ever value k n proved in Ex. 4 1 from being the case if only we find the terms of the first sequence 2 1 . ) case in which every term of one series 3. Thus in Ex. In fact. 4 as our comparison series. But this is first . provided r&. . Further tests for convergence and divergence. by comparison with which the tests not only convergent but very rapidly convergent.. m for which lm = n. (2/3) =16/81&.170..u .. which was in essence that of using the series of Ex. The examples on pp. 7 that n*r *0 as tt^oo. converges to the .... . . LXVII.lt. 5 these tests failed. then so is + uQ v2 + . then the series 2wn where n and 2v n converge to sums s and the summation extending to all pairs I. 4 the series considered are of course series of positive terms. whose general terms are () n and n~ 12 decrease far more rapidly. and k is large. 171] AND INFINITE INTEGRALS 1 315 Examples LXVIII. where ~ fc ?&. . sum 4.
. No. * Five terms suffice to give the sum of 2n~ 12 correctly to 7 places of decimals. Then it is possible to find a number d such that nu n 2i$ for an infinity of values of n.31(5 THE CONVERGENCE OF INFINITE SERIES [VIII so that the 1000th term of the first sequence is less than the 10 130 th part of the corresponding term of the second sequence. But before we we shall state t 173. 12). and prove a simple and important theorem. very much more 172. . . Jackson) to the author s tract Orders of Infinity (Cambridge. assumed only that steadily with n: all sufficiently i. positive Abel s (or Pringsheim s) Theorem.* We shall proceed to establish two further tests for the s convergence or divergence of series of positive terms. so long as we are may for in virtue of Dirichlet s theorem dealing with series of positive terms above we may rearrange the terms without affecting the question of con This condition it point of view : vergence or divergence and there is nothing to prevent us rearranging the terms in descending order of magnitude. or at any rate large values. but it is essentially of a more elementary character than the two theorems mentioned above. From one be regarded as no restriction at all. are Cauchy s) sufficiently general for our needs in this chapter. Math. and applying our tests to the series of decreasing terms thus obtained. (or Maclaurin Integral Test and Cauchy s Condensation Test.000 are needed to give an equally good approximation to 2/i~ 2 A large number of numerical results of this character will be found in Appendix III (compiled by Mr J. Thus the series 2 (2/3) n is far more rapidly convergent than the series 2w~ 12 and even this series is . and decreasing terms. that it gives a sufficient test for divergence only and not for convergence.000. 2 rapidly convergent than 2?i~ .e. If 2w n is a convergent series of Suppose that nun does not tend to zero. / we proceed to the statement of these two tests. it is positive. We assume that u n decreases that v n+l ^ un for all values of n. as to the nature of the function u n In applying either of these tests we make a further assumption about which we have so far . though very far from being completely general. which. f This theorem was discovered by Abel but forgotten. Tracts. which This is a onesided theorem in shall call Abel s Theoremf. Let n be the first such value of n n 2 the next such value of n which is more than positive . then Km nun = 0. and rediscovered by Pringsheim. whereas some 10. is satisfied in all the most important cases.
^n 3 .. &. n 3 . nz ni&. since * nu n =1. 2^1? ^3 &. 1 [The series + 2^ + 32+4+52 + 62 + 2 72 + ^ + 9 + ^Q2 + &.gt.. 1 Ji &. since may according as n is or is not a perfect square. the next eight by J.gt.n 2 . .gt. convergent. 2 ja.d. . since its terms are greater than those of i+iHii+iH.gt.gt. in which u n =ljn or it 1/ra .gt.~ &. series is convergent.] The converse of AbeVs theorem is not true.lt. the second by . 2^ 2 .] is (!/TI) and 2. a new series whose terms are severally greater than those of the divergent and so series and therefore 2w n is divergent. since &. n 3 the next such value of n which is more than twice Then we have a sequence of numbers n^ n%. are divergent. the next two by . On grouping in brackets the terms of the new series thus formed we obtain and this series is divergent. decreases with n and lim nu n =Q. . But. the next four by J. we have &.gt. Bub it is easy to sec that the terms of the series i+KJ+i/i+ii+M+iHsatisfy the condition that nu*+~Q. 3. 2 {I I (an + b)} 1. 2".gt. such that w 2 &. . and so n 2 . . whenever u is a it is clearly not true that 0. &. Show that Abel s theorem not true if we omit the condition that u n decreases as n increases.. which is divergent.lt. it is not true that.... and also niu ni n 2 um &.. u n decreases as n increases.e. . .. i. Use Abel s theorem to show that 2 [Here nu n +l or nu n *lja. Examples LXIX. as large as ?i 2 and so on.". Thus we can bracket the terms of the series 2 u n so as to obtain on. if un [Take the series 2 (Ijn] and multiply the first term by 1.171173] } AND INFINITE INTEGRALS 317 twice as large as n . be rearranged in the form is and each of these perfect square. then 2?/ is convergent.n 1 But... and so on. and .gt. In fact nu n =\\v if 2 2".
to whom usually attributed.(!) + .1&./&.+^ _ O(n) 2 1 tends to a positive limit.. J\ (f ) is a continuous and steadily increasing function of l . steadily as n increases. n ^ oo .318 THE CONVERGENCE OF INFINITE SERIES 174..0) when v 1 ^x v.u v is (1).lt. and decreases steadily as x increases. as &. in the first case.gt. &. 2 3 or tends to a positive limit as n^oo .0)^&.gt. and so v + v + . Let !). was discovered by Maclaurin and rediscovered by Cauchy. i*5 a (a.. not greater than ^.. we have &.gt.lt. &.gt. and.I&.. since (n) increases steadily] as tends to a limit or to infinity s f * oo Hence (f) according .gt.(2)+..l&. if v is any positive integer.lt..lt. Hence convergent or divergent according as (n) tends to a limit or to infinity as n^oo and therefore. (1) sum of The test the series is not greater thanj^ + 1. +v n v2 s +v + .) function of x which is positive and continuous for all iy values ofx greater than unity.lt.. by a con tinuous and steadily decreasing function (x) of the continuous Maclaurin s (or Cauchy s) decreases &.I&.0i) = &.lt. is +v n ^0(l)0(w) ^ convergent.lt.gt. does or does not converge according as ./&.lt. variable x. when x n. then the series &.gt. and Hence Sv. &.j&./&.lt. .JIvl v so that Ol f 7 vl ) 9 \V 9 \ V) V y . Then u1 +H + . [VIII V\ iv v I * If u n Integral Test. Then 2)^ is a series of positive terms.lt.. &. 4&. (f ) = I (a. Let us write so that &. does or does not tend to a limit the * it is I as % * oo . we can write u n (n) and suppose that (n) is the value assumed. .gt. (/&. &. Then..gt.lt.gt.) c?^?.lt.(!).
(f ) 1. 1. VII. divergent is when = 1. And if s &.(vl)&.gt. i &. and it is easy to see that *()* as *~ oo . 1) that the series If s s &.) 3.lt.lt. where s is any rational number. then We have seen already is (77 5 and f 7 f  Exs. 1 s Thus ^ then f is 1 ~**.lt. series the first case its ^n~ sum is less convergent ifs&. LXVII. Ex. and so &.gt..gt. and we can apply the Here unless 5=1.. Prove that TT&. follows from (6) of unless &. Prove that 2.lt.lt.* say.lt.174. *&. By s far the most important applica s tion of the Integral Test to the series l. *.(#) = &.lt. result might have which we already know the series 2 however interesting to see how the Integral Test may be applied to In this case (1/w).(!)+ For and Ch.+3. x .l&.gt. Examples LXX.l.gt. divergent if s than s/(s 1). Misc. It is (1/ra). s&. then __ (mf l) 2 (m + _ 175. The series S~ is s s . throughout the interval (v 1. that v v v) . 175] AND INFINITE INTEGRALS in fact be less 319 it The sum must 160.gt.. j&. then u n decreases as n increases.2 then .+.lt. 14. For if n &. 41.oo .0^y&.(*) and this cannot be true for all values of v.& QJ )?i (Math. Trip.l ^ and in So far as divergence for is concerned. If s &.. 1909.lt. when the preceding analysis fails. Prove that if m _L &.lt.. than &. this been derived at once from comparison with 2 to be divergent.s + 2.+n +.gt. it is obvious that the series divergent.lt. GO as f  !).(v). LXIX. If test. 1 then g ** 1 as f ^ x and .
. 11 2?i~ 11/10 are convergent. 3&. If it is (n) ^1&.l+s or t &. n+1 and 2 n+1 n+1 22 ((2 (2 ) diverges then so do 22 ) and then the inequalities just obtained show that 2&.gt.gt.) 6.gt. 5. a&. 3. which shows that $  oo as *oo . tends to infinity with The series 2w~ 2 2. / y i # . I+s. is convergent or divergent accord 2?i*~*.gt.(n) a is convergent or then the series decreasing function of as S2 n (2 W ) is convergent or divergent. respectively. Trip.lt. and without 2. ss and s are rational Prove that (Math. I . Prove by an argument similar to that used above. Examples LXXI. + & n\ z where all the letters denote positive numbers and the and arranged in descending order of magnitude. 3 /2 2&.0.i~ sums are 10 n are / The series 2?is /&. where s&. 3. x = ~~ J [*du u 1 () ()&.". 176. 1 then divergent. 1911.lt.l.lt.] Discuss the convergence or divergence of the series 2 (! 7i s i + a 2 n s* + .gt. If &.lt.lt.lt.gt. ./&.J&.gt.gt. If w n 2". &.(n) can prove this by an argument which we have used already ( 77) in the special case of the series 2 (I/ft).X+ a ng as 4.lt. . + 2) . Cauchy s Condensation Test. 172 is The second if un of the is two tests mentioned in n.In the first We place f(3) + $(4) 5 2* (4). + a k nSk }l(b^ + b 2 n tz + . .j&. .lt. .(2 )&.gt.(ft) diverges. that () = . where [Compare with )&. divergent according 2&.gt. t&.gt. Sw.320 THE CONVERGENCE OF INFINITE SERIES Zr [VIII But by putting x = u we obtain [*+ dx l ]* and so &. as follows: = (f&. 1.lt.lt. &. and their The series 27i~ 1/2. integration.. not greater than divergent.I then the series 2 n ~ $ (n ) is convergent. (n) * 1 &. . .gt.
gt. we have But is it there any special point in supposing the lower limit to be unity. and =1. if series &.lt. are accordingly led to formulate the following definition If(f&. Nor (f&.gt.gt.gt.] 2&. For our present purposes the field of application of this test is It enables us practically the same as that of the Integral Test. tend to a limit as x * x .gt. And from n ) this set of inequalities it follows that is 2 2n (/&. and if so on.175177] AND INFINITE INTEGRALS 321 On the other hand 40 (4). than 1 I: Show that if a is any p6sitive integer greater convergent or divergent according as 2an 0(an ) is convergent or divergent. [Use the same arguments as above.lt. : We (t) is a function oft continuous when lim x * I &..gt.lt. 2. then the (n) is convergent or divergent according as the integral (x) does or does not it function 4&. to discuss the series 2 n~ s with ease. natural to extend our definition to other cases.e. J 1 Then we shall say that the integral $ is (t) dt convergent. cording &.gt. Hence deduce Abel s Theorem of 177. If 22 7i W (2 ) converges then 173. (x) is 2 &. and that rx lim X**X&.gt.lt. (2 converges then so does 2 &. Thus the theorem established. terms.gt. So far l\ and we shall call the is supposed (t) positive and decreasing.f&.(n) then is . that. (f&./&.(t)dt t = a. as s 1 or s 1.. ac Examples LXXII. The Integral Test of 174 shows a positive and decreasing function of #. it is obvious that lim 2 W ft (2 ) = 0. For 2 n~* will equal converge i. . a 2 a 3 . Infinite Integrals. taking groups of a./&. . Let us suppose that does tend to a limit.gt. W ns or diverge according as 2 2 2~ converges or diverges. and has the value integral an infinite integral./&. (w).lt. 21 J a H.gt.(t)dt = l.
..J&. VII.gt... when 00.... but is quite independent of t. The notion of the more complex than that of the finite . Infinite integral is in fact essentially integral.. no way altered by the substitution of any other letter for t (cf. or what is known as a repeated limit. (iv) Of course the reader infinite integral to 2 or is \ir.lt.. (a). The ordinary integral between Ch.. of which it is a development. &. tends to oo (or to If (x) tends to a 3&. as x oo a limit... may if &. (1) is convergent and has the value I..(f)dt was defined finite in 156 and 157 as a simple infinite integral is therefore the limit of a certain sum. .. More (a?) integral is (ii) $ (oo )  any integral function of $ (x\ then the value of the 3&.e. when none of these alternatives occur. /a limit.. is always positive it is clear that an increasing function of convergence and divergence to oo &... as defined in contrast a finite integral.gt. finitely or infinitely.gt.gt..l&... denote by is $ (oo ).lt.. .se If we write I J a then the integral converges. we shall sometimes call in limits a and A. or oscillates. On the other hand.lt. &. we shall say that the integral diverges to oo and we can give a oo similar definition of divergence to Finally.gt. These definitions suggest the following remarks.. diverges. . 4&. In the special case in which is (] x. or oscillates according as (x) tends to oo )..gt. then the value of the integral is * (oo ). i. we shall say that the integral oscillates. which we generally.322 then THE CONVERGENCE OF INFINITE SERIES we shall say that the infinite integral t [VIII .(#) .. will not be puzzled by the use of the term denote something which has a definite value such as The distinction between an infinite integral and a finite integral : similar to that between an infinite series and a finite series is no one supposes that an infinite series necessarily divergent. limit... as x * oo . f. Hence the only alternatives are (iii) The in and is integral (1) of course depends on a. The the limit of a limit.
and in any case could only be formulated have investigated in greater detail the theory of the when we function 212 .gt.lt.lt. has no analogue for integrals and that the analogue of Cauchy s test is not of much importance.gt. (x) we have the theorem K&. That A is true of integrals as well as of series we have already seen in Corre 177.(t)dt&. Thus the (6) of k r* i (x) dx is convergent. and fy (x) for all values of x greater than a. depending does on the notion of successive terms. We may as it .K for all values of x greater than a.gt. : rx I if $ (x) dx is convergent.lt. 178] (vi) AND INFINITE INTEGRALS : &. vergence or divergence of the infinite integral (1) of 177. The to consider case in which It is natural (#) is positive.a. /&.f&. then the infinite series 2 $ (n) and the positive r oo infinite integral I &. then r I &. We leave it to the reader to formulate the corresponding test for divergence. analogous to theorems A D of 167. J b (. and rx I \lr Ja (x) dx ^ K \ Ja (x) dx.lt./&. observe that D Alembert s test ( 168).lt. (x) dx converge or diverge together. b 178. then I J a ^r (x) dx is convergent r oo d&. (ii). and .) / dx + I J a J $ (x) dx. and b&.f&. V V (vii) The reader will find no difficulty in formulating and proving theorems for infinite integrals analogous to those stated in (1) 77.lt.. what are the general theorems.gt.lt. 323 ^ 1 The Integral Test of 174 may now be stated in the form if (x) is and steadily decreases as x increases. a constant K I J a &.f&.gt.177.r) dx = rb J a / (/?. (x} dx is convergent r I &. corresponding to C.gt. concerning the con &. Similarly. sponding to B we have the theorem that the necessary and sufficient condition for the convergence of the integral (1) is that it should be such that possible to find.gt.lt.
gt.0.324 THE CONVERGENCE OF INFINITE SERIES r x as . lira xs = (x) where I &.gt. [VIII $ (x) = we shall tests are obtained The most important by comparison with the integral do in Ch. &.lt..lt. is in each case unity.gt. &.(n}*0. divergent according as s &. . and the breadth of the peak corre x\ t 1 2 Fig. The area . when x^a.gt.J&. There is one fundamental property of a convergent infinite series in regard to which the analogy between infinite series and infinite integrals breaks down. (as) dx is a convergent integral . is is greater than the greatest root of the convergent if s&.gt. 3 X and spending to x=n is 2/(rc + l) 2 /". (a) &. that if / &. then (x) . 50. when x =a .r\\ and otherwise divergent. Kx~s . &. .gt. r oo &. then (j&. where s &. if divergent then the integral is convergent or or s ^ 1. for any value of I 1 $ (#) cfa? .lt. but it is not true. Here the height of the peaks corresponding to the points 2.gt. of the peak is l/(n+l) 2.&. (f) (x) dx is convergent 1.f&.gt. but it is not true that &. $ (a?) is always positive.lt.gt.lt. always even when &. it is evident that.f&.gt. the I.l. o /oo (x} Examples LXXIII.gt.l and in particular. 3.gt.lt. and are as follows if Kar where s&. . 1./&./&. special /:* : &. .lt.gt. whose convergence or divergence we have investigated in 175. and is 0.lt. IX.lt.gt.gt.(x)dx is convergent then Consider for example the function $ (x) whose graph is indicated by the thick line in the figure. 2 3.(#).(n) is convergent then (j&.lt..&. If 2(j&. The integral where a and A are positive and a denominator. if &.
I J a cos (ax + ft) dx J a ^oo . an J o /co ^(^*)d!rifl convergent then ^ I 2 x&. Prove generally that / provided that the integral I &. oscillate finitely as 4.gt. (#) ^ tends to a limit as * oo .gt. ( J a $(x}dx respectively.gt.5. 3.lt. oscillate infinitely as rm say that I /a $(x}dx is (f&.r. Integrals from oo to a f oo .lt. I we say that J oo is f %(*)&.*+^ are convergent? In the first two integrals it is supposed that a&.0.t&. / #2 sin#o?. J a where n is any positive integer. 7.178] 2. 2 (. : &.lt. Such J oo integrals possess properties in every respect analogous to those of the integrals discussed in the preceding sections the reader will find no difficulty in formulating them.gt.v = Q.(x}dx.f)(x J oo )d.r is convergent.lt. sinxdjc. The integrals I a /cos#cff.lt.lt.r ) o?.^&. and have the values k.J* JaX* 13 r dx r )*+*". AND INFINITE INTEGRALS Which [ of the integrals 325 r~ Ja&. then we convergent and equal to I. If the integrals J oo ( &. v 1 /a^H? r x dx ^ r* Ja ]a &. / J a J a xn cos (or f /3) dx.* convergent and has the value k + 1. . The integrals I ^costfcfa?. Prove that dx J 8.gt. then are both convergent. and in the last that a is greater than the greatest root (if any) of the de nominator. *~oo .
in [That the converse not true in shown by the example which &..{f(t)}f(t)dt ...lt.326 10. as in b where a=f(b). and # oo . (3) r&.gt.. Finally suppose that the functional such that #^oo astf^oo..lt.(x}*~0.t&. Prove this and the Integral Test and 173.lt..(#) = cos #. ..t&. for any value of f greater than we have.gt. by arguments analogous to those of 11. Further suppose 161. .. and so oo we see that the integral .t&. may be extended so as to apply to infinite integrals.u f I x n+l $ (x) is dx^ then the /*. relation x=f(t) tend to is =/(T). oo In the second case we obtain I J = &..gt. a. #2 &.t&.(%) always positive is the converse statement general is also true..$&.lt..lt.. If $ (x] is positive and steadily decreases.gt.j&..{f(t)}f(t) dt &...(x)dx is by means J a of Abel s Theorem convergent.{f(t)}fG)&.. then x&. Suppose that (1) is convergent.lt.lt..xJb  lim T&.lt.gt..gt.lt.$&. THE CONVERGENCE OF INFINITE SERIES Analogue of Abel s Theorem of rcc [VIII 173.lt.lt.... On the other hand it may happen that f * In the first case we obtain or as T*. making r in (2).gt. and u n = ".lt. convergence of J a (x)dx involves that of is n. T oo as r~* oo J ^ $(x)dx= a r lim T* ( &. \ substitution 161 Application to infinite integrals of the rules for and integration by parts. xn = nir. r*cJ b (4X We shall return to this equation in 181.... (1) Transformation by substitution.. (a) and I &. .2. $i &.. Then. that.c. . The rules for the transformation of a definite integral which were discussed in 179.b J b is convergent and equal to the integral (1). (6) directly.. If a = XQ I &..gt.lt.(x)dx a \imr&... If &..
are &. J (a0)/a J according as a 3.gt.cc . I (b(at+fl)dt.t&. a f(aft/a.a J 6 $(t)dt= f~ / . . by means of the substitution x = t .178. r J  (#) dx.] 4.lt. I oo J $ (as) oo doc.&?.lt.m [It follows at once. If / J a $ (x] dx f convergent then it is equal to one or other of a I $(o*+)&. Show that Jx / [Put 6. * [For f^ J _ {d)(xa}d)(xb}}dx= *~ a ~b &.gt.lt.gt. Now use the Integral Test.(t}dt( I d)(xa)dx\ ? ~ J &.(&.lt. 2&. 179] AND INFINITE INTEGRALS for the integrals 327 There are of course corresponding results ra I &. 2.lt.gt./&.gt./&. a Show. 2 dx = TT. x=$ and integrate by x&. then and verify the result /oo by calculating the value is of each integral directly. then the convergence of the series 2 (ri) implies and is implied by that of the series + /3). which it is not worth while to set out in detail for himself.] (^ )&. 1.lt. parts. is positive or negative.gt. that if s &.t&. j8 If $ (x) is a positive and steadily decreasing function of #. then {$(xa}$(xV)}dx=(al&. Ja converge or diverge together.gt.gt. : the reader will be able to formulate them Examples LXXIV.lt.}(h]c). on making the substitution x=at+j3. and a and any positive numbers. that the integrals  &.(xb}dx J ? = ( J .ft.lt.(#)&. 1 and a &.gt.lt.gt.^ as If (x)*k as [ J _oo and $ x oo . a&.
and we obtain the result . in Other types of infinite integrals. T I 00 y&m jf\ . &).gt. J o n UT# then .. 161) is The formula for integration by parts ( J [*/(*) a f (*) dx =/(? ) &. (a?) dx = lirn /(f) &. 180. ordinary or finite integral given in Ch. or from tO 00 .] Integration by parts. Examples LXXV. the definition of the definite It is possible. to extend the notion of the so as to apply to many cases in which these conditions integral . (*) cfo. (?) /(a) (a) . &. The second integral may be discussed (2) similarly. however.lt. n= W(w4. 3. Hence prove that j m r . Show that /* nI T ^A OT+n ( j ^m^  then &..!)}/_!. where K is the greatest value of p throughout the interval  \ ( I a.lt. that (1) the range of integration is finite and (2) the subject of integration is continuous. Verify the result by applying the substitution ^7 = ^2 to the result of Ex. 4./&. lx + .lt. Then if any two of the three terms Suppose now that f*oo in the above equation which involve f tend to limits.lt. It was assumed. (f) f(a) $ (a) oo There are of course similar results 00 for integrals to . Hence !*(&)*. J f a where p*0 as *oo. n =m ! (tt Show similarly that if /m n = ..*/ f J a (*) &. so does the third. and the modulus of the last integral is less than or equal to a b K. VII. 1.gt.328 The first THE CONVERGENCE OF INFINITE SERIES of these [vill two integrals may be expressed in the form (ab)k+ / pdt.
A) comes at one or other of the limits a. A) while *oo as x*a where in particular. A) into a finite number of subintervals It is evident that each of which contains only one and. We shall suppose that . consider here. (x) x~ s .lt.lt./&. if it comes between a and A.179. When there is more than one such (a. . . define the integral over the whole interval as being the sum of the integrals over each subinterval. We that it is the second of the conditions (1). we can then fA define I &. by &. A) except for a finite number of values oo as x tends or of x. Further. to assuming each of these integrals fined. Let us then suppose except for x &.gt.gt. we need only consider the case in which contains one such point f.gt. We = the case in which f = A. differ from those of Ch. .gt. if a therefore consider how we can define &. The infinite integrals which we have discussed in the preceding sections. for example. A s through values greater typical example of such a function is given by &. (2) that is not satisfied. it is evident that the definitions to which we are led will apply.gt. (x) to than a.f&. It is natural to try to cases at any rate. 0. .(x)+tt f2 (x) &. 180] AND INFINITE INTEGRALS 329 are not satisfied. if the value of the integral over each of these subintervals has been defined. be continuous throughout (a. we can suppose that the one point f in (a.(#) = a.gt. to have been satisfactorily de shall suppose.lt.lt.lt. (x) dx as r Ja fA 6 (x) dx f J a J 6 (x) dx. then. or. say x = . that (f a . (x) is continuous throughout the range of integration (a. frame definitions applicable to some such There is only one such case which we shall (f&.. we can then . = 0. VII shall now suppose in that the range of integration is infinite. and that (j&. A. . with trifling changes. to any of these exceptional values from either side. Let us when s &. A) point we can divide up (a. For.
A) contains any finite number of infinities of (as). But if we make the 1}*OC / substitution y = 1 /#. (x) dx : and then.. CA I Ja (x a )~~ s dx = rA lim (x e^+Qj a + e I s a)~ dx. we obtain I y or. . . We are thus led to the following general definition: if the integral a+e tends to a limit I as e^f 0. Thus lim C / A J \\A dy = is I x~ s dx. what lim e*.lt. Nearly all the remarks (i) (vii) made at the end of 177 apply to those of the infinite integrals of the second kind as well as to first. GO integral in which the subject of integration tends to included in the range oo as x tends to some value or values or to of integration will be called an infinite integral of the second kind : An the first kind of infinite integrals being the class discussed in 111 et seq. Similar considerations rA s lead us to define (x a)~ dx by the equation &. we can extend our definitions to cover the case in which the interval (a. J x~ s dx.lt. s ~". Similarly. dy is convergent if s &. I J a.(#)^oo as x tends to the upper limit A t we define J (f)(x)dx as being rAe lim / d&. f {* e x~ s dx.330 THE CONVERGENCE OF INFINITE SERIES r [VIII The lim I integral I y s ~2 . exists provided that s 1 and it is natural to define the value of the integral (1) as being equal to this limit. 1/r. 1 ( 175) and means y I/A s ~2 dy. as we explained above.+ 0^ the same thing. we rA I shall say that the integral (x) (f) a. when rA a &.gt. dx  is convergent and has the value I.lt.
178). and the integral on the righthand side kind. first The integral . i.2. 4 ).lt. &.m&.lt. arises if 2 in the example considered a moment m^ ago. while / (#). It is clear that &. c=l. and that /(*) = */(! ) Thei1 6 = 0.co We . Suppose for example. If &. and (1) becomes . . whole interval (a. In this case the (10) of substitution x=*f(t) transforms an infinite into a finite This case integral. that (#) where l&. if (/&. as an infinite integral of the second kind. A) then we can of course identify A and A. integral on the righthand side is defined as the limit. and therefore tends to a limit or to oo and the truth of the result stated becomes evident.jy 1. In this case 7) is a finite integral.v) do? + CA I 7 ^1 on the righthand side of the above equation increases as f decreases.180. as T*C.(#) is continuous except for x=a. oo as o.lt. dx=\ ae j (. however small (cf. such that If $ (#) is positive throughout positive throughout (a. is an infinite integral of the second On the other hand it may happen that $ {/ (*)}/ ". Examples LXXVI. may discuss similarly the case in which &. and a=0. (#) 7 a dx diverges to oo It is clear that. 0(#)d#*oo . and lira J in {/(/)} / (*)* = virtue of the corollary to Theorem 160.lt.lt.(#). . 181] 181. then the necessary and sufficient condition that is (#) dx should be convergent that we can A a+e find a constant K such J a that for all values of e. we can choose a number A between a and A. Now (#) is the [ 7 A (#) C A &.e. of the corresponding integral over the range (6.gt. And when {/(*)} / (t] has an infinity at t=c the integral is The = (!+ ^)~ m essentially an infinite integral.oo as d?a.lt.lt. (#) . AND INFINITE INTEGRALS write the equation (4) of We may now 179 in the form (1). then convergence and divergence to oo are the only alternatives for the integral. If the condition is not satisfied then I *".(t)dt f (0 is continuous for t=c.*a+0.lt. We shall then say FA that the integral I &. r).
332 2.r dx. and the value of J il/x 7aix/(^) each 6.] integral 10.gt. Such an under any where 5&.2. &. attains its greatest value (3/aM.7  c are convergent.K(x 3.lt. 1.] is 4.  ^^ cfcp. when A =TT.gt. 2.lt. f^sinx dx. For the range of integration is infinite integral as ^ X . if 2. 1. and &. while the integral is divergent if s ^ 1. is zero. merely a particular case of a general [This parison theorem analogous to that stated in 178. J that.lt. /i* 9.&. 1 The integrals / =r .^&. / .lt. it dx is convergent if t &. does not fall directly .lt. sin^ ] ^ I J 2 .gt.) The integral I 7o (cos x) (sin x} 1 m dx is convergent if and only if I &. 11 1911. where s &.] The integral cf T \i ig convergent if and only if s&.. . as x tends r TT I to either limit.1.lt.gt. Xs . 3.p that &. then the integral divergent. 12.. If (f)(x}*~cc as is x * a+ if and $ (x) a)~ t 8 . XP J7 alternate in sign and steadily decrease in absolute value. where o A &. is convergent p &.".lt. THE CONVERGENCE OF INFINITE SERIES Prove that [VIII if 5 &. Show that / ^^dx.. 7^n&.lt..f&. then CA J a is \ &.lt. 1. K(x a)~ where s^l.1. Show also the integrals X^ /".(#)&. [Transform the whose limits are kir and ( + !) by the substitution x = /C7r+y. s + 1. [The subject of integration tends to 7. if Show &.r /". The QO X integral I is convergent.lt.(x}dx convergent. com Are the A integrals f dx dx A dx dx [A ( A dx A [A dx [ dx convergent or divergent 5. / * f ^ J o of our previous definitions.gt.lt. .gt.lt. where 0&.sinx T d^. ** 0.lt.
1 then JO 19.ts as x^l from either side so that the subject of integration becomes a continuous function of x if we assign to it the value t s when x=\. s &.AND INFINITE INTEGRALS and the subject of integration tends to oo define this integral as being equal to the sum as x ^.gt.t. s &.1 )/(l #)*. 1.lt.] 15.lt. f ".) .s&. but (^8 ~ 1 ^.lt. on the lines of 16. [ I sn mx sin J o & dx.lt. and can be removed by attaching a particular value to it at that point.lt. 1909.^fll^ r*^* l+# + + Jo l ! Jo l t ip. The second . . In this case it is usual to suppose the definition of the is subject of integration completed in this way.". 333 It is natural to l+x provided that these two integrals are both convergent. t &.lt.gt.lt. 1. Substitution and integration by parts. 1. It should be noted that finite integral oo is a convergent infinite integral of the first kind if when s 1 the first integral is an ordinary but then the second is divergent.+ 0. first integral is a convergent infinite integral of the second kind 0&.lt. 1. &. ^. 179.l.*&. &. [It should be noticed that the subject of integration is undefined when #=1. then 18. s &.lt. integrals sin m~ x .lt. The ~x integral / J &. Thus the integral from to is &.lt.gt. Thus the ".lt. 0. convergent if and only x*~ l if &.lt. are ordinary finite integrals. Prove by integration by parts that  if s &. If 0&. The reader should formulate the general theorems for himself. o i x dx is convergent if and only if &. The if s 1. t ~t d% is convergent if and only if &. Jo x dx . /* T 14. if the subjects of integration are regarded as having the value in when #=0. The formulae for trans formation by substitution and integration by parts may of course be extended to infinite integrals of the second as well as of the first kind. It often happens that the subject of integration has a discontinuity which due simply to a failure in its definition at a particular point in the range of integration.
and p are all positive then o .&7. * . illustration of this. ". Prove that J 2 [Put r=sin ^ and use Ex. t . which is 4)V2.334 THE CONVERGENCE OF INFINITE SERIES if I [VIII 20. Let J=r 2 (. certainly wrong whichever sign we . * Prove that f . tt f 2 1 2 1.= * and jf^r. and so we obtain the value choose. If s &. The following example affords a good for transformation by substitution.J (J/a^A TVip. Since y = 8 when x appear to be led to the result 1 and y = 20 when x = 7. Establish the formulae 25. 8.7=48.a] 2 (ii) by means of the substitu (i) by means of the substitution x tion (6 x}l(x a) = and (iii) by means of the subs itution x= a cos + b sin 22. Show. t LXIII. we The indefinite integral is 1 (y_ 4)3/2 + 4 (y^. Show. by means of the substitution x=ptJ(p}\ t\ that if I. 1 then 24.  /". by means of the substitution at=t/(I . Now let us apply the substitution which gives x = 3 *J(y 4). 1912.gt. m.lt. 23. /^{(*)(6~)} JaV{(#a)(6tf)} = a + (b . We find by direct integration that . 6^+13)&.= *ir (a + 6).) Some care has occasionally to be exercised in applying the rule 182.1\ that both positive then and m are 21.
Example. Thus a formula which will be found to lead to the correct result. and the value of an infinite whether of the first or the second kind. or for all values of x within some definite interval (a. if we transform the integral / J o dx=ir by the substitution #=arc sin y. apply to series of terms or integrals of functions whose values be either tions of the sum may positive or negative. it has always been explicitly or tacitly assumed that any conditions imposed upon un may be violated for In the case of a series all that is necessary is that such a the terms are positive) should be satisfied from some definite term onwards. Our defini of an infinite series.lt. 1/V(1 y 2 )or dx/dy = . so that dx 1 As x increases from 3 to 7.lt. y increases from 4 to 20. Similarly in the case of an infinite integral the conditions have been supposed to be satisfied : a finite number (e. Verify the results of transforming the integrals f / I (4aP o . and the other sign must be chosen. Of course the u n or case in which they are all negative is not essentially different. and the But the examples by which we have illustrated them.1/^(1 ?/ 2 ) &.x + Jgy f ) 77 cfo. as it can be reduced to the former by changing un into &.ir y y o 2 by the substitutions 4x x + ^=y #=arc siny respectively. of terms all condition that for all values of x greater than some definite value.181183] AND INFINITE INTEGRALS 335 The explanation is to be found in a closer consideration of the relation The function &3 6x+I3 has a minimum for #=3. $ (x) into (x). when y. Series of positive and negative terms. according as we must observe that ck/cfy TT or x TT.lt. divergence which we special tests for convergence or have established in this chapter.lt. have had reference almost entirely to the case in which all these values are positive. As x increases from 1 to 3.g. / cos 2 ^ &. Similarly. and y dx\dy is between x and negative. a +8) which includes the . y decreases from 8 to 4. &. = 4. integral. ^# ^TT &. 183.lt.
But when the changes of sign of w n persist throughout the series.gt./&. / J a \x 1 dx . Further. changes sign as x * oo . as in the series 1 (a?) continually . . when &. Thus our tests apply to such a series as n2 since n 2 10 to such integrals as l 10 &. general problem of series con chapters. a series negative. let v n or wn be equal to an according let vn . as in the integral f sin a.wnt an = vn + wn . and wn = u n or wn = 0.** [ s 3 Jo o l2x V# 7 .lt. . as in the integral [ A sm .gt.x a \ .gt. Let us then consider in which any term may be either positive or Let so that an = u n if u n is positive and an = . or when +J J+ finite. when x &. Ji & 7 (& . and &.gt. as u n is positive or negative.gt.gt. according as un is positive or negative.336 THE CONVERGENCE OF INFINITE SERIES [VIII value a near which the subject of integration tends to infinity. J. . ^&. what is the same thing. have to consider certain simple examples taining an infinite number of both positive and negative terms.gt. For now we have to consider the possibility of oscillation as well as of We shall not.lt. *.or. 2# &.lt. or as a. x &. since 3# 7 &. when the number of both positive and negative terms is in i. where a is a point of discontinuity of &. a/ then the problem of discussing convergence or divergence becomes more difficult. w that un = vn . .un if un is negative. = v n or vn = 0. the other being in either case equal Then it is evident that v n and n are always positive. . (/&. convergence or divergence. f and 1 . (a.). in for this volume. when n ^ 4. in But we 184. ^un Absolutely Convergent Series.e. and to zero.lt. shall. according as u n is negative or positive or. have to consider the more the ensuing integrals.
2w n or : S (vn wn ) is convergent and equal to %v n ^w n . l) and an = l/ri wn = l/n 2 while v n = I/n 2 or v n =0 according as n or w n = according as n is even or odd. + 2 + + UnJ &. convergent when l^. and it is by no means evident a priori that this precludes oscillation on J the part of 2 u n . case. We is are thus led to formulate the following definition. by theorem (6) of 77.o 4 fortiori I *. /n 2 odd or even and We A. l(l/2) 2 + (l/3) 2  ~l .v^l. then 2& n is absolutely convergent. 18) any selected from the terms of a convergent series of positive terms is convergent. in the Suppose that the series 2 n is convergent. so are the series . formed and negative terms taken separately and the series is equal to the sum of the positive terms plus the sum of sum of the negative terms. 22 . AND INFINITE INTEGRALS our series is 337 then u n = ( is n for example. example above. the series %u n absolutely convergent. xxx. + 1 + n.lt. said to be When 5or n or S un  is convergent. we can. \ Ka n . Examples LXXVII. DEFINITION.gt. ^i = ".]  If ". is convergent. 184] If. and bn &. And hence.2a n is a convergent series of positive terms. choose n Q so that when w 2 &. theorem that an absolutely convergent series is con 2 u n is convergent.. where 2 n is This is the Then both ^vn and ^wn series are convergent for (Ex. If 3. ( 1. 2a n is a convergent series of positive terms. when any positive number d is [Since \ assigned. 5. then the series 2 w .lt. Employ the general principle of convergence 84) to prove the  vergent. can now distinguish two cases. 2 . and therefore 2ww 2. for instance...r n is absolutely H.183. The reader should carefully guard himself against supposing that the statement an absolutely convergent series is convergent is a mere tautology. When we say that 2w n is absolutely convergent we do not assert directly that 2w n is convergent: we assert the convergence of another series 2zn. proved above amounts to this it is : And what we have by its positive if the %u n is absolutely convergent then convergent.
338 THE CONVERGENCE OF INFINITE SERIES [VIII If 2 a n is a convergent series of positive terms. . It is now easy to see that any abso lutely convergent series has the same property. then N and therefore tends hypothesis that to is 2u n = 2v n oo 000 N 2wn ... Any series selected from the terms of an absolutely convergent series absolutely convergent. have now to case indicated above. if In the first place we note 2&. then the series For they obviously cannot both converge.. We viz. 184 must both diverge to oo n ^w n of vergent. [Examples are afforded by the series 2rn cos?i0... the original series is said to be conditionally convergent. vn ./. f so rearranged as to become ^u n and let . . say . is convergent. [For the series of the moduli of its terms is a selection from the series of the moduli of the terms of the original series.] is 6. 2r n sin$ of 88. For . were formed from u n Then Sa n and so are Sv . Conditionally convergent series. 4. and that the only case to which the sign of equality can apply which every term has the same sign... then the series 2 a n cos w0... (1). is that in convergent Extension of Dirichlet s Theorem to absolutely Dirichlet s Theorem ( 169) shows that the series. Zwn .. that. terms of a series of positive terms may be rearranged in any way without affecting its sum. con Sttfo j ..lt... and so 186.. consider the second series of B. 2 a n sin nB are absolutely convergent for all values of 6. 2vn = %vn and 2w n = Zwn . from u n as ct n . It is clear from equa sum of a conditionally convergent series . Prove that if 2 un is \ convergent then 2 UH  (^2 Un .u n be wn be formed is ..w convergent.. as it is a rearrangement of 2crn which are rearrangements of 2vw 2w n . is N with N.] 5.lt. as this would involve the convergence of 2 (v n + w n ) or 2cr n And if one of them. that in which the moduli 2 n diverges to oo  DEFINITION... v n . Hence 2v w tion (1) ..v %u n conditionally con . wn vergent. n are above that the both divergent... but 2 wn is I divergent. which contrary to the %un 2&.. and 2v w divergent. 185. by Dirichlet s Theorem.. If 2ttn is convergent. let 2.  Also.
LXXVII.. f . and property all absolutely convergent series (Ex. It is naturally a much more difficult matter to formulate tests of convergence for series whose convergence. completely. . + Vn &.lt. + Vn . U + Mj f . .lt. . be expected that we should be able to find convergence as simple and general as those tests for conditional 167 et seq.184187] is AND INFINITE INTEGRALS 339 the limit of the difference of two functions each of which tends n. . Tests of convergence for conditionally convergent It is not to series.. absolutely convergent. and every v less than the corresponding we could VQ at once infer that . depends of the positive by the negative terms. we are now considering. U  + .. +  Vn &.. *2&. 18). as it depended essentially shall see in a on the convergence of %v n and ^. and that the theorem fails rate the proof of 185 We is not true for series such as 187. v were positive. .. that any selection to oo with And it seems more itself forms a convergent series.. UQ + .lt. . .w n separately.+\Un\. as is of shown by equation (1) above. than likely that the property prescribed by Dirichlet s Theorem at any will not be possessed by conditionally convergent series from the terms . both unrestricted as to sign. all that we and v s are But + V. u s only were positive and we could infer % +K +   . It is obvious too that %u n no longer possesses the of convergent series of positive terms (Ex. . tests essentially on the cancelling instance there In the first for convergence of conditionally convergent For suppose we wish to infer the convergence of 2v n from We have to compare that of ^u n V +V + l . + Unt and so that Sv n is that V when the us can infer is in the general case. If every u and every u.lt. xxx.. 5). + Unt and so that %v n that I is convergent.u \ 99 9 .  U +. are no comparison series. If the every v numerically less than the corresponding u. moment that this conjecture is well founded. + V: + . +  Vn &. + Un . This would enable us to infer the absolute convergence of 2v ro but if Sw n is only con from the absolute convergence of n ditionally convergent we can draw no inference at all.
although each of its vergent. ss ..lt. terms is numerically less than the corresponding term of the former series. Since s = _ )t . between &. s2 ..f&. conditionally convergent if and oscillatory if .lt.0.!.j&..(0)..lt.+( l) Then Hence s . and ?l .lt./&..j&. 1. (0) and &. and &. &. Examples LXXVIII.lt. .fa + &. &.gt. THE CONVERGENCE OF INFINITE SERIES [VIII is We shall see shortly that the series series l*+i . (1).. is oo a decreasing sequence..lt. .. from which it follows that both sequences must tend to limits.. a&.gt.0.lt. is an in . and that the two limits must be the same. sn = . conditionally convergent series./&./&. and therefore . The convergence of the most important series of this type established by the following theorem. . Alternating Series.gt.gt. .lt./&..lt.gt..s n . &.^1.gt..gt./&. its sum . to *(0)$ (!) + is convergent. series series are The simplest and most common what is known as alternating is whose terms are alternately positive and negative... are conditionally convergent. But the + $ + + + ..lt. Let us write .gt. a&. . lies between &.lt. where 0&.lt..gt.o for &. the sequence s slt . fa.k . series The 2 ( l) n (% + )..lt.0. s4 ... s zn+1 .l) 2n+1 (j) 2n+l = 0. is It is therefore only natural that such tests as we can obtain should be of a much more special character than those given in the early part of this chapter. If $ (??) is a positive function of n which tends steadily zero as n^ oo then the series .p and &.gt.gt... (1).340 Example. (j) &. tends to . . &. (0) let &. &. GO But creasing sequence.lt..lt it is clear that this limit lies s1 = a limit.. The series where 2. s2n .. That is to say.lt. con divergent..+ .lt. .. 188. (2). is absolutely convergent if s&. tends to a limit or to and s lt s3 .#&..gt. &. and therefore tends to a limit or to lim (Szn+iSzn) = lini (.
_(z2&. Let s be the sum and S2 n the sum of its first 2n terms.l) n n oscillates finitely. &. 4. although it positive. Then let t3n 11 2n~+I 1 Now since the sum of .lt.gt. and sum of the first 3n terms. 188 are satisfied except that $ n tends steadily then the series 2 ( . Alteration of the sum rearrangement of the terms. should tend steadily to zero is essential to the truth of the The reader will easily verify that ^(2/1 + 1) 1 \/(2?i) + 1. this condition is not satisfied. This example shows that the say. so that of a conditionally convergent series by of the scries 1 1 Now in consider the series denote the which two positive terms are followed by one negative term. to avoid any This series may be difficulty as to the definitions of the first few terms. the terms inside the bracket is clearly less than and l{m ^n\2 + ^ Hence + + ^ lim l^ by 156 and 158. Consider the series which we suppose to begin with the term for which w = 2. as all its terms are and limw^ n =l. The . If the conditions of I.gt. so that $n &. 188] 3. (i)nn Jn J series 2 fy n is convergent but 2 xn is divergent. The sum of the series of 188 lies between s n and * n + 1 for all values and the error committed by taking the sum of the first n terms instead of the sum of the whole series is numerically not greater than the modulus of the (w + l)th term. AND INFINITE INTEGRALS 341 of n . Hence the original series is divergent.lt.187. condition that theorem. to a positive limit 6. ~ . is of the form 02 03 + 04where n *~0. written in the form SN or (1)". &. 5.
we can determine a 2 (&. W For has a limit as Test. less Abel s Test.342 and two it THE CONVERGENCE OF INFINITE SERIES follows that the [VIII the last equation.lt. and so tends to a limit as n *.+_+. tends to zero And therefore tends to a limit.lt. 188. 68. Also n.. or to diverge to co or x For a proof we may refer to Bromwich s Infinite Series.gt. and therefore sn &.lt. constant K so that  sv &. sum to n terms Also since 2 series n . Misc. There test. then 2 a n n is convergent. i. Hence the $. Then we have the theorem if n is a positive and 2 n is convergent. that it is convergent. but the righthand side of Later on we shall give the actual values of the sums of the 213 and Ch. n&. +a n ..oo .) is convergent. is any Dirichlet s Test.. though of than Dirichlet s.lt. &. the series is 2a v (f) v is convergent. j. which tends to a limit as n*~ oo . and all its 0o~&.K not actually convergent. + 1 ) is absolutely convergent.oo is not necessarily zero. 19.e.f) and 2a n is convergent. n *cc. and as 2 w is convergent it follows that 2a n convergent. Ex.j&. IX. postulate less about 2a n viz. since the terms are positive. frequent application another Suppose that &. by : (j&.lt. Now the series is vergent. If n satisfies the same conditions as in series which converges or oscillates finitely. It can indeed be 7.lt. The reader will easily verify the identity where sw =a + ai + . then the series &.  K for all values of v. p. oscillates finitely.lt.lt. .. is a positive and decreasing its limit as n *...+ . function of n. Thus we n but to make up for this we postulate more about and decreasing function of n. more general the following.gt. but that &.lt. due to Abel. . Hence.i s con an d lim0 B =0. say I and Km ($ w . test of A is which includes the 188 as a particular test case.. is sometimes useful. series : see sum of the series (1) is not s. if $i) + ($i $ 2 ) + . diverges to x . The series 1 + __ + _. proved that a conditionally convergent series can always be so rearranged as to converge to any sum whatever.gt.] 189.j) n. at any rate ((/&. which.lt. [Here where s 2n = 1 . as in Dirichlet s Test.j)n (&. test.. Abel s and Dirichlet s Tests of Convergence. to .1) = 0. : Dirichlet s 2 an is &...gt.
1.lt. . . Let us suppose..gt. nn = ( n l) .n . since cos 3. If 2 n is convergent then both series are absolutely convergent (Ex. The sine series converges to zero if 6 is a multiple of ir IT. 6. when  m 21 m and so when 2. =[T^=fTT . where h and s m&. n is a positive function of n which tends steadily to then the series 2 ft cos ?id.lt. 4) for all values of 6. the second series vanishes identically. And in this case the series above written are con ditionally and not absolutely convergent. l ..gt. AND INFINITE INTEGRALS may be stated as follows its : 343 series a convergent remains con decreasing vergent if we multiply terms by any sequence of positive and Examples LXXIX.gt. H lefthand side of (1) therefore lies between are the algebraically least and greatest of sm The h&.j) n is convergent.m&. that the conditions of Abel s Test are satisfied.. If we put 6 = rr in the cosine series we are led back to the result of 188. LXXIX. xxiv. 2 cos nd and 2 sin n& if oscillate multiple of series For. 0t given any positive number 8. We have identically l + . we denote the sums of the first \z\ and and write 2=Cis0. 189] This theorem factors. . Dirichlet s and Abel s Tests may also be established by means of the general principle of convergence ( 84). m .f) m and s m&.188.lt. . except perhaps the first series when 6 is a multiple of 2?r. The if cosine series oscillates finitely if 6 is an odd multiple of and diverges 6 is an even multiple of TT. unless (in the case of the first series) a multiple of and &.gt. 8). we have . s&. finitely The by sn series ?r.mQ. 2$ n is divergent. as will be proved in Ex. It follows that if zero as n*.+a v . The series 2n~ 8 cosn0.s . and the whole interest of the result lies in its application to the case in which gent : (f&.lt. Thus the series 2a n &. LXXVII. for example.lt. so that =l when 6 is not a n terms of the two and =+=!.z That the series are not actually convergent follows from the fact that their nth terms do not tend and so   [ \ to zero (Exs. n&. 2n~ 8 smn0 6 is are convergent 2n if 1. In this case the first series reduces to w which may or may not be conver 2^&. . tn .0.lt. we can choose m so that m &.lt. 2 ff) n sin nQ are convergent..oo . 1 \ * and tn are also not greater than 2/ 1 .gt. &. But...m + s H(f) m . 7.
lt.344 4.lt. [For suppose. of course.(v n + iw n ). it follows that 2 &. said to be absolutely THEOREM. vn absolutely convergent if convergent. as n *. we have confined We shall now consider the series %un = ?. introduce anything really novel. conditionally convergent 0&. noting where when 6 is a multiple of rr. the series The series is There is however one class of such are separately convergent.s.lt.lt. z For if  2w n j is 2  vn  . then the series convergent.0). 2 wn : absolutely convergent.oc n is a positive function of n which tends steadily to w cos?i#. and so S { \vn + w n } is con \  \ vergent but  Un  = VOn + Wn) ^\Vn + \Wn 2 \ \ t . unless 6 is a multiple of ir.g. And in this case it is obvious that The reader should write n cos?i0 is divergent. is an obvious extension of DEFINITION. 5. the series 2&. THE CONVERGENCE OF INFINITE SERIES The series of Ex. out the corresponding argument for the sineseries. Mention any exceptional cases finitely. e. convergence The necessary and ** sufficient condition for the absolute of 2w n the convergence of 2 un  \ or S */(vn + w n *). and only if. except the sineseries when 6 is a multiple of tr. which that of 184.J it fails 190. by Dirichlet s Test.s 1. that 2$ w nB or cosn() \ is convergent.lt. Since cos 2 $ ^ cos nd j \ .lt. The consideration of such series does not.gt. series so important as to require special treatment. The series ^u n where u n = . if [VIII if s&.l.v n and + iwn is 2wn are . is 2&. then both of the series are convergent. If 2a tt ?i~ a is convergent or oscillates then 3.a n n~ l is convergent when 6.lt. since 20 n is divergent and 2$ n cos 2??0. where vn and convergent wn are real. Accordingly we give the following definition. and 20 n is 2&. convergent.f) 2 n cos (1+oos 2*0) convergent.. and oscillatory if s (finitely if 5 = and infinitely if s&. 2$ TC sin7i0 are not absolutely divergent. if. Series of complex terms. But this is impossible. 3 are in general absolutely convergent &. If (f) . all So far ourselves to series of whose terms are real. t&.gt.
gt. if The series 2 r is convergent for all values of yri~^ For un = then as n &. A power series %an zn may be convergent for all values of a certain region of values. There. z. We have already come across some cases of expansion in series of this kind in connection with Taylor s and Maclaurin s series ( 148). It is obvious that any power series converges when 2=0. now z is we were concerned only with a real variable x. deduced directly from the general principle of convergence this as an exercise to the reader. the theory of the ordinary functions which occur in elementary analysis (such as the sine and cosine. For if  1/ t \ . We leave Power Series. by d Alembert s Test. whatever value all all is z may have. Exs. and S w n are convergent whenever 2 w n  \ 2  I  is  con   vergent. except when 2 = 0. which will be discussed in the next chapter) is that Such a series is called a power series in x. for It is sufficient to give 1. when \ and the generally absolutely convergent. 2 u n is original series is absolutely con shall see later on that a power series. It is obvious that aft absolutely convergent series is convergent. The convergence of an absolutely convergent series (cf. which ^anxn is concerned with their expansion in series of the form however. since its real and Dirichlet s And imaginary parts converge separately. Hence (cf. 2. 5) the modulus of the nth term tends to GO with n. . The series 2n\ zn is not convergent for any value of z except 2 = 0. may also be 1). LXXVII. 185) may be extended at once to by applying it absolutely convergent complex series to the separate series 2u n and 2wn . shall where a complex variable. One of the most important parts of 191. or for no values except z 0. values of z. consider a few general properties of power series in We z. u n = n\ zn then ? B + 1 u n = (n + V)\z\. which tends to oo with n unless 2 = 0. On the other hand V n ^ V0 2 + Wj\ Wn ^ V(V + Wn ). z. convergent for vergent for convergent. xxvu. Theorem ( 169. QO . and the logarithm and exponential. values of We Hence. an example of each possibility. Ex. 2. 1. A. and so the series cannot converge.189191] AND INFINITE INTEGRALS 345 and therefore so that 2  vw  2 un is convergent.
when 2z n is always convergent when \z\&.K Show that the result is true even if the series oscillates when z=zlf [If s H = Oo + aiZ l + .lt. circle of convergence. say the point z = R.. THE CONVERGENCE OF INFINITE SERIES The z  \ &. finitely sn \&.] The region of convergence of a power series. But a n zf = ^vi ^ ^i +  \ \ I   and the argument can be completed as 193. the class at which the series converges and the class at which it does not. Then it is clear that every point of the first class lies to the left of every point of the second class. Example. before. K \&. &. on the other hand.gt. and therefore  we a n z^ n &.lt. series ticular value of say z l = ^ (cos 2a w ^ n 2 a n zn is + i sin 1 convergent for a ^). \ K for all and the result follows at once by comparison with the convergent geometrical series 2 (r/r^ if 11 .+an z 1 n then we can find K so that for all values of n.l and never convergent This was proved in 88. and that the first class of points does include points besides z = Q. if z = r r1} we have But. converges when z = r then it converges r. 192.r par it is absolutely convergent for all values of z such that \z\ l .lt. it The second class.lt. The first class must contain at least the one point z exist. = 0. B. Hence there is a point. Now let us divide the points r of the positive real axis into two classes. Let z = r be any point on the If the power series positive real axis. If a power z.%K&. need not all values of z. which divides as the series may converge for however that the two classes.346 3. and may itself belong to either one or the other. In other words. \ K convergent. Thus we have an actual example of t each of the three possibilities.lt.lt. . since is can certainly find a constant such that values of n. Suppose does exist. then &. [VIII series 1. the series converges at P then it converges absolutely at all points nearer to the origin than P. at all points inside the Then circle the series is absolutely convergent \z\=R.. In absolutely at all points inside the circle z particular it The converges for all real values of z less than r. For lim a n zf = 0.gt.
P . where a 0. or (2) (3) converges absolutely for all values of z or converges absolutely for all values of z within a certain circle of radius R. For centre less let is P be any such point. For if it converged at absolutely at all points nearer to as it does not converge at Flg 51 * * it would converge and this is absurd. it converges absolutely at points on its circle of .. by Theorem B.. and that in case (2) it is absolutely convergent everywhere. 51). The examples which follow show that as a matter of fact there are very diverse possibilities as to this.. so as to include P it. therefore.gt. diverging when 2 = I/a and oscillating finitely at all other z . than P far we have excepted the cases in which the power series does not converge at any point on the positive real axis (1) = Q or (2) converges at all points on the positive real except z So nowhere except when z that in case (1) the power series converges 0. . 1.. and whose radius is We can draw a whose p/ inside in Q. has a radius of convergence equal to I/a. and In case (3) the circle is called the circle of convergence its radius the radius of convergence of the power series. It should be observed that this general result gives absolutely no information about the behaviour of the series on the circle of convergence. It is clear power (1) series either converges for z and for no other value of z . &. than R. Thus we obtain the following result: a axis.191193] AND INFINITE INTEGRALS 347 circle. any point between A and Q (Fig. The series l+az + a?z +. points on the circle. all z3 has its radius of convergence equal to convergence. Let this circle cut OA \ Then the and series is con vergent at Q. On the other hand the series cannot converge at any point P outside the circle. Examples LXXX. and does not converge for any value of z outside this circle. 2. It does not converge anywhere on its circle of convergence. The z series ^ + z2 % + 02 + . 1 . absolutely conver gent at P.
A . question of its convergence on the circle./z. \/\ In the case a n + iZ n+ l an n z =X. which or greater than unity according as 1/X. than the radius of con where A is a constant 192). If & is an odd 1 1. is less 4. Consider the series . When and the z is series on the circle of convergence we assumes the form may write z=cosd\i sin#. then n /z  less &. vergence of the (cf. Hobson. pp. convergence except the point z= . all points 5. The real and imaginary parts are both convergent... Plane Trigonometry . can use Cauchy s Test ( 168. \a n so that the radius of convergence is unity. 3 that its radius of convergence unity. unless 6 is an odd multiple of ?r (Exs. so that we can use D Alembert s Test ( 168. The series is called (for reasons which will appear later) the logarithmic is series. X..lt. is If 2a n z n is where fa *0 as z *0.8) O If ! J3 + ___ m is a positive integer then the series terminates. LXXIX. or  an l/n \ &.. 3. is less  or greater than In the second case we The logarithmic series. if  an + l \ j\ a n * \ X.lt. * et seq. pp. 4). .* a power series which 194. Infinite Series. is any number n and ^j&. as n  .gt. and the series assumes the form multiple of ir then z  J . series. and so l/G*l*l) Sec Bromwich. For if p. 225 (3rd edition). 2) similarly.348 3. and f(z) is its then it is easy to see that/ (2) can be expressed in the form sum. U 3). The binomial series. though not absolutely onvergent. which We is shall not discuss here the a little more difficult. 1ms 1/X as lira i its radius of convergence. then the first series a + a 2 + a 2 2 + . Uniqueness of a power series.. = convergent for some values of z at any rate besides 2 0. and so diverges to of its circle of oo Thus the logarithmic series converges at 1. 268 et seq. In general = mn\ I ~". THE CONVERGENCE OF INFINITE SERIES More 1 [VIII oo generally. It follows from Ex.
] if Prove by multiplication of series that and z  &. If z is less than the radius of convergence 2a n z n 2bn zn then the product of the two series is  \ .lt. ~~ as is 6.2 . 1 &B _ 1 + . z)f(m . where wn = u vn + u^n! + Q . .. Prove similarly . then R.n all [Assume the truth of the theorem for 1 1(1. which we have already ex tended to all absolutely convergent series. 2c n z n where . by squaring the series for 1/(1 z\ that 1/(1 z) =I+2z + 3z 2 + ... series is The [This equation forms the basis of coefficient of z n in the product .. 170 that if Multiplication of Series. 1.z) indices m + J = 2 *. by Ex. z} =f(m+m 9 z}. the f(z)/(lz) = 3..+.lt. 15 that whose modulus is less than some number ^ then an = b n for all values of n. 2an&*=2bn &* values of It follows from Ex.. cB =a &n +a &. function f(z) cannot be represented by two different power series. then/(w... can this result to all cases in . Prove..lt. which ^u n and %v n are absolutely convergent for our proof was merely a simple application of Dirichlet s Theorem. where . a positive integer.l. n %(n + l)(n + 2)z The Binomial Theorem .. up to m. %u n and 2vn Stt n. + aB & . Euler s proof of the Binomial Theorem. Then.. 195. and \z\ is less than either 2s z n where s a  \ R 2 tl .. We now extend of either of the series Examples LXXXI. LV. This result is capable of considerable It shows that the same generalisations into which we cannot enter now. easily proved by induction. and f(z) is the sum of series when z or unity. n =a + () 1 + .2. 4. We saw in are two convergent series of positive terms. then If _ (lz) m 1.  1. and m is for a negative integral exponent.193195] where if AND INFINITE INTEGRALS a 349 K is number independent for all of z z. 2... if If the radius of convergence of 2a n z n is /?. the general term being 5. that 3 1/(1 s) = l+32+62 2 + . then x 2v n = 2w. + u nv . . 1. z\&.
". is Aw n = wn . ?^ ?l =A A ( n ). Misc. \&. 9. 11. a positive [The result of Ch. (Math. ^ 2w r A*(w). and so \wn (n + 1 so that 2 wn is certainly not convergent. except when convergent than k.350 This is THE CONVEKGENCE OF INFINITE SERIES a polynomial in [VIII m and m to : integers this polynomial must reduce r but when n+m j ? m and m are positive j n v i r tue of the Binomial Theorem equal for for a positive integral exponent. values of z : and it is zn . [For the series for easy to see that if f(z) is absolutely convergent for .". _ _5_ 36 (JfafA. 1912. which tends MISCELLANEOUS EXAMPLES ON CHAPTER 1. f ) and {CW + {S (*)}* = !. ^= zn . 2Jn+J(n where 2. That the theorem is not always true when 2un and 2vn are not absolutely convergent may be seen by considering the case in which _ ".. VIII. (2n + 2)/(w + 2).) Show that where and so on.w n + if 1 . If f(z) = l + z+ . order n**. (1)".] 7. Then But J{(r + 1) to 2 .gt. Failure of the Multiplication Theorem. shows that A fc 8 (?i ) is in general of Show that 00 1 (7i+l)(24 + 3)(2rc + 5)(ra+4J _ n2 + 9n + 5 ". TWp.) [Resolve the general term into partial fractions. 1890. when every term of the series is zero. VII.. .)} ^$(n + 2).  then f(z)f(z }=f(z + z all n . Ex. then wn =  a then if C(z+z } = C(z}C(z }S(z)S(z \ S(z+z ) = S(z} C(z + C(z} S (z ). Discuss the convergence of the series Sw*{V(^ + l) is real. Trip. 1)}.] }.] .] 3. A2 &.gt. s is integer less and only if k r + s + 1. all positive integral values of m and and if m two such polynomials are then they must be equal identically. + .
. if a polynomial P (n) convergent. 8. is a divergent series of positive and decreasing terms. i \ 2 3 w (Math. . convergent but that the corresponding series in which the groups contain 2. 4. shows that the first series is convergent the second we leave to the reader.. is also a This decreasing sequence whose limit is zero (Ch.) .AND INFINITE INTEGRALS 4. Discuss the convergence of the series . Trip. which successive terms of the same sign form groups of 1. . 1a [ / 1 and it is easy to deduce that 2?t~ ~a 1 lies between I/a and (l/a) + l.v n then ^. 2(l)sm. [It follows from 1a * 174 that % lI*". Prove that lim a 2 n~ l ~a = l.. is zero. 3. 9.. 2^1cosV * ... \Yn (l cos \ V / where a 7.. 1899. Prove that the series in is 1. is a decreasing sequence of positive numbers whose limit then the series are convergent. 8. then 11..* ". If ?i.gt. [For if (ui + u z + . If Uo + Ui + u 2 + .) where 6 and a are 8. +u n )ln=.. we can determine and a constant A such that 2{# (n)P(n)(A/n)} is Consider in particular the cases in which R(n) is one of ( Show that the series 1 l J is convergent provided only that z 6. \ / 2 (.0 then lim p * c 2 n= 12. Prove that if a&. Trip.. (Math. terms. s.. 5.!. is real. real. Misc. . In particular the series are convergent.J . is not a negative integer. . 4. terms oscillates finitely. 27).. v2 v3 . 9 n 2sin. 1908. 2 &.] 10. . the functions l/(an + b).. 2.gt.*". . Exs... 351 Show that. IV. Investigate the convergence or divergence of the series 2 sin. R (n) is any rational function of n.
. can have a root whose modulus when z = l Cis ( .] If \a n &.lt.. its sum then 1/(1 being if  z/(l &... we obtain But if .... it follows from the sum of a polynomial and certain partial fractions of rational function as the p and from the Binomial Theorem for a negative integral the type Al(za) ...lt. Thus the recurring series this that \a n is certainly convergent for values of z whose modulus is less than !/(?. . THE CONVERGENCE OF INFINITE SERIES Find the sum of the series [VIII 2 u n where . then the series has the sum xj{(x . relation (k. (1).ff) is a root.] Find the sums of the series (in which [The all the indices are powers of converges only if  2). 16... ...1) (x 2 + 1)}. first series z\ &.1 for all values of n then the equation than cannot have a root whose modulus it is less ^..+p k a n _ k =0 ... finitely. is equal to  is and the only case in which that in which a n = . . The polynomial 1 +piz+p 2 z* + called the scale of relation of the series. If x=l then w w =(l)".. #n_#ni for all real values of 1/1 (Math. [If x is not equal to unity and the sum is 0.. Recurring Series..352 13. + 1 and If 3.=1 then  M=0 the series oscillates 14. is For let certainly convergent for values of z whose G be the greater of the two numbers modulus where a n is the it follows from the equation (1) that \a n \^Ga n modulus of the numerically greatest of the preceding coefficients and from KGn where is independent Of n. the second series converges to 15.. A power series 2a nzn is said to be a recurring series if its coefficients satisfy a relation of the type a n +p 1 a n _ 1 +p 2a n _2+.l and to z) z \ 1. known results as to the expression of any Conversely.gt.. 1901. 1. Then .. .p k z and add a new series in which all the coefficients after the the results.) x for which the series is convergent.l)th vanish in virtue of the (1). K k z we multiply the series f(z} = ^a n zn by p& p 2 z . . \&. zf(l z)if\z\&. z). whenever they are convergent. where n&.k the expansion first place that the series is sufficiently small. P*_i are constants. . pk are independent of ra. so that where is P P19 ....lt.gt...Cis(n0). and p l} p z . Any recurring series is To prove this we observe in the of a rational function of z. Trip.lt.
. Trip. a&. that any rational function whose denominator is not divisible by z can be expanded in a power series convergent for values of z whose modulus is sufficiently small. /?! = . 16 Such called a linear differenceequation in a n with constant coefficients. 18_et seq. Thus &. + un _ 2 = ig B series whose If u n is a polynomial in scale of relation is (1 n of degree . as in Ex. is 1 or  22. in fact if [2 p. is A relation of the type of (1) in Ex. 18. that the series is a recurring series. coefficient of z n is The values of J n A%. Exs. 1904. 16.] H. B depend upon the which may of course be chosen arbitrarily. 17.J)}. (2 + ( is and (Math. (Math. 1 the necessary and series is that it sufficient condition that a power scries should be a recurring should be the expansion of such a rational function of z. where p is the least of the moduli of the roots of the denominator (cf. that its sum is a )z* () = A 122 2 + l22 A B 2 o. in terms of a lt and 2. first three coefficients a . where 19. 1913. if f(n) is coefficient of z n in the expansion of zj(l + in powers of (1) then (2) /(N)+/(l)+/(n2HO. a^ 2. And it is easy to see. 1898. Solution of DifferenceEquations. then 2w n 2 n is . We l _ find. IV. B are numbers easily expressible Expanding each fraction separately we see that the where A lt A 2 and . equations may be solved by a method which will be sufficiently ex plained by an example.}.z) k + l a recurring (Math. Misc.lt. Expand 9/{(s 1) (z + 2) 2 } the in ascending powers of z.AND INFINITE INTEGRALS 353 exponent.) 21. Trip. Suppose that the equation is Consider the recurring power series 1a n z n . The u n =A cos nB + B sin n6. Deduce that f(ri) is equal to according as n is of the form 3& or 3 + 1 or 3k + 2.) 20. Trip. 23 . solution of the differenceequation u n 2 cos 6 u n _ j A and are arbitrary constants.gt. by reversing the argument above.) [If p n is the probability then p n =% (pn \+Pnz) also p = 1. A n. 3 /() = (3B  where or 1 a complex cube root of unity. and verify 3 this by means of the identity z/(l + z+z 2 )=z (1 z)j(l z ). Ch. player tossing a coin for every tail. Prove that 2. is to score one point for every head he turns up and two a total Show to play on until his score reaches or passes that his chance of making exactly the total n is .
26n are convergent and have the sums then Deduce that if 2cw is convergent then its sum is A B.. . + C n =^i if Hence prove that A. [This follows from splitting fractions. n... the series 2a n . : one or both are not absolutely convergent.** # 24. Ex. 23.lt.. i 1 l term tends to AB (Ch. IV. ^ l . + Bn c Bn + 2 Bn _ + . The modulus of than /3{ ^ + e 2 + . The result. being merely an algebraical identity.. o n ^ ! ( i l).. then [Let ^4 n = ^4 + f n Then the expression given is equal to B + B2 + . Prove by multiplication of series that ..] Jo 1 /V !_&.] If 25. We have already seen that we can multiply the series 2a n 2& n in this way if both series are Abel s Theorem shows that we can do so even if absolutely convergent . provided only that the product series ..] than the greatest value of $ greater The first the second is less  ) 1 : j  26.. n. B.. 2. 1. When a &. This result is known as Abel s Theorem on the multiplication of Series. An^A and Bn ^B as w^oo . . Misc. ! / 1 I A 2 3 */ * ng [The u coefficient of z will be found to be \fn\ \fn Now use Ex. 27).. 2.^ =5 n n . must be true for all values of a save 1. + n B A .gt.+an b l and 4n then and C1 + C2 f. Prove that if c n =a l bn + a 2 b n _i + . THE CONVERGENCE OF INFINITE SERIES Prove that [VIII if n is a positive integer and a is not one of the numbers 1. taking a = 0.. up each term on the righthand side into partial the result may be deduced very simply from the equation P(iff){i(i*)}^ x Jo and l(l#) n in powers of x and integrating by expanding (1 #)/(! a?) each term separately.. + e B }M where /3 is any number and this expression tends to zero..354 23. so that A n *~A... Bn +B.
from J(b/ct) to 0.gt. m + 2n + 1 then are positive.] 31. Show that if creases steadily from  ^y = ax(bjx\ where a and b are positive. 9 to establish the convergence of the 28. x increases from *J(b/a) towards oo 32.] Prove that a 1 30. then y in co to GO as x increases from to oo . / and that 232 . where a and b are positive. values of .] For what values of ? m and n is the integral / sin w x (1  cos x} n dx convergent 29. . Establish the formulae In particular. Hence show that .AND INFINITE INTEGRALS 27. series. as y increases from and x2 decreases *J(ab) towards co. 1 then [In this and the succeeding examples it is of course supposed that the arbitrary functions which occur are such that the integrals considered have a meaning in accordance with the definitions of 177 et seq. prove that if n &.gt. then two x correspond to any value of y greater than *J(ab\ Denoting the greater of these by x and the less by #2 show that. Hence show that Show that if 2y = ax + (&/#). 355 Prove that [Use Ex. [If m + if 1 and &.
integrals of the first and second 37. The is /(^Brl/^+y Also deduce the last result from Ex. _ ~ 2 f* J a V(2y^) where proof 35. by putting 2 but their last two results remain true when /3 &. Misc.fa? then 00 p J Q TT r (38 + a a) (^2 + 6 2) if a. 31. then d^g J o a^H.lt. Prove that [ if 6 is positive then tfdx _TT_ afidx f * Jo 36.ay. A=p + *J(ay}. Ex. then not quite so simple.lt. INFINITE SERIES Prove the formula AND INTEGRALS [VIII /o 34. 2 6 (a + 6) positive. (Math. TT Jo and (^ 2 2 + &. 1913. /3.2/3^ + y 2 ).356 33. If a and b are &.gt. (^^?+6^~ 26 J {(^infinite Extend Schwarz s inequality (Ch. Establish similar results in which the limits of integration are and 1. Trip. 42) to kinds. 2 ) (* + &2 = ) 2 (a + 6) Deduce that T 00 and y are ay. Prove that if &.(#) is the function considered at the end of 178 then 38. Prove that . 7(seo positive.) .lt. VII.
Generally . algebraical functions. explicit functions. rational functions. Attempts have been made to integrate some function /(#) in terms of functions already known. it has been proved that this is so but as a rule such proof has not been forthcoming until later on. when it was found that certain algebraical equations could not be solved by means of the numbers One of the most fruitful sources of new already recognised. to has happened that mathematicians have taken the impossibility granted as soon as they have become reasonably convinced of it. functions has been the problem of integration. These new functions have generally been introduced because it appeared that some problem which was occupying the attention of mathematicians was incapable of solution by the functions already known.CHAPTER IX THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS OF A REAL VARIABLE 196. THE number of essentially different types of functions with which we have been concerned in the foregoing chapters is not very large. The process may be fairly means of compared with that by which the irrational and complex numbers were first introduced. and have introduced a new function (oc) defined by its for F . Among those which have occurred the most important for ordinary purposes are polynomials. or implicit. has begun Sometimes strict it a appear probable that the problem is insoluble. and trigonometrical We are however far from having exhausted the list of functions which are important in mathematics. direct or inverse. These attempts have failed and after a certain number of failures it . The gradual expansion of the range of mathematical knowledge has been accompanied by the introduction into analysis of one new class of function after another.
when in One such case Ch. not indeed a strict proof that log x a reasonable presumption that it is. 4) that it cannot be a rational a rational function is a rational function function. 197. Lxxvi.358 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS viz. But it is a few experiments that differentiation will very easy to become convinced by For example. and the original problem could not possibly be solved has been finite established. now consider We towards shall oo * . never get +#) any number of times is always the product of J(l+x) The reader should test the a rational function. to be a really function. how log x behaves as x varies from It follows at once from the definition that log For such a proof see the author s tract quoted x is a on p. since the derivative of whose denominator contains only repeated factors. definition log 1=0. one Similarly. new Let us consider what grounds we have for supposing logo. by correctness of the statement by experimenting with a number of examples. [iX possessing the required property. the logarithm of x. 236. 2) the if the range of integration includes integral has no meaning We might have chosen a lower limit other the point a?=0. \ must suppose that x is positive. a function which involves sin x or cos x. The question whether it can be an algebraical or trigonometrical function is more difficult. and so generally. With this than 1 . profess to give* but We shall therefore treat it as such. the result of rid of algebraical irrationalities. the function log x by means of the equation (dx log*]. since (Ex. VI we defined occurred in the preceding pages. We have seen already (Ex. but 1 proves to be the most convenient. that F (x) f(x). if we differentiate differentiating v/(l or other of these functions persists in the result. and we shall find on examination that its are quite unlike those of any function which we have as yet properties encountered. that we do not is a new functiontherefore. We have. Definition of log x. XLII. no combination of the functions previously known could thus the correctness of the assumption that possibly have. they have investigated the properties of F(x) and it has then appeared that F (x) has properties which . . by the equation log*^ We . We define log rdt a?. Starting from this definition.
1.gt. Examples LXXXII. obtain if we make the *dt substitution t = l/u in the integral.lt. /(! . u &.196.lt. The general form of the graph of the logarithmic function is shown in Fig. Prove from the definition that if &. Since the derivative of log x is I/a?. then [For log (l + u)= I J o * T* .lt. and very steep u when x is very small.] . the curve is very gentle when as is very large. 1. log (1  w) &. If &. . the slope of Y Fig. 4. 52.w). 1 then u &.lt. we Thus log x tends steadily to oo as x decreases from 1 to 0. then logx is negative.j tdf /u 3. 52. For dt l Moreover. and the subject of integration 11 lies between 1 and 2. Prove that [Use Ex. 197] OF A REAL VARIABLE 359 continuous function of x which increases steadily with x and has a derivative and it follows from If x is 175 that log x tends to oo as x ^ oo positive but less than 1. Prove that log (1 + u) lies between u 2 and u 1 ti ^j2 (1 r when u f is v.
.. Show in the same way that there is no solution of the equation which possesses a differential coefficient arc tan x. are of It will be defined certain different the was said function orders of greatness. . but whose order of greatness are such functions... we du Ji/y u see that x log xy = x [** dt [ I 7:= Jj * = f I J l l du .. .. 1. third. is where a .. whole series of functions is which tend to first. Hence x =6 (x} V this is true for every pair or / (x} = C/x. is of the ath We may suppose a as small order of greatness when x is large. obtained by taking (7=0. distinguishing between functions which. (1) log x + log y .. Thus there is no solution fundamentally from log#.. 199.(x] =yf (#) But if of values of x and y then we must have xf a constant.. .. smaller than the We may say generally that a a any positive rational number. (i). first... when x is large. $x t . remembered that in Ex. (x) to be of the fcth order of greatness when /(&)/#* tends to a limit different from zero as as tends to infinity.360 198.log (l/y) = which proves the theorem. making the substitution t . eliminating / t (xy\ xf . %/x.. we obtain the two equations the equation and is so. xxxvi.. = yu. except the trivial solution /(#) = 0. . second. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [iX The functional equation satisfied by log equation a?. first with respect to x and then with respect to y. The function log a? satisfies the functional /(^)=/()+/(y) For. For when we differentiate the functional equation. . where C . 6 we ways in which a function of x may tend to infinity with x. 2. It can be shown that there is no solution of which possesses a differential coefficient and is fundamentally distinct from log x. A f It is easy to define a infinity with x.[ /ydu u u Jl = log x .. and it is distinct easy to see that (7 =0... Thus *Jx. Examples LXXXIII. and is fundamentally distinct from The manner in which log x tends to infinity with x....
positive number. we could always find a value of a so small that X* would tend to infinity more slowly still. This fact is small . which seems so natural. it follows from the y log y theorem proved above that lim ZH^+OO lim y a log y y*&.gt. Proof that (log#)/#*^0 as #^00. Thus .lt. loosely . The behaviour of log x as x a &.198201] as OF A HEAL VARIABLE than 0000001. since a&.lt. that if/(#) tends to infinity with x. we could always find a value more rapidly still. conversely. (sol). 200. T A or loga. At any rate it might be possible supposed that if / (x) tends to infinity with x.a?/p t &. but more slowly than any positive power of of x.gt. and therefore 201. oo and logos tends to log (1/x) logx to GO as x tends to zero by positive values. integral or fractional. 361 it we please. Since = (log x)\afif x= 1/y. and. + 0.lt. e. &. It shows that the presupposition stated above. The logarithm x tends to infinity with x. of a so great that x a would tend to infinity however rapidly.a //3 But. but log (I/a?) tends to oo more slowly than any positive power of I/a?. /5 be any dt &.&.gt.gt.lt.gt. and so l &.. Perhaps the most interesting feature of the function log x is its behaviour as x tends to infinity.lt. is unfounded. any positive number we can choose a smaller positive value of ? a And &. + (log #)/#".(aPI)/ft&. [* dt .&. oo but for all positive values of by saying that but the reader will hardly require at this stage to be warned against such modes of expression. Let Then 1/t l/t ~P when t 1. In other words loga? &. then (log x)lx* op. integral or fractional.gt. however slowly.gt.g.lt.. if a is ft. Now &. less And might be thought that by giving a all possible values we should exhaust the orders of infinity of f (x).ft. = 0. x /ft*Q as x + oo . sometimes expressed the order of infinity of log x is infinitely a. when x 1.
Let us consider once more the which possesses the property that. [The function rapidly than Jxf(log #) . v/(log x\ #(log#).f) . Math. is not all. . or to oo according as f(x} occurs to the right or the /(. #(log log x\ . [Jx/(log x} is rapidly than x is any positive rational number.lt.infinity 12. new terms We Thus we obtain a series formed of two simply infinite series arranged one after the other. v/(log log x\..] How does the function ft/ /(*){* behave as (loga?) oo ? (log log [If a=t=8 xf}l{x* (log*)* (log log xf } x tends to then the behaviour of * For fuller .lt. . (x} 2. we can prolong of infinity for functions of y which putting x = \\y we obtain a similar scale values. . Then ^(logr) tends to oo more slowly than log#. or ^F/(log#)*. But this Since Consider the function log log x.362 202.] such a function. where is such a function. then/(#) and (x) both tend to oo as o. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Scales of infinity. is an inherent characteristic of the logarithmic scale of infinity. It will be gathered from these examples that 3. . No. 4fo. Thus log log y tends to oo with y. log x. Find a function which tends to oo more slowly than Jx. but more a where a is any rational number less than 1/2..lt. #6/12 1.. which tends to infinity more slowly than any of the old terms. series of functions [iX The logarithmic scale.(&*) in the series. . &. By log log log log #. the series to any extent we like. $(logx) than J(log x\ and of so on.r)/0 (x) tends to can now continue this series by the insertion left of &. Examples LXXXIV. where . we can insert a new term . .. of log y. log log x.*oo while &. with log #. and it will by now be obvious that by introducing the functions log log log x. ij{f(x) F(x}} satisfies the conditions stated. but more slowly than any power Hence we may continue our series in the form x.r. . &. &.lt.r/(loglog#) incompleteness 4. v/. information as to l scales of .lt. but more a a is any rational number. . . v/.. We can begin to the right of all those already written down. F(x) of the series than (x} tends to oo more slowly (x} such that we could insert f(x) and more rapidly than F(x\ [Thus between Jx and $x between v (log x} and x/(log a?) we could insert (log x) 5/12 And. Tracts.* tends to tend to oo as y by positive Between any two terms /(a?). it follows on putting x=logy that (log log y)/(log y) a = (log x)/x a ^ 0.. see the author s tract Orders of Infinity Camb. a (log#)/# *0. if f(x) and (x) are any two of the functions contained in it.] Find a function which tends to oo more slowly than v/#. the logarithm of log x. generally. for all positive values of a. .
x/{^sino. Now where n log xy 2 = log x + x. it can only pass once through the value 1. it is dominated by that of (loglog^) a Thus /(#). (1 cos^)log(l/^) according to the rapidity with which they tend to zero as 8.r) 203. Arrange the functions #/^(log #). any positive integer. according to the rapidity with which they tend to zero as 7.gt. If a=/3 then the power of x disappears and the behaviour of/ (x) is dominated by that of (log^) a ~^ unless a = /3 when . Arrange (I/*)..ao if a a = ft a ora = fta = and /(#) ^0 if a&. . # log log ^/{log(l/^)}. .co . al(x (log ^p) Dx (log log ^) a = al{x log 1 ~ }. The number by It e. In other words e is defined by the equation Since log x is an increasing function of x. 6. a".~^". immense importance in higher mathematics.202. of analysis.log(l/^)}. log x n = n log x. . &. log y and so o(? = log x is 2 log log = n 3 log x. (# log log log . 203] is OF A REAL VARIABLE 363 dominated by that of x a ~^. #^ + 0. TT.gt.gt. ar (log log . 1 a }. j8". Show that 1 Dx log log x= \j(x log x\ Dx log log log x= and so 9. &. ft or &.lt. \(x log x log log x\ on. Show that = Dx (log ^) a and so on. . . . log e Hence n log e = n./3 . e We is usually denoted which like of shall now introduce a number.^/^/(log log #) according to the rapidity to infinity as #900. in the stricter sense of Hence our 95.ft or a = ft a ". {J(x + 1 }}!x (log xf x&. 5. definition does in fact define one definite number. #v/(lg #)/lg log # # log log #/\^(l with which they tend Arrange (log x)lx.gt. .  is.. log log xl(x log x\ x log log xlJ(x* +1 ). one of the fundamental constants We define e as the number whose logarithm is 1.
and it follows from likewise a continuous function of y. real values of y as the inverse of In other words we write x=& if y = log x. as so varies from towards oo . positive integers. and ey Thus.lt. and epfq denotes the positive qth root of we have e*&. ey all We now define the ex for ponential function the logarithmic function.. (1)... e. positive or negative.. of a power such as ey in which the index is irrational.....0.364 Again..... p = log so that loge p!q = log (eP if )v =q log e^. we have loge^ = 2/ .. (2) is are consequences of one another so long as y rational and ey At present we have not given any definition its positive value. log e rational values of y.0.lt. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS if [iX p and q are any ep .... has Example..lt. The exponential function. denotes the positive yth power of e. For if x = ey and x+ g=ey+r) then Thus T? is  greater than very small must &..... from one value of x corresponds one value of y.. Prove that 2 &.gt.lt. We saw that.] 204.....lt.*} f Jo 3. y has any positive rational value.lt..... Also JiT^Ji so that e &. and the function e y is defined for rational values of y only. e &. in the stricter sense.. and if T) is ....... /(# + ) if also be very small.lt. It is easy to give a direct proof of the continuity of the exponential function.. and conversely. increases oo towards oo Thus to steadily... pd? 72 i* P = P J* t~ Jo 2^ &. Also y 109 that x is is a continuous function of x. [In the first place it is evident that r and so 2 &. p/q. and than /# if &. 1 &... x=e y . y . 3. In other words the equations y = \ogx......gt.. and log e~ y = all y = Hence the equation (1) is true for y..
In particular ey form of the graph of ey is as shown in Fig.. from the ordinary rules we can choose two sequences yi.. are irrational then . . The (1) principal If properties so of the exponential function. 1 xx = 6* x e^.203205] y OF A REAL VARIABLE 365 Thus e is a positive and continuous function of y which increases steadily from towards oo as y increases from oo Moreover ey is the positive yih power of the number towards oo . e y *. and zlf z. then dy/dx=I/x. . or e~ y Q lim l/e y. since the exponential function is continuous. or both. X Fig. The general a rational number. .+zn = ey+z . Then. We may also deduce the functional equation satisfied by e from that satisfied by logx. 53... whenever y is 1 when y = 0. If y or . function (2) More of the exponential function is equal = eay then dx/dy ae ay generally..... y n . Thus the derivative itself. 205. we have ey x x ez In particular ey = lim e y* x lim ezn e~ y = e = 1. The exponential function functional equation This follows. e. when y and z.2 = e y2 then y + y2 = log x^ + log x2 = log x^xz and . of rational numbers such that lim yn = y. lim zn = z. of indices.. For if y1 = logx1 y2 = logx2 so that x = e yt x.y2 . x= e y} that y = logx. z are rational. .z nt .2 . in accordance with the elementary definitions. y l . and . . 53. if x satisfies the to the.. dx dy = x = ey .
.lt.* The y x.e..366 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [lX Examples constant. ..... (3) av } (e )/?/. x. value of a however large.e. .. aX e .. log^. nfi /v3 x x . &.a av e I ayew. The result follows on putting any x = ey It is clear also that ei y tends to oo if 7 if 0. 207. we obtain and so f(y]\f(y] Thus (z)lf(*\ and therefore each is constant.. 1. 7 &. or when there is some other special reason to the contrary. except in the particular case into x = e v The exponential function was introduced by inverting the equation y = \ogx and we have accordingly..] The function y.lt. for any positive value tt see that (log#) /#0 1//3. to this exponential scale also... follows that we can construct a scale of infinity similar to that constructed in a scale of functions which scale is 202. as y*Q. we obtain 206. = lim er^y* = ^ co for all values of a however great. denote e( x? ).... We shall now revert to the more natural plan of taking x as the independent variable. up to the present. 7. e z&. and in each case more rapidly than any power of this result it y. where K is a distinct There is no solution of the equation f(y + z)=f(y}f(z) fundamentally from the exponential function. The reader should try to apply the remarks about the logarithmic scale. = [Applying the Mean 1 where &. is reversed) be combined into one scale .. however small.. and to fi of Writing a for we for . .. . e( e *\ .. If a dx\dy = ax then x = Ke v.2x .. 1). The two scales may of course (if the order of one or.e.gt... used y as the independent and x as the dependent variable in discussing its properties. except when it is * .. We saw that (loga?)/0?0 as #^co. From i. 2. . x = e v simultaneously. but extending in the opposite direction ... .... then dx\dy~ax^ where a is a constant.... . e e*. eG\ ..lt. /jZ pX3 e. y and z in turn...log log . . ey tends to infinity ivith y more rapidly than any power of or lim y*/e* as y . necessary to consider a pair of equations of the type y = logx.. tend to oo more and more rapidly as #oo. 0. &.. LXXXIV. The x general power a ..  \y \ ..] 3. f if x=f(y] Value Prove that Theorem. e* .. . LXXXV.. [We assume that f(y] has a differential Differentiating the equation with respect to coefficient. so that x = Ke ay (Ex. The function a x has been defined only for rational values of x. fl 2 Cx where of course e x *.gt. made in 202 and Exs.
power of (3) I/a?. The graph then similar in shape to that of ex but reversed as regards right and left. take this as our definition of a* when x is irrational. and that log a log a.l then a* where a is positive. LXXXV. is denned only for positive values of a. and 1 . Moga as x^0. Then . and a*^co 1 then a* e e~^ where is positive. but the particular * result (Ex. ax is a continuous function of x. = a x+y and (a x )v = axv. for rational indices. in the first place. ax log a. . (4) ax is also a continuous function of a. = P lg lg y = (p/q) y= e xlosa . In the course of the preceding chapters a great many results involving the function ax have been stated with the limitation that x is rational. If a is = xlosa = x . If a&. log a = a log a. D a a x = Da e xlosa = e xlosa (as/ a) = ara*(5) (a* l)/a.205207] OF A KEAL VARIABLE e. The definition and theorems given in this section enable us to remove this restriction. and Dx ax = Dx exlosa = exloga log a = ax log a. ax x ay = exlosa x )y x e yl 8a and in the second (a (2) = evloga * = exlosa = e x . is This of course is a mere x corollary from the fact that D x a form of the result is often useful it . .* l)/a? of course equivalent to the a as x * 0. and a*^0 as #^oo. of a x .gt.lt. .gt. We shall now consider the case in which a any positive number. of the function ax are as follows. more rapidly than any &. The x x graph of a is in this case similar to that of e as x * oo more rapidly than any power of as. ax x a y For. In other words the laws of indices hold for irrational no less than (1) Whatever value a may have. 367 is when a p/q. Thus 10^ 2 = eV2logl It is to be observed that a x when a? is irrational. = ap yi and so Suppose that x is a positive rational number the positive value y of the power a p/q is given by it from which 2 lo g y follows that &. and is itself essentially x=x The most important properties positive. 3) that (e". . We .
... the equations and x &. that / lim If only. (2) 1 + )= 7?\f / lim 1 + oo r\^ = we suppose that f * oo or f ^ we obtain the result expressed by If through integral values (1). IV. 1. If we put h = 1/f. We shall now identify this limit with the number e of the preceding sections. that expressed / by nf (/r l + ) = lim(l) n*&.lt. we have or for ..gt.\ /vA H =e* nJ (1)....oo . as n^ao. We can however establish a more general the equations fl result.. As the result is of very great importance. we proved that (l+(l/7i)j tends. 73. to a limit which we denoted provisionally by e. native lines of proof. (1) we shall indicate alter Since it follows that lim^^^=a?.368 208. n(lxVn so that l n ] &. viz. In Ch. y is = v positive and x e we obtain..logx &.n(x l) .gt. however large.. we see chat lim f log as f oo or f ^ tinuous it follows that *... THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [iX The representation of n e x as a limit.. n is any positive integer... logA Writing y simple transformations... (3).lt..e. after some Now let .... f 1 +^ =x is Since the exponential function con as f^ooor * x : i.
lt. Deduce from (1) of 208 the theorem that e* tends to infinity more rapidly than any power of y. The reader is probably its familiar with the idea of a logarithm and use in numerical calculation. . denned by the equations x = ay. any rate for sufficiently large values of 1 and by 74.770 =y The (2) VM now follows from the be proved in the same way. He will remember that is in elementary algebra loga #. of course applicable only when y is rational. then (1 + n) n . result !) tends to a limit ( 75) and now follows from the inequalities (3) of 1/ 1 208. 10). The more general result if result may we replace l/n by a continuous variable h. 208. (1 For n (x l n I}n(l.x~ The n ] = n (x lln . since a. then + ^n )". and + and ^*oo.gt. by taking e 2 9. limit. y = l and n=Q in the in &.lt. (cf.e z . . n ] = lim n (x li li n = log x. Examples LXXXVI.208210] Then (9) of 0&. The representation of log# as a lim n ( We can also prove 75) that I x~ l &.772. then 0.. 210. that 2 if 5 Prove.0 If rc n ^co.lt.gt.^. equalities (4) of 2. Common logarithms. is 24 . 1.lt.] n &.*oo . 5. we (l see that n log (1 + if l n) ^. [Writing n log and using Ex. y= loga x. results of If n is a function of (1f  n ) in the w such that n%n form J as rc ^ oo . &. OF A REAL VARIABLE at 369 n . Idt I/ I Hence deduce the 3.gt. xxvil. 4. the logarithm of x to the base a.gt. 1 dt * dt I*/ 209.1) x~ ll n \ which tends to zero as tt^oo. inequalities (4). this point is often passed over in silence. (173 . Prove that then x t&.7 1 &.l / n to 1 ^O (Ex.lt. 1 then (*V _* /)/(* *i) &. which evidently tends to as n^ao. and so that if x &. 4. 1) 209. l/n. LXXXII. This definition though H.
Integrals containing the exponential function. sin0 = 6/r. vol. The value of log e 10 is 2302. Show that y has an values form a geometrical progression and which on the curve  (Math. 3. xxi. cos6 = a[r. vatives of the functions . ax cos bx al= e + bJ. then * See for example Chrystal s Algebra.. e ax sin for. . 2 and the x form a geometrical progression. J. Prove that if a &.. i. J. and that their b 2 sum is l+e a /b ". Z = Iog also x = 10* = e logl so that = (lg^)/(log 10). . we obtain [Denoting the two integrals by 7.bl. If then x For =& is and = log X = \0g X. and integrating by parts. a cos bx + b sin bx x f x 7 x cos bxdx = sin bxdx= bx e e J ^^/and .370 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS e. Trip. Prove that a sin bx b cos bx ..lt. + 62 !_&.. e".* Examples LXXXVII./* 5. Trace the curve infinity of lie y=e~ ax smbx. . [iX Our logarithms are therefore logarithms to the base numerical work logarithms to the base 10 are used. Ioge Thus it when once It is no part of our purpose in this book to go into details If the reader is concerning the practical uses of logarithms. e * cos bx.} of the axis of positive half Prove that the successive areas bounded by the curve of Ex. and that of its reciprocal 434.. Show that Dx e ax cos bx = reax cos (bx + 0).gt.. logio# y e 10 X. ch. Hence determine the nth deri and show in particular that Dxn e ax =an e ax 2. Dx e ax sin bx = re ax sin (bx + 6} where r=J(a? + b 2 ). where a and maxima whose b are positive. them he should consult some textbook on not familiar with Elementary Algebra or Trigonometry. . 1. ax sin bx . z . aJ= e Solve these equations for 4. e easy to pass from one system to the other 10 has been calculated.
where X true of &..fc. x =bx. or b ae.00 is e* x2 +wdx.ij) . positive integer.*(rfif_!r r . e~ x x n dx = n\. has two real roots. e to integrate e bx. and the integral is transformed ex. . Show how P(x. dx/du = l/u. sinfo. The result now becomes obvious when we draw the line y = bx. Show that the equation ea&. [Put into that of a rational function of 9. Draw the graphs of e* mining any maxima and minima 14..... where a and &. siumx. n. according as b ae. one. It follows that 7. deter of the functions and any points of inflexion e~* .0 and a greater than the greatest root of the denominator of (x).lt.~] 7n = fe ax x n dx then aln = e ax x n nln _ l calculated for all . or none...lt.^&. In can be if positive integral values of Prove that. b are positive. 0. 12.gt. . coslx. ax . so that the line y=aex touches the curve at the point (I/a.] 13. is and is not equal to b.gt. e ax e bx . 15.] 11. [This follows from the fact that e^ x tends to infinity more rapidly than any power of x. xe x xe~ x.] which passes through the origin 242 . P . is .xe x*. xe~ xZ and xlogx. [This follows from the fact that P can be m kx where m is a expressed as the sum of a number of terms of the type Ax e ...gt.lt. 2 .] Integrate 2 x (c e + a2 e ~ x ) ( distinguishing the cases in which a 10. OF A REAL VARIABLE If 371 [Integrate by parts. if =!. /oo . e). . &.). o /oo 8. ! . is = ae. denotes a polynomial. Prove that / J a e^R(x}dx\ where X&.lt. ft.. e&. is convergent for all values of p. Show how to find the integral of any rational function of # = logtt. where Prove that we can integrate any function of the form P(x. R Prove that  I . = [The tangent e&. any function of the form .lt. is convergent.). on their graphs..210] 6.*) jr ****(*). then / 2 n\ / . when ex =it. b to the curve y at the point (. The reader should discuss the cases in which a or b or both are negative. and that the same r I e^ +^x z 2 n dx^ where n is J oo any positive integer.. cosmx. n is a positive integer.
Express cosh# and sinh x in terms (a) of cosh 2^ (6) of sinh 2o?. * : for an explanation of this phrase see Hobson s Trigo . 2^. The reason of this analogy will appear in Ch. Verify tho. tanh ( . cosech 2 ^?=1. log f e ax COS 2 18. sinh 2 o7=l. 1 Z)a. Determine roughly the positions of the real roots of the equations e * sin log {07 + ^ . X. tanh J 07 = coth = cosech x.r. and 17. may occur. Discuss any ambiguities of sign that 23.) Prove that jD x cosh#=8inh#.] Hyperbolic cosine nometry. cosh 2 o? sech 2 + tanh 2 o?=l. [All these formulae may of course be transformed into formulae in inte gration.27 ?/. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Show Draw ex [iX that =l +07f i^o?2 that the equation has three real roots. + 1)} = ^..gt. (Math. /)x coth#= . /)x sech 07 07. The hyperbolic cosh functions. Z^ tanh . ^7. ?/ 21. tanh x (sinh o7)/(cosh coth o? = (cosh #)/(sinh #).cosech 2 #&.x) = cosh 07.=7. xvi. Dx cosech x ^xlg Z)z log  cosech x coth #. the graphs of these functions. Draw 20.. sinh 07. sinh ( x) o? = sinh x. o?. ex = l+x has no real root except #=0.logcosho?=tanho7. Trip. 07).tanh coth 2 x 07.372 16. cosh = cosh 2 ^ + sinh 2 #. sinh 07 1 arc tan =\ sech #. The hyperbolic functions cosh x* x = \ (e x ~ e *). Establish the formulae ( cosh .] [It follows that the sin wo? 22. are denned by the equations x x=^(e + e~ *). e _ = _l_ sinh j .x) = .r= sech 2 #. . sinh 2o. = 2 sinh ^7 cosh cosh (x+y) = cosh ^7 cosh y + sinh sinh sinh (^7 +y) = sinh x cosh + cosh x sinh y. Z^j.t these formulae may be deduced from the corresponding formulae in cos x and sin #. the graphs of the functions log {x + v/(^ 2 + 1)}..sech # tanh ex . Dx sinh x= cosh = . cosech 07= I/ (sinh 07). e^sin^lOOOO. sech 07= l/(cosh 07). ch. by writing cosh x for cos x and i sinh # for sin x. 19. 1908. same is true of all the formulae involving cos nx and which are deduced from the corresponding elementary properties of cos# and siiio?.
*&.lt.) 32. then I dx 30 ^ ^ is equal to (I/a) arg tanh (xja) or ] (I/a) argcoth (#/).valued. 33. x.lt. of the function for positive values of x.gt. and arg tanh x when !&. 2 1)}. shall denote the functions inverse to cosh x. showing that it &. of opposite signs. according as a? is less than or greater than a. \ r TT and y is positive. Discuss the case in which b 2 + a 2 = 0. Prove that (Math.a) + ^(* . arg sinh is and that arg cosh in general twovalued. x l i* ^ 1 as #*oo. cJ or both negative. 27. or no real roots according as a + b and a. if^^sinha? then y = cosh. 0. I ^ or is equal to arg cosh 29. [The results of Exs. We #.0 it has two.xfa) &. sinh x sinh y = b.#&. and (log #)/#*.lt. and if y tanh. and according as #&.lt. 1. OF A REAL VARIABLE Prove that cosh x&.+ 0. tanh x by arg cosh #&. ^91 as #*. where 2 if 6 2 f c 2 while if 6 2 + c 2 a 2 &. many of the formulae of Ch.a&.gt. Account Prove that if for the 26. xxvu. .lt. 11.gt.lt.!. x\ /. Show that if  \ ir &. and cos x cosh y = 1. then 28.lt. = 2 log y (a: . VI. Trip.6). x &.b are both positive.b\ J7 31. 34.210] 24. Sketch the graphs of the functions. and both of the two latter functions are one.lt.gt. __^__ = 2arct an x/(^) ( (a&. Cf.0.#/a).gt.lt. sinh x.lt.1. c&. or to . Prove that if a (&. x l x has a maximum when t &.gt.6)} x}} (a &. f x &. arg tanh Show x is is defined only when defined for all real values of #.] Show also that the function x = e. and draw the graph 35.lt.lt.gt. Solve the simultaneous equations cosh x cosh y = a. tanh x &. one. b &.arg cosh ( .0. 1913. ambiguity of sign in the first case.lt. then ^=^ log {(1 +?/)/(! y)}.lt. Prove that if a&.] Prove that dx /.lt. then I j = ^ a*  arg sinh (#/). 0. 25. 28 and 29 furnish us with an alternative method of writing to  a good 30. 373 1 and 1 &.gt. while arg sinh x #. has no real roots Solve the equation a cosh x + b sinh x=c. (x&. l x [For x =e( i l sx)/* Ex. then x=log{y^(y 2 = or=log{?/ + v/(3/ +l)}.
Misc. as ra^oo.gt.] 37. divergent.oo .gt. and that x= 10 6 (1382 +log log x) = 106 (1382 + 2 63) = 16450000 gives a tolerable approximation to the root.374 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [IX If {/(n + !)}/{/ (?i)]W. [It is by general graphical considerations that the equation has two positive roots. then {/&. which is less than 200. VIII ( 175 et seq. = n~ n n\ [Iff(ri) Ex. provided of course that is For example l/(?zlog?i) positive. Discuss similarly the equations e x =1000000 #1000000^ 6 ** = iooooooooo 211. To determine roughly the size of the negative root a little greater than large positive root we may proceed as follows. The phrase a little greater than must be interpreted similarly. ex Discuss the approximate solution of the equation easy to see =x imm . 27). where Z&. . can construct functions which tend to zero.)/ 13 82 or (10 6 x 2 63)/13 82.] 38.] 40. Now use #{(2^)!/(n!) 2 }^4asw^oc. and so (l/*)log/()^logJ (Ch. since 13 82 and 2*63 are approximate values of log 10 and log log 10 It is easy to see from these equations that the ratios log x 13*82 respectively. 39. IV. 36. If ex =ximm then 6 6 roughly. yet less rapidly .0. We showed in Ch. Ex. but suffice to give us a good idea of the scale of magnitude of the root. : and log log x : 2*63 do not differ greatly from unity.gt. one a little greater than 1 and one very large*.gt. more ~ than n~ l a however small a may rapidly than l/n. We saw however in 200 that with the aid of logarithms we &. $(n \}/n ^ \\e as n *. and one 1. Thus S(l/w) is ~ but S n~ l a is convergent for all positive values of a. + !)}/{/ (w)} = (1 + (1/n)} &. [For log/(rc + l)log/O)^log.000. the error involved being roughly measured by 10 6 (log log x. l ^l/e.*. then 36. It means a good deal larger than the roots of such equations as usually occur in elementary mathematics . and the question as to whether the series it is n log n * The phrase very large is of course not used here in the technical sense explained in Ch. IV. Logarithmic tests of convergence for series and integrals.) that are convergent if s 1 and divergent if s^l. be. as n^ao.263) or (10 6 log log. such a function. The approximations are of course very rough.
of course. 1. as f to GO  GO if 5 1. The . ^101/100 are convergent.lt. first series is a direct consequence follows from That of the second the fact that (log 7i) 100 is less than n^ for sufficiently large values of n. and divergent if and greater than K/(n log n) for all values of n greater than some And there is a corresponding theorem for integrals definite value. convergent s &.] .l. n. .(log a) * 1 if a&.gt.lt. &. 211] is OF A REAL VARIABLE 375 convergent or divergent cannot be settled by comparison with s any series of the type S n~ .gt. first &. a)(\ogx) we have * xlogx dx ~ dx _ (log f y* .lt. n (n) is positive greater than some definite value.gt.210. l s /(l s) integral tends to the limit (\oga) and to oo if s The second integral tends 1.l.gt. taking /3= 1/209. The same is true of such series as j__j_ ?*(logtt) y ". which we may leave to the reader.gt.(??) is s ]. and such tests are easily deduced from the Integral Test of 174 For since Dx (log x) ~ = l s s D x log log x = [ : .j&.lt. The series ". (n) is positive and less than K/{n(logn) where I. for all values of &. And so. where n n(log ogn) tf(lo~g^ and a are greater than unity.101 / 100 is less than n ~ The (logtt) mm / convergence of the third series follows from the comparison test at the end of the last section. how ever small j3 may be. divergent if s It follows. that &. provided that it is positive. Examples LXXXVIII./&. if &. [The convergence of the of the theorem of the preceding section.gt. 100 for sufficiently large values of n. log log n 2 7lV(logw) It is a question of some interest to find tests which shall enable us to decide whether series such as these are convergent or divergent. are convergent if 1./&.gt. Hence the series and integral __i_ s) r Ja _d_x__ s&.
. the rate of increase of the higher logarithmic functions is extremely slow.000. The reader should starting from the equations ". 4. #&.ooo &.. x &.gt.gt.693. Dx (log*.lOg*.1. The same and e e exponential functions.0. For such series we need a still more delicate test. &.../(log log n) cannot be settled by the theorem of p.gt. since in each case the function under the sign of summation tends to zero more rapidly than l/(?ilogn) yet less rapidly small. are convergent for ~ all values of p and &. X) l ~a = X log X Iog 2 X. divergent /&.lt. Thus to make I we have to suppose x a number with over 8000 figures.0 requires x&.. &* e 1Q &. to prove i fc r n (log* n} B _ dx log logo n.gt. the series arid integral theorem : no n I _ log n . and it is easy to see that ee &.\ J a % x Iog 2 x .gt. Iog3 #logloglog#.gt.e.gt.] X (log Tl .376 2. ^ and a being any numbers great to ensure that logA 7i and log # are positive when n^n$ These values of n and a increase very ra pidly as k increases and fc : thus log#&.0 requires . j . 1 where a is any positive number however than n~ l (logn)". log _ fc _ the l &.* log log log log x figures. while 1010 10 has of course 10. following x (log* x)* are convergent if sufficiently or A ^a. log _ i s&. Conversely. .lt. * See the footnote to p. be able..gt. ee ee &. 362. such as e * where a has functions as d a and d aa remark of course applies to such 9 It has been computed that 99 has 369. e increase with x.gt.. .log log #. fc X}* where Iog 2 x =. and so on e 2o.gt. Iog2 ^&.gt. loglog^&.gt.0 requires 20.100 value greater than unity any . (log TI) (log log w) ?i 1 (log w) (log log yi) The question of the convergence or divergence of such series as n log n log log n n log w &.gt.!.000.000.? .000..gt. 10. . . IQSOOO^ The reader should observe the extreme * rapidity with which the higher . ee. 3.gt. ".*. similarly the series s nl are divergent. (log nfrt w ioo/ioi ^ og n yw  ^ i og ^+ j The series (log log n) where s&. 375. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS The series [iX n are divergent.
lt.lt. and y is usually called Euler s 174. 1. +ii n then 2(wn /s /l _i) is convergent or divergent according as 2?in is convergent or . and ..] 6.gt. oo . Show that the series +. &. 1. OF A KEAL VARIABLE Prove that the integral if s &. [Compare the general term of the series with l/ from 2 n~ 8 in the same way that the above 1 and otherwise divergent. s &. I log ( where )! dx. This result also may be refined upon by the introduction of higher logarithmic factors. 1.] 2. 7. 377 a con / 1. tends to a limit y as noo. derived from 2 (Ifri). dx has no meaning for any value of s. Euler s limit. negative. divergent if s2i [Consider the behaviour of as f*fO.gt. Prove generally that if 2w n is a series of positive terms.211] 5. and that 0&. [This follows at once from The value of y is in fact 577. . is divergent.y^l. constant. If &. 1 and so the integral diverges at the upper limit when sufficient 0.. &. &. is convergent if s .lt. If a and b are positive then tends to a limit as n*co 3. is vergent 1.lt. 1 then tends to a limit as n*~ 4. if s : . s &.gt. f The last at the lower limit is 1 is a necessary condition for convergence example shows that s but {log(l/#)} 8 tends to GO like (1 x} B as #^10..lt. Prove that I jlog(H &.lt.. .lt..] The necessary and conditions for the convergence of I x a~1 fW dx jlog are a &.lt..1. 8^ = ^ + 112 + . Show that Examples LXXXIX. Show also that the series derived series is 5. s &.
[The proof the same in the case of convergence.lt. Hence. 212. 1. If 2un is n_l divergent. the derivatives of the exponential function are equal to the function itself. / and so 2(ttK_i) convergent.lt..gt.lt.&.lt.gt. whatever be the value of x oo . to zero as y denoting Euler s constant.lt. it is evident that ) . But x n \n !9 as e ex &. log {1 + ( = log (* B n _ j) and . .+ . . 2u n is divergent then w n /sn _ 0} and w n /* . Since all s and log Theorem.lt. and oo Subtracting and making n^&..s Prove that the same result holds for the &. log S ( 2 / Sl ) oc . x we have 2 _ + #_ xn 4 f ~l 2! (n + l)! x ?i! e 0x where (Ex.1 (Ex. f n being 7i*oo numbers which tend we see that the sum of the given series 8. n . 6&..378 divergent. n^ GO . and e x . 1) .. &. &. when it converges. Series connected with the exponential x Expansion of e by Taylor arithmic functions. 2u n is convergent then If &. . 1. 12).gt.] Sum the series !. is &. then *n n _ 1 and the divergence of 2 (*/*) follows from that of 2 (ujs^. and u n from a certain value of n onwards. as might happen with a rapidly divergent series. 0&. [We have by Ex. See also 213.] Prove that the series oscillates finitely except when C= y. If on the other hand u n n _ l for an infinity of values of n. making n tend to we have The series on the righthand side of this equation the exponential series. oo as n*~ series 2 ( un /sn ).s then u n js n 7. 5 n _j tends to a positive limit * rt _ioo. xxvu.gt. + ^j + ( &. is log 2. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [If is [IX I.gt.2* . for all these values of w. In particular we have + and so 2] is known as + ". LXXXII.] + log (*3 /* 2 + + log (S n /Sn _ j) = log (*/*!) tends to 6.
..gt.gt... Employ the exponential series to prove that e* tends to infinity more n rapidly than any power of x.x .gt. From this the equation (1) follows E n*&. n.. = e *^". Examples XO. The reader reproducing will observe that the exponential series has the property of itself when every term is differentiated. 3..gt. unless x is then the greatest term in the exponential series is the an integer. and deduce that n ! lies between 2 n (w/e?) and 2 (n + 1) (n/e} n .2 l. Then 1. : of powers of x would possess this property connection see Appendix II.x) =/(0) = 1.  . 1. its truth when x is negative follows from the fact that the exponential series. LXXXI. And.*. provided n&. Show that _ 2.(n{e} is one term in the series for e n. by The power series Let and suppose that x &.x&.] 4. If x is positive [#] + l)th.211. and when x is positive . 7. we have also. E Thus But ( 208) the first and last functions tend to the limit ex as therefore n (x) must do the same..x. satisfies the functional equation f(x)f(y] =f (x+y\ so that / (#)/ ( .= for all positive values of a.. where and i/ = l/. .gt.lt.n \n) which is less than n (x). as was shown in Ex.212] a result OF A REAL VARIABLE as the 379 Also known a* exponential theorem. .] 5. Prove that n = (nn jn !) (2 +S + x 2 ).. [Use the inequality ex /n !. &. by the binomial theorem for a negative integral exponent. for and that no other series some further remarks in this it for ex is so important that it is worth while to investigate an alternative method which does not depend upon Taylor s Theorem. n [For n /n I Show that n e \&. 0. when the preceding term is equal to it.
and proceed as in Ex. Deduce the exponential theorem. Sum the series I express Pr (n) . or. Prove that $l.] Determine a. = 2 {(^ 3^ + 3) e* + ^2 3}/*2 ..380 6. is a positive integral multiple of ( ( = e..+w 3 3 . c so that {(x + a}e x + (bx + c)}/x 3 tends to a limit as ~kv _L / evaluate the limit. [Multiply numerator and denominator by ft+1.# 4. Draw the graphs of of ex . = 50. Sfj=i (4^when 3 last series is equal to zero x= . [We can Pr (n} in the and 8. since the lefthand side is integral. and the righthand side less than {!/(? + !)} + {!/(? + 1)} + 2 .. 10. ( !).. Trip. and draw the graph of the function x+a . 7. l + #. (Math.2.. Prove that 2 ^^. b. . THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Show that e is not we must have a rational number. 11. 3 then 2 In particular the 9. multiplying up by q \ . ^ where form Pr (n) is a polynomial of degree r in n.. Show that 2 ^= and that if /S n = ! + 2 + .) Prove that 2 (nfn is where any 2 (w /^ !) l} !) positive integer. 2 (w 2 /^ = 2e. and that 2 e. (n /n k !). l+x+%x\ l + tf + J^ + ^B 3 and compare them with that 13. tf^0. [If [iX e=p/q. 1904. where p and q are integers.__ s positive or negative according as n is odd or even. 7. 2! q\ and this is absurd. 12.
~~ __ /v. . .e Jo Jo ^ xdx=~e\ * and generally 15.\p log a + %p 2 log a 213. ! JT2 = e*lA.. when 1 x 1.lt. Xi(t) I* Hence prove that t then t t Jo XQ dx&. A = e*l. where Hm = r* J m ^ ^ y.lt.212.1).. For / + 1) = and so. 213] 14... 1909. require to show that the limit of This is almost obvious when Rm &.gt. x = 1 ..lt. &.. e*.# ..lt. . X v (t) &. so that  1 ~~ * See Appendix II for some further remarks on this subject.) The expansion in powers of Another very important logarithmic series. dXvldx**Xv i. OF A REAL VARIABLE 381 If X then = e*.gt. Show that the expansion in powers of terras (2 x 2+ P = a 2 begins with the p of the positive root of a {I. to the series obtained by integrating each term of the series 2 1 from =0 to t=x. Xi(t}={ X l dx&. And this is in fact the case. (Math. i. it is natural to expect* that log (1 f x) will be equal.t*. when m tends then to oo &. + ( 1) _ L 1 \m fm ~h t x &. if i 1+1 2 .e.1 . Trip. . and 1/(1 + 1) = 1 . Deduce the exponential theorem. . If on the other hand we put = and x = f. .tf unity.0 .m+i i m+1 1 and therefore less it than l/(m 4. Jf3 = e* if &. We is zero.lt.l Jo xe*dx&. + log a}} .lt. 1. for Rm is positive and less than x /mj/ /. &. Since if t is numerically less than 1\. to the series x t + 1 #2 + J& 3 . .gt..^ + L . x is that for log (1 + x). .lt.(#2/2 !).
1 as well as when x = reader. a? ^ 1 . VII (Ex. for.lt. represented by Ch. If 1 # lies outside these limits the series If x= we obtain a result already proved otherwise (Ex. 214.)dt + Jo = a? provided that l^x^l. since arc The only difference tan. log (1 1 &. 1911. we need is values of only consider positive when x = . LXIII. and which we saw the integral. The series for the inverse tangent.  jrf /* and so Rm * 0.f). since the in the range of integration is 1/(1 . 1. LXXXIX. &. i Rn . a little simpler. 3. And the series is convergent We leave the discussion to the is The value of arc tan x lies which represented by the series \ir of course that which in between . is x. difficulty presents itself in the discussion of the remainder in the .) 4 [A Obtain the series for tan x by means of Taylor log(l+ff) and arc s theorem. . It is easy to manner that prove in a similar rx arc tan * = fit 1 rx i I J o f^l (lt* + t*... Hence provided that is not convergent.lt.lt.?? is that the proof is an odd function of a?. Trip. 7).u) has the sign of ( l) Also.\TT and when  1 ^x^ 1. ni we have o &.382 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [IX which shows that Rm greatest value of 1/(1 . we obtain the formula ExamplesXCI. Prove that if x is positive then (Math. 3) to be the value If x = 1.
.r 2 )} 1) ! 1 ^ (# 2 + 1) ~/2 sin {n arc tan (!/#)} is (Ex. fails when x = Q.l) ( l {1 1(1 1 +. 1910. if x &. obviously If y&. 1912.213. by putting ^=10 in the second formula. 6=arg tanh (1/49). easy to see (cf.9 log 2 (Math. then the formula log 13 = 3 log 11 +log 5 . 0. as arc tan (I/a. = arg tanh 3.) is then Ex..*] 5. c 2. Find log 10 to 3 places of decimals from the formula = 3 log 2 log 10 7.gt.ey^x".2. 163). Show that if log 2. Trip. the corresponding discussion for the Binomial Series.. which not greater in absolute value than 1/n. + log (1 + ). Prove that log \T^/ 2 ". 1 1)". .. 11) that arc tan (l/) must then be interpreted as meaning . Show that where a = argtanh (1/31). and log 11 are known.0 then [Use the identity y = (\ calculate log 2. J+ slowness of its practically useless. that log 11 = 2397895.) 8.. and there is no difficulty about the remainder. 11).] 6. (Math. #=:(should be used Ex..] * log and log 5 rapidly and with The formula It is for D xn ?r. In the case of the second series we have Dxn arc tan x = D xn ~ = .) gives log 13 with an error practically equal to 00015.. LVI. 214] first series is OF A EEAL VARIABLE when x s is 383 n ( negative.gt. +^~^ J /( 1 ~^~y) 1 is s series may be used to a purpose for which the series convergence. and log 3 = 10986123. 2 and (cf. (i . owing to the Put y = 2 and find log 2 to 3 places of decimals... show. and that if ^&. 9.. log 5. Cauchy form.gt. (1/161). XLV.i(\ +6xy. arc tan x undefined.. viz. Given that log 2 = 6931471. if Lagrange s form Rn = ~l l} xn l{n (\ + 6x} n] used . XLV. Trip. [These formulae enable us to find log any degree of accuracy.
[Proceed as in 17.. Using the logarithmic series and the facts that Iog 10 2 3758 = 3758099.) 13. Show that f J f i ^ ". Trip. that i x log \i _= & / if  1&.lt. 1896. 1910. (Math. show that an approximate solution of the equation #=100 Iog 10 # is 23758121. to this order. Trip.J 214 and use the result of Ex.) 15. (Math.384 10. and 215. Show x &. Prove similarly that 18.. a: Show 1 that the expansion of (1 in powers of x begins with the terms 12. We have already ( 163) investigated the Binomial Theorem . and log sin x = log x . when #=10. + x + x 2 + %x\ Show that (Math.+ . and estimate the accuracy of the result. x as far as #*. 1910... to 6 places of decimals. (Math. obtain an approximate value of Iog 10 e.~J o + b and so that the sum of the series can be found.. [Use Ex. Trip. sums of 1. 1.. Calculate in this way the The Binomial Series.. Trip. Prove generally that 1 l if a and b are positive integers then 1 + a~~a~+b a + 2b~ _ f l t a~l l dt t ". for large values of x.. LXXXI.T ? r =^^ + ^ 9  if l^tf^l. Deduce that l_^ + i_. 1908.) 7.lt. ir = 4 arc tan + #) 1 + (1/5) arc tan (1/239).) * 16..) Apply the formula. to approximately. = { 7r + 2log( N/2 + l)}/4 v/2. 2. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Show that (1/3) [IX = arctan (l/2) + arctan ?r and calculate 11. Trip. 1910.fa log cos x + log cos %x. XLVIII. and Iog 10 e = 4343. (Math. Expand log cos x and log (sin x\x) in powers of verify that.] 14.
1. Apply the process to [Expanding by the binomial theorem.#&. Let N* be the square z and let nearest to x x being positive. xjN 2 is comparatively small and the surd 2 ij{l (x/JV )} can be expressed in a series M MN^ MN t JM=N which is at any rate fairly rapidly couvergent r and may be very rapidly so. 385 irrational assuming that !&.#&. =:x_ /.(the value given by first two terms) as an approximate value. 215] OF A REAL VARIABLE ra is rational.g. ct seq.214. and the proof of the theorem given in 163 retains its We shall not discuss the question of the convergence validity. Prove that if !&. e. Approximation to quadratic and other surds.lt. Infinite Series. is less than 003. and the is less than 3 2/64 2 which . pp. viz.* so that the rule for the differentiation of (1 Examples XOII. After the second term the terms alternate in sign and decrease. Let N/J^ be a quadratic surd whose numerical value is required. Hence the error is one of excess. p.lt. 25 .] . the error being less than #*/32# 7 The resul t follows. Thus Let us consider the error committed in taking 8j3g. .! then 2. The same method may be applied to surds other than quadratic surds.lt. Since x + x or . 271. c.lt. * See Bromwich. of the series when x = 1 or # = !. 3 If x is small compared with N 2 then the error being of the order x^fN 1 .! and that we have When m is Dx (1 f x) m= {m/(l + x)} + x) m remains the same. to 4/1031. we have the error being less Also fix 2 than the numerical value of the next term. 150 H. Plane Trigonometry (3rd edition).mnot be greater than N. Hobson.
.. then a good approxi mation for $M is 56 56 A3 this Show that when ^V=10....... by less than 1 per from f N+% (M/N 2 ) by less than AT/90000. We know that this series convergent for all values of x. as A ^.. numerically less than so that p (A) * as A*~0. exp # x exp y = exp Again where p (A) is expA1 ^ =l+2 + 2] + l h h2 ..... xn 2 w 2 .. exponential and logarithmic functions.. /vi2 starts is from the exponential series l +#+ gj + ...) to Show how sum the series where Pr (ri) Sum is a polynomial of degree r in (ri) n.. we have Writing y = ex.. Pr in the form /771\ ( ) A + Ain+A 2 n (n 1) + . 5.0. (Math.. We that exp have now a choice of procedure.. And so exp (x+h) i exp# = /exp A IN ^exi ft exp# \ Ej ti J (3)... [Express 7. and we may therefore define the function exp x by the equation + ..... = 1.. Trip. as in Ex.. LXXXI. (1). that (#!?/) . approximation is accurate to 16 places of decimals.. of either then %M differs (Math. 7....... 1882... or Dx exp ^=exp x we have proved that exp# is Incidentally a continuous function.... (2). 6.386 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [iX If differs from 4...... as in Ex... #=1... M N 3 cent. and x is small compared with N.... Trip.px and observing vdt . We shall now give an outline of a method of investigation of the properties of ex and log x entirely different This method in logical order from that followed in the preceding pages.] the series 2rt o \nj o W I /7/2\ ) ^n and prove that An alternative method of development of the theory of the 216. We then prove... xc.) If M=N*+x... 1886. 7......
.) 2.m x} ~ MISCELLANEOUS EXAMPLES ON CHAPTER IX*. in a similar / (m.x &. n From (2) it follows that if n is a positive (exp #) If =exp nx. e x. (Ex. 7 . x is a positive rational fraction m/?z. calculate Iog 10 2 and Iog 10 3 to four places of decimals. is the greater. and \og 10 n=plq. we is when x is irrational.lt. since 10 P ends with and therefore is not divisible by 10. have 10 p = ?i 9 which is impossible. to exp x or ex . Trip. lo g 2 J [Take logarithms &.] * A considerable number of these examples are taken from Bromwich s Infinite Series. x} f (m x) = / (m 4.. Show integer not a ct . N N log 10 tt=a+log IO A r cannot be rational. [If n is not divisible by 10. n (exp n l) =exp n. starting from the equation t where  1 &. in this chapter. 252 . 1905. then {exp (m/n)} = exp m = (exp 1 m ) . then Iog 10 If n 10 ^\ where 3. Develop the theory of the binomial series 1 + manner. that Iog 10 n cannot be a rational number if n is any positive power of 10. 1.215. may exp. LXXXI. .lt. and so we have say. The logarithm then defined as the function inverse Example. 1..# exp ( &) = ! .lt. Given that Iog 10 e = 4343 and that 2 10 and 3 21 are nearly equal to powers of 10.lt. for all rational values of x. as being equal to exp x. and so exp (m/ri) m This result equal to the positive value of (exp 1 ) /w be extended to negative rational values of x by means of the equation is . 216] and. ( s^2) &. Determine which of (^e) and observe that V 3 /(V 3 + i0 and I 7r ".lt. define the logarithmic function as the function inverse to the we are brought back to the point of view adopted earlier But we may proceed integer then differently. where e = exp 1 = 1 + 1 + Finally + define + . if OF A REAL VARIABLE 387 we exponential function. we and n q does not. 6). (Math. V3 692 9 &.
and the function itself tends to zero as x * cc .gt.lt. not defined ? Consider also the same question (6) equal to 1 (c) equal to the functions to..) Show that 11 Show 1 that oo .*+ !)".] that the function (log 12.] as x that (log ( 1 + #)}/# and #/{(! towards GO increases from Show + x} log (1 + x)} both decrease ..2vto.x for (a) 5.gt. 1910.&. f [It is easy to prove that log#&. [The derivative of the function is r*+r as is *(&.  1 then w&. [Put 8. steadily . .  N log(lKr) . where to . 1909.gt. and !(*) + (*} . but for a?=0. where 6. 0.gt.log #)/(. = 0. towards co that. x is Mx N where that ex than the greater of 2 log greater Show &. log J/.gt. (Math.lt. 1906.gt.gt.. 9. This expression since up the righthand side into partial fractions. Show . Trip. log log #. 2 (1 +x) (log (1 + #)} .) 7. Show that is negative and increases steadily towards as x increases from towards co . and (Math..gt. +x=e*&. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS [iX For what values of x are the functions log x. where to = log. M and N are large positive numbers. attribute to the value when #=0 is becomes continuous we Use Ex. llx. cx 0.. Trip. and use the fact that sinh when &.decreases x is it steadily from 1 to as x increases if from towards it [The function undefined when #=0. . 7 to show that the derivative negative. M and 16 A 72 . log (tf+r)=log *?+.gt. decreases steadily as x increases from 13. . 2#V#. . Trip. is and find its limit as P ositive #*.] \i/ \*/ Prove that (Math. &.. %x&. If 1 x&. lllx.gt.388 4. Show llx (I+x)~ 10. log log log a?..a?). as x increases from 1 assumes once and only once every value between the function 1. and so the inequality given is certainly satisfied if and therefore certainly satisfied if %x&.) &.(*+*) easily seen by splitting is positive.
&. .lt. Misc. VI. co . + .gt. en and *n tend to as n &. use 209 and Ex. . ra Prove that .. 4..] Prove that and b are positive then [Take logarithms and use Ex.lt.] Show that where y 19.QO [We have 1 where 17./&.gt. Ex. if Now a.. 18.. LXXXII. [Cf. _ i og x as *. Ex.gt.gt.lt. (Math.) Show that . w=4 (1 .] Show that li_ 21.^ /). (Cf. 6)..+. [The sum of the first 3n terms is where 20.(#) 389 (#)/&. Prove that where Sn = !+ + . is Euler s constant (Ex. Ex. ^ iv 2^ = 1 +O 0+. x .] 16. if a? is 6.lt. a f(x}=x 15. If/(#) and then /(#)/# (or) * oo. LXXVIII.(#) = log #.OF A REAL VARIABLE 14. positive then n log {(1+#V)}^. &. as n &...? tend to infinity as #*oo. 16.. If/&. ^?i log 3 f . and are positive integers then as ft  oo .i + . I Hence prove that the sum of the series when continued to infinity is 3+ 22. Show that the series being formed from the series 1 ..gt. 1905. &. LXXXIX. 2 log 2. 33. by taking alternately two positive terms and then one negative.] By taking prove that (log #)/#". LXXVIII. Trip. 1) and fn &. and / (#)[Use the result of Ch.*() for all positive values of a.
for all values of x. member Then. of a.1 then 2! (Math. 461 et seq. B. Prove that % (a/n) n tends to or to according as a &. Show that the sequence al = e t a 2 =e e\ a3 = e e&. then the term [If a is the algebraically greatest & outweighs all the rest as . which is rather + shown beyond the scope of the theorems of this chapter. n It can be (l/7i)}". Find the limit as # oo of \b +b x+. [The logarithm of the function if 2 log 1 + _27\ L] 27.lt... (Math. . THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Prove that the sums of the four series 2 _IX 747i are .a if x is positive then (+*)/! * n is / f as 7i a.gt.lt. . 1886.a/e.] if e k (x} is any member of the when n &. e.. *.. pp.oo .] . [If un =n\(ajri) then ^n+1 /i/tt =a{l that the function tends to oo when a = e: for a proof. x = e x \ and so on. .# oo . different real numbers.gt.lt.. Prove that x &. . &.) [The series is difference between l/(# + l) 2 and the sum of the first n terms of the 28..) Prove that diverges to oo .] 25. . see Bromwich s Infinite Series.. .gt. . a n . of the exponential scale.. J. Trip. 24.. e or a&.] 29. .. tends to infinity more rapidly than any [Let e l (x} e e% (x) e \( ek (n} exponential scale. log 2 ! respectively. &..390 23.+ 1 distinguishing the different cases which 26. can hold are polynomials and a.4e aa. 1908.gt. Trip.i Deduce that &. No equation of the type where A. [Compare with 2 (x/n). ! 7 47i 2 l 7 (2w+ !)*! _ oo 7(2*1+1)*! JTT J. may arise. .
lt. behave as x+ + y = sin x log tan ^ ^7. a ^/ *.log x ( \ a centre of symmetry. The equation then it &. or three real roots if a 0.0 or a = 0.OF A HEAL VARIABLE 30.lt.a.lt. *. n (x) then (i) 2. y from to 1. of the result for n= 1. dy 1 . has no real root unless that of a. It follows that /(0) = tend to zero with y.] .lt. or two if a how to distinguish between the different cases. 6 has two real roots or none. / (0) = 0. &. ^ is &. none.0n and (iii) all the roots of n =0 are real and n+i= (ii) and separated by those of n _ 1 = 0.lt.gt. &. has one..gt.i &. How do the functions * 8in (1/x) * sin2 {l/x \ a.\[In the 0&. ^io g x J2^ an odd function of x.lo fe x T one. and that as steadily increases x J showing x increases through ) .f&.lt.(*)=/". Then where ey and t v /".] 32. that the point all (0. (j) . according as a log a b ex &.r after differentiation. .0.0. ^&. Trace the curve y = . Prove that (f) if Dx n e~ x&. [To prove (iii) assume the truth distinct. Show by graphical considerations that the equation 36.gt. a &. x&.# $.(&.&. . i) is real values.(0)} or a = J{f".i&.)/(#).lt.gt.{. = K and for large (positive or negative) values of #. 35.lt.lt. The general solution of f(xy}=_ f(x)f(y\ where a where a is a constant and that of is x . one.00 *** Trace the curves y = tan x e tan x . [In proving the second result assume that / has derivatives of the first three orders.(0) is positive or negative. is clearly ( fe  l Also 1. and then whose sign /sinhi.gt. according as/". 1.gt.l. 391 Prove that where a is to be put equal to ty (x} and /3 to (a?) .* ( \ is the same as first place x I. (0)}.lt. ? 33. of x for which &. Deduce that the equation 1. If a or a lo a 6 . =ax + b has one real root if a&.i = e. and show 37.r. two. so that a = ^{/".gt.gt. &.a&.. Establish a similar rule for the differentiation of 31. and consider the signs of K for the n values l $(#) a polynomial of degree n. function.lt. 2. 34. : / is a differentiate is cosh ax or cos.
a glance at a figure will show that these results are contradictory.J(x Hence llx for all values of #.&+2 A&. + jjyl xn \ in powers of x are *&.gt. since its degree is odd.gt. .lt. and its x which has the sign of 38.lt. I then 42. . . If it has (#) or /2A: + i (#) niust vanish once at least between ) ( f&. 0. the error being about J {(a b}/a} This formula is interesting historically as having been employed by series. . Trip. ~~ f 1 * ^2 ^21(71 43) 4./. + Vn ^ = has one real root if odd and none n even.lt. + ^ .lt. .gt.zk + \( x } Glc fzk( x } has just one root. / The n + 2 terms in the expansion of log ( 1 + # + ^. then f zk + z a and /3. and so two.. = has at least [Assume this proved for n=l. and it cannot have more since. Then 2k + i (x) one real root. Prove by multiplication of series that if 1 &.. positive roots and one negative if a if &. and Hence Prove that if a and b are positive and nearly equal then s [Use the logarithmic approximately. one negative root if a. fx * with first x. (Math.] 41. if it f had. would have to vanish once at least. say at y.l^ .) .lt. f &. Napier for the numerical calculation of logarithms.gt.a &. 2. &. no real roots. say a and (3.0 2 &.lt. Hence/2fc+ i(#) = = cannot have more than two. \fflnh dy/dx&.392 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS inside the large [iX The function derivative is bracket tends to zero as x &.gt. n is negative.] Trace the curve y = e + 2#). x. . 2.] /2 t + 2 (V)=0 has 40. And / But * 2 (y) =/ H i (y) is + r~ &. Show n is that the equation if / is (x) = I +x f 1C* + . x &. /2t + 2(A ) is a ^ so positive when # large (positively or negatively). Prove that where 43. and show that the equation a = e1 has no real roots and two 39. .lt.gt. 1899.
o J while if a is positive and 2 ay&. The results are more easily obtained by problem of the differentiation of an infinite series is beyond our range.lt. 1 then differentiation.. Are there any other really different cases in which the integral is convergent? 48.OF A REAL VARIABLE 44. and 2 /3 &. then y)) . /3.x &.gt. Prove that if a. of *./3 the value of the integral is that value of the inverse tangent being chosen which lies between and TT. J p Deduce. Show that if 1 &. 6.log a.gt. 393 Show that the expansion ezp. Prove that if a&.] 46. but the [Use the method of Ex.. and verify independently. that provided that a and 5 are positive. y are all positive.1.lt. n lo Jo a \] J J /". each of the functions a. 47. alogais positive for all positive values of a. Prove that o x+a)(x Jo (x+a)(x + b) P f* = a log(Y ab b * \bj a  z. I then du Ji .gt. n+ ~ 2 n u*u^.^ ( Math TriP 1909  ) 45. xcn.l in powers of x begins with the terms ".ay.
[Use the substitutions x=\jt and ^=aw.lt. b&.0 then r io g = .gt.a&. Discuss the case in which a = 1. 1.0. Prove that I log 1 ( +". 1 .gt. and if a&.lt.J(a  Prove that and deduce that if &. 2 &. [Integrate by parts.gt.394 THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS is [iX and deduce that the value of the integral if 1 &.2 ) J o \ xJ dx = ira if a&.lt. 49. where a&. in the same T 1J y_ S2 value is ~ 2 arg a+1 arc tan xdx /i 51. Prove that if a&. showing that ! / ^ .lt. Transform the integral its JO ^*T*^^^ ways.gt.gt. a &./3&.] 54.lt.0. 0&.l.lt.lt. If 0&. 2a log a.gt.] . JTT \ log 2. then dQ cosh 6 + b sinh 53. then 1 1f log  52.1.
such as the polynomial P(z).y) + ig(x. and conversely same thing as to assign a : . We no general definition of what It is meant by a function of z. as that in which we defined a function of the real variable i. AND CIRCULAR FUNCTIONS 217. natural to describe such expressions as functions of z. am z = arc tan (yfx) functions of z\ The definition.y\ For example variables x and y. it In this chapter we shall generally find convenient to write x + iy rather than x + yi. although perfectly legitimate. Ill we defined the complex variable = x + iy *.e. EXPONENTIAL. that this definition is can be derived. Functions of a complex variable. between z and sponds to is a function of z if any relation subsists to say that corre in virtue of which a value or values of Z Z Z some or all values of z. so are x and to assign a value of z is precisely the be found. z \ 2 *J(x + 2 ^/ ). same might seem natural to define a function of z in the way x. and in fact we did describe the quotient (z) and Q (z) (z)/Q (z). is precisely the same thing as a complex the definition suggested. on closer not one from which any profit it will But y. z In Ch. For if z is given. x are * iy. xy. . where have however given are polynomials. function of the two real f(x. Thus a function of z according to pair of values of x and y. examination. as a rational function P P .CHAPTER X THE GENERAL THEORY OF THE LOGARITHMIC. classes and we considered a few simple properties of some of It is expressions involving z.
or in other words complex therefore . from y) A toB. because it [X all. and ty and suppose that. we should be led far beyond the limits of this book. does not really define a new idea at It is more convenient to use the expression function of the variable z in a more restricted sense. The question which will occupy us in this chapter is that of defining shall call the we and determining the principal properties of the logarithmic.f) and (f) ty are functions of . 0) . by the equations Let AB where &. Real and complex curvilinear be an arc C of a curve defined integrals. a special class to which the expression shall be restricted. to We shall begin with the logarithmic function. and roots of z ( 47). in the same direction.396 is futile THE GENERAL THEORY OF THE LOGARITHMIC. explicit and implicit. 219. complex number z.lt. the argument ( 44) of the the argument of the function f {z) under consideration. t with continuous differential coefficients the point (x. There is no difficulty in extending to the complex variable the definitions We of algebraical functions. rational functions of z ( 46). this selection is made. It is natural attempt to define it by means of some extension of the definition log x = t*dt I (x &. which we gave 26 In all these cases ( 27) in the case of the real variable x. But if we were to attempt to explain how to pick out. functions are of course so far defined for real values of z only. as t varies from td to t l} moves along the curve. ex These ponential. the logarithm indeed for positive values only. have already defined polynomials in z ( 39). point z. and what are the characteristic properties of the special class of functions selected. attempt but shall confine ourselves entirely to special functions defined directly. 218. and in order to do this we briefly shall find it necessary to consider some extensions of the notion of an integral.gt. . from the general class of complex functions of the two real variables x and y. W^e shall therefore not to give any general definitions. and trigonometrical or circular functions of z.
y (/&. a polynomial in z or rational function of Then we define f(*)d* (2) I c (u which is itself + iv) (dec +idy). where g and h are continuous functions of x and y. (c) The value are paths such as are contemplated in the definition. Suppose example that f is real and positive. The definition of Log Log f. the general logarithm of by the equation where C is a curve which starts from 1 and ends at f and does not pass through the origin. of Log Log z has been chosen.gt.be any f. Now let f=fM*7. of when the particular path of integration at present it is not clear how far the value z resulting from the definition depends upon what path is is thus defined But for chosen. AND CIRCULAR FUNCTIONS define the curvilinear integral 397 Then we (1). We define f. to substitutions x = (t). Further let us suppose that f(z) is = u + iv z. i. We call C the path of integration. complex number. 54) the paths (a). Thus (Fig. 220. as being equi valent to the ordinary integral obtained by effecting the formal = ^r (t). defined as meaning vdy) I (udx c +i I J J c (vdx + udy). say . (6).217220] EXPONENTIAL. Let us suppose now that so that z describes the curve C in Argand s diagram as t varies.e.
is log But in this case. the equations x = t. 54. log f Log f is (possibly at any rate) a many valued function. of Log f. and in point of fact we is equal to log I every value of Log . shall see that this is not the case. value of Log f. is in . when f is real and positive. we may get a different path . Then one . by whatever (2) different value corresponding to every (3) we may get a number of different values each of which class of paths : corresponds to a whole and the truth or falsehood of any one of these alternatives no way implied by our definition. so far as we can see at present.398 equal to line THE GENERAL THEOEY OF THE LOGARITHMIC. and with this particular choice of the path of integration. (1) viz. in the general case. the logarithm of f according to the Thus one value at any rate definition given in the last chapter. the path of integration can be chosen in an There is nothing to show that infinite variety of different ways. y = 0. three alternatives are equally possible. and log f is only one of its And in the general case. values. f. as . from 1 to a path which possible path of integration is the straight we may suppose to be denned by Y Fig. This is why we have adopted . Log Log f instead of log f. In this case. we have so that Log f is equal to log f. the notation f. that we may always get the same path we go from 1 to f.
or rather one of the amplitudes of 1.log 1 = log p /&. EXPONENTIAL.j&. us suppose that initially = 2kir. 0). It is clear that [log r] = log . 221] 221.t&. Also x+ in virtue of the definitions of iy \dt dt 219.. y = r sin 0. It is evident from the figure that increases from 2&7T to as t 2/C7r + &.TT TT. r \ dt d6\ smv ^r) + i (sm dt \ . y) of any point on the path C are functions and so also are its polar coordinates (r.gt. may have any value other than zero The coordinates of t. any positive value.gt. AND CIRCULAR FUNCTIONS 399 polar The values of coordinates of the point z We suppose for the Log f.). &.lt. . . viz. But x = r cos 0. Let us of integration is the straight line from suppose first that the path value of is the The initial 1 to f. Let 2&7T. f Q dr v dt 7 + db r cos n jdt so that where [log r] denotes the difference between the values of log r at and = and [0] has a similar the points corresponding tot = tj_ . when the path of integration is a straight line. while p may present that (/&. moves along the line. Let us suppose that the = % are p and so that = p (cos + i sin f . &. Thus and. &.lt. have t&. (#. f ( cos v ~dr r.220.lt. amplitude of 1. Log f = log p + i&.. meaning. .lt. but the value of [0] requires a little more consideration.lt.gt.gt.gt. . where k is any integer. Thus f or a real negative value. and dx dt T dy + * r dt .
. Along the equal to [0] shown in the figure by a curve continuous initially line.gt. so that [0] = 47T + 6 arid Log f = log p + i (47T + &. If is initially equal will final value will be 2kjr + 4?r + &. creases to the value 2/J7T de .. equal to 2&?r. Hence it be convenient in general to denote the principal value of f. Fig.. . We value./&. imaginary part lies between fact that its Next let us consider any path (such as those shown in Fig. and the principal value is characterised by the TT and TT./&. and its it Fig. Consider./&. the principal so that the Log f is the ordinary logarithm log f.XOP at Q. and then increases again.gt. c?. then and the Log f = log = log p f if.gt. neither of which includes the origin. Log f by log Thus log f = log p + {(/&. Y On the other hand it is easy to construct paths of integration such that [0] is not equal to &./&. first 0.. is still It is easy to see that &.lt. and finally but slightly more the curve bound Thus if the path it of integration line is from 1 to such that the closed curve formed by does not include the origin.).lt.gt. for example.gt. to 2&7T. will shall call f principal value of When is this particular value of real and positive. 57. 56) such that the area or areas included between the path and the straight line from 1 to f does not include the origin. being equal to 2&7T The dotted curve shows a similar to 2&7T 4case in which the straight line and complicated two areas./&. &..lt. the curve indicated by a continuous line in have increased 2?r when we get to P and by by 4?r when we get to Q.400 THE GENERAL THEORY OF THE LOGARITHMIC.gt. f p = Logf and $ = 0. for example.lt.lt.
There is however no reason now integration why we should not revert to the natural notation in which z is used as the argument of the function Log z.f&.lt. and so and negative. and is therefore not admissible as a path of Both IT and n. The values log ( z} + iri and log ( z} ni correspond to paths from 1 to z lying respectively entirely above and entirely below the Either of them may be taken as the principal value of Log z. 6) to denote the following examples. where k is any positive or negative integer. in a precisely similar way. 1.lt. The values of Logs are still the values of \ + i am 2. In what precedes we have used f to denote the argument of to denote the coordinates of Log f and (f. Examples XCIII. and we shall do this in t and z (x. as real axis. a value of Log and it is that every value of log f + i am ? is clear from the preceding discussion  that every value of Log f must be of this form. H.6&. Here k is positive. and 4. to the first We shall choose the value log( z) + 7ri corresponding path. are the different  values of am f.rr.are values of am 2. By making the path wind round the origin in the opposite direction (as shown in the dotted path in Fig.i Log f = log/? (2&7T + $). . and 6 is equal to one or integration. Since  \=p.). AND CIRCULAR FUNCTIONS 401 In this case the path of integration winds twice round the If we had taken a path winding origin in the positive sense. (r. k times round the origin we should have found.gt. If this path is a = and straight line then k the function f . The value of k is determined by the path of integration chosen. &.lt.gt. : We may value of summarise our conclusions as follows is the general Log f log  f  + i am = log p + i (2/for I &. 57). that [$] = 2kjr + &.lt. 26 . In this case the straight line from 1 to z passes through 0. we obtain a similar series of values in which k is negative. is an integer.221] EXPONENTIAL.lt. We excluded the case in which z is real other of them: also log z  r= z.lt.) an arbitrary point on the path of and its coordinates. we conclude f . supposed above that ir&./&. 97) or (p. where k   convenience dictates. viz. and the different angles 2k TT f &. y).
. The equation Log logz.. (1).. The function Logz satisfies Log^^Logsi+Logf] .. . the graph of the function I (Log x) of the real variable x.... if z lies outside the triangle formed by the three points 0. 1 =log0 =i2 an(* logZi+log^^Trt. or an equation which completely 4. true : m is an integer.... except for # satisfied The functional equation the equation by Log z. The equation Log 2m = w Logs.402 2. THE GENERAL THEORY OF THE LOGARITHMIC. y.. is equal to p when x is positive and to q when x is negative. and extending to infinity through these points parallel to OX in the negative direction. The real [x functions of 3..... where is every value of the righthand side the converse is not true.... b. Draw lines graph consists of the positive halves of the halves of the lines y = (2k + 1) TT.. but completely true.g... which is one of the values of LogZiZ2 but not the principal value. we shall call a complete equation.. = z =$( I+ i N/3) = cos ITT f i sin z rfc jj TT. is not completely a value of the lefthand side.. Zl = TI (cos 61 4. y = 0.... It is however not true that ... true..... The equation is true if z lies =6... e. The equation is true 8... and lines outside the region bounded by the line joining the points z = a.. In fact Iogz 1 z 2 = 7n.. x and and imaginary parts of any value of Log z are both continuous = 0. (1/2) = Log z is that log (1/2)= 6..] 9...... in the sense that every value of either side of this equation is one of the values This follows at once by putting of the other side.i sin 61).. (2) Iog0 1 32 =log0 1 + log$2 in all circumstances.. is An equation such as (1).. except when z is real and negative.. and applying the formula of p. 7. It is also true 5. [The y = 2kir and the negative The function /(a?) of the real variable #. then Iog .. zl If.. a..... 401. in which every value of either side a value is of the other.. z2 = r% (cos 6% + i sin 2 ). defined by T/ O) =pn + (qp)I (log a?)..
exp It is certain that to any given exp f It z correspond infinitely many different values of f. to any given would not be value of f correspond infinitely many values of z. The ey exponential function. however not the case.^7&. IX we defined of the real variable y as the inverse of the function a function = log x. indicating which of them are values of log Log (1 + tV3) and which of Log log (1 +i v/3). 2?r. or in other words Thus z if f = Log This that exp f is an infinitely manyvalued function of f theorem. #0. function of f. log r*! = log and ?. The exponential function exp f is a onevalued For suppose that Zl  T! (cos 0! + i sin f. when (a) real For what values of z is (i) logz (ii) any value of Logz or (6) purely imaginary? 12. It is naturally suggested that we should define a function y of the complex variable z which is the inverse of the function Log*. Plot roughly the doubly infinite set of values of Log Log (1 + 1^/3).. + i sin 2) are both values of exp Then and so log r x + * (0! + 2??i7r) = log ra + i (0 + 2 2ri7r).1. AND CIRCULAR FUNCTIONS defined 403 The function / (a?) by and to r is equal to 11. where is and 6 the least positive angle determined by the equations cos : sin 9 : 1 : : log r : + 27r 2 : x/{(log r) + (0 + 2for) 2 }.gt. Ifz=x+iy then Log Log 2 = log .2 0j 4 2w7r = 2 + 2n7r. In Ch.1. we call z the exponential of % and write = ar.lt. where m and n are x integers. 222] 10. If any value of Log z z is equal to f. DEFINITION. EXPONENTIAL. p when #&. Thus r =r 2) and 0! 2 differ by a multiple of Hence 262 . This involves . ^2 = ^2 (cos 6. 222.221. as is proved by the following is THEOREM. value of unnatural to suppose that.lt.ft + i (e+2#r). conversely. to q when 0&.
z Then + = Log s = log r + i (6 + 2m7r). i. an equation which we have proved already ( 205) to be true for real values of fj and fa .gt. many values of which We have already defined the meaning of the symbol a^ in a considerable variety of cases. According to the definitions there given a* has at most two values.. We shall not follow this course because we shall have to give a more general definition of the meaning of the symbol e* : we shall find then that e$ represents a function with infinitely exp f is only one. Hence f = log r. [X If %is real then exp f defined in Ch. and accordingly exp (f + If 77 117) =e (cos 77 + i sin 77).2nnr r = e*. f. = ? + Wi ? = f + ^ Then exp x exp f = ef (cos ^ + z sin 77^ x e& (cos + i sin 773) = e* {cos (77 + ) i sin + 772)} = exp (?. Let 2 2 2 . .e. The exponential function therefore satisfies the functional relation /(fi + ?*)=/(?i) /(&). = + 2mir = . ). It is defined in elementary Algebra in the case in which a is real and positive and f rational. = then exp f = e*. as we have and the imaginary parts of exp (f + irj) are continuous functions of f and 77 for all values of and 77. as exp f =e* when f is real. where m is an integer. i a 77. 224. 225. + i . It is clear that both the real ?i The functional equation satisfied by exp f. X 77 2 { (77. i+f&. function of For if z = e* then logs Hence 2 = exp f. IX. The general power a*. i*rj 77 t or 77 .404 THE GENERAL THEORY OF THE LOGARITHMIC. already inferred in 222. COROLLARY. or a real and negative and f a rational fraction whose denominator is odd. The value of exp Let f = f + ^ and = exp f = r (cos + i sin 0). one of the values of Logz is f. the real exponential = f. f=e f . It might seem natural. to adopt the same notation when f is complex and to drop the notation exp f altogether. 223.
IX. 2 . (1) If a is flog a. (ifr exp {(p/q) Log a\ = &* * Cis {(p/q) + 2m7r)}. we saw then. then (^r 4. real or complex. viz. AND CIRCULAR FUNCTIONS 405 In Ch. is real: . give a general definition of a$ which applies to all values of a and f. with the one limitation that shall We now a must not be equal to DEFINITION. then one value of f Log a.lt. . must first satisfy ourselves that this definition all consistent with the previous definitions and includes them as particular positive and f real. such expressions as in one way or another. which are precisely the values of a * Hence our new definition is also consistent with 48. IX is. III. It is easy to see that if ra assumes all possible integral values then this expression assumes q and only q different values. (2) If a = e r (cos ty + i sin Log a = r v/r). and exp (flog a) = e* loea which agrees with the The definition of Ch. zero. IX we gave a new definition. as definition adopted in Ch.i + 2 WITT). Ill we extended our definitions to cover the case in which a is any real or complex number and f any rational number p/q. found in pl&. consistent with the definition given in elementary Algebra and so our new definition is so too. (3 + 20 2+3i .222225] EXPONENTIAL. where m may have any integral value. expressed by the equation which applies whenever f is real and a real and positive. attached a meaning to but we have as yet given no definitions which. is where Log a is any value of the logarithm of We cases. is The function at a* defined by the equation = exp (f Log a) a.enable us to attach any meaning to such expressions as (l+i)^. and in Ch. . Thus we have. that of Ch.
. f where L = f log a. {r) In general cf is an infinitely I many. unless 77 = 0. then f = k/(m + + m conclude that a* is infinitely n) is rational.r. real.lt.lt. Let a = o.r) [ cos {^ log o + (^ + 2w7r)} + i sin log o. On the other hand we unless %is real and rational. f^r. by a multiple of 2?r.e.lt.rr. Then f Log a = (f + 117) + {log &. [x The TT general value of G f= + irj.lt. . a^= exp (f Log a) = e L (cos M + i sin M).+ f (^ + 2m7r)}]. so that.(cos A/T + i sin ^) the notation of where r &. 17 which &.lt. =f. of a* is Thus the general value e fiog&.r+2n.and rj (ijr 2m7r). principal value.W+2mn) function.r + i ty + 2wwr)J = 3f + O/.406 226.lt. then the moduli of all the different values of a&. manyvalued is real and rational.r.= are the same. particular cases are of especial interest.r = a. by supposing m = is / principal value of a* t log 07^ Two and f e^ loga . then is &. = 97 log &. For a I _e has a different value for every value of m. where But if shall differ. = 0. a= e r or = log &. Thus the i. 17 v^ = 0. is then = l. This = 0. If a is real and positive is real.lt. THE GENERAL THEORY OF THE LOGARITHMIC. when finite We number of values. If on the other hand 77 = 0. are But any two values differ the same or differ by a multiple of unless their amplitudes This requires that 2?r.valued log&. in 225. r/^Tr.49). The principal value of a$ exp (f Log a) is obtained by giving Log a its in the general formula. if at all. and n are different integers.lt.r + f (^ 2??i7r) . =. f (^r 2wi7r) and f (^r 2w7r). and the cos i sin fi/r + is principal value of (cos^ + isin^ is a further generalisation of De Moivre s Theorem (45.r the value defined in and the principal value of c If a = l and ( the last chapter.lt. a$ has but a have already seen that.
2m7n (cos (77  2i7r) + i i. 3): also true of the principal values. so that e*= e f .gt.] What are the corresponding results as regards the equations Log a 10. is + not as a rule a value of af ? unless 9. is The principal value of e^ if is real.).226] EXPONENTIAL.r+2miO lie 6 = r} log a + ^ (^ + 2W7r) fr)2)/ . a + Zmiri) + (loga+2niri)} . sin (17 + 2i7r)}. so that e^ (cos = 0. AND CIRCULAR FUNCTIONS 1.r = l. IX. are the vertices 3. and all the points on the spiral If ?W^ ^~^. we obtain 2mjr+ 1 sin 2w7rf ) * as the general and as the principal value. The values of a .lt. Hence All the values of 2.] (1 Find all the values of + i)*. 407 Examples XOIV. Trip. m=n. 7.] 4. denoting here the positive Show that Log e^ (1 + 2wiri ) + 2wwz. i 1 + \ (l+i) l + i. is (iii) (a?f = (a?)* = a# ? For what values of (ii) (a) real purely imaginary of unit any value (6) the principal value of modulus ? e* . Log any integer. 5. + The equation a^xa^ = a^ ^ principal values. hand side. i But where k is =cos^7r4isin^7r. f Find all the values of i\ (i [By definition i*=exp Log i). and that 6. when plotted in the Argand diagram. i* are therefore real and positive. (i) f = Log a. = 0. is not completely true. is completely true but not always true of the principal values. xcni. it is The equation \ja=a~^ completely true (Ex. where m and ^ are any integers. &.lt. we obtain A// we write e for a in the general formula. 1899.) independent of a. value of the exponential defined in Ch. is e^ (cos 17 77+ isin which equal to exp ( 223). The function e^. [If as = r (cos 6 + i sin 6] we have r = e f logri. log &. In particular. 8. in general Log c has a double is infinity of values. but is true of the value of the righthand side is a value of the left[Every . but the general value of a^xa^ exp { (log viz. of an equiangular polygon inscribed in an equiangular spiral whose angle is (Math. The equation a*xb * = (ab) &.
any values of the logarithms being taken. 7n/(l + 2w7rO. The principal value. therefore. where &. And = generally. =Log according to the definition. where x is real. equal to z. we obtain e14. The general value ~ e~ (m n) of  #*+#*!.gt. and so Loge But. the principal value of at is equal to we may say (ii) that Thus if a=e then [s Loga z any value of a equal to first z._ g so that is a doubly infinitely many. where .7 + l). for which m = 0. What are the corresponding con ditions that all the values should be of unit modulus ? log \ 12. Logarithms to any base.gt. where am a denotes any value of the amplitude. according to the second definition. is always If x is real value. We may define f = Loga z in if two is different z . should both be integral. Log g (l) = (2m + l) and n are any integers. raising each side to the power i 13.0 then case in which any value is real is that in which m The =jo/(2&. or We may say (i) that f= = Loga z if e 2.().exp (x log x\ a rational fraction pj(2q + l\ or is irrational. THE GENERAL THEORY OF THE LOGARITHMIC.valued function of to this definition. viz. z is if if the principal value of with Log z. Thus Io f=Lo? . .lt. The necessary and and {rj [x sufficient conditions that all the values of af should  be real are that 2 a + am a}/?r. being real. Log a 2 according (Logz)/(Loga). z. when q gives the real value cases of reality are illustrated by the examples 15.lt. value. ltx= The only &. #* = exp (x Log x) = exp (x log x) Cis 2i7r# the real. then there is one other real given by m = q.0 then #*. ways. then there is no other But if x is of the form p{2q. is V[2 (cosh 2 (m + w) TT+COS (2 log#)}]. first factor . or if exp=z. 16. : Explain the fallacy in the following argument since e^ miri =e^ n = I where m and n are any integers. Zm7r =e~ 2n \ themselves real In what circumstances are any of the values of ? [If x&. =Loge identical or =(Log s)/(Log e). Log e l=2m7r*7(l + 2tt7n).408 11.
exp The lefthand sides of these equations are defined. virtue of the results of 227. we define the other trigono tan? = sin ? J. by the ordinary geometrical definitions adopted in elementary Trigonometry. real or complex. in of cos ? and sin ? for all values of ?. cos? cot?=^ sm cos ? . we have cos ? = J {* a + (1/0). Also. EXPONENTIAL. cosec ? cos? ?=  1 . (1 6). cot ?. sin? It is evident that cos ? sin ?. only for real values of ? The righthand sides have..?) + exp (.. AND CIRCULAR FUNCTIONS 409 The exponential values of the sine and cosine. we can deduce a number formulae. From the formula exp ( + irji) = exp of f (cos ij +i sin 77).)} (.1 i{t  (Ijt}}. and sec ? are even functions of ?. sec? = 1 . 3 cosa ?+sm r=J[{*+(VO} {*(VO! ] = l (3). ?=. ?.226228] 227.. Taking = 0. on the other We are hand. Having defined cos ? and sin metrical ratios by the equations ?. been defined for all values of ?. any of the trigonometrical 228. if exp (i?) = t. when ? is real.(2). subsidiary 77 . and and cosec ? odd functions. with the elementary definitions for real values of ?. therefore naturally led to adopt the formulae (1) as the definitions These definitions agree.?)} (la). 77. We ?+ r can moreover express the trigonometrical functions of in terms of those of ? and ? by precisely the same formulae ? . tan ?. and.. we changing the sign of cos sin ij extremely important (irj) i 77 obtain exp itj) = cos + i sin sin i i 77. cos? = sin {exp (.{exp(. sin ? = 2 .?). We Definition of sin? and cos? for all values of saw in the last section that. exp ( = cosrj (177) Hence = {exp + exp ( exp for ( TI ^ i {exp (irj) We ratios of can of course deduce expressions 77 in terms of exp (itf).
They then give expressions for the real and imaginary parts of the cosine and sine of a complex number..... We . (6).i sin sinh sinh /... functions..sin sin iq sin irj ( + 117)= sin irj + cos = cos = sin cosh rj . 19. sin i for sin and cos 2^ for cos 2.. exp (if) =t t = cos and similarly we can prove that fcos f sin sin". 1 229.... defined in this section.. = sin sin(f+j7r) = cos f . such as the formula cos 2^= cos 2 sin 2 remains true when is allowed to assume complex values. ^ sinh f for sin . The generalised hyperbolic and sinh f. (5).. 230.. f. It follows from the addition formulae that cos sin + ir}) = cos  cos cos ir) . (1). LXXXVII. LXXXVII^ 21 between the formulae for the hyperbolic and those for the identity. is . we defined cosh for real values of by the equations cosh f = J {exp f + exp ( fl}. Hence 2 cosh 2= cosh ^+ sinh 2 ^.. if we write cosh for cos cos cosh . ordinary formulae of elementary Trigonometry are corollaries of the equations (2) and so all such algebraical (6) relations hold also for the generalised trigonometrical functions ... cosh q + i cos rj... . (4).. sin(+^). It remains true therefore if we write cos i for cos f... These formulae are true for all values of and 77. . For = t we have exp (if) . : may be applied to any trigonometrical which explains the correspondence noted in Ex. if [x which hold in elementary trigonometry. sinh f = J {exp fexp (f)} . or.. In Ex.. ( etc.e.. Formulae for cos(+^).. The reader will easily verify the following : can now extend this definition to relations #= & sin#=isinh& cosh i= cos f. The interesting case that in which and 77 are real. we can agree that the equations (1) are to define cosh and sinh for values of real or complex..410 as those THE GENERAL THEORY OF THE LOGARITHMIC.. sinh i{=i sin have seen that any elementary trigonometrical formula. We i. The same process of transformation It is of course this fact ordinary trigonometrical functions.. all complex values of the variable. and cosh 2f for cos 2^. sin(+ fHsinfcosf + cosfsinf In particular cos(f + j7r) All the . in other words.
 any multiple cos  ( f irf) \ sin  ( + 777) = ^/(cos 2 + sinh 2 = V(sin 2 + sinh 2 cos   77) 77) = V{ = V{ ( (cosh 277 + cos 2)}. and equating real and imaginary parts.228230] EXPONENTIAL. 2 + cos/. 1 &. If a= 1 77 ( l) m a.cos 2)}. where sin  sinh 77 is real. the general solution of cos = ^ is =( + !) 7rilog3. (1 + 177)= cosh 2r.. is Hence either 77 = or impossible unless  1 a multiple of TT.irf) sin 2f sin 2^ cos which leads at once to the result given. cos (I + ?) cos (g^iy) + irj) cos ( .} . then ( sin (arn cos 7.lt. = mir then cosh = m is odd. . we obtain cosh 77 = a. . = 0. of TT.+ ir).gt. or and we are led back to our first case.cos 2 /  .. = sinh 2 17.)}. . If (i) 77 = then cos = a. (+* ?) = v/{cos cot (1 + z?7) cos ir. are (i) Determine the values of [For example cos for which cos f and sin real is (ii) purely imaginary.a^l. (e. . 77 so then . 411 Examples XCV. Prove that Log cos ^1 ( + =^ + { where = log J (cosh 277 + cos 2)} 1 and jB is any angle such that cos B 77 sin 5 77 cos I cosh sin  sinh for s/li (cosh 2?7 + cos 2)} a Find a similar formula 8. [For example si ". = 0. Putting (=t. + ^)} = 177) sin 2 1 5. 6. If (ii) . Solution of the equation cos cos=a. If  cos ( \iri)  = 1.lt.] is real when 77=0 or when 2. is =2/(7rarccosa. AND CIRCULAR FUNCTIONS 1. ( (cosh fy [Use 3.] tan ( + 177) = cosh 2?. which This hypothesis leads to the solution &. Log sin ( 4177).  If  cos (% + ir)} =1 1 then sin 2 = sinh 2 77. the equation . and if sin (1 + ^) =1  then cos 2 1 = sinh2 ?. then cosh  77 = ] a and we are led to the solutions For example. If a where arc cos a lies a&..] cos sin  cosh i (cosh 5.g. between and %TT.1 and that either a ^1 and m is even. v TJ I cos sinh 77 co 2) .
1 )}].... [arc cos M i log (L + V( 2 .... Then A2 &. these two signs must be different. 10.0 and a = in the 12...lt.. = then Z =  a + 1 +  a. where a is real.. L = A +A 2 M=Ai~A l ... however. W( AI&..0 We may suppose /3=^0.. cos we must take The general solutions of these equations are = 2&7rarccos where M&.gt. a  Work If /3 out the cases in which a &. Solution of cos = a +2/3. as well as those of the equations (1). Solve sin [x =a. we put cosh 2 i)=x we find that or x = (A i A 2) 2 where .0.gt...= /3 .lt.. same way. 8insinhi.. and arc cos M lies between and TT... may be deduced by merely changing the sign of i. the solutions of cosh 77 = a.gt.  Verify that the results thus obtained agree with those of Ex. since the latter equations after squaring it...0 and cosh rj=A iA 2. Show is that if a and /3 are positive then the general solution of sin =a + i where arc sin J/ possible cases. sin sinh77=/3 .... lies between and ^TT.] .gt.gt.. solutions we have only used the second To distinguish the two sets of of we observe in the first of the equations ambiguous sign in the second. 13... 2 ..... Obtain the solution in the other Solve tan =. ^l = Suppose a &... Hence the general solution required is = 2/br 11..gt.. . and the sign of sinh ?/ is the same as the Since /3 0...... The values the equations of 77 and cos thus found above include...0.412 9. THE GENERAL THEORY OF THE LOGARITHMIC. 14.gt. In this case cos cosh?7 and If (a/cosh + (/3/sinh 77)2 = 1. that the sign of sin is the same as the ambiguous sign (2).. (1). [All the roots are real. rj= log (Z + X/(Z 2  1)} . (3). where a is real. 2 i7) since the results when /3 &..1  and J/= + 1  a1 8.... Also cos^=a/(cos and since cosh 77 &. where = a.. &. /3&... (2).
1 7. is the inverse of the exponential function. That there is such a connection may also be inferred from the facts that we have ex in terms of exp i. Let us consider more particularly the equation *Lllc 2n2~ 9. i or to if i if f i moves away to infinity the line lies above the real axis and 231. 0. often assumes different forms according to the values of it . .. . and c is also real. These facts suggest the existence of some functional connection between the logarithmic and the inverse circular functions. and that the logarithm pressed the circular functions of &. is where 6 is the numerically least angle such that 16.. but ^ x (1) dx if a 0. and to if moves away along a similar line making an angle greater than TT with OX. Prove that cos and jsin tend to infinity along the real axis. 18.230. a. The connection between the logarithmic and the inverse We found in Ch..4 COS 2?rC COS (?r v/2) cosh 17). VI that the integral of a /3. and sometimes by means of inverse trigonometrical functions. then the modulus of cos 2rrz V[i (1+ cos 47rc + cos (27rV2) + cosh (27rV2) .. where /3=}=0.. sinh 77). EXPONENTIAL.lt. 19. /3. to to i if it lies below. sometimes can be expressed by means of logarithms.. are constants. trigonometrical functions. dx *M^ if = 1 arctan v7a a &. where cos 2?rc is is real. Prove that  exp exp r7)} ( + iij) \ = exp (exp cosh cosh 77) 77) cos R (cos cos (B I (sin sin ( = cos (cos + IT)}} = cos (sin  cosh (sin sinh (cos sinh 77). . Thus. for example. any straight line co if ^ moves away towards through the origin other than either half of 20.). where a.. (TT s/2)}]. If z exp (TTI). . a. rational or algebraical function $ (#. Prove that tan tends to along the straight line of Ex. /3.. Prove that exp f tends to oo if moves away towards infinity along any straight line through the origin making an angle less than TT with OX.gt. 19. AND CIRCULAR FUNCTIONS that the general solution of tan 413 Show =af ij3. 231] 15.
Obtain the remaining 232. when the .. and this does in fact differ by a constant from any inverse circular value of arc tan (x/a). where .# 2 )}.. u=i Log y = .i Log [x which is we obtain y = xi^/(\ x2 ). The standard formula connecting the logarithmic and functions is (4).. Now where k Log (x ia} = \ log (# 2 + a2 ) i (0 + 27r). arc cos x= . It is most hand side reduces to 1 T . Thus is &. where & 3). we should be led to the formula where C is a constant. ( 221) where I is an integer.lt. is real [x write which holds when a and ($a)J(x + a) is positive. Thus: equations equivalent to the first of the equations (4) and (5) by similar reasoning. where x is real. 2i Log (exp 2^v) =y + TT. easily verified by putting ^=tany. right 2^ Log /cos V + i sin ?A . ( is any integer.^^1 each of these formulae also completely true. an integer and is the numerically least angle such that 2 2 cos0 = ^/v/(^7 + a ) and sin $ = a / v/(^2 + a 2 ).1 &. f : arc sin x=i Log {ix is J(I .#2 )}. The power z is series for exp^*. . so that the equation (4) is completely true The reader should also verify the formulae (Ex. #2 )}. xcm. We saw in 212 that when real Moreover we saw in * It will be convenient 191 that the series on the righthand side now to use z instead of f as the argument of the exponential function.i Log {xi j(l .lt.414 THE GENERAL THEORY OF THE LOGARITHMIC. this equation will not be found to be actually true.(5).. If we could ia instead of a in this equation. i*J(l (5). and the question is suggested whether. Solving the equation where # = exp (iu\ with respect to y. now that we have denned the logarithm of a complex number. \cosy ismy/ ) = 1 r. Example.
Hence {/(&) .. Suppose that F when y changes to y + k. Let the sum of the series (1) it be denoted by F(z). and so {&..I\/k*i as A. and F (z)=f (y). F is a function of y such coefficient. F^ that TT Hence changes to zero with k.. Then K tends to and K _ fcos(F+/QcosF) /fcos(F+/QcosF] nr TT j/t j{~ r Of the two quotients on the righthand side the first tends to a ..#&.lt. and we shall now prove that this is the case.. (2).. Then andso and so... its real a differential differential TT... if k \ &. and are a fortiori continuous functions &. 7) that satisfies the functional (z) equation follows F F(g)F(h) = F(z + h) . It is naturally suggested that the equation (1) also remains true. (y) * = Vf^(^)J (*} I and therefore cos where have of y. The series being absolutely convergent. and so  Now f(y) where = F(iy) = 1 + (iy) + an even and (y) is ^ (i/) an odd function of y.gt. 0. 1. F is F+ K a continuous function of y..231.lt. by direct multiplication (as in Ex.lt. F + i sin F. LXXXI. Since/ (y) has and imaginary parts cos Fand sin F coefficients.. 232] EXPONENTIAL.. Now let z iy.. where y is real.. AND CIRCULAR FUNCTIONS 415 remains convergent (indeed absolutely convergent) when z is com plex...
(1) of Z* 228 follows at once that ooM1jj + jj. Prove that cos z cosh z and sin  \ &. \ and  sin z \ z. since cos Y has a differential coefficient with sin Y. F. in Ex. 1. cos 2z = 2 cos 2 z . But we have seen already that Hence where G is ay and a constant. for all values of z.lt. so that C is a multiple of But /(O) / (y) = cos y f i sin y. The power series for cos z and sin z. Derive similar identities from the equations cos 2 z + sin 2 z = 1. Thus jP (iy) = cos y + i sin y for values of y.. 27T. &. for real values of z Z* These results were proved 1. 4.2 sin 2 z.416 limit THE GENERAL THEORY OF THE LOGARITHMIC.. / (y) = cos (y + 0) + i sin (y + O).1 = 1 . Hence K/k respect to y. we have (cos F(x + iy) = F (x) F (iy) = exp # or for all values of z. 3.  Z* mwjj + gp. Y = y + C. \ Prove that Since sin if  z \ &..lt. From it the result of the last section and the equations Z* . [x when &*0. if # also is real. f(y) = tf(y} = .sin F+ i cos =1. Examples XCVI. Further f (y) ( sin Y\.lt. and real all And. y + i sin y) = exp (x + ty). 233. = 1 when y = 0.. by means 2. sinh j \z\. and the second tends to the limit has a differential coefficient with respect tends to a limit.i cos Y) = .lt. 2z=2 sin z cos z we have Prove by multiplying the two series on the righthand side equating coefficients ( 194) that 2 2n ( 195) and 3 Verify the result by means of the binomial theorem. 2 and sin z f z  .lt. LVI. . so that Y to y. Calculate cos i and sin i to two places of decimals of the power  series for cos z &.. I then cos z  \ &.
1) are valid for all values of that the expansions of cos (at+Ji) and sin (x + K) in powers of h x and h. = exp (cos 2) sin .232234] 5. . =i {cos (cos 2) cosh (sin 2) + cos (sin 2) cosh (cos 2)}. cos z cosh 2= J 2 2 in {l + (!)} cos J W7r z.. 6. and sin a cosh z..... (7= exp (cos 2) cos (sin 2).. ..} =2 2** exp (**) z.. Show 4! that 1 + + g + ... EXPONENTIAL. 8..i sin 2 sinh = cos ( 1 + z i) Hence 7... Show (Ex. l co l^ + ^.* ^= sin 2 +~ sin 22 + sin 32 [Here = exp (cos 2) and similirly (cos (sin z) + i sin (sin 2)}. 71 sin 2 2 =1(1 It is clear that the cos =(3 sin s.. log(l 2 We 3 found in 213 that + *) = *i* + ^  ..] Sum Sum 1 11. ] Expand Expand sin 2 sinh sin 2 cos sinh z. z and sin 3 2 in 22). Expand cos z cosh z in powers of [We have [exp {(1 +i) 2} + exp { cos z cosh z + i sin 2 sinh 2= cos {(1 . real or complex.... is any integer. in powers of 2.. 27 .. LVI. indeed absolutely righthand convergent. series involving sines... powers of sin 3 2 [Use the formulae . when z real H. to same method may be used expand cos z and sin n 2.. The is logarithmic series..i) z} =  (1 + 1) 0}] and similarly cos z cosh 2 . ^ = l^ + ^ z 8 . . the series cos z ~ cos 22 ~ cos 32 ".] Sum . AND CIRCULAR FUNCTIONS that 417 Show exp {(!+*&. where n 9....gt. (sin 2). 2?L and the corresponding 12. The series on the side is convergent. (1) when and numerically less than unity.iS= exp {exp ( ..sin 82).. {cos (sin 2) C .. 234...i sin (sin z)}.it)} = exp (cos 2) Hence 10. 13.. .
[x z has any complex value whose modulus is less than unity..lt. not real. may suppose that z is not plane the formula (1) has been proved already for real values real. = rif ifJo dt where It follows from (1) of 164 that Cr fm \*m\*l o J n^ 1 + BT. C cannot pass through draw a figure to illustrate the argument.lt. with the exception of the value 1. If we put z r (cos 6 + i sin 6) = + r. 1 &. and u = I & to r. and C is the straight line joining the points 1 and 1 + z in the of the complex variable u. viz. that (1) is true for all values of z such We that  z  ^ 1.lt. 1. modification of the argument of shall in fact prove rather more than this. Hence 1 f I i /m nj . as where We of z.418 THE GENERAL THEORY OF THE LOGARITHMIC. Now 1 1 + %t or   w  is never less than r w+1 the perpendicular from on to the line 7? C* &.lt. The reader is . then u will describe r G as t increases from And /du c u ~ f dt 1 J + &. It is naturally suggested that the equation (1) remains true for such complex values of z. so that  r \ &. It will be remembered that log (1 that + z) is the principal value of Log (1 + 0).. That this is true may be proved by a 213. * Since z is recommended to when produced.
lt. the line from . when r These equations hold when = .. log (1 Now + z) = log {(1 + r cos 6) + ir sin 6} + 2r cos 6 + r ) + i arc tan 2 i log (1 / \1 r sm e s + ? COS That value of the inverse tangent must be taken which lies between For. We j \ Changing z into z we obtain 235. cos0:J cos 2^ sin(9.. I and conditionally z convergent when z series is : &. 2(9 + 2r cos 6 + r ) = r cos 2 r 2 cos + arc tan (r^ ~#) i? 7 3 cos 3(9  . .. and for all values of ^. It follows from (2) that a 3 log(l+*) = *J* + J*  .lt.. AND CIRCULAR FUNCTIONS as 419 and so R m * m * oo .0&. that. See the preceding footnote. The series is in fact absolutely convergent when (Ex..234.. . A log particularly interesting case is that in which r = 1. l... 235] EXPONENTIAL. It is easy to see that they also hold when 1 ^ r ^ 0..z is the vector represented \TT and \ir.gt. except that.1 to z.... the principal value of am (1 + z) always by = 1. = r s^ n ". since 1 4.lt.. 272 . have of course shown in the course of our proof that the this however has been proved convergent already LXXX. obtain.... = 0.. = 1. we ^ &..J sin 2^ * + J cos 3^ + J sin 36&.. on equating the 3 and imaginary parts log (1 in equation (5) of 234. 4). . 4 r2 s ^ n 2 ^ +i S1 n 3^  . ^ r 1. In this case we have 6) (1+ z) = log (1 + Cis and so = J log (2 + 2 cos (9) + i arc tan sm ( * ) (// \1 4~ cos if 7r&. (5).lt. when r = 1 6 must not be equal to an even multiple of TT... . except 6 must not be equal to an odd multiple of TT. =ilog(4cos 2 0).* lies between these limits when z lies within the circle z\  Since z m = rm (cos mO real + i sin w#).7r..
%r2 sin 40 + ^r3 sin 60  .. The graph of the function represented by the sine series is shown The function is discontinuous for 6 = (2k + 1) TT.lt. = arc tan the inverse tangents lying between TT and ?r...r3 cos 30 + .lt. to the formula arc tan z=zz in 214. we obtain If z is real and numerically less than unity. . &. = arc tan ( r cos . values of 6 save odd multiples of TT (for which values the series sum of the sine series is (6 2kir) if (2/c (2k + 1) TT. [X The sums of the series. while the !)TT &. that if r sin0 + r2 cos 20r cos 6 + \r* sin 20 then r3 siri 30. =4 log (1 + 2r sin + r 2 ).. and zero if 6 is an odd multiple of TT.] then 2.. r4 sin 40  ^ cos 30 + . 2 2 [Use the formula log c= log (a +6 2ab cos C).gt. by the results of 231. = 6 .420 THE GENERAL THEORY OF THE LOGARITHMIC. already proved in a different manner Examples XCVII. we are led. and subtract. in Fig. 1. If we write iz and iz for z in (5). for other values of 0.lt. ..r&. Prove that if  l&. are easily found from the consideration that they are periodic functions of 6 with the 2 is ^ log (4 cos \6) for period 2?r.l and \Tr&.l of log (l+iz) and log(lw) in powers of z.lt. if pJ tan 0! .arc tan ^TT. . a&. .6&.lt. 2a* cos 2(7. 3. Fig. 58. Thus the sum of the cosine series all is divergent). Prove that.lt. 58..r&.b.lt. . the inverse tangent lying between series for all other values of 6. in any triangle in which 7 7 2 logc=loga a cos C PT^.\ir r sin 20 . ^n and Determine the sum of the Prove.%I A cos 40+.lt. by considering the expansions l&.
.. (3).lt... applications of the logarithmic series..l.f&. 236] 4.. .9 6. Hence which tends to the limit z/(l te )} =  . The Let z be any complex number.235..... Some exponential limit. AND CIRCULAR FUNCTIONS 421 Prove that cos 0cos sin &.lt..... . (1) we can deduce some other If t results which real. the inverse cotangent lying between pressions for the sums of the series cos sin 2  cos 20 sin \ir and % TT sin and + . 236. number small enough log (1 Then 3 + hz) = hz% (hz)* + i (hz) and so where so that &. cos ^ sin 20 cos 2 + . If in particular we suppose h = lim nft we obtain log (1 \ + ) W/ z...gt... ..\) ) W/J \ = exp z . (2).. EXPONENTIAL. .. (h. and h are and h is sufficiently we have log (1 + tz + hz) h log (1 + tz) _ as 1 . It follows that lim log(L+fe) fc^O =^ where n is a positive integer. values of From small. = arc cot (1 + cot 0+ cot 2 0). n 1/w..... z)*0 as /i^0... = log (1 + 3 cos 2 6]... / hz ~h + tz) h^O.lt.. we shall require in the next section.$ cos 20 cos 2 6 + 1 cos 3d cos 3 6 sin 20 sin 2 . 208 for real This is a generalisation of the result proved in z. find similar ex sin + 1 siri 30 sin 3  . and so / lim ( 1 w^x\ +w \ / ) n = f n**A lim exp \n log ( / 1 V + . and h a real to ensure that hz\&.
+ te) m where t.. then (exp ) = = ((cos x+ * i sin % ) exp v/ f } {(cos %+ sin ^) f ( sin % + ^ cos x) % } ex p ^ = (xj/ + i^ ) (cos % f * sin x) ex P ^ = (ir + iy ) exp (i/r + i x ) = exp $. exp ( is the same as when &.. is = Ytz exp m log (1 + **) ^ = mz(l Here both 237..&... the 236 that if t is real then ~ (1 + tz) m = mz (1 + tz) m ~\ . ( We have proved already 215) that the sum of the series is (1 + z)m = exp fm log (1 + z)}. 1 for all real values of m and all real values of z between and 1. . and we exp {m log (1 + z)}.lt.... [x of We (1 shall also .lt.422 THE GENERAL THEORY OF THE LOGARITHMIC. 3) the series is less sum .e.. i........ + tz)m and (1 + m ~ l have their tz) principal values The general form of the Binomial Theorem.. If a n is the coefficient of z 11 then m whether is n m is real or complex.. is still + z) m than unity.. always convergent if shall now prove that its principal value of (1 It follows from Hence the modulus of z (Ex. with respect to We observe t. if &.. (1 +tz) m.() = v/r (t) + ? % (t) is (t) a complex function of whose and imaginary parts $ and x (t) possess derivatives.1 .lt. (4). so that the rule for differentiating This being so we have real. LXXX. require is a formula for the differentiation m any number real real or complex. first that.
The sum of the binomial series . 1 d( Jo \m(mwhere &. But pn n and so Hence we (Ex. if (t) = (1 + tz) we have principal value. so that (cf.".. .(*) = m (m (t) still 1) . But if + 1 sin 6) then 1 + tz = VO + 2r cos + V) ^ 1 (9  &..lt. of z such that more complete discussion of the binomial account of the more difficult case in which z = 1. of Bromwich s Infinite Series. I say. real or complex. where the logarithm has all values principal value. &. as n THEOREM.gt. l). m &. we have where Rn = =r (cos # ^ftftfc (1 ~ On ~ l &.. so that n (0)_/m\ ?i! ~w 164.gt. 6) /? n *0. ..  A series.236.. . (n) (0 dt. (m  n + 1) zn (1 + m ~n tz) .lt.(m7i + l) ~(/r^i)T~ 1 r f (l^&. and therefore arrive at the following theorem. and z\&. ".lt. in virtue of the remark made at the end of .l. This formula holds if t = 0. (/&. for all values of m. xxvii. AND CIRCULAR FUNCTIONS 423 and its having any real or complex values and each side having m Hence.lt.i=i?y 1 fl)".i/ is V2 its exp {m log(l + z}}. 237] z EXPONENTIAL.r. on pp.lt.lt. Now. Rn ^Q.  rn  1 .l&. will taking be found . and therefore m(ml)..lt. 225 et seq.(mn + 163) l) (1 ".
424 THE GENERAL THEORY OF THE LOGARITHMIC. z=ir.] /(m. Suppose m real. where j 2 &. n I by direct multiplication of series. 6 1 . general result of p. z} = exp {m log (1 +z)} &.. that z] f (m z) =f (m + m . r &.1. [Of course when m is 3. [X Examples XCVIII. If z and p.J TT &.4.. 6. assuming the / (m. by an argument similar to that of if m is real and 1 z 216. Then since we obtain 2 *=exp (m log (1 + 2r cos 0+r )} Cis m arc tan 2 + 2r cos 6 + r all m Cis is im arc tan (  the inverse tangents lying between TT and \ir.lt. [These results follow at once from the equations cos 5. are real.lt. and equate the real and imaginary parts. then . 6. for all real values of m. 2 2. I J . 9 r i5 7&. that j .] /J. 2(l+r 2 ) [Take 4. and &.lt. 423. .] m = l.#&. 1 when a positive integer the series is finite.lt. if we suppose 6=%ir. m& + i sin md = (cos 6 + i sin 6} m = cos m 6).5.6. 1. 7 . We =2 2) proved (Ex. I then * _ 1.1. Prove that l if &. 4. In particular. satisfies the functional equation / (m. 3. rr then COS W10 = COSW sin w(9 = cos^ tan 5 tari 3 ^ + .3 f*4. Deduce. LXXXI.10 VI 1. z). / ( I 1 r 13. we obtain T~ (?) T* + \5 r5 ~ ) = (* + rrf H sin m arc tan r ( ) 2. 1. m= ^ in the last two formulae of Ex.lt.lt.3.7 ? d 2.5 2.lt. 2. 4. if Prove that .8 = _ V /(. Verify the formulae of Ex. and without rational then . 8. 5 . 3 . (1 + ^ tan 0) m .4 2.lt.
(a f t 6) #} c?. then the sum of the eccentric angles of the points of intersection of the ellipse and the curve differs by a multiple of 2?r from * {log/ (.] . we have and imaginary parts. AND CIRCULAR FUNCTIONS 425 MISCELLANEOUS EXAMPLES ON CHAPTER 1. according as c is positive or negative. =0 or by /{*( is equal to one of the values of the product of the roots of this equation. and equate or real sin 2#/(cos so that. 6 sin + . and /(#.lt. equation tan z=az. c log 2}.EXPONENTIAL. = (k + TT Hog cot (for + 0).lt.. a/&. #)}. Determine the number and approximate positions of the roots of the is real.p =I then &. then where m is any odd or any even integer.gt.^= ^ ^. where P is 7. y) 6. where 2.gt. where are real and c2 2 &.gt. If a cos 0+6 sin 0+c=0. where u= exp ia and 2a i Log P. Jc is any integer. already (Ex. and deduce the LXXXVIL 5. 4) that the equation has infinitely many real Now let z=x+iy. + &2 .gt. y is zero. 4. Prove that if 6 is real and sin 6 fc) sin &.f&. a) [The eccentric angles are given by f(a cos a. Show that if (#/a) 2 + (y/6) 2 = l i s the equation of an ellipse. exp /CO results of Ex./(a 3. We obtain sinh 2y/(cos 2# + cosh 2y) = cry.#). xvn. according as sin 6 is positive or Show x is real then 7ttvt/ exp {(a + ib} x] = (a + ib] exp {(a + ib) x]. LXXXVIL 3.J(a 6/&.log/ (a. 6.6 ). .. is any even or any odd that if integer. unless x 2x + cosh 2y) = ax. 2 2 2 and a is an angle whose cosine and sine are + 6 2 ) and 4. { . where & negative. X. /exp Deduce the results of Ex. denotes the terms of highest degree in the equation of any other algebraic curve. where a [We know roots. Show that the real part of e i log (l+i) is TT (+D/8 cos {%(4k+ 1) a.
U. To one value Z corresponds one of z. is real.a&. z as n^x then (1+0 (z. If If n $ (z.gt. has no complex roots.) we considered some simple examples of the geometrical relations between figures in the planes of two variables 2. but no complex Show that if a? is real then e ax cos&#=2 Find where there are ^(n + l) or ^(n + 2) terms inside the large brackets. 21 et seq. of real 11. xxi. X= (a/rr) log where )fe is any integer. but to one of z infinitely many . If x = Q then tanhy=a?/.426 THE GENERAL THEORY OF THE LOGARITHMIC. a similar series for e ax sin bx. y. or circular functions. then &=0. has an infinity roots. Z connected by a relation Transformations. Exs. roots if 8. ri)} n * exp . has no complex roots if a^O. and JT. and 6^ that suppose ?r&.0 then the real parts of all the to zero. z=f(Z).lt. 6 are the coordinates of z the relations (irY/a). z. r. Thus x=0 or j/ = 0.] The equation tanz=az + &. n) &.lt.lt. c are real. has two purely imaginary roots 10.lt. Y.0. of Z. tan z = a/z. lo = k lo g (^ + X 2 + i arc tan ) In Ch. Suppose firstly that z = exp (irZIa). where a a&.l .1. We shall now consider some cases in which the relation involves logarithmic.  The equation 9. those of Z.lt. It is or easy to see that this equation has no real root other than zero if a ^ Thus there are two purely imaginary and two such roots if a&. If y = we come back to the real roots of the equation. where a and  b are real and b is not equal If a&. that Log z has its to a strip of its plane parallel principal value logz.a &. 13. we have If x.gt. . ^ (if) is a complex function of the real variable then [Use the formulae = ^ + *X 14. 12. complex roots are numerically greater than 6/2a . 6 Xsse XI* C os y= e". If we F= (0/7r) + 2ka. but if The equation tanz=atanh cz. 22 et seq. 0&. one point to the axis OX . TT. [X This is impossible. and the righthand side numerically greater than unity. &.. where a and and of purely imaginary roots.l.lt. otherwise all the roots are real. x/a sin ( ff F/a). Ill (Exs. the lefthand side being numerically less. and ^T is confined and extending to a distance a from it on each side. and Misc. Z= (a/ IT} Log z of where a is positive.gt. exponential. r.
Misc. where is the longitude (measured from the plane = 0) and 6 the north polar distance of the point on the sphere. may Verify that the results of Ex. parallel to the axes of and TJ. taking a value of Log z other than the principal value relation 427 and conversely. and to the lines Y= YQ the orthogonal system 19. To one of the latter lines corresponds the whole of a parallel to OX. showing in particular that the whole zplane corresponds to any one of an infinite number of strips in the ^plane. (f) 77 . coordinates of a point on the sphere are and Cartesian axes OX. Show also that to the line Q corresponds the ellipse OX X=X f __ _ _l X$ 2 . 0.lt.. f \C cosh (jya)J [c sinh _L_1 =1 2 (TT JT /a)/ and that these ellipses form a confocal system and that the lines correspond to the associated system of confocal hyper Trace the variation of z as bolas.EXPONENTIAL. we must make 15. Y= Y Z XX How does Z vary as z describes the degenerate ellipse and hyperbola YQ formed by the segment between the foci of the confocal system and the remaining segments of the axis of xl 17. By we obtain a similar plane. between the zplane and another strip of breadth 2a in the lines in the Z.plane for To the circles which X and Y are constant correspond the and radii vectores in the 2plane for which r and 6 are constant. z move continually round and round the Show that to a straight line in the Zplane corresponds an equi angular spiral in the zplane. of a OY. 77. are taken on the tangent plane. The 1) on to the tangent plane at the north pole. Show that . AND CIRCULAR FUNCTIONS of this strip corresponding to one of the whole 0plane. describes the whole of a line Q or for different values of . parallel to To make Z describe the whole of the latter line circle. [The transformation z = c cosh (nZj be regarded as compounded from the transformations z2 = exp (nZJa). each parallel to the axis and of breadth 2a. 16 are in agreement with those of Ex. of length 2a. to the lines Discuss similarly the transformation z=ctdinh(nZ}a) showing that Q correspond the coaxal circles X=X {x .c coth (27rX /)} 2 +# 2 = c a cosech 2 (2vX /a\ of coaxal circles. Y the coordinates of the projection of the point are and that #+M/ = 2tan^$Cis&. Ex. 25. . 16. Discuss similarly the transformation z=ccosh (wZ/a).] ) 18. but to a circle for which r is constant corresponds only a part. 14 Ill. Y and those of Ch. The points of a unit sphere whose centre is the origin are projected from the south pole (whose coordinates are 0. The Stereographic and Mercator s Projections.
a) + J(Z b) = v/(6 ..a) exp ( x + iy\ ..... .lt. J(Z...TT and TT.. then R can vary only from exp ( TT) to exp TT..a) .J(Z.Zb\ = ba\ cos 2y.. same point 2... 0=log R\. Thus the Zplane corresponds to the ring Z is : TT to n r=exp( is TT). r^expTr.) to exp TT and 6 real values.. we obtain another map in the plane of Z. generally known as the Stereographic Projection.... F=logcot0. [x This projection gives a map of the sphere on the tangent plane.J \ has its principal value.. Discuss the transformation showing that the straight lines for which x and y are constant correspond to sets of coufocal ellipses and hyperbolas whose foci are the points Z= a and [We have *J(Z. moreover... \Za\. from  ranges through all bounded by the circles infinitely often. but this ring is covered allowed to vary only between TT and ?r. covered when R varies through all positive then r has the range exp (..Zkrr. parallels of latitude and and straight lines parallel to the axes of In this map X Y Discuss the transformation given by the equation Za showing that the straight lines for which x and y are constant correspond to two orthogonal systems of coaxal circles in the ^plane.. As all so that give the logr= 0. 20. 21..$&..428 THE GENERAL THEORY OF THE LOGARITHMIC. so that the variation of Z is restricted to a ring similar in all respects If however 9 to that within which z varies. longitude are represented by respectively.. must be regarded as having a barrier along the negative real axis which z (or Z) must not cross. The whole plane values and of 0=log# .gt.... If now we introduce a new complex variable Z= JT+ iY= ilog^z=i log \ (x + iy) so that X=&. where k is an integer. so that values of k logr=e. usually called Mercator s Projection. so that the ring is covered only once. The transformation z = Z\ we have exp (log r If z=Z\ where the imaginary power + 10) =z = exp (i log Z) = exp (i log R .. 22.. we shall suppose that &=0. as its amplitude must not transgress the limits .0). (1).exp ( x iy) will be found that \Za\ + \Zb\ = \ba\cosh2x.b) = v/(6 and it a) .. Each ring.
lt. 4 53 of \f(z) j. 2+ x/3 = 3 73. Suppose each plane to be divided up into an infinite series of rings by circles of radii Show how other. given by the pair of z equations = Z\ Z=z~\ where each power has To circles whose centre is the principal value.n z na \ where a do &.EXPONENTIAL. (z a) (z. 2 00. behave as z approaches the origin the real axis. If f(z) the curves for which /(z) the forms of the level curves ot 29. 24. or cover the whole plane. parts of 27. Trace the variation of z when Z.b) exp z (Cartesian ovals). will 40. then z moves along an equiangular spiral whose pole is the origin in its plane. starting at the point exp ?r. 2. and the real and 26. origin in one plane correspond straight lines through the origin in the other.a)l(z .] is is Level Curves. moves round the larger circle in the positive direction to the point exp TT. is real. a region bounded by inequalities of the type [The angle may reduce to a line. its 23. (straight lines). AND CIRCULAR FUNCTIONS 429 We thus obtain a correspondence between two rings. 28. Sketch the forms of the level curves of (za) (zb) (z c). 100. Z both imaginary oscillate finitely. 59. a (concentric circles).v/3 = 27. round the smaller circle in the negative direction. If l z=Z How does Zz ai where a is real. we level curves of Sketch f(z). probably find but little difficulty in arriving at a of the level curves of any given rational function . value of the power being taken. . [Some of the level curves of the latter function are drawn in the curves marked ivn corresponding to the values 10. 25. i. Show that the region of convergence of a series of the type 2  X a. . and round the remainder of the larger circle to its original position. The reader general idea of the forms but to enter into details would carry us into the general theory of functions of a complex variable. along the barrier. any equiangular spiral whose pole is the origin in its plane.lt.] Discuss the same question for Z=z a + bi .b) (z 30. call z constant the a function of the complex variable 0.] . \Z moves round and round a circle whose centre is the along ai origin (the unit circle if z has its principal value). back along the barrier. (1 f z/v/3 (coaxal circles).e. +zz )/z. Fig. where a and b are any real numbers. am z 61 an angle. and Amoves along an . to make any ring in one plane correspond to any ring in the by taking suitable values of the powers in the equations z=Z\ Z=z~ i . is &.
59. Fig. 60.430 THE GENERAL THEORY OF THE LOGARITHMIC. Fig. 61. Fig. .
87. to whether c&. Tn Fig. 431 sinz.lt. except that the origin and scale are different. sin# + sin 3&. 61 shows the level curves of [expz corresponding to the values of k given by log 05. while the sum of the second series is if 2k7r&. 63. where c is a real 1.0&. 150. It follows that the first formula holds for all values of 6 save multiples of TT (for which the series diverges). 60.7r.] . 2 60. 63. 2 83.lt.l curves of sin zc. 50. Sketch the forms of the level curves of exp zc. 62. 50.l. 520.. constant. (i) zexpz. 900.(2k+I) TT.gt.] 33. In Fig. 37. 63 we have taken c = 2.lt. 173. = 100. TOO.(2+2) r. 779. AND CIRCULAR FUNCTIONS 31.lt. Fig... 0&. =J. The level sketched in Figs.lt. 450. \TT if (2+l) ?r&.lt.. where c is a positive constant. 100. 62 we have taken c=5. Prove that if then cos 6+$ cos 30 + cos 50 + . series for all other values of 6 for and determine the sums of the are convergent. the curves ivn [Fig.lt.  05. 1 41.] correspond to 32. = 35.0&. which they [Use the equation where 2 = cos 0+r sin 6. 62. 1559. and the curves ivn correspond to &=58. 20. 71. O OO. and if 6 is a multiple of TT. I OO. 20. 60. Sketch the forms of the level curves of which represents the level curves of sin z.] Fig. = log cot 2 \6.6&. (ii) [See The curves marked ivin 4 00.lt. If c = l then the curves are the same as those of 300. Fig.9+sin 50 }. 2 00. When & is increased by IT the sum of each series simply changes its sign. are [The nature of the curves differs according as or c&.EXPONENTIAL. and the curves ivni correspond to =29. Fig.7r 34.
30 + * sin 50 . 0aor0 + aisa Prove that if multiple of 2n.. J n then 77.gt.cos a) 2 }.0 ? Prove that.. if the roots of Ax 2 + 2J3x+C=Q TTI have their imaginary parts of opposite signs. and determine the sums of the series for all other values of 6 for which they Prove that log (4 (cos cos 6 cos a + \ cos 26 cos 2a + i cos 30 cos 3a + . then dx rri /oV. [x 6 &. = sin sin0are convergent. 6... c is positive..lt.lt. What 39.0. Discuss also the cases in which a or b or both are real and negative.= .. neither a nor b is real then dx Io fa (xa)(xV) each logarithm having its = log (a) log (6) ab b= . &.432 35. and 8&. Prove that if a and ft are real. where principal value.= log (see + tan 0) 2 . then dx I the sign of is positive.. 36.ci.J(IPAC} being so chosen that the real part of . 38. is the value of the integral when /3&.lt.gt.^ cos 30 + i cos 5(9 . unless 37. THE LOGARITHMIC AND EXPONENTIAL FUNCTIONS Prove that cos if &. Verify the result when a=ci.
* We can select a particular value of am corresponding to each point Z Z Fig. of r. the real and imaginary parts of Z. B. It is evident that if z describes a closed Z describes a corresponding closed Z correspond origin. by first selecting a particular value corresponding to the initial value of Z. V) The Proof that every Equation has a Root LET be a polynomial in the values of z z. here that we assume that r does not pass through the origin. + a n with real or complex coefficients. in the argument which follows. differing by describes multiples of 2?r..APPENDIX (To CHAPTERS I III. and each of these values varies continuously as r. A. To any value of an infinity of values of am Z.valued and con tinuous function of * It is X and Y. denotes a one. which we may the 2plane and the ^plane in the 2plane. shall assume for the present that the path r does not pass through the respectively. We shall. n. and then following the continuous variation of this value as moves Z along r. Z=P(z) = a Q z + ai z n i + . use the phrase the amplitude of Z^ and the formula am Z to denote the particular value of the amplitude of Z thus selected.. Thus amZ". Fig. then We path y path r in the ^plane. We can represent call and Z by points in two planes. IV. 28 .
obtained by supposing equal to the sum of all the increments of z to describe each of these smaller contours separately in the same sense as y. on the boundary of any one of these contours. when . may therefore suppose We Fig. shall its We now prove Z Z We can divide y into a number of smaller contours by drawing parallels to the axes at a distance 81 from one another. B. but to any closed contour in the ^fplane which does not pass through the origin. winding once round any multiple the origin in the positive direction. D. when we add 6 its increment when z moves from Q to P is up the increments of am Z due to the description of the various parts of the smaller contours.434 APPENDIX I When may be Z. not merely to r. after describing r. that this y. Thus if its path is like (6) in Fig. a point at which Z=0. what we wish to prove is already established. For if z describes each of the smaller contours in turn. Then the increment of amZ. then its amplitude will have increased the origin. in the same sense. as will certainly be the case if r does not enclose it may differ from its original value by of 2rr. Fig. . As z moves from . all cancel one another. description of parts of y * itself. save the increments due to the pass through the origin . C. returns to its original position.* If there is. There parallels: no difficulty in giving a definite rule for the construction of these the most obvious course is to draw all the lines x = kdlt y = kSlt where is k is an integer positive or negative. C. D) have described the boundary of y once. Associated with any such contour there is a number which we may call the increment of am Z Z describes the contour a number independent of the a particular value of the amplitude of Z. like path (a) in Fig. then the path of z must contain inside or on at least one point at which Z=0. then so that. when z describes is not the case. once from P to Q and once from Q to P. and each part of each of the dividing parallels twice and in opposite directions. am Z varies continuously. its amplitude the same as before. is amZ it will ultimately (see Fig. since Z does not and if the increment of am Z is in this case 6. B. as in Fig. Thus PQ will have been described twice. P to Q. initial choice of returns to it that if the amplitude of is not the same when original position. or by 2?r. These remarks apply.
Now 5m .lt.. we can .0. sides are parallel to the axes and of length z S.1 . . common XQ . the one and only point situated inside every square A m How ever small d may be./&.. Hence the amplitude of Z is unchanged.gt. inside Am . and a series of contours y^ sucn that (i) A m + i lies entirely inside A OT (ii) ym lies entirely ym ) &. . %p.. and we can find a contour y 2 entirely included in a square A 2 of side S 2 and itself included in A 1} such that sunZ &. whose limit us take an infinite sequence of decreasing numbers d A S 2 is zero.gt. . Now z let us take any closed contour lying entirely inside this square.. .gt. A TO . hand corners #1 ym) and (x m + 8m y m + Sm ) are the lower lefthand and upper rightof A. parallels.. * We may. and whose (#01 yo) is 0&.fr also describes a closed But the latter contour evidently lies inside the circle whose centre is a and whose radius is p... In any case every point of the contour lies in or on the is am Z of the smaller contours. contour. At all such points where  ] &. viz.n it is clear that xlt x^. and that they have a Similarly ym and ym +8 m have a common limit yQ and .g. where a Since P(x + iy] is a con tinuous function of x and y. at all points x + iy inside the square or on its boundary..* By repeating the argument used above.. . (iii) amZ . there must be at least one such that am Z is changed when z describes This contour may be a square whose sides are parts of the auxiliary yi. 282 .lt. . z describes ym .gt. and this circle does not include the origin. that inside each square A ?n we can find a closed contour the description of which by z changes Hence P (XQ + v/ ) = 0.. take S m =5 1 /2 w*.. we can draw a square whose centre is (# yo) and whose sides are parallel to the axes. is changed when . can now be shown that For suppose that P(xQ + iy^ = a. . .gt.gt.gt. 72&. + 81. we can draw a square which includes (# yo). say yi .gt.&.gt. &. if I 435 changed when z describes y. a decreasing sequence.. xm . or may be composed of parts of these parallels and parts of the boundary of y. boundary of a square AI whose sides are parts of the auxiliary of length 8 lm parallels and We can now further subdivide yl by the help of parallels to the axes at a smaller distance 5 2 from one another. But this contradicts what was proved above. is an increasing and . &. &. e.gt. and which is such that  =/&.. If (#m&... determine a series of squares A l5 A 2 . let .lt. As describes this contour. .APPENDIX Hence. #2 + 8g limit #m+8. is changed when z describes the contour. Z=a + &. and inside this square a closed contour such that It am Z is changed when describes the contour..
that a continuous function upper and lower bounds in any such region. that when z describes any contour y in the positive sense the increment of am Z is 2&7r. that We can then show.436 All that remains is APPENDIX to z I show that we can always describes y. We leave it as an exercise to the reader to infer. for to suppose the contrary. however small and then. either inside or on the boundary of y but if y is is the origin.gt. is to admit the truth of the theorem.  we have proved what we want. find some contour such that am Z is changed when Now Z= We can choose &. We define. precisely on the lines of 102. at any stage of the argument. however. y} attains its Now is \Z\ = \P(x+ifi\ a positive and continuous function of x and y. and whose radius  R as is  large enough. at all points on y. R so that Kl where is any positive number. . and points on and inside y. similar to that used above. by an argument am(l+p) y in the positive sense. and the proof that Z=Q r. then P(^ ) = 0. is another proof. on an extension to functions of two or more variables of 102 et seq. while am zn is unchanged as z describes on the other hand is increased by 2ft7r. \ co * &. then there must be a point If w=0.0. much as in 102.x&. then z \ the last hypothesis is untenable. if y whose centre is the origin and whose radius is R.lt. The point ZQ must a circle whose centre lie We may therefore suppose that : i&. the results of There It depends. P (z) &. is the circle where p&. which is often given. Hence amZ is increased by Qmr. passes through the origin.gt. And we can prove. since therefore suppose that ZQ lies inside y. We have assumed throughout the smaller contours into which it is argument that neither nor any of the This resolved. has a root is completed. where k is the number of roots of Z= inside y. we have . y\ for all pairs of values of x and y corresponding to any point of any region in the plane of (#. / (#.5. the upper and lower bounds of a function /(#. proceeding on different lines. If m is its lower bound for 2 for which Z\ = m. multiple roots being counted multiply.lt. assumption is obviously legitimate.lt. and this must be the least value assumed by \Z\. We rnay . from the discussion which precedes and that of 43.gt. y) bounded by a closed curve.
0 then the numbers are . which have their real parts positive and negative.p^ and passes k times through the point in which this circle is intersected by  the line joining P(z ) to the origin. Show that if R is any lBj It number such that + l?.rt is equal to the increment of when 2.0.lt.gt. l 2.gt.L ^ IP ". b&.lt. they are p and p. + .0 they are p + 1 and p.gt. In particular show that all the roots of 2 5 .!&. z [Trace the variation of am + az + 6) as describes the contour formed is by a large semicircle whose centre is the origin and whose radius the part of the imaginary axis intercepted by the semicircle. Discuss Verify the results when p = l.i&. \P(*)\&.*". z describes the contour. Determine the numbers of the roots of the equation z 2P + az + 6 = where a and b are real and p odd. and rearrange P(z) according powers of .APPENDIX If I 437 to we put Z=ZQ + .gt. described by z at which P (z } + A k  * \ Hence there are k points on the = \P (z ) /zp* and so  circle /ip p=/&. Show that if a&. p.0.tpw.lt.] 4.gt.\P(*d + A k t*\ + tnft. and if 6&.. I lie Show that the number of roots of f(z) =Q which within a closed contour which does not pass through any root {log/(*)}/2. and this contradicts the hypothesis that m is the lower bound of  P (z) . k moves k times round the circle whose centre is P(zQ ) and radius A k ^ k \=p. then + .lt. and let Ak = /*. We can choose p so small that Then and \P(*)P(*o)A t {*\&.. I Let ( Ak  be the first of the coefficients which does not vanish.1 and p+ 1 if &&. R..132 7 = are in absolute value less than all the roots of zn +a n l z .lt. EXAMPLES ON APPENDIX 1.0 the particular cases in which a =0 or 6 = 0.1 (z%&.gt. \ It follows that m must be zero and that P z = 0. 3.* R* l^i &. we obtain say. &. =p.lt.+ a n =0 are in absolute value less than R. suppose that z Now Then moves round the circle whose centre is z and radius p.$pp*. and Consider similarly the equations .
Trip. (z) and the result of Ex.) = 1. ez = az z . 8 to equations of any degree. z=z3 is denned by and the parts exterior of the sides of the triangle formed by z lt 22 3 parts to the triangle of three small circles with their centres at those points. . 71.gt. and (z) 2. Cesaro s Elementares Lekrbuch der algebraischen Analysis.) 6.  &. TT. lie inside y are the same.] Show that the roots of f = the sides of the triangle (z lt z^ z3 ) at their middle points. z3 are the roots of /(=0. 1 . that when z moves along the z l to ( z = z^ = z 2 from a point near a point near straight line joining the points 22) the increment of am is z z + nearly equal to 7.] Extend If f(z) the result of Ex. (Math. according as n is odd or even.gt. Show that the equations ez = az.gt.e. /3 numbers of the roots of the which have their real parts positive and or n and n. [Use where 9.lt. 1891. A . e*=az^ where 2 1 a&./() at all points on r which 12. APPENDIX Show 22n I that if a and 1 /3 are real then the equation negative are + a 23 2n ~ f 2 = nl and w + 1.gt. are the foci of the ellipse which touches [For a proof see p. 10. 352. 1910. Show . Show that when z describes the contour the increment of z l is z equal to 8.(z) &. contour enclosing the three points 2 = 2^ z = z2 . Trip.438 5.lt.j&. negative root and one have respectively (i) (ii) one positive and (iii) one positive and two complex roots within the circle one positive root (Math. 7. and y  are two polynomials in does not pass through any root of f(z\ and then the numbers of the roots of the equations 11. glt 2&. Prove that a closed oval path which surrounds all the roots of a cubic equation f(z)=0 also surrounds those of the derived equation the equation / (z) = Q. is a contour which 7&.
difficulty and of the utmost interest and importance Let us consider some special instances.APPENDIX II (To CHAPTERS IX..lt.l) are performed on the function. when treated in their most general forms. Again. by integrating the equation between the limits and x. . problems of great in higher mathematics..r + is itself + . in ponential function we proved that the differential coefficient of the ex /%*2 exp^=l+. terms taken between the same limits.. that it does to x. taken between the limits and x. or that . equal to exp #.#^1. is sum of .e.. X) A Note on Double Limit Problems IN the course of Chapters IX and X we came on several occasions into contact with problems of a kind which invariably puzzle beginners and are indeed.. 216. In 213 we proved that where 1&. are commutative when applied to the function (. What we x t f dt proved amounted to this. not matter in what order they . that r _dt__ J or in other _ infinite series 1 I+~t~Jo the integral words that of the sum of the equal to the .t+t 2  . Another way the integrals of its of expressing this fact is to to oo and of integration from say that the operations of summation from n w i.
. and in this volume shall not prove. of the sum of a finite . And it is natural to suppose that. y*Q X*Q lim (#+#)} x**Q limy =0 y*0 .. The operation of taking the real part of z is never commutative with (2) that of multiplication by i. number of terms is equal to the sum of their differential and in 160 we stated the corresponding theorem for integrals. or in other words that ..e. or that the operations of summation from to co and of differentiation with respect to x are commu tative when applied to xn/n\..y). in certain circumstances which it should be possible to define precisely.cZ* are commutative with respect to the operation of summation of a finite number of terms.gt. f.440 that is APPENDIX II to say that the differential coefficient of the sum of the series is equal sum of the differential coefficients of its terms. xxxvn.... to the Finally exp x is we proved incidentally in the same section that the function a continuous function of x. Thus lim y*Q { lim *.. 1 we saw that the sum of a finite number of con tinuous terms is itself continuous..a. In Ex. and in 113 that the differential coefficient result. except when 2=0 for . or that the sum of the series is sum of the series is to oo equal to the sum of the limits of continuous for #=. (3) variables The x operations of proceeding to the limit zero with each of two and y may or may not be commutative when applied to a function f(x. they should be commutative also with It is natural to respect to the operation of summation of an infinite number. Thus we have proved that in certain circumstances the operations symbolised coefficients by lim. suppose so but that is all that we have a right to say at present. i x R (x + iy] = ix.. that the limit of the the terms.. : A few further instances of may Multiplication commutative and noncommutative operations help to elucidate these points. Z&. a? i. .() { lim (#+y)} = lim # = 0. and making x tend to are com In each of these cases we gave a special proof of the correctness of the We have not proved. or that the of operations of summation from mutative when applied to xn \n\. for 2x3x. any general theorem from which the truth of any one of them could be inferred im mediately. R {i x (x + iy}} y.t = 3x2x^ for all values of x. (1) by 2 and multiplication by 3 are always commutative.
We can always.gt. find z so that But they are equal generally. oo (4) The operations 2. IX) is that in which each operation is one which involves a passage to the limit. ) a?*! The preceding examples suggest that there are three possibilities with respect to the commutation of two given operations.. of a little ingenuity. such as a differentiation or the summation of an such operations are called limit operations. by the exercise z and^Z Z^ shall differ from one another. if we use the word in a more practical sense.*U ^i n j ( 1 (1)* S 1 ^00 ! n = log2. The general infinite series : question as to the circumstances in which two given limit operations are commutative is one of the most important in all mathematics. 1 x*~I may then or may Thus if x*~l through values less than lV* 1 limJ 2 x&.. x ^ lira ilm ^ = lim = lim(l)=l. very special circumstances (3) they cases ivhich occur practically. . not be commutative. but on the other hand lim Is (#n #n + 1 )l=lim{(l#) J x+l fh*lU + (#^2 + . except in : .. . lim. as meaning LL naturally or in ordinary cases. (1) the operations may always be commutative (2) they may never be commutative. viz. in the strict theoretical sense of the L word general . We may however remark that the answer to the general question is on are two limit the lines suggested by the examples above. may be commutative in most of the ordinary The really important case (as is suggested by the instances which we gave from Ch.. But to attempt to deal with questions of this character by means of general theorems would carry us far beyond the scope of this volume. i **! im^.APPENDIX but on the other hand lim (lim II 441 ^ ) = lim X +Q X = Km 1 = 1...} = liml = l. in a great majority of such cases as are likely to occur viz.~i (acf V ( x n IV* ^1 n \=\\m J log (!+#) log 2.. If L and operations then the numbers LL z arid LLz are not generally equal.
we cannot be satisfied and in the higher branches of mathematics the . a result obtained by assuming that two limitoperations are commutative is probably true it at any rate affords a valuable suggestion as to the answer to the problem under consideration. . Detailed investigations of a large number of important double limit problems will be found in Bromwich s Infinite Series. such as in the instances which occurred in Ch. in default of a further study of the general : question or a special investigation of the particular problem. detailed investigation of these questions is an absolute But for necessity. in APPENDIX II with an answer of this kind an exact science like pure mathematics. IX.442 Of course. be regarded as we gave suggested only and not proved. In practice. But an answer thus obtained must. the present the reader may be content if he realises the point of the remarks which we have just made.
.lt. define TT also continuous and steadily increasing. We write further (4) cosy = L==.\ TT so that the functions shall remain continuous for those values of y. cos (yf TT)= cosy. siny = = ( .APPENDIX (To III 158 AND CHAPTER IX) The circular functions THE reader will find it an instructive exercise to work out the theory of f x the circular functions.y &. starting from the definition (1) y=yO) = arctan^= dt . .  du fx i du fx * du = arc tan xi + arc tan xz * . (5) tan (y + 7r) = tan y. Finally we define cosy and siny. We write ^=tf(y) = tany. . and tany for all values of y other than odd multiples of ?r. there is an inverse function (2) x=x (y). outside the interval JTT). when y is . The cosine and sine are continuous for all values of y.~s . Df. u=  We find arc tan ~. If we (3) by the equation *is then this function defined for ^TT &. We have The further development of the theory depends merely on the addition formulae. by Df. is positive. These letters at the end of a line indicate that the formulae which it contains are definitions. Df. where the square root or TT. . Write and transform the equation (1) == by the substitution . thus defined cosy and siny for all values of y. sin (y + 7r)= siny. Df. the tangent except at the points where its definition fails.%TT.* The equation (1) defines a unique value of y corresponding to every real value of x.lt. and we define cosy and siny. As y is continuous and strictly increasing.
alternative theory of the circular functions is based on the theory of An account of this theory.2 and 3/1+3/2 lie in 77 (".. and the power series derived from Taylor s Theorem. = cos y must be chosen for at least one value of y2 sin(y t . in which. The formulae for differentiation of the circular functions may now be de duced in the ordinary way. will be found in ". #2=0. From (4) and (6) we deduce (cos 3/j cos (3/1+3/2)= cos 3/2. An infinite series.gt. ".i if)? but immediately extensible to all values of y l and y 2 by means of .VThittaker and Watson s Modern Analysis (Appendix A). It follows from considerations of continuity that the positive sign must be chosen in all cases. &. The corresponding formula for 1 a +y ) nrny be deduced in a similar manner.) the equations (5). for example.2.tan yi tan 3/2 . To determine The equation reduces to cos y.sin yi sin 3/2) the sign put 3/2 =0. is defined by the equation . which shows that the positive sign viz.. cos . .444 From (6) APPENDIX this III we deduce 1 . an equation proved in the first instance only when y lt y.
2) varies on the first line. been regarded as an incomplete symbol a symbol to which no independent meaning has been attached. throughout this book. those points are special for which 2=0. . systems of analytical geometry contain infinite The elements. and there detail. The line at infinity is a line in precisely the same not a limiting infinite. and no rational account of the geometrical The confusion about these matters so prevalent infinite can be based upon it. and it is often useful for purposes of illustration. ax + by + cz=Q. triads being classed together when their constituents are proportional. In others certain special points and lines are regarded as peculiarly distinguished. may be called common numbers lines (#. though one has been attached to But the infinite of geometry is an actual and certain phrases containing it. where is point or line Thus. 2=0. When (#. limiting . is no occasion to develop the matter in is The point of importance is this. and so on. is historically important. the point for which 2 = 0. in the commonly used text books of among analytical geometry. the special elements excepted. This special line is called the line at infinity the line This a is not a treatise on geometry. y. In a system of real homogeneous geometry a point is a class of triads of real numbers (#. 6. though not all. for Every point of the first line viz. It is how ever no more than an illustration. y). and there is one special line. except one. = example. the corresponding point on the second line ax + by + cz = 0. may have no point in common. The line This correlation varies so that its distance from the origin tends to infinity. have correlates in the second. the circular points at infinity. The infinite of analysis and not an actual infinite. the illustration is taken for the reality. The symbol oo has. in such a manner as to tend in the limit to the special point for which 2=0. A line is a class of points which satisfy a linear relation In some systems one a. It is possible to set up a correlation between all homogeneous and common first Cartesian geometry in which elements of the system. has a correlate on the second. and it is on the relations of other elements to these special elements that emphasis is laid. z\ not all zero. y. students arises from the fact that. A line is Cartesian geometry a point is a pair of the class of points which satisfy a linear relation ax + by + c= 0. on exactly the same footing as another. object of this brief note is to point out that these concepts are in no way dependent upon the analytical doctrine of In what real limits. the line at infinity. corresponds to the line ax + by + c 0. for it is from it that the vocabulary of the subject has been derived. . . sense in which other lines are lines. in which a and and two 6 are not both zero.APPENDIX IV The infinite in analysis and geometry SOME. There are no infinite elements. . in what may be called real homogeneous Cartesian geometry viz. c are not all zero.
AT THE UNIVERSITY PRESS .CAMBRIDGE J. B. : PRINTED BY PEACE.A.. M.
.
.
.
.
.
5m. DD3. 3/80 BERKELEY.RETURN Astronomy Mathematics Statistics/Computer Science Library 100 Evans Hall 642338 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS DUE AS STAMPED BE UNIVERSITY OF CALIFORNIA. CA 94720 . BERKELEY FORM NO.
.BERKELEY LIBRARIES ii \ J ^ \W.C. .U.