CHAPTER BASIG PROPERTIES OF NUMBERS ‘The title of this chapter expresses in a few words the mathematical knowledge required to read this book, In fact, this short chapter is simply an explanation of what is meant by the “basic properties of numbers,” all of which—addition and multiplication, subtraction and division, solutions of equations and inequalities, factoring and other algebraic manipulations—are already familiar to us. Nevertheless, this chapter is nota review. Despite the familiarity of the subject, the survey we are about to undertake will probably seem quite novel; it does not aim to present an extended review of old material, but to condense this knowledge into a few simple and obvious properties of numbers. Some may even seem too obvious to mention, but a surprising number of diverse and important facts turn out to be consequences of the ones we shall emphasize. (Of the twelve properties which we shall study in this chapter, the first nine are concerned with the fundamental operations of addition and multiplica- tion. For the moment we consider only addition: this operation is performed on a pair of numbers—the sum a ++ 4 exists for any two given numbers a and (which may possibly be the same number, of course). It might seem reason- able to regard addition as an operation which can be performed on several numbers at once, and consider the sum a; + -*- +a, of x numbers i, . . . y aq a8 a basic concept. It is more convenient, however, to consider addition of pairs of numbers only, and to define other sums in terms of sums of this type. For the sum of three numbers a, anc, this may be done in two different ways. One can first add 4 and c, obtaining # + c, and then add a to this number, obtaining a + (b + 0); or one can first add a and 6, and then add the sum a + 6 to ¢, obtaining (a + 6) + ¢. Of course, the two compound ‘sums obtained are equal, and this fact is the very first property we shall list: (PL) Ia, 5, and care any numbers, then atGta= td) te ‘The statement of this property clearly renders a separate concept of the sum. of three numbers superfluous; we simply agree that a+ d+ c denotes the number a+ (b+) = (a+ 6) +e. Addition of four numbers requires similar, though slightly more involved, considerations. The symbol a + + e+ dis defined to mean @) (atHy+o+e or 2) @tGta)ta or 3) at (bbo+d), or 4) at b+e+d), or 6) (a+) +42).4 Prologue ‘This definition is unambiguous since these numbers are all equal. Fortunately, this fact need not be listed separately, since it follows from the property Pl already listed, For example, we know from PI that Gt temat Oto, and it follows immediately that (1) and (2) are equal. The equality of (2) and (G) is a direct consequence of P1, although this may not be apparent at first sight (one must let 6 +c play the role of # in PI, and d the role of ¢). The ‘equalities (3) = (4) = (5) are also simple to prove. It is probably obvious that an appeal to Pt will also suffice to prove the equality of the 14 possible ways of summing five numbers, but it may not be so clear how we can reasonably arrange a proof that this is so without actually if these 14 sums, Such a procedure is feasible, but would soon cease to be iff we considered collections of six, seven, or more numbers; it would be totally inadequate to prove the equality of all possible sums of an arbitrary finite collection of numbers ai, .. . » ax. This fact may be taken for granted, but for those who would like to worry about the proof (and it is worth worrying about once) a reasonable approach is outlined in Problem 23, Henceforth, we shall usually make a tacit appeal to the results of this problem and write sums ai + +++ tq with a blithe disregard for the arrangement of parentheses. ‘The number 0 has one property so important that we list it next: (P2) ‘Ya is any number, then a+0=0+0=4 An important role is also played by 0 in the third property of our lis (P3)_ For every’number a, there is a number —a such that a+ (a) = (-a) +a = 0. Property P2 ought to represent a distinguishing characteristic of the number 0, and it is comforting to note that we are already in a position to prove this. Indeed, if a number x satisfies ateea for any one number 2, then x = 0 (and consequently this equation also holds for all numbers a). The proof of this assertion involves nothing more than subtracting a from both sides of the equation, in other words, adding ~a to both sides; as the following detailed proof shows, all three properties P1-P3 ‘must be used to justify this operation. Ir ate=a, then (a) + (a+x) = (—a) +a= 05 hhence ((—a) +a) +x = 0; hence O+%=0; hence x=0.Basic Properties of Numbers 5 ‘As we have just hinted, i is convenient to regard subtraction as an operation derived from addition: we consider a — # to be an abbreviation for a ++ (—#). It is then possible to find the solution of certain simple equations by a series of steps (each justified by P1, P2, or P3) similar to the ones just presented for the equation a + x = a. For example: Ir x$3=5, then («+ 3) + (—3) = 5 + (-3)s hence x+ (3+ (—3)) = 5-3 hence x+0=2; hence xa? Naturally, such elaborate solutions are of interest only until you become con vinced that they can always be supplied. In practice, its usually just a waste of time to salve an equation by indicating o explicitly the reliance on proper- ties Pl, P2, and PS (or any ofthe further properties we shall ist), Only one other property of addition remains tobe listed, When considering the sums of three numbers and ¢, only two sums were mentioned: (e+ 6) “pe and a+ (6-+0. Actually, several other arrangements are obtained it the order ofa, 8, and cis changed, ‘That these sums are all equal depends on (P4) Ta and 6 are any numbers, then atbabte ‘The statement of Pé is meant to emphasize that although the operation of addition of pairs of numbers might conceivably depend on the order of the two numbers, in fact it does not. Itis helpful to remember that not all opera tions are so well behaved. For example, subtraction does not have this prop erty: usually a ~ b x 6 — a, Im passing we might ask just when a — 6 does qual b — a, and itis amusing to discover how powerless we are if we rely only on properties PI-P4 to justify our manipulations. Algebra of the most, tlementary variety shows that a ~ 6 = # — only when a = 6. Nevertheless, itis imposible to derive this fact from properties P1-P4; it is instructive to examine the elementary algebra carefully and determine which step(s) Cannot be justified by PI-Ps. We will indeed be able to justify all steps in detail when a few more properties are listed. Oddly enough, however, the crucial property involves multiplication “The bast properties of multiplication are fortunately 20 similar to those for addition that litle comment will be needed; both the meaning and the conse Gquences should be clear. (As in elementary algebra, the product of a and & will be denoted by a8, oF simply ab) (P53) Ifa, 5, and ¢ are any numbers, then (b+) = (ade (P6) Ife is any number, then alet6 Prologue Moreover, 1 x 0. (The assertion that 1% 0 may seem a strange fact to lst, but we have to list it, because there is'no way it could possibly be proved on the basis of the other properties listed—these properties would all hold if there were only one number, namely, 0.) (P7) For every number a x 0, there is a number a~! such that (P8) Ifaand 6 are any numbers, then ab One detail which deserves emphasis is the appearance of the condition 4 ¥ Oin P7. This condition is quite necessary; since 0° 6 = 0 for all numbers there is no number 0 satisfying 0 + 0~! = 1. This restriction has an impor- tant consequence for division. Just as subtraction was defined in terms of addition, so division is defined in terms of multiplication: the symbol a/6 ‘means a * 5~!, Since 0“? is meaningless, a/0 is also meaningless—division by 0 is always undefined. Property P7 has two important consequences. If a+b = ac, it does not necessarily follow that & = ¢; for ifa = 0, then both a 6 and a"¢ are 0, no matter what 6 and c are. However, ifa ¥ 0, then & = c; this ean be deduced from P7 as follows: ir ach arcandax0, then als (a+b) =a (ads hence (a™*+a)-b = (+a) “<5 hence Tebeteg hence bee It is also a consequence of P7 that if a = 0, then either a = 0 or 6 In fact, if a:b =Oanda x0, then a1 (a-8) hence (a~*+ a) hence 1b = 0; hence 0. (It may happen that «= 0 and 4 = 0 are both true; this possibility is not excluded when we say “either a= 0 or 6 = 0"; in mathematics “or” is always used in the sense of “one or the other, or both.”) ‘This latter consequence of P7 is constantly used in the solution of equations. ‘Suppose, for example, that a number x is known to satisfy 0. (1-2 ‘Then it follows that either y — 1 = Oorx +2 = 0; hence x= lors = 2. On the basis of the eight properties listed so far it is still possible to proveBasic Properties of Numbers 7 very little, Listing the next property, which combines the operations of addi- tion and multiplication, will alter this situation drastically. (P9) Ifa, 5, and ¢ are any numbers, then abt aabtae (otice that the equation (b +o) ‘a = 6a +e ais also true, by P8.) [As an example of the usefulness of P9 we will now determine just when a — & = bai 1 then hence hence Consequently and therefore |A second use of P9 is the justification of the assertion a~ 0 = 0 which we have already made, and even used in a proof on page 6 (can you find where?) ‘This fact was not listed as one of the basic properties, ever: though no proof \was offered when it was first mentioned, With PI-P8 alone a proof was not possible, since the number 0 appears only in P2 and P3, which concern addi- tion, while the assertion in question involves multiplication. With P9 the proof is simple, though perhaps not obvious: We have a0Fa0=a-(0+0) 20; as we have already noted, this immediately implies (by adding (@-0) to both sides) that a-0 = 0. ‘A series of further consequences of P9 may help explain the somewhat mysterious rule that the product of two negative nurabers is positive. To begin ‘with, we will establish the more easily acceptable assertion that (—a) "6 = =(a" 8). To prove this, note that (-a)-b barb [(-a) +4] -b =0-6 =0. It follows immediately (by adding —(a A) t0 both sides) that (—a) *# =a" 6). Now note that, (-a)* (0) + (Ce BY) = (0) (8) + (a) = (=a) +8) (-2)-0 =o. Consequently, adding (ab) to both sides, we obtain (=a) + (=8) = ab.8 Prologue ‘The fact that the product of two negative numbers is positive is thus a conse- quence of P1-P9. In other words, if we awant PT to PO to be true the rule for the product of two negative numbers is forced upon us. ‘The various consequences of P9 examined so far, although interesting and important, do not really indicate the significance of P9; afterall we could have listed each of these properties separately. Actually, P9 is the justification for almost all algebraic manipulations. For example, although we have shown how to solve the equation (= 1) = 2) we can hardly expect to be presented with an equation in this form. We are more likely to be confronted with the equation, aoe 2 = 0. The “factorization” x4 — 3x + 2 = (x 1)(x — 2) is really a triple use of P9: = De@=-2 += 2) + (-1)-@-2) sxe tx (-2) + (<1) ext (=1)+(-2) = att al-2) + (-D) +2 txt 2. ‘A final illustration of the importance of P9 is the fact that this property is actually used every time one multiplies arabic numerals, For example, the calculation 13 x24 2 26 32 is a concise arrangement for the following equations: 13-@-10-+4) 13-2-104 13-4 26-10 + 52. 13-24 (Note that moving 26 to the left in the above calculation is the same as writing 26-10.) The multiplication 13» 4 = 52 uses P9 also: 13-4 = (1 1043)-4 S 110-4 $304 410-412 = 41041-1042 =(41)-10-42 =5:1042 = 82.Basic Properties of Numbers. 9 ‘The properties PI-P9 have descriptive names which are not essential to remember, but which are often convenient for reference. We will take this opportunity to list properties P1-P9 together and indicate the names by which they are commonly designated. (Pl) (Associative law for addition) a + (6-40) = (@ +A) be (P2) (Existence of an additive a+0=0+a=0 identity) (P3) (Existence of additive inverses) a + (=a) = (a) +a = 0. (P4) (Commutativelawforaddition) a+ = 6 +a (P5) (Associative law for multiplica- a (bc) = (a*b)*« tion) (6) (Existence of a multiplicative a1 =1-e= a; 10. idesitity) (P7) (Existence of multiplicative a a= 1, fora 0, inverses) (P38) (Commutative law for multi-, a= = 6 plication) (P9) Distributive law) abo aabtare, ‘The three basic properties of numbers which remain to be listed are con- cerned with inequalities, Although inequalities occur rarely in elementary ‘mathematics, they play a prominent role in calculus. The two notions of inequality, a < 6 (ais less than 8) and a > 6 ais greater than 6), are intimately related: a < bmeans the same as b > a (thus 1 < 3and 3 > 1 are merely two ways of writing the same assertion), The numbers a satisfying a > 0 are called positive, while those numbers a satislying a < 0 are called negative. While positivity can thus be defined in terms of <, it is possible to reverse the procedure: a < 4 can be defined to mean that § — a is positive. In fact, it is convenient to consider the collection of all positive numbers, denoted by P, as the basic concept, and state all properties in terms of P: (P10) (Trichotomy law) For every number a, one and only one of the following holds: @) 2=0, })_ a is in the collection P, (iii) —a is in the collection P. (P11) (Closure under addition) Ifa and b are in P, then a + 6 isin P. (P12) (Closure under multiplication) If @ and 6 are in P, then a+ b is in P.10° Prologue ‘These three properties should be complemented with the following defini a>6 if a-bisinP: a ay a2b if a>bora=s; asd if a 0 if and only if@ is in P. All the familiar facts about inequalities, however elementary they may seem, are consequences of P10-P12. For example, if. and § are any two numbers, then precisely one of the following holds: @ «-5=0, (ii) a — bisin the collection P, (iil) —(@ = 8) = b = ais in the collection P. Using the definitions just made, it follows that precisely one of the following holds: @ a=4 Gi) a>, Git) b> a. A slightly more interesting fact results from the following manipulations. Ma < 6,40 that # ~ ais in P, then surely (6 + 0) ~ (a +) isin P; thus, if 0. The only difficulty presented hy the proof is the unraveling of defini- tions. The symbol a < 0 means, by definition, 0 > a, which means 0 — a = =a is in P. Similarly — is in P, and consequently, by P12, (—a)(—8) = ab is in P. Thus ab > 0. ‘The fact that ab > 0 if @>0, 4 > 0 and also if ¢ <0, 6 <0 has one special consequence: a? > 0 if a 0. Thus squares of nonze:. numbers are "There sone lightly perplexing feature ofthe symbols > and <, The statements 14is3 itis? are both true, eventhough we know that < could be replaced by < in the fist, and by = in the second, This sort of thing is bound to occur when < ised with specie numbers: the tsefulnes ofthe ayiba reveled by a statement Fhe Theorems there equality hal for some values of «and , while inequality hold for other values,‘THEOREM 1 PROOF Basic Properties of Numbers 14 always positive, and in particular we have proved a result which might have seemed sufficiently elementary to be included in our list of prope >0 (since 1 = 19. "The fact that —2 > 0 if ¢ <0 is the basis of a concept which will play an extremely important role in this book. For any number a, we define the absolute value |a| of « as follows: a a20 =a, a0. Note that lol is always positive, except when a = 0. For example, we have [-3) <3, T= 7, V2 — V3) 1 + V2 V3, and [t+ V2 - V0] = V0 — V2 = 1. In general, the mos straightforward approach to any problem involving absolute values requires treating several cases sepa- rately, since absolute vales are defined by cases to begin with. This approach, may be used to prove the following very important fact about absolute values For all numbers a and 6, we have la+ 8] < lol + il. We will consider 4 cases: () 20, 2) a20, @) aS0, 620; @) 20, bso. In case (1) we also have a + 6 2 0, and the theorem is obvious; in fact, let slat b= lal t lh, so that in this case equality holds. In case (4) we have a + 6 <0, and again equality holds: b 20; 6 <0; la +b) = (a+ 6) = -a + (8) = [al + OL In case (2), when @ 2 O and & <9, we must prove that ja+bjisanb. ‘This case may therefore be divided into two subcases. If a + 6 2 0, then we ‘must prove that atbsa ie, 6 Sh Which is certainly true since 4 is negative and ~6 is positive. On the other hand, if a + 5 <0, we must prove that eae key -aSa, which is certainly true since a is positive and —a is negative.12 Prologue Finally, note that case (3) may be dispoted of with no additional work, by applying case (2) with a and 6 interchanged. Although this method of treating absolute values (separate consideration of various cases) is sometimes the only approach available, there are often simpler methods which may be used. In fact, itis possible to give a much shorter proof of Theorem 1; this proof is motivated by the observation that Vai, (Here, and throughout the book, Vx denotes the pasitive square root of x; this symbol is defined only when x > 0.) We may now observe that (a+ Bt = (0+) = at + Dab + 3 Sat + Dal + 0 lal? + 2lol ol + [alt = (ll + [ape From this we can conclude that ja + #| < lal + |b| because a? < * implies x <), provided that x and y are both non-negative; a proof ofthis Tact is left to the reader (Problem 5) One final observation may be made about the thaorem we have just proved: a close examination of either proof offered shows that la +4) = lol +16 ifa and # have the same sign (ie., are both positive or both negative), oF if fone of the two is 0, while la + Bl < Jal + [6] if a and b are of opposite signs. We will conclude this chapter with a subtle point, neglected until now, whose inclusion is required in a conscientious survey of the properties of numbers. After stating property P9, we proved that a — b= 0 ~ a implies a = b, The proof began by establishing that ati) =o +1), from which we concluded that a = 6, This result is obtained from the equation (1 +1) = 6: (1 +1) by dividing both sides by 1+ 1. Division by 0 should be avoided scrupulously, and it must therefore be admitted that the validity of the argument depends on knowing that 1 + 1 0. Problem 24 is designed to convince you that this fact cannot possibly be proved from properties P1-P9 alone! Once P10, PI1, and P12 are available, however, the proof is very simple: We have already seen that 1 > 0; itfollows that 1 + 1 > 6, and in particular 1 + 1 0. This last demonstration has perhaps only strengthened your feeling that it is absurd to bother proving such obvious facts, but an honest assessment of our present situation will help justify serious consideration of such details. InBasic Properties of Numbers 13 this chapter we have assumed that numbers are familiar objects, and that PI-P12 are merely explicit statements of obvious, well-known properties of numbers. It would be dificult, however, o justify this assumption. Although cone learns how to “work with” numbers in school, just what numbers ar remains rather vague. A great deal of this book is devoted to elucidating the concept of numbers, and by the end of the book we will have become quite well acquainted with them. But it will be necessary to work with numbers throughout the book. It is therefore reasonable to admit frankly that we do not yet thoroughly understand numbers; we may still say that, in whatever way numbers are finally defined, they should certainly have properties PL-PI2. “Most of this chapter has been an attempt to present convincing evidence that PI-P12 are indeed basie properties which we should assume in order to deduce other familiar properties of numbers. Some of the problems (which indicate the derivation of other facts about numbers from P!-P12) are offered as further evidence. It is sil a crucial question whether P1-P12 actually ‘account for all propertics of numbers. As a matter of fact, we shall soon see that they do not In the next chapter the deficiencies of properties PI-P12 will become quite clear, but the proper means for correcting these deficiencies is not so easly discovered. The crucial additional basic property of numbers which we are seeking is profound and subtle, quite unlike PI-P12. The dis- covery ofthis erucial property wil require al the work of Part II ofthis book. In the remainder of Part I we will begin to see why some additional property is required; in order to investigate this we will have to consider a litle more carefully what we mean by “numbers.” PROBLEMS 1. Prove the following (@) Wax = a for some number a 0, then x Bayt = Naty) (ii) If? = y%, then x = yor x = ~y (iv) #8 — 98 = (Nia bay 99. () at myth = Ge Me bE tag by. (wi) x8 +57 = (x+y) — ay +593). (There is a particularly easy way to do this, using (iv), and it will show you how to find a fac torization for x" + y" whenever n is odd.) 2. What is wrong with the following “proof”? Let x = . Then aoe +9 —)) =e), xty ym Qe14 Prologue 3. Prove the following: @ f= Frith exo. ae ad +e i) Op ib d 160. ora Gi) (a8)! = a9, if a, 6 £0. (To do this you must remember the defining property of (a6)~") = Sita no. Gane a fe _ad w ¢ it 6, ©) S/ Ga Bit gano. (vi) If 6, d #0, then = jifand only if ad = bc. Also determine when 4, Find all numbers + for which Gil) 5 — a8 < -2, (iv) (= 1 — 3) > 0. (When is a product of two numbers posi- tive?) &) 8 2rt2>0, (i) eet >2 (vil) xt = x + 10 > 16, (il) 28 ++ 1> 0. Gx) (= we FS) 3) > 0. &®) @— Ve VI >0. Gi) <8, (i) e+ 3< 4. city) be e+ >o (xiv) 5. Prove the following: @) Wa d,thena~¢ 0, then ac < be Ifa <4 and ¢ <0, then ac > be a> 1, then a? >a, WO Oand a < 6%, then a 0, without the additional assumption « < 2. A generalization of tis fact Occurs in Problem 2-20 97, (@) Prove that if x" = y" and » is odd, then x = y. Hint: Fist explain why’ suffices to consider only the case sy 7 > 0; then show that x - x are both imposible. () Prove that if x" =)" and nis even, then x - #8, Although the basie properties of inequalities were state! in terms ofthe collection P ofall postive numbers, and < was defined in terms of P, this procedure can be reversed, Suppose that Pi0-PI2 are replaced by (P'10) For any numbers « and 6 one, and only one, of the follow- ing holds: @ a=6 Gi) a 0. Why is it then obvious for « <0?) (b) Prove that —b signs. This manewver is valid, but we are not yet in a position to prove this (see Problem 7-9). Parts (b) and (4) of this problem provide a direct demonstration that the > signs hold (b) Using the fact that BH Dy ty = ty) 20, show that the assumption x? ++ ay +32 < O leads to contradiction, (©) Show similarly that if v and y are not both 0, then Ast + Gy + 48> 0, Bit + Say + 39> 0. **(@) Show that if « and y are not both 0, then shay bay agh byt > 0. “16. (a) Show that (+9)! = tty? only when x = O ory = 0, (+3)? = 29-49? only when x = Oory = Oorx = ~y. (b) Use Problem 15 to find out when (x +)* = x4 +4 **(0) Find out when (x +)" = xt +98 From the assumption (x+y) = x + you should be able to derive the equation 384 2xly + 2ay? + 92 = 0, if ay x 0. This implies that (x +9)* = ay tay? = aye +9). You should now be able to make a good guess as to when (x +3)" = x" +3 the proof is contained in Problem 11-41 17. (a) Suppose that 6* — 4¢ > 0. Show that the numbers 0 for all x. Hint: “Complete the square,” ie, write x° + bx +e = (x 8/2)? +? Use this fact to give another proof that if x and y are not both 9, then x? + ay +)? > 0, (@) For which numbers ais it true that xt + axy + y? > 0 whenever and y are not both 0? (c) Find the smallest possible value of x" + bx + cand of ae? + by ++ 6, for a = 0. (Use the trick in part (b).) 18. The fact that a? > 0 for all numbers a, elementary as it may seem, is nevertheless the fundamental idea upon which most important in- ‘equalities are ultimately based. The great-granddaddy of all inequalities isthe Sehwarte inequality say tony: S Vint at V5 tet The three proofs of the Schwartz inequality outliued below have only one thing in common—their reliance on the fact that a* > 0 for all a. (a) Prove the Schwartz inequality by first proving that (et +)? +328) = Gay + ay)? + (aye — a) (b) Prove that if x) = Ay; and xp = Ays for some number A, then ‘equality holds in the Schwartz inequality. Prove the same thing if J = y2 = 0. Now suppose that yr and yz ure not both 0, and that there is no number X such that x; = Ay: and az = Ays. Then. 6 < On — an) + Oy — m)* = Mit + yet) = 2NGrya sb aye) ob Gr tan). Using Problem 17, complete the proof of the Schwartz inequality. (c) Prove the Schwartz inequality by using 2ay < x* + y* (how is derived?) with first for ¢ = 1 and then for i = 2. (d) Deduce, from each of these three proofs, that equality holds only when yi or when there is a number such that x1 = Aya and x; Nye In our later work, three facts about inequalities will be erucial. Although proofs will be supplied at the appropriate point in the text, a personal assault fon these problems is infinitely move enlightening than a perusal of a com pletely worked-out proof. The statements of these propositions involve some ‘weird numbers, but their basic message is very simple: if is close enough to xi, and y is close enough to ye, then x + y will be close to xe -+ yo, and xy will beBasic Properties of Numbers. 19 close to tayo, and 1/y will be close to 1 /yp. The symbol “2” which appears in these propositions is the fifth letier of the Greek alphabet (“epsilon”), and could just as well be replaced by a less intimidating Roman letter; however, tradition has made the use of € almost sacrosanct in the contexts to which these theorems apply, 19. *20. “2. +22. Prove that if ie <0] <£ and ly — yo } 5 y-w <¥ then (x +9) = rtrd)! < ly) —@— yl &, then fart ta) + (eugat s hag) arb tay Hint: Use part (a) to give a proof by induetion on &: (©) Let s(ay, « . . a4) be some sum formed from ai, . » . . ax. Show that Slay os. yaa) arb + op Hint: Theremustbetwosums.(ay,.. . yaande"(arj1, « sax) such that lary mn) = slang ss 5 yan) Fargas «a, Suppose that we interpret “number” to mean either 0 or 1, and ++ and «to be the operations defined by the following two tables. eee Ona oe 0nstd OF ele 0 psa 1 A OT ER Check that properties P1-P9 all hold, even though 1 + 1 = 0.CHAPTER, NUMBERS OF VARIOUS SORTS In Chapter 1 we used the word “‘number” very loosely, despite our concern ‘with the basic properties of numbers. It will now be necessary to distinguish carefully various kinds of numbers. ‘The simplest numbers are the “counting numbers” 1,203 ‘The fundamental significance of this collection of numbers is emphasized by its symbol N (for natural numbers). \ brief glance at P!-P12 will show that ‘our basic properties of “numbers” do not apply to N—for example, P2 and. 3 do not make sense for N. From this point of view the system N has many deficiencies. Nevertheless, N is sufficiently important to deserve several com= ments before we consider larger collections of numbers. ‘The most basic property of N is the principle of “mathematical induction.” ‘Suppose P(x) means that the property P holds for the number x. Then the principle of mathematical induction states that P(x) is true for all natural numbers x provided that (1) Pa) is true (2) Whenever P(A) is true, Pl + 1) is true, Note that condition (2) merely asserts the truth of P(k + 1) under the assumption that P(A) is true} this suffices to ensure the truth of P(«) for all x, if condition (1) also holds. In fact, if P(1) is true, then it follows that P(2) is true (by using (2) in the special ease & = 1). Now, since P(2) is true it follows that P(3) is tue (using (2) in the special case £ = 2). It is clear that each number will eventually be reached by a series of steps of this sort, so that P(k) is true for all numbers & A favorite illustration of the reasoning behind mathematical induction envisions an infinite line of people, person number 1, person number 2, person number 3, . If each person has been instructed to tell any secret he hears to the person behind him (the one with the next largest number) and a secret is told to person number 1, then clearly every person will eventually learn the secret. If P(x) is the assertion that’person number x will learn the secret, then the instructions given (to tell all secrets Iearned to the next person) assures that condition (2) is true, and telling the secret to person number 1 makes (1) true. ‘The following example is a less facetious use of mathematical induction. ‘There is a useful and striking formula which expresses the sum of the first numbers in a simple'way: a22 Prologue n(n +1) thers te tov tna tt To prove this formula, note first that itis clearly true for n = 1. Now assume that for some integer & we have Ve be GED Then Wear HEED ERE? 2 B+? 2 ke +E +2) 2 Pho bee HT s0 the formula is also true for k + 1. By the principle of induction this proves the formula for all natural numbers », This particular example illustrates a phenomenon that frequently occurs, especially in connection with formulas like the one just proved. Although the proof by induction is often quite straightlorward, the method by which the formula was discovered remains a mystery. Problems 4 and 5 indicate how some formulas of this type may be derived The principle of mathematical induction may be formulated in an equiv- alent way without speaking of “properties” of a number, a term which is sufficiently vague to be eschewed in a mathematical discussion. A more precise formulation states that if A is any collection (or “‘set”"—a synonymous mathe- matical term) of natural numbers and () Lisin a, (2) kt isin A whenever fis in A, then d is the set of all natural numbers. It should be clear that this formulation, adequately replaces the less formal one given previously—we just consider the set 4 of natural numbers x which satisfy P(x). For example, suppose 4 is the set of natural numbers x for which itis true that (+0), 2 Tt tam ur previous proof of this formula showed that A contains 1, and that £ + 1 is in A, if kis, It follows that A is the set of all natural numbers, ic. that the formula holds for all natural numbers » ‘There is yet another rigorous formulation of the principle of mathematical induction, which looks quite different. If 4 is any collection of natural num-Numbers of Various Sorts 23 bers, itis tempting to say that A must have a smallest member. Actually, this statement can fail to be true in a rather subtle way. A particularly important set of natural numbers isthe collection A that contains no natural numbers at all, the “empty collection” or “null set,”* denoted by 9. The null set $ is a collection of natural numbers that has no smallest member—in fact, it has no ‘members at all. This is the only possible exception, however; if A is a nonnull set of natural numbers, then 4 has a least member. This “intuitively obvious” statement, known as the “well-ordering principle.” can be proved from the principle of induction as follows. Suppose that the set 4 has no least member. Let B be the set of natural numbers m such that 1, .. . , mare all not in A. Clearly 1 is in B (because if 1 were in 4, then A would have 1 as smallest member). Moreover, if f, ... , are not in 4, surely # + 1 is not in A (otherwise f + I would be the smallest member of 4), so 1, ... ,£-+ Tare all not in A. This shows that if fis in B, then & + 1 is in B, It follows that every number x is in B, ie. the numbers 1,... , mare not in A for any natural number n. Thus 4 = 0, which completes the proof. It is also possible to prove the principle of induction from the well-ordering principle (Problem 9). Either principle may be considered as a basie assump- tion about the natural numbers. ‘There is still another form of induction which should be mentioned. It sometimes happens that in order to prove P(k + 1) we must assume not only P{&), but also P(.) for all natural numbers ! < 4. In this case we rely on the “principle of complete induction”: If 4 is a set of natural numbers and () tisin a, Q) E+ Lisin Aif1,...,karein A, then is the set of all natural numbers. Although the principle of complete induction may appear much stronger than the ordinary principle of induction, it is actually a consequence of that principle. The proof of this fact is left to the reader, with a hint (Problem 10). Applications will be found in Problems 6, 16, 19, and 20. Closely related t0 proofs by induction are “recursive definitions.” For ‘example, the number n! (read “n factorial”) is defined as the product ofall the natural numbers less than or equal to n ale de2-. 0a den ‘This can be expressed more precisely as follows: a wet Q) aban myn ‘This form of the definition exhibits the relationship between n! and ( ~ 1)! * Although it may not strike you 82 collection, in the ordinary sense ofthe word, the nll set arises quite naturally in many contexts, We frequently consider the set A, consting ofall Satsying Some property often we have no guarantee that Ps atid by any number, so that A might be Brin ft often one proves that Pie always fale by sowing that 4 = 824 Prologue in an explicit way that is ideally suited for proofs by induction. Problem 21 reviews a definition already familiar to you, which may be expressed more suceinetly as a recursive definition; as this problem shows, the recursive definition is really necessary for a rigorous proof of some of the basic properties of the definition, ‘One definition which may not be familiar involves some convenient nota- tion which we will constantly be using. Instead of writing abo tae, ‘we will usually employ the Greek letter 2 (capital sigma, for “sum”) and write In oer words, $y denotes the sim of the numbers ob Fane Yiettatec tan nin +1) Notice that the letter really has nothing to do with the number denoted by But only purveyors of mathematical austerity would insist too strongly on such precision, In practice, all sorts of modilications of this symbolism are used, and no one ever considers it necesary to add any words of explanation.Numbers of Various Sorts 25 ‘The symbol for example, is an obvious way of writing arbartastastant ss tay or more precisely, at Son ‘The defcienis ofthe natural numbers which we dicovered at the begin ning ofthis chapter maybe partly remedied by etedng tis stem tthe stot integer =1,0,1,2, This set is denoted by Z (from Germa: PI-P12, only P7 fails for Z. A still larger system of numbers is obtained by taking quotients m/n of integers (with n # 0). These numbers are called rational numbers, and the set of al rational numbers is denoted by Q for “quotients”). In this system of numbers all of PI-P12 are true, [tis tempting to conclude that the “properties of numbers.” which we studied in some detail in Chapter 1, refer to just one set of numbers, namely, Q. ‘There is, however, a still larger collection of numbers to which properties P!-P12 apply—the set of all real numbers, denoted by R. The real numbers include not only the rational numbers, but fother numbers as well (the irrational numbers) which can be represented Zahl,” number). Of properties by infinite decimal; x and V2 are both example of irrational numbers. The proof that is iratonal isnot easy -we shall devote all of Chapter 16 of Part IIL toa proof ofthis fact. The irationaliy of V2, on the other hand, is ahite simple, and was known to the Greet, (Since the Pythagorean theorem ‘hows that an isosceles ight triangle, with sides of lengh I, has a hypotensee of ength V3, itis not surprising thatthe Greeks should have investigated thi auestion) The proof depends on afew observations about the natural im ters. Every natural number a can be writen either in the form 2k fr some integer fo ele in the form 2+ I for some integer k (his “obvious” fact has a simple proof by induction (Problem 7). howe natural number ofthe form 24 are caled even; thoe of the form 2 +1 are called odd. Note that even numbers have even squares, and add numbers have odd wquares (2K (2k #1) 4k = 2- (28, HEP AEH 2+ WH Im particular follows thatthe converse must algo hold: if seven, then nit evens ifs odd then is od, The proof that V2 is irrational is now quite26 Prologue simple. Suppose that V2 were rational; that is, suppose there were natural numbers pe such ht ) ‘ We can assume that p and q have no common divisor (since all common divisors could be divided out to begin with). Now we have = 2 ‘This shows that p* is even, and consequently must be ever for some natural number £. Then that is, p AR = 248, " "This shows that gis even, and consequently that q is even. Thus both p and ¢ are even, contradicting the fact that p and have no common divisor. This 2 contradiction completes the proof. Tt it important to understand precisely what this proof shows, We have demonstrated that there is no rational number x such that x* = 2. This aser- tion fs often expresed more briefly by saying that V2 is ierational. Note, however, that the use ofthe symbol V7 implies the existence of sme number ecenariy itrational) whose square is 2. We have not proved that such number exist and we ean assert confidently that, at present a proof is Inpeutle for us. Any proof at this stage would have to be based on PI-PI2 (the only properties of R swe have mentioned); since PI-PI2 are also true for Que exact same argument would show that there sa rational number whose sguare is 2, and this we know is false, (Note thatthe reverse argument will not work—our proof that there is no rational number whose square is 2 cannot be used to show that there is no real number sehose aquare is 2, because ouF proof used not only PI-PI2 but alo a special property of, the fact that every umber in Q can be written p/ for integers p and 4.) “This parculardeiiency in our list of properties of the real numbers could, ofcourse, be corrected by adding a new property which asserts the existence ot squate roots of positive numbers, Resortng t such a measure i, however, neither aesthetically pleasing nor mathematically satisfactory; we would st pot know that every Humber has an ath root ifm odd, and that every positive tuum has an mth root itn is even. Even if we astumed this, we could not prove the existence of a number » satlying x° + x1 = 0 (even though there does happen to be one), since we do not know how to write the solution of the equation in terms of nth roots (in Tac, i€ is known that the solution tannot be writen in this form). Andy of course we certainly do not wish € ‘Svume that all equations have solutions, since this fase (no real number < tatinfes + 1 = 0, for example). In fact, this retion of investigation is tot ruta one. The most useful hints about the property distinguishing R from Q. the most compelling evidence for the necessity of elucidating thisNumbers of Various Sorts, 27 property, do not come from the study of numbers alone. In order to study the [properties of the real numbers in a more profound way, we must study more than the real numbers. At this point we must begin with the foundations of calculus, in particular the fundamental concept on which calculus is based— functions PROBLEMS 1. Prove the following formulas by induction. na Qn + @ te bee ENED i) PH tte tm 2. Find a formula for @ YQ H-143454--+ + @e-1) GY Gaye este see + Qn Hine: What do these expressions have to do with I+ 24344 055 + Qnand 1223 sf Cn)? 3. 10. <£ * + + r*, multiplying this equation by r, and solving the two equations for S. 5. The formula for 1? + + + + +n? may be derived as follows. We begin with the formula (+1 A 9 BF SRL Writing this formula for = 1, .. . , n and adding, we obtain Boas +3-141 B23 HS-DHL ($1) = nt P4+3-n4$1 @ +11 = 3 +e eo +a] tnNumbers of Various Sorts 29 ‘Thuswe ean find 5 Awe already know Yk (hich could have been found in a simitar way). Use this method to find @ vtee tat ce ee ee cae ' ite el ) Sataat tT 3 5 nth win IP {Use the method of Problem 5 to show tat $A am alwys be writen in the form Fit a eae por (The first 10 such expressions are + +e tan ant thet fant tat ae He tht = pent dot tat det tale tt Van — fant + ant bet 4 gat a + Bet — doe Stow be $ dt = vont dnt = ae? Be 180 = hy bl fe nt ptt = te Notice that the coefficients in the second column are always $, and that after the third column the powers of x with nonzero coefficients decrease30 Prologue 10. a. 12, 13. 14, 15. “16, by 2 until x? or mis reached. ‘The coefcients in all but the fst two columns scem tobe rather haphazard, but there actually some srt of pattern: finding it may be rewarded as a supersperpcaciy test. See Problem 26-16 forthe complete story.) Prove that every natural number is ether even or od Prove that if set of natural numbers contains and contains + 4 whenever it contains fy then contains all natural numbers > fo Prove the principle of mathematical induction from the welLordering principle Prove te principle of complete induction from the ordinary principle of induetion, Hint: IF contains 1 and contains n+ 1 whenever it con- tains I,» - consider the set 2 ofall such that 1... Fare al int (a) I ais rational and bis irrational, is a + 6 necessarily irational? What if and bate bot irrational? (b) Ita is rational and is irrational, ib ab necesariy irrational? (Careful?) (©) Is there a number a such that os rational, but as rational? (a) Are there two irational numbers whose sum and product are both rational? (a) Prove that V3, V3, and V6 are irrational, Hint: To teat V3, for example, use the fat that every integer is ofthe form 3r or Se + 1 or 3n+ 2. Why doesn't tis proot work for V4? (b) Prove that V2 and V3 ar irrational Prove that (a) V2 + V3 is irrational. (b) V6 ~— V2 — V3 is irrational. (a) Prove that it = p+ V7 where p and q are rational, and m is a natural number, then 4" = a + § V for some rational a and d. (b) Prove also that (p — Vg)" =a — 6 Vg. (a) Prove that if m and n are natural numbers and mt/nt < 2, then (m + 2n)*/ (m + n)? > 2; show, moreover, that (m+ 20 (m +0) -2<2- (©) Prove the same results with all inequality signs reversed. (©) Prove that if m/n < V2, then there is another rational number Jn’ with m/n < mi in! < V2, 1c seems likely that Vn is irrational whenever the natural number is not the square of another natural number. Although the method of Problem 12 may actually be used to treat any particular ease, it is not“7. 18. 19. Numbers of Various Sorts 31 clear in advance that it will always work, and a proof for the general cease requires .ome extra information. A natural number p is called a prime number if it is impossible to write p = ab for natural numbers ‘and b unless one of these is p, and the other 1; for convenience we also agree that 1 is not a prime number. The first few prime numbers are 2, 3, 5,7, 11, 13, 17, 19. I'm > 1 is not a prime, then = ab, with a and 8 both < ns if either « or bis not a prime it can be factored similarly; con- tinuing in this way proves that we can write n as a product of primes. For example, 28 = 4-7 = 2+2+7. (a) Turn this argument into a rigorous proof by complete induction. ‘To be sure, any reasonable mathematician would accept the informal argument, but this is partly because it would be obvious to him how to state it rigorously.) A fundamental theorem about integers, which we will not prove here, states that this factorization is unique, except for the order of the factors. ‘Thus, for example, 28 can never be written as a product of primes one of which is 3, nor can it be written in a way that involves 2 only once (now you should appreciate why 1 is not allowed as a prime). (b) Using this fact, prove that Vn is irrational unless n = m* for some natural number m (©) Prove more generally that ‘Vn is irrational unless n = mt (a) No discussion of prime numbers should fail to allude to Euclid’s beautiful proof that there are infinitely many of them. Prove that there cannot be anly finitely many prime numbers fy, Py Pay a by considering pi=pa° + *Pa +h (a) Prove that if x satisfies Ppa tt + tay = 0, for some integers dis « - + » ay then x is irrational unless x is an integer. (Why is this a generalization of Problem 16?) (b) Prove that V2 + V/2 is irrational. Hint: Start by working out the first 6 powers of this number. Prove Bernoulli's inequality: If h > ~1, then (+A) BT tn Why is this trivial if 4 > 0? ‘The Fibonacci sequence a, a3, ax « «is defined as follows: ami, a=, ay aban forn > 3. ‘This sequence, which begins 1, 1. 2, 3, 5, 8, . . . , was discovered by Fibonacci (cirea 1175-1250), in connection with a problem about32. Prologue 20. 2. rabbits. Fibonacci assumed that an initial pair of rabbits gave birth to fone new pair of rabbits per month, and that after two months each new pair behaved similarly. The number a, of pairs born in the nth month is ay-1 + qn, because a pair of rabbits is born for each pair born the previous month, and moreover each pair born two months ago now «gives birth to another pair. The number of interesting results about this sequence is truly amazing—there is even a Fibonacci Association which publishes a journal, Tle Fibonacci Quarterly. Prove that (st ‘One way of deriving this astonishing formula is presente 23-8, ‘The result in Problem 1-6 has an important generalization: Ifa, - - - 5 aq & 0, then, Problem att Var aS (@) Why is this ue if ay = © + + = ay? Suppose not all a: are equal, say a; aj, If a, and a; are both replaced by (a; + a,)/2 what happens 10 the “arithmetic. mean” Ay = (a+ * "+ “+ an)/n? What happens to the “geometric mean” G, = Vaq>s.. *an? Why does repeating this process enough times eventually prove that G, < dy? (This is another place where itis a good exercise to provide a formal proof by induction, as well as informal reason.) ‘The reasoning in the previous proof is closely related to another interest ing proof, (b) Using the fact that G, < dy when n = 2, prove, by induction on 4, that Gy < Ay for n= 24 (©) For a general n, let 2" >», Apply part (b) to the 2 numbers Oy ees 4 tay Ang vee a An to prove that Gy S An. ‘The following is a recursive definition of a": Prove, by induction, that22, 23. Nambers of Various Sorts 33 (Don't try to be fancy: use either induction on 7 or induction on m, not both at once.) Suppose we know properties PL and P4 for the natural numbers, but that multiplication has never been mentioned. Then the following can. be used as a recursive definition of multiplication: 1b=b, @t baa bs, Prove the following (in the order suggested"): a: (6-46) = a°b +a: (use induction on «), al ab «a (you just finished proving the case 6 = 1). In this chapter we began with the natural numbers and gradually built up to the real numbers. A completely rigorous discussion of this process requires a little book in itself (see Part V). No one has ever figured out how to get to the real numbers without going through this process, but if we do accept the real numbers as given, then the natural numbers can. be difined asthe real numbers of the form 1, 1 1,1 +1 +1, etc. The whole point of this problem is to show that there is a rigorous mathe- ‘matical way of saying “ete.” (a) A set 4 of real numbers is called inductive if () Lisin a, Q) E+ Lisin A whenever & is in A. Prove that (Ris inductive. di). ‘The set of positive real numbers is inductive. (iii) ‘The set of positive real numbers unequal to $ is inductive, ‘The set of positive real numbers unequal to 5 is not inductive. If A and B are inductive, then the set C of real numbers which are in both 4 and B is also inductive. (b) A real number » will be called a natural number if n is in every inductive set. (Prove that 1 is a natural number. Gi) Prove that & + 1 is a natural number if k is a natural number.