Professional Documents
Culture Documents
PISKUNOV
differential
and
intégral
calculus
VO i n
Æ H O G E P E H U H A J Ib H O E H H H T E rP A / Ib H O E
H CM HCJlEHM fl
TOM II
DIFFERENTIAL
AND
INTEGRAL
CALCULUS
VOL. I I
M I R P U B L IS H E R S MOSCOW
First published 1964
Second printing 1966
Third printing 1969
Second édition 1974
Fourth printing 1981
H a OH3AUÜCKOM R3blKê
P r in te d in ih e U n io n o f S o v ie t S o c ia lis t R e p u b lic s
CONTENTS
3.1
Line intégrais ......................................................................................... 216
3.2
Evaluating a line in té g ra l....................................................................... 219
3.3
Green’s formula ...................................................................................... 225
3.4
Conditions for a line intégral to be independent of the path of inté
gration .................................................................................................... 227
3.5 Surface in té g r a is ...................................................................................... 232
3.6 Evaluating surface in té g ra is................................................................... 234
3.7 Stokes* formula ...................................................................................... 236
3.8 Ostrogradsky’s f o r m u la ........................................................................... 241
3.9 The Hamiltonian operatorand some ap p lic a tio n s................................ 244
Exercises on Chapter 3 247
CHAPTER 4 SERIES
4.1 Sériés. Sum of a s é r i é s .......................................................................... 253
4.2 Necessary condition for convergence of a sériés . . . . ! ! ! ! ! ! 256
4.3 Comparing sériés with positive terms ............................. ! ! ! ! ! ! 258
4.4 D’Alembert’s t e s t ...................................................................................... 260
4.5 Cauchy’s t e s t ....................................................................... ! ! . ! ! ! ! 264
4.6 The intégral test for convergence of a s é r i é s ......................................[ 266
4.7 Alternating sériés. Leibniz* th e o re m .....................................................’ 269
4.8 Plus-and-minus sériés. Absolute and conditional convergence . . . . 271
4.9 Functional s é r i é s .............................................................................. 274
4.10 Dcminated s é r i é s ........................................................ ! ! ! ! ! ! ! ! ! 275
4.11 The continuity of the sum of as é r i é s .................................... 277
4.12 Intégration and différentiation of sériés .......................... * 280
4.13 Power sériés. Interval of convergence ............................................ ’ . 283
4.14 Différentiation of power sé rié s ........................................................... 288
4.15 Seriez in powers o f \ —a ................................. . ! . ! . ! . . ! ! ! 289
4.16 Taylor*s sériés and JVu>çlaurin*ssériés ....................................... 290
4.17 Sériés expansion of f u n b tio n s ................................................ 292
4.18 Euler’s formula .................................................... ! . 1 ! ! 294
4.19 The binomial sériés . 295
Contents 7
.
6.11 Solution of the Dirichlet problem by the method of finite différences 405
Exercises on Chapter 6 ...................................................................................... 407
CHAPTER 9 MATRICES
9.1 Linear transformations.Matrix n o t a t i o n ............................................... 519
9.2 General définitions involving m a t r i c e s ................................................. 522
9.3 Inverse transformation................................................ 524
9.4 Operations on matrices. Addition of m atrices...................................... 526
9.5 Transforming a vector into another vector by means of a matrix 529
9.6 Inverse m a t r i x .............................................................................. 531
9.7 Malrix in v e r s io n ...................................................................................... 532
9.8 Matrix notation for Systems oflinear équations and solutions of
Systems of linear e a u a tio n s .................................................................... 534
9.9 Solving Systems of linear équations by thematrix m ethod................. 535
9.10 Orthogonal mappings. Orthogonal m a tr ic e s ......................................... 537
9.11 The eigenvector of a linear tran sfo rm atio n......................................... 540
9.12 The matrix of a linear transformation under which the base vectors
are eigenvectors .............................................. * .................................... 543
9.13 Transforming the matrix of a linear transformation when changing
the basis ...................................................................................................... 544
9.14 Quadratic forms and their transformation .......................................... 547
9.15 The rank of a matrix. The existence of solutions of a System of
linear équations ...................................................................................... 549
9.16 Différentiation and intégration of m a tr ic e s ......................................... 550
9.17 Matrix notation for Systems of differential équations and solutions
of Systems of differential équations with consta/it coefficients . . . 552
9.18 Matrix notation for a linear équation oforder n ................................ 557
9.19 Solving a System of linear differential équations with variable co
efficients by the method of successive approximations using matrix
n o t a t i o n ..................................................................................................... 558
Exercises on Chapter 9 ...................................................................................... 563
a p p e n d i x ......................................................................................................... 565
i n d e x ................................................................................................................. 567
CHAPTER 1
D1FFERENTIAL EQUATIONS
This relation connects the unknown function v and its dérivative we hâve
at
a differential équation in the unknown function v. This is the équation of mo
tion of certain types of parachutes. To solve a differential équation means to
find a function v = f( t) such that identically satisfies the given differential
équation. There is an infinitude of such functions. The student can easily verify
that any function of the form
( 2)
12 Ch. 1 Differential Equations
satisfies équation (1) no matter what the constant C is. Which one of ttiese
functions yields the sought-for dependence of v o n /?T o find it we take advantage
of a supplementary condition: when the body was dropped it was imparted an
initial velocity v0 (which may be zéro as a particular case); we assume this
initial velocity to be known. But then the unknown function v = f(t) must be
such that when t = 0 (when motion begins) the condition v = v 0 is fulfilled.
Substituting t = 0, v = u0 into formula (2), we find
whence
-o ,
C -t» —
From this formula it follows that for sufficiently large t the velocity v dé
pends but slightly on u0.
It will be noted that if k = 0 (the air résistance is absent or so small that
we can disregard it), then we hâve a resuit familiar from physics: *
v = v0+ g t (2")
This function satisfies the differential équation (1) and the initial condition:
v = v0 when t = 0.
Example 2. A flexible homogeneous thread is suspended at two ends. Find
the équation of the curve that it describes under its own weight (it is the same
as for any suspended ropes, wires, chains).
Solution. Let Af0 (0, b) be the lowest point
of the thread, and M an arbitrary point
(Fig. 1). Let us consider a part of the thread
M 0M. This part is in equilibrium under the
action of the résultant of tnree forces:
(1) the tension 7\ acting along the tangent
at the point M and forming an angle <p with
the x-axis;
(2) the tension H at M0 acting horizontally;
(3) the weight of the thread ys acting ver
tical 1y downwards, where s is the length of the
arc M0M and y is the linear spécifie weight of
the thread.
Breaking up the tension T into horizontal
and vertical components, we get the équations
of equilibrium:
Tcosqp = //, Tsinq) = YS
Dividing the ternis of the second équation by the cojresponding terms of
the first, we obtain
tan çp= ~t- s (3)
* Formula (2") can be obtained from (2;) by passing to the limit:
A". [('-TK-+Ï]=-+"
l.î Statement of the Problem 13
Now suppose that the équation of the sought-for curve can be written in
the form y — f(x). Here, f (x) is an unknown function that has to be found.
It will be noted that #
tanq> = / ' (■*) = •£"
Hence,
dx a (4)
where the ratio — is denoted by a.
y
Differentiate both sides of (4) with respect to x:
d2y __1_ ds
dx2 a dx (5)
But, as we know (see Sec. 6.1, Vol. I),
£=V^+(af)
Substituting this expression into équation (5), we get the differential équa
tion of the sought-for curve:
S -7 «
It expresses the relationship between the first and second dérivatives of the
unknown function y.
Without going into the methods of solving the équations, we note that any
function of the form
j, = flc o s h ( i - + C 1) + C , (7)
satisfies équation (6) for any values that Cx and C%may assume. This is évident
if we put the first and second dérivatives of the given function into (6). We
shall la ter on indicate (Sec. 1.18), without proof, that these functions (for
different Ci an<j C2) exhaust ail possible solutions of eauation (6).
The graphs of ail the functions thus obtained are called catenaries.
Let us now find out how one should choose the constants Cx and C2 so as
to obtain precisely that catenary whose lowest point M has coordinates (0, b),
Since for x = 0 the point of the catenary occupies the lowest possible position,
the tangent here is horizontal, ~ = 0 . Also, it is given that at this point the
ordinate is equal to b y = b.
From (7) we find
*',=sinh (■ j+ c *)
Putting x = 0 here, we obtain 0 = sinhC!. Hence, Ci = 0. If the ordinate of
the point Af0 is bt then y — b when * = 0 .
From équation (7) we get b = (1 + 1) + Ca, assuming x = 0 and Ci = 0,
whence C2= b — a. Finally we hâve
y = acosh
14 Ch. 1 Differential Equations
Equation (7) assumes a very simple form if we take the ordinate of M0 equal
to a. Then the équation of the catenary is
y = a cosh (x/a)
1.2 DEFINITIONS
Définition 1. A differential équation is one which connects the
independent variable x , unknown function y = f(x)t and its deri-
vatives y' , t f , . . . . yM .
Symbolically, a differential équation may be written as follows;
F{x, y, y ', y \ . . . . «/<">) = 0
or
f* y *ny \ _ a
’ dx* ’ • • • ’ dxn J
If the sought-for function y = f{x) is a function of one indepen
dent variable, then the differential équation is called ordinary.
In this chapter we shall deal only with ordinary differential
équations. *
Définition 2. The order of a differential équation is the order
of the highest dérivative which appears.
For example, the équation
y '— 2*ÿa + 5 = 0
is an équation of the first order.
The équation
y"-\-ky'—by—sinx = 0
is an équation of the second order, etc.
The équation considered in the preceding section in Example 1
is an équation of the first order, in Example 2, one of the second
order.
Définition 3. The solution or intégral of a differential équation
is any function y = f(x), which, when put into the équation,
converts it into an identity.
* ln addition to ordinary differential équations, mathematical analysis also
deals with partial differential équations. Such an équation is a relation between
an unknown function z that is dépendent upon twoorseveral variables*, y ........
these variables x, y, . . . . and the partial dérivatives of z: ^ , etc.
The following is an example of a partial differential équation in the unknown
function z(x, y):
dz dz
X Fx= y Ty
It is easy to verify that this équation is satisfied by the function z = x 2y2
(and also by a multitude of other functions).
In this course partial differential équations are discussed in Chapter 6.
1.3 First’Order Differentiai Equations 15
C> (3 )
Since only one curve of the family passes through each point
of the plane, for every number pair x and y, a unique value of
C is determined from équation (2). Putting this value of C into
(3), we find as a function of x and y. This is what yields
the differential équation that is satisfied by every function of the
family (2).
\
20 Ch. 1 Differential Equations
<r >
I
* These transformations are permissible only in a région where neither N l (y)
nor Afa (x) vanish.
22 Ch. 1 Differential Equations
which is
ln | y | = —ln | x | + ln | C | * or ln |* /| = ln
£
x
ç
whence we get the general solution: y = — .
Putting the value of C into (5) we get the desired mass of radium as a function
of time (Fig. 6): #
m = m0e - * f (6)
The constant k is determined from observations as follows. During time tQ let
a% of the original mass of radium . Hence, the following relationship is
fulfilled:
whence
—kt0= \n ^ 1 Tôô)
or
Thus, it has been determined that for radium k = 0.000436 (the unit of time
measurement is one year).
Putting this value of k into (6) we obtain
m = m0e-°'OOQA36i
Let us find the half-life of radium, which is the interval of time during which
half of the original mass of radium decays. Putting in place of m in the
latter formula, we get an équation for determining the half-life T:
= m0e “ 0-000486r
whence
—0.0004367 = — ln 2
or
$=/(*. y) (1)
/(* . y ) = f ( i > i )
. du u du u9
U+ X d x ~ T = I ? ’ Xd x ~ l- u *
Separating variables, we obtain
<1 U9 X \ U9 U J X
“ 2 ^ ~ Ul , “ 1= ,n|jC , + , n | C | OT - ^ ? = l n \ uxC \-
- £ = l n |C ÿ|
fact that the ratio of two homogeneous functions of the same de-
gree is a homogeneous function of degree zéro.
ExampJe 5. The équations
(2* + 3y) dx+ (x —2y) dy = 0
(x2+ y2) dx — 2xy dy = 0
are homogeneous.
i .e.,aby = ayb. But in this case ^f- = ~ = %, that is, üy = %a, by = Kb,
and, hence, équation (1) may be transformed to
dy (a x + b y ) + c
dx X{ax-\-by)-\-Cy
1.6 Equations Reducible to Homogeneous Equations 27
Then by substitution
z = ax-\-by • ( 6)
arctan u == In \C xx Y 1+ u2 |
or
Cxx Y ï + t f = eaTci*nu
P u ttin g — in place of u, we obtain
*i
arctan —
c V x \+ y \= e
Passing to the variables x and y , we finally get
.___________________ arctan u-
-—-1
C /( * - 2 ) * + (y -l)» = e
Example 2. The équation
2x+ y— \
y' = 4 x + 2 y + 5
Since z = 2 x + y , we obtain the final solution of the initial équation in the form
- H - ( 2 * + y ) + ^ ln | 1 0 * + 5 iH -9 |= x + C
or
lOy— 5jc+7 ln | 1 0 * + 5 y + 9 | = CX
that is, as an implicit function y of x.
1.7 First-Order Linear Equations 29
u { f x + P v ) + v Tx =C^ <3)
Let us choose the function v such that
X + p° - 0 (4)
Separating the variables in this difîerential équation with respect
to the function v, we find
^V ----- Pdx
Integrating we obtain
—In| Cx1+ ln 11>| = — J Pdx
or
- f Pdx
v = Cje J
Since it is sufficient for us to hâve some nonzero solution of
équation (4), we take, for the function v(x),
- f Pdx
v(x) = e J (5)
where J Pdx is some antiderivative. Obviously, o (x )^ 0 .
30 Ch. î Differential Equations
Putting the value of v(x) which we hâve found into (3), we get
^notingthat ^ + P i> = o j:
or
du _ Q (x)
dx v (x)
whence
5
“- U i d x+ c
Substituting into formula (2), we finally get
or
y = v (x) § T $ d x+ c °(x) (6)
The C in the first term cancel out; in the second term the product
CC is an arbitrary constant, which we shall dénoté by C, and we
again arrive at expression (6). If we dénoté dx = qp(x), then
expression (6) will take the form
y = v (x)y (*) + Co(x) (6')
It is obvious that this is a complété intégral, since C may be
chosen in such a manner that the initial condition will be satisfied:
when x = xu, y = y0
The value of C is determined from the équation
1/q= ï '( * o)<P(* o) + C ü (*„)
^ ~ F f T ^/=(JC+1),
1.7 First-Or der Linear Equations 31
Solution. Putting
y — uv
we hâve
dy dv . du
dx dx 1 dx
Thus, the complété intégral of the given équation will be o! the form
_ (* + l)4 ~ C (x + 1)2
y=
The family obtaihed is the general solution. No matter what the initial con
dition (x0, y0), where x0 ^ — 1, it is alway6 possible to choose C so that the
corresponding particular solution should satisfy the given initial condition. For
example, thehe particular solution that satisfies the condition y0 = 3 when x0= 0
is found
' * as folio
' Mows:
(0 + 1 )4 - C ( 0 + l) a; C=y
s-1 — +1> r ‘ l
* This équation results from the problem of the motion of a body provided
the résistance of the medium F dépends on the velocity: F — X1v + \ 2vn.
The équation of motion will then assume the form m or
dv . Xi X.
1.8 Bernoulli*s Equation 33
u { ê - 2xv) + v Tx= - 2*
Equate to zéro the expression in the brackets:
dv A A dv .
------2xv = 0, — = 2 x d x
dx v
l n |o | = jt2, v = ex*
For u we get the équation
rî du n«
ex ^ - = — 2x*
dx
Separating variables, we hâve
du = — 2e~x*x3 dx, « = — 2 J e~x*x? d x+ C
Integrating by parts, we find
u==x2e~x* + +
z = uv = x2-t-l-i-Cex*
8 2082
34 Ch. / Differential Equations
y ~ 2 = x2-{-1 -{-Ce*2 or y=
V x 2+1 + Ce*2
Note. Just as was done for linear équations, it may be shown that
the solution of the Bernoulli équation may be sought in the
form of a product of two functions:
y = u(x)v (*)
where u(jc) is some nonzero function that satisfies the équation
v' + Pv —0.
that is, (2) is a necessaiy condition for the left side of (1) to be
an exact differential of some function u (x, y). We shall show that
this condition is also sufficient: if (2) is fulfilled then the left
side of (1) is an exact differential of some function u(x, y).
From the relation
we find
X
u = [ M (x , y)dx + <p(y)
*o
where x0 is the abscissa of any point of the domain of existence
of the solution.
When intégrâting with respect to x we consider y constant, and
therefore the arbitrary constant of intégration may be dépendent
on y. Let us choose a function <p(y) so that the second of the
relations (4) is fulfilled. To do this, we difïerentiate* both sides
of the latter équation with respect to y and eauate the resuit to
N (x%y):
| J = l i ü d x + v ' ( y y ^ N (x ' y )
. . . dM dN
but since = , we can write
X
or
N (x, y) — N (x0, y) + q>' (y) = N (x, y)
Hence,
<p' (y) = N (x0, y)*3
3*
36 Ch. 1 Differential Equations
or
qp(ÿ)= J N (x0, y)dy + C1
yo
Thus, the function u(x, y) will hâve the form
* y
u= (.v, y)dx+ 5 N (xt , y)dy + Ct
x. y»
Here P (x0, ya) is a point in the neighbourhood of which there
is a solution of the differential équation (1).
Equating this expression to an arbitrary constant C, we get the
complété intégral of équation (1):
x y
l M (x, y)dx + J N (x0, y)dy = C (5)
xo yo
Example. Given ihe équation
2x , . y2 — 3x2 .
y3 X 1 y4 y ~
Let us check to see whether this is an exact differential équation.
Denoting
yJ y4
we hâve
dM 6x M 6*
ày ' 'y * ' d x z ’P
For y £ 0, condition (2) is fulfilled. Hence, the left side of this équation is
an exact differential of some unknown function u (x, y). Let us find this function.
Since ~ = , it follows that
dx y3
« = J •p -< te + « p (ÿ > = p -+ ç (y )
where <p (y) is an as yet undetermined function of y.
Differentiating this relation with respect to y and noting that
~3x2
ày
we find
3x* , , . . y2 — 3x2
— r r f q » (y)=-
y4
hence
9 (y) —— - 4 -
•2
“ (*• y ) = ÿ “ 7 + Cl
Thus the complété intégral of the original équation is
x2 1
1.10 Integrating Factor 37
( 7 + x ) d x~ — dy = Q
J + T +c =0
or
2x
x2+ 2C
1.11 The Envelope of a Family of Curves 39
-H
Fig. 7 Fig. 8
given (x , y): C = C(x, y). Thus, for ali points of the envelope the
following équation holds:
<D[x, y, C (x1, y)\ = 0 (2)
Suppose that C(x, y) is a différentiable function that is not con
stant in any interval of the values of x and y under considération.
From équation (2) of the envelope we find the slope of the tan
gent to the envelope at the point M (x, y). Differentiate (2) with
respect to x assuming that y is a function of x:
, d® a c , p<D . cXD <3C] , _ n
dx + dC dx • [ d y dC dy \ y ~ U
or
v * + v j+ ® h [ § + § y '] = o (3)
The slope of the tangent to the curve of the family (1.) at the
point M (x, y) is found from
—0 (4)
(on this curve, C is constant).
We assume that «D^^O, otherwise we would consider x as the
function and y as the argument. Since the slope k of the envelope
is equal to the slope k of the curve of the family, from (3) and
(4) we obtain
*[§+£']-•
But since on the envelope C(x, y)=^= const, it follows that
dC . dC , , n
Since for any points the equalities <J>i = 0, <I>^ = 0 are fulfilled,
it follows that for them the equality G>c = 0 is also satisfjed.
We hâve thus proved that the coordinates of singular points
satisfy équations (6).
Summarizing, équations (6) define either the envelope or the
locus of singular points of the curves of the family (1), or a
combination of both. Thus, after obtaining a curve that satisfiés
équations (6), one has further to find out whether it is an envelope
or the locus of singular points.
Example 2. Find the envelope of the family of circles
( * - C ) 2 + £/2- / ? 2 = 0
that are dépendent on the single parameter C.
Solution. Differentiating the équation of the family with respect to C,
we get
2 (x — C) = 0
Eliminating C from these two équations, we obtain the équation
y2 — R2 = 0 or y = ± R
It is clear, by géométrie reasoning, that the pair of straight Unes is the
envelope (and not the locus of singular points, since the circles of a family
do not hâve singular points).
Example 3. Find the e;ivelope of the family of straight lines
xcos a -t y sin a —p = 0 (a)
where a is a parameter.
Solution. Differentiating the given équation of the family with respect
tu oc, we hâve
—x sin a + y cos a = 0 (b)
To eliminate the parameter a from équations (a) and (b), multiply the
terms of the fîrst by cos a, and of the second, by sin a, and then subtract
the second from the first; we then hâve
x = p cos a
Putting this expression into (b), we find
y — P sin a
42 Ch. 1 Differential Equations
Squaring the terms of the last two équations and adding termwise, we get
x2 + y 2 = p2
This is a circle. It is the envelope of the family (and not the lccus of singular
points, since straight lines do not hâve singular points) (Fig. 9).
Example 4. Find the envelope of the trajectories of projectiles fired from
a gun with velocity u0 at different angles of inclination of the barrel to the
horizon. We shall consider that the gun is located at the coordinate origin and
that the trajectories of projectiles lie in the jq/-plane (air résistance is disre-
garded).
Solution. First find the équation of the trajectory for the case'when the barrel
makes an angle a with the positive x-axis. In flight, the projectile participâtes
simultaneously in two motions: a uniform motion with velocity v0 in the direc
tion of the barrel and a falling motion due to the force of gravity. Therefore,
at each instant of time / the position of the projectile M (Fig. 10) will be
defined by the équations
x = v0t cos a
at2
y = v0t sin a — — ~
These are parametric équations of the trajectory (the parameter is the time t).
Eliminating t , we get the équation of the trajectory in the form
gxi
y = * ta n a —
2v\ cos2 a
a
Finally, introducing the notation tan a = k t = a, we get
This is the équation of a parabola with vertex at the point (^0, , the axis
of which coïncides with the y-axis. It is not £ locus of singular points [since
parabolas (8) do not hâve singular points]. Thus, the parabola
Solution. DifTerentiate the given équation of the family with respect to the
parameter C:
2 (x —C) = 0
Eliminating the parameter C from the two équations, we get
y=o
The x-axis is a locus of singular points —a cusp of the first kind (Fig.. 12).
Indeed, let us find the singular points of the curve
y3— (x—C)2 = 0
for a fixed value of C. Differentiating with respect to x and y, we find
F; = - 2 ( x - C ) = 0
Fy — 3y2 = 0
Solving the three foregoing équations simultaneously, we find the coordinates
of the singular point: x = C, y = 0; thus, each curve of the given family has
a singular point on the x-axis.
For continuous variation of the parameter C, the singular points will fill the
entire x-axis.
44 Ch. 1 Differential Equations
Example 6. Find the envelope and locus of singular points of the family
(y ~ C )* -± (x -C )* = 0 ( 10)
—2 (y _ C ) + y 3 < * - C ) a = 0
or
y - C — (x-~C)*=: 0 00
or
y —x — (jc-x )2= 0
*-*+i-[*-*+ r=° 4
* “ •*— T
1.12 Singular Solutions 45
2
We hâve obtained two straight Unes: y — x and y - x — -^-. The first is the locus
of singular points, the second is the envelope (Fig. 13).
Note 2. In Sec. 6.7, Vol.I it was prtfved that the normal to a
curve serves as a tangent to its evolute. Hence, the family of
normals to a given curve is at the same time a family of tangents
to its evolute. Thus, the evolute of the curve is the envelope of the
family of normals of this curve (Fig. 14).
This remark enables us to point out another method for finding
evolutes: to obtain the équation of an evolute, first find the family
of ail normals of the given curve and then find the envelope of
this family.
1.12 SINGULAR SOLUTIONS OF A FIRST-ORDER
DIFFERENTIAL EQUATION
Let the differential équation
± V R 1- y 2
Whence, integrating, we find the complété intégral:
( * - C ) 2 + (/2 = fl2
y = x%+^{%) (h
It is integrated by introducing an auxiliary parameter. P u t-^ = p;
then équation (1) will take the form
y = x p + * (p ) (V)
Differentiate ail the terms of this équation with respect to x,
bearing in mind that p = is a function of x :
P = X% + P + * ' W %
or
[ x + Ÿ (p)]d£ ~ o
Equating each factor to zéro, we get
and
* + ♦'(/>) = 0 (3)
(1) Integrating (2) we obtain p = C(C — const). Putting this
value of p into (T), we find its complété intégral:
y = xC + y{C) (4)
which, geometrically, is a family of straight lines.
(2) If from (3) we find p as a function of x and put it into
(T), we obtain the function
t/ = x p (x )+ i)> [/?(*)] (1")
The solution (1") is not obtained from the complété intégral (4)
for any value of C. This is a singular solution; it is obtained by
élimination of the parameter p from the équations
y = xp + Mp(p) ^
x + i|>'(/>) = 0 j
or, which is the same thing, by eliminating C from the équations
y = xC 4- ib (C)
* + tc (Q = 0
Thus, the singular solution of Clairaut’s équation defines the
envelope of a family of straight lines represented by the complété
intégral (4).
Example. Find the general and singular solutions of the équation
dy
a
dy dx
y=x
dx 2
/■+(=)
4a Ch. 1 Differential Equations
(1 + C2) 2
The singular solution (the équation of the envelope) is obtained in parametric
form (where the parameter is C):
(1 + C 2) 2
aC3
3
(1 + C2) 2
Eliminating C, we get a direct relationship be-
tween x and y Raising both sides of each equa-
2
tion to the power and adding the résultant
équations termwise, we get the singular solution
in the following form:
2 _2_ 2
x 3 + y, 33 = a '3
This is an astroid. However, the envelope of the
family (and, hence, the singular solution) is not
the entire astroid, but only its left half (since it is évident from the parametric
équations that jt< ;0) (Fig. 15).
p = v ( p ) + [ * ? ' ( p) + %
or
p — *P (p )~ N > ' (p) +♦' (p)] % (O
Let us find the singular solutions. Since p = p2 for p0= 0 and px= 1, the
solutions will be linear functions [see (I')]:
y = x -02 + 02, that is, y = 0
and
y= x+ 1
When we find the complété intégral, we will see whether these functions are
particular or singular solutions. To do so, write équation (T) in the form
dx 2 2
dp x \ — p i - p
and regard x as a function of the independent variable p. Integrating this
linear (in x) équation, we find
«“ - ' + 5 ^ <">
Eliminating p from équations (T) and (II), we get the complété intégral
y = (C+V"x+1)2
The singular intégral of the original équation is
y= 0
since this solution is not obtainable from the general solution for any value
of C.
However, the function p = * + 1 is not a singular but a particular solution;
it is obtained from the general solution when C = 0.
F { x>y* ï )= 0 ( i f)
f (* • ». - - 5 V 0 <3>
ta" » - T ! r l b r <6>
dx
We obtain the slope of the tangent to the equipotential lineby
differentiating relation (5) with respect
to x :
du_ | du dj£_n
dx ‘ dy dx
whence
du
dy_ ÈL
dx~ du (7 )
àÿ
Thus, in magnitude and sign, the slope
of the tangent to the equipotential line is
the inverse of the slope of the tangent to
the flow line. Whence it follows that equipotential lines and flow
Unes are mutually orthogonal.
In the case of an electric or magnetic field, the lines of force
of the field serve as the orthogonal trajectories of the family of
equipotential lines.
Example 1. Find the orthogonal trajectories of the family of parabolas
y — Cx2
Solution. We Write the differential équation of the family:
y' = 2Cx
Eliminating C, we get
y x
Substituting — ^ for y \ we obtain a differential équation of the family of
y
orthogonal trajectories
Î.15 Orthogonal and Isogonal Trajectories 53
that eut the lines of the given family at an angle a , the tangent of which
equals k.
Solution. Let us Write the differential équation of the given family. Difife-
rentiating équation (8) with respect to ac
we find
Fig. 20 & L -k
dx y
A + 1 *
dx 1
whence, dropping the subscript 7\ we find
dy
dx
i-k ±
which defines the family of isogonal trajectories. To find out precisely which
curves enter into this family, let us change to polar coordinates:
~ = tan<p, 'ffx 2+ y 2 = p
ln p = y (p + lnC
p = Cek
Consequently, the family of isogonal trajectories is a family of logarithmic
spirals (Fig. 20).
1.16 Higher-Order Differential Equations 55
the function f(x, y, y ', . . . , yln~1)) and its partial dérivatives with
respect to the arguments y, y', . . . . y(n~l) are continuous in some
région containing the values x-— x0, y —y0, y ’ -- y'0, .
yin-i) —yin-D' (fien there is one and only one solution, y —y(x), of
the équation that satisfies the conditions
i/*=jc0= y»
y'x~x,=y«
= y ln- l)
These conditions are called initial conditions. The proof is beyond
the scope of this book.
If we consider a second-order équation y" ~ f (x, y, y'), then the
initial conditions for the solution, when x ^ - x 0, will be
y=- y0’ y ' = y»
where x0, y0, y ’0 are given numbers, these conditions hâve the fol-
lowing géométrie meaning: only one curve passes through a given
point (x0, y0) of the plane with given tangent of the angle of
inclination of the tangent line y'0. From this it follows that if we
want to assign different values of y'0 for constant x0 and y0, we
get an infinitude of intégral curves with different angles of incli
nation passing through the given point.
We now introduce the concept of a general solution of an équa
tion of the nth order.
56 Ch. 1 Differential Equations
/ ■ - " = S / ( * ) < * * + c»
*•
1.17 An Equation of the Form ÿ(n) = f (x) 57
jd x ( x — x0) + C s
X, 'X ,
/ r . . . , ^ cos k x — i . ~
y = \ sin kxdx + Cx = ------------------
o
y =~ î ( c~—£ — - ) <* * +* * +( : ,
0 0
or
sin kx » x . ~ _
y= — jp— b y + C i x + c ,
centrated. Let the x-axis be horizontal along the axis of the girder in its un-
dèrformed state and let the i/-axis be directed vertically downwards (Fig. 21).
Each force acting on the girder (the load of the girder, and the reaction of
the supports, for instance) has a moment, relative to some cross section of the
girder, equal to the product of the force by the distance of the point of appli
cation of the force from the given cross sec
tion. The sum, M (jc), of the moments of ail
the forces applied to that part of the girder
situated to one side of the given cross section
with abscissa x is called the bending moment
of the girder relative to the given cross sec
tion. In courses of strength of materials,
it is proved that the bending moment of the
girder is
EJ
R
where E is the so-called modulus of elasticity
which dépends on the material of the girder,
J is the moment of inertia of the cross-secti-
onal area of the girder relative to the horizontal line passing through the centre
of gravity of the cross-sectional area, and R is the radius of Curvaturé of the
axis of the bent girder, which radius is expressed by the formula (Sec. 6.6,
Vol. I)
p _ il± Æ *
R~ I y" I
Thus, the differentiaf équation of the bent axis of a girder has thê fortn
y' _ M (x)
( 2)
(l + „'*)V. EJ
If we consider that the deformations are small and that the tangents to
the axis of the girder, when bent, form a small angle with the x-axis, we can
disregard the square of the small quantity y '2 and consider
\_
R
y"
Then the differential équation of the bent girder will hâve the form
M (x)
( 2 ')
EJ
The initial conditions are: for x = 0 the deflection y is equal to zéro and the
tangent to the bent axis of the girder coïncides with the *-axis; that is,
0 *=o = °- y'x=o ^ °
Integrating the équation, we find
y = 2ËT{lX - T ) (3)
% = p- Then S = i -
Putting these expressions of the dérivatives into équation (1),
we get a first-order équation,
Set
then
d2y _dp
dx2~~dx
and we get a first-order differential équation in the auxiliary function p of x :
ÿdx= -La VT+p*
Separating variables, we hâve
dp _dx
whence
in { p + /T + 7 ) = -+ C t
p = s i n h ( 'z r + c ‘ )
But since p = ^ , the latter relation is a differential équation in the sought-
for function y. Integrating it, we obtain the équation of a catenary (see
Sec. 1. 1):
y = a cos h ^ - - 4 - C j^ + Ca
Let us find the particular solution that satisfies the following initial con
ditions:
yx=o =
yx=o = °
The first condition yields Ct = 0, the second, C j= 0 .
We finally obtain
y = ocos h
3yn = y 3
dy dp
Solution. Put P ~ -fc and consider p as a function of y. Then y" = p j î
and we get a first-order équation for the auxiliary function p:
Pa= Ci —y or p = ± K c l - sr ,/»
_dy.
But p = — ; consequently, for a détermination of y we get the équation
dy y 1* dy
. = dx, or = dx
V c .- y 'l -
62 Ch. 1 Differential Equations
whence
__ y1/3 dy
x -)- C2
V C tft'-X
To compute the latter intégral we make the substitution
C,
Then
1
dy = 3t (** +
Consequently,
y u dy
V C , / '* - !
Final 1y we get
cî
I 3< (<»+D
t
* + C 2 = ± - L K c . ÿ ‘^ - 1
J F K c l!/,/>-l(C1ÿ*/’ + 2)
^1
( C l ÿ *'> + 2)
w
Example 3. Let a point move along the jc-axis under the action of a force
that dépends solely on the position of the point. The differential équation of
motion will be
d2x
m = F(x)
dt2
The first term of this équation is the kinetic energy, the second term, the
potential energy of the moving point. From this équation it follows that the
sum of the kinetic and potential energies remains constant throughout the time
of motion.
The problem of a simple pendulum. Let there be a material point of mass mt
which is in motion (due to the force of gravity) along a circle L lying in the
vertical plane. Let us find the équation ot motion of the point neglecting résis
tance forces (friction, air résistance, etc.).
1.Î8 Second-Order Equations Reducible to First-Order Equations 63
Putting the origin at the lowest point of the circle, we put the *-axis along
the tangent to the circle (Fig. 22).
Dénoté bv / the radius of the circle, by s thç arc length from the origin O
to the variable point M where the mass m is located; this length is taken with
the appropriate sign (s > 0 , if the point M is on
the right of O; s < 0 if M is on the left of O). y
Our problem consists in establishing s as a function
of the time t.
Let us décomposé the force of gravity mg into
tangential and normal components. The former, equal
to —mgsin <p, produces motion, the latter is can-
A
celled by the reaction of the curve along which the
mass m is moving.
Thus, the équation of motion is of the form
d2s \
M (xÀ J>
m d t * ^ ~ mg Sinq> S ,
5 0
Since the angle qp= y f°r a circle, we get the équation *
d2s s
— S sm T Fig. 22
This is a Type II differential équation (since it does not contain the independent
variable t explicitly).
Let us integrate it in the appropriate fashion:
ds d2s
T r ~ p> dS
Hence,
or
dp__
P ds ■f
p dp — —
T*
whence
p2--=2gl COS y + CX
Let us dénoté by s0 the greatest arc length to which the point M swings.
For s = s0 the velocity of the point is zéro:
ds
= P = 0
It s = s0
This enables us to détermine C1:
0 = 2gt c o s ^ l+ C ,
whence
Ct = —2gl COS y -
Therefore,
fd s \* „ , ! s «o
— ' — 2g/ ' ™
64 Ch. î Different ial Equations
or, applying to the last expression the formula for the différence of cosines,
/ d s \ 2 . , . s + s 0 . s0— s
\ d t ) = 4& s in —^21 2 sin-°----
21 (5)
or *
j {= 2 Y g l ^ s i n s- ± j± sir ( 6)
21
This is an équation with variables separable. Separating the variables, we get
* =2 V g ld t (V
S “f - S a . S a -----S
Y s ,n ^ r sm V
We shall assume, for the time being, that s =é s0, then the denominator of the
î/
fraction is different from zéro. If we consider that s = 0 for /=^0, then from (7)
^ye get
* -= 2 Ÿ g ït (8)
s -f sfl . s0— s
sin - n r smJLw -
This is the équation that yields s as a function of t. The intégral on the left
cannot be expressed in terms of elementary functions; neither can the function
s of t. Let us consider this problem approximately. We shall assume that the
angles and y are small. The angles ~^ S° and S° ÿ S will not exceed-—
In (6) let us replace, approximately, the sines of the angles by the angles
—s
^dt= 2 Ÿ~gi î / ’ 21
r 8 V 2T
<6 '>
Separating variables, we get (assuming, for the time being, that s 7^ s0)
FT=T- Yï*‘
Again we consider that s = 0 when / = 0 . Integrating this équation, we get
(8 ')
.2 .2 f l
or
whence
(9)
We take the plus sign in front of the root. From the note at the end of
the solution it follows that there is no need to consider the case with the mi
nus sign.
1.18 Second-Order Equations Reducible ta First-Order Equations 65
Note. When solving, we assumed that s ^ s0» But it is clear,. by direct sub
stitution , that the fonction (9) is the solution of équation (6 ') for any value of t.
Let it be recalled that the solution (9) is an ^pproximate solution of équa
tion (5), since équation (6) was replaced by the approximate équation (6 ').
Equation (9) shows that the point M (which may be regarded as the extre-
mity of the pendulum) performs harmonie oscillations with a period T — 2n
This period is in dépendent of the amplitude s0.
Example 4. Escape-velocity probfem.
Détermine the smallest velocity with whfch a body mnst be thrown verti-
cally upwards so that it wi11 not return to the earth. Air resistar.ee is neglectedi.
Solution. Dénoté the mass of the earth and the mass of the body by M
and m respectively. By Newton’s Iaw of gravitation, the force of attraction f
acting on the body m is
where r is the distance between the centre of the earth and the centre of gra-
vity of the body, and k is the gravitational constant.
The differential équation of motion of this body with mass m will be
d2r M •m
md T '= - k ~ W
or
A*
~ * r2
The minus sign indicates that the accélération is négative. The differential
équation (10) is an équation of type (2). We shall solve it for the following
initial conditions:
for / = 0 r = R, ^=*V
Here, R is the radius of the earth and v0 is the launching velocity. We dénoté
dr_ d2r _dv _dv dr dv
d t ~ V> dt2~~dt~~dr " dt~~Vdr
where v is the velocity of motion. Putting this into (10), we get
dv M
k
dr r*
Separating variables, we obtain
vdv = —kM —•
Intégrâting this équation, we find
v2 1
-= k M ^ C t («O
From the condition that v — v0 at the earth’s surface (for r = R)f we détermine Cx*
vô
66 Ch. 1 Differential Équations
or
It is given that the body should move so that the velocity is always positive;
v2 kM
hence, > 0 . Since for a boundless increase of r the quantity — becomes
v2
arbitrarily small, the condition -y > 0 will be fulfilled for any r only for the
case
(13)
or
(14)
where
k = 6 .66 - 1 0 - 8 cm3/g-sec2
R = 63-107 cm
At the earth’s surface, for r = /?, the accélération of gravity isg(g = 981 cm/sec2).
For this reason, from (10) we obtain
or
Dénoté by <p the angle fprmed by the positive x-axis and the
tangent to the curve; then
( 2)
1.19 Graphical Method of Intégrâting Second-Order Equations 67
y ~ R
But
y ' = tan<p, 1 +y'* = 1 -f-tan*<p = sec*<p,
1
(1 + y nYu = | sec3(p | | COS3 <p I
therefore
y R | cos3 cp | (3 )
Now putting into (1) the expressions obtained for y and y", we
hâve
T T \ l à ^ = f ( x ’ y ' tan<P)
or
| cos8 <p | •/ (*, y, tan<p) ' '
It is thus évident that a second-order differential équation déter
mines the magnitude of the radius of curvature of an intégral
curve if the coordinates of the point and the direction of the
tangent to this point are specified.
From the foregoing there follows a method of approximate con
struction of an integraLcurve by means of a smooth curve com-
posed of arcs of circles. **
To illustrate, let it be required to find the solution of équation
(1) that satisfies the following initial conditions:
yx—x0—yoi yx—xa—ya
Through the point M0(x0, y0) draw a ray M0Ta with slope
y' = tan<p„ = y'9 (Fig. 23). From équation (4) we find the value
R = R0. Lay off a segment MtC0, equal to R0, perpendicular
to Af0T0, and from the point C0 (as centre) strike an arc
* Up till now we hâve always considered the radius of curvature positive;
in this section we shall consider it a number that can take on both positive
and négative values: if the curve is convex (y" < 0 ), we consider the radius
of curvature négative (R < 0); if the curve is concave (y* > 0), it is positive
(R > 0).
** A curve is called smooth if it has tangents at ail points and the angle of
inclination of the tangent is a continuous function of the arc length s.
m Ch. 1 Differential Equations
with radius /?„. It should be noted that if /?„ < 0, then the seg
ment M„C0 must be drawn in that direction so that the arc of
the circle ïs convex upwards, and for R„ > 0, convex down (see
footnote on page 67).
Then let x lt yx be the coordinates of the point M x which lies
on the constructed arc and is sufficiently close to the point Af„
while tan q>j is the slope of the tan
gent M XTX to the circle drawn at M x.
From équation (4) we fînd the value
R — Rx that corresponds to Af1. Draw
the segment M XCX, perpendicular to
MXTX, equal to Rt, and from Cx (as
centre) strike an arc M^M2 with ra
dius R x. Then on this arc take a point
Af, (xt, ÿa) close to Mx and continue
construction as before until we get
a sufficiently large piece of the curve
consisting of the arcs of circles. From
the foregoing it is clearthat this curve
is approximately an intégral curve that
passes through the point M0. Obviously, the smaller the arcs
M^MX, AÇM2, . . . . the doser the constructed curve will be to
the intégral curve.
Example 1. Let there be an équation y " — y ~~0. It is easy to verify that the
functions ex , e~x t 3ex , 5e~x are solutions of this équation. Here, the functions
ex
and e~x are linearly independent on any interval because the ratio — - = e 2x
6~ x
does not remain constant as x varies. But the functions ex and 3ex are linearly
3ex
dépendent, since = 3 = const.
70 Ch. 1 Differential Equations
ln W = —J at dx+ InC
X.
1.20 Homogeneous Linear Equations 71
or
x
ln w = — r\ al dxm
whence
X
- \ a' dx
W = Ce *° (7)
It will be noted that we can write the function (7) and say
that this function satisfies équation (6), which is clear if we sub-
stitute this function directly into (6). The assumption 0 is
not required.
Formula (7) is called Liouville’s formula.
We define C so that the initial condition is satisfied. Substitut-
ing x-=;t0 into the left and right sides of (7), we get
r 0=c
Consequently, the solution that satisfies the initial conditions
takes the form
-Î-*
W = W0e x* (T)
By hypothesis, W0^=0. But then from (7') it follows that IF=^0
for any value of x, because the exponential function does not va-
nish for any finite value of the argument. The proof is complété.
Note I. If the Wronskian is zéro for some value x —x0, then
it is also zéro for any value x in the interval under considération.
This follows directly from (7): if 1F = 0 when x = x0, then
( r ) , =, . - c = o
on the interval [a, 6], which is to say y.t and yx are linearly
dépendent. #
But this contradicts the assumption that the solutions y2 and
yY are linearly independent. We hâve thus proved that the Wrons-
kian does not vanish at any point of the interval [a, b\.
Theorem 6. If yx and y, are two linearly independent solutions
of équation (3), then
y ^ C iy i + C2y2 (8)
where Ct and C2 are arbitrary constants, is its general solution.
Proof. From Theorems 1 and 2 it follows that the function
t'ii/i ~t~
is a solution of équation (3) for any values of Cx and C2.
We shall now prove that no matter what the initial conditions
yx=x0 y'x- *0 :r= y'o» it is possible to choose the values of the ar
bitrary constants Cx and C2 so that the corresponding particular
solution C1yl + C2yl, should satisfy the given initial conditions.
Substituting the initial conditions into (8), we hâve
y<) ~ ^ i^/io 4“Cai/io g
I/o “ Cxyï0 -|- C2y2Q j
where we put
(yi)x~x0 ÿ10; {y?)x-r„ ~ y»o» (*/i)*=*0 i/io> (*/2)*=*0 -- i/2o
From .system (9) we can détermine Cx and C2, since the determi-
nant of this system
y io y 20
~~y loi/îio i/ioi/ao
ylo y»o
is the Wronskian for and, hence, is not equal to 0 (by vir-
tue of the linear independence of the solutions yx and y2). The
particular solution obtained from the family (8) for the found
values of C1 and C2 satisfies the given initial conditions. Thus,
the theorem is proved.
Example 2 . The équation
whose coefficients ax= - and = ----- ^ are continuous on any interval that
does not contain the point x —0 , admits the particular solutions
yi = *. y*= -x-
(Ihis is readily verified by substitution). Hence, its general solution is of the
form
y = C1* + C 2 —
74 Ch. 1 Differential Equations
or
whence
( 10)
1.21 Second-Order Homogeneous Linear Equations 75
*1 = - } + YT - * ’ k*=-T-Yf-ï
The following cases are possible:
I. and k 2 are real numbers and not equal (k 1^ k î ),
II. fe, and k 2 are complex numbers,
III. k, and k 2 are real and equal numbers ( k ^ k ^ .
Let us consider each case separately.
I. T he ro o ts o f th e a u x ilia ry é q u a tio n are real a n d d is tin c t,
k 1^ k î . Here, the particular solutions are the functions
# ,= £ * '* , y 3 =-<;***
These solutions are linearly independent because
— =
t/L e*.* ^
const
Hence, the complété intégral has the form
y H-
Example 1. Giveri the équation
y" + y' — 2y=r.O
The auxiliary équation is of the form
—2 = 0
We find the roots of the auxiliary équation:
= —y ± Y"4"“^2
^ = 1, 2
The complété intégral is
y = C,ex -j-Cte~2x
I I . T he ro o ts of th e a u x ilia ry é q u a tio n a re com plex. Since com
p le x roots are co n ju g a te in p airs, w e W rite
kl =a-\- ip, kt ■=a —«p
w h ere ______
a = —T* P=Yq~7
The particular solutions may be written in the form
yl = eia +if t x, = (4)
These are complex functions of a real argument that satisfy the
differential équation (1) (see Sec. 7.4, Vol. I).
1.21 Second-Order Homogeneous Linear Equations 77
y = Y e~x sin2*
Its graph is shown in Fig. 24.
Example 3. Given the équation
y ”+ 9y = 0
Find the general solution and a particular solution satisfying the initial con
ditions
y x —Q— y x=o = 3
Solution. Write the auxiliary équation
*2 + 9 = 0
We find its roots to be
k 1= 3/\ kty = —3t
The general solution is
y = Ci cos 3jc+C 2 sin 3x
1.21 Second-Order Homogeneous Linear Equations 79
III. The roots of the auxiliary équation are real and equal. Here,
kx= k2.
One particular solution, y1= ek*x, is obtained from earlier reason-
ing. We must find a second particular solution, which is linearly
independent of the first (the function ek*x is identically equal to
ek*x and therefore cannot be regarded as a second particular solu
tion).
We shall seek the second particular solution in the form
y2= u (x) ek*x
where u(x) is the unknown function to be determined.
Differentiating, we find
y2= u'ek*x + kxuek*x = ek*x (u! -f kxu)
y2= + 2klu,ek*x + k\uek'x = ek*x (u" + 2kxu' +
Putting the expressions of the dérivatives into (1), we obtain
ektx [u" + (2kt + p)u' + (k\ + pkx+ q) u] = 0
Since kx is a multiple root of the auxiliary équation, we hâve
k\ + pk1+ q = 0
78 Ch. î Dijferential Equations
III. The roots of the auxiliary équation are real and equal. Here,
kx — k2.
One particular solution, yx = ek^x, is obtained from earlier reason-
ing. We must find a second particular solution, which is linearly
independent of the first (the function ek*x is identically equal to
ek*x and therefore cannot be regarded as a second particular solu
tion).
We shall seek the second particular solution in the form
y2= u (x) ek'x
where u(x) is the unknown function to be determined.
Differentiating, we find
y2= u’ek*x + = ek*x (u' + kxu)
y2- + 2klufek^x + k\uek'x = ek*x (u" + 2kxu' + k\u)
Putting the expressions of the dérivatives into (1), we obtain
ek'x [u" + (2kx+ p)u' + (k\ + pkx+ q) u] = 0
Since kl is a multiple root of the auxiliary équation, we hâve
k\ + pk1+ q = Q
80 Ch. î Differential Equations
Examples:
1. The functions y 1= ex , y2 = e2x, y 9= 3ex are linearly dépendent, since for
C1 = l , C2 = 0, Ca= — we hâve the identity
CXex + Cte2x + C93ex ^ 0
2 . The functions yi = l, y 2 = x, y3 — x2 are linearly independent, since the
expression
Cil -\~C2X-\-C9x2
will not be identically zéro for any Clt C2, C3 that are not simultaneously
equal to zéro.
8 . The functions y l =ektXy y2=ek*x, . . . , y n ~ e knx, . . . , where k it k2, . . . , kn, . . .
are different numbers, are linearly independent. (This assertion is given without
p roof.)
Let us now solve équation (1). For this équation» the following
theorem holds.
Theorem. If the functions yx, yt, . . yn are linearly independent
solutions of équation (1), then its general solution is
y = Ci!h + C2y2+ • ■• + CnlJn (2)
where Cly . . Cn are arbitrary constants.
If the coefficients of équation (I) are constant, the general so
lution is found in the same way as in the case of second-order
équations.
(1) Form the auxiliary équation
kn + a1kn~1+ a 2kn- 2+ . . . + an = 0.
(2) Find the roots of the auxiliary équation
• • •» hn
(3) From the character of the roots Write out the particular
linearly independent solutions, taking note of the fact that:
(a) to every real root k of order one there corresponds a parti
cular solution ekx\
(b) to every pair of complex conjugate roots &(1) —a + fp and
k{2)=^a— t’P there correspond two particular solutions e^cospx
and ea*sinpx;
(c) to every real root k of multiplicity r there correspond r
linearly independent particular solutions
ekx, xekx, . . . , xr~'ekx\
(d) to each pair of complex conjugate roots &(1) = oc-ftp,
k(1) = a — /p of mult iplicity \i there correspond 2\i particular so
lutions:
eax cos p*, xeax cos p*, . . . , x*- 1eax cos Px,
eax sin Px, xeax sin Px, . . . , x*- leax sin px
82 Ch. 1 Differentiai Equations
Here, y10, y2». yl. y'io, y’io. yl dénoté the numerical values of the func-
84 Ch. 1 Differential Equations
and C2; in other words, there exist values Cx and C2 such that
formula (3) defines the solution of équation (1) which satisfies
the given initial conditions. The theorem is completely proved.
Thus, if we know the general solution y of the homogeneous
équation (2), the basic difficulty, when integrating the nonhomo-
geneous équation (1), lies in finding some particular solution y*.
We shall give a general method for finding the particular so
lutions of a nonhomogeneous équation.
The method of variation of arbitrary constants (paranteters).
We write the general solution of the homogeneous équation (2):
y = Cxyx + C2y2 (7)
We shall seek a particular solution of the nonhomogeneous
équation (1) in the form (7), considering Cx and C2 as some (as
yet) undeterm ined functions of x.
Differentiate (7):
y' = Cxyx+ C2y2'1- C1y1T C2y2
Now choose the needed functions Cx and Ct so that the following
équation holds:
C%yi “HC2t/2= 0 (8)
If we take note of this additional condition, the first dérivative y'
will take the form
y —Cxyx-J- C2y2
Diiferentiating this expression, we find y":
y "— Cxyx + C2y2-f- Cjÿi 4- C2y2
Putting y, y' and y" into (1), we get
Cxyx-f- C2y2-{- Cxyx-f- C2y 2-f- ctl (Cxyx -f- C2y2) -|- ct2(Cxyx-f- C2y2) = f (.v)
or
C x (lJ \ + û iÿi + fl2ÿi) + C 2 (y*. + axy2+ a y ù + Cxyx-f- C„y2— f (x)
2
*y’ — —
X
=0
Since
u” —
1 we hâve ln y' = ln *-|-ln C, y' = Cx
y x
and so
y = Clx2-\-C%
For the latter expression to be a solution of the given équation, we hâve
to define Cx and C2 as functions of x from the System
CiX2-f- C2 • 1 = 0 , 2Cjx -J- C2 •0 = x
Soiving this System, we find
Putting the functions obtained into the formula ^ = Cpcï + Ca, we get the
general solution of the nonhomogeneous équation:
— — Jf® x®
i r - c 1*«+ci + Ç - ^
_ — jf3 —
or y — C1x2 + Ca + “5- , where Cx and C2 are arbitrary constants.
ü
When seeking particular solutions, it is useful to take advan-
tage of the results of the following theorem.
* If we put C i= C, = 0 , we get a particular solution of équation ( 1).
86 Ch. 1 Different ial Equations
* It is clear that the theorem holds true for any number of terms in the
right member of the équation.
1.24 Nonhomogeneous Second-Order Linear Equations 87
* We remark that ail the results given above also hold for the case where a
is a complex number (this follows from the rules for differentiating the func
tion emx, where m is any complex number; see Sec. 7.4, Vol. I).
88 Ch. î Differential Equations
Consequently,
ÿ*==* ( — eX
and the general solution is
90 Ch. 1 Differential Equations
Solution. The auxiliary équation has roots k 1= 2i, k2= —2/; therefore, the
general solution of the homogeneous équation is of the form
y = Cj cos 2 * + C2 sin 2x
Putting these expressions of the dérivatives into the given équation and
equating the coefficients of cos 2x and sin 2x, we get a System of équations
for determining 4 and B:
4 3 = 1 , —44 = 0
whence 4 = 0 and 3 = -^-. Thus, the general solution of the given équation is
y* = e2x ^ c o s x + - | - s i n x ^
Note. Ail the arguments of this section hold true also for a
linear équation of the first order. To illustrate, let us consider
a first-order équation with constant coefficients (this équation is
frequently encountered in engineering applications):
% + ay = b (io)
where a and b are constants. We find the general solution of the
homogeneous équation
% + ay = 0
Form the auxiliary équation
£ + a = 0 , k = —a
1.25 Hlgher-Order Nonhomogeneous Linear Equations 93
General remarks on Cases 11 and III. Even when the right side
of the équation contains an expression with only cosp* or only
sin^jt, we must seek the solution in the form indicated, that is,
with sine and cosine. In other words, from the fact that the right
side does not contain cos|fo or smp*, it does not in the least
follow that the particular solution of the équation does not contain
these functions. This was évident when we considered Examples 4,
5, 6 of the preceding section, and also Example 2 of the présent
section.
Example 1. Find the general solution of the équation
yiv -.y==x3+ i
Solution. The auxiliary équation k* — 1 = 0 has the roots
fci = 1, fc2 ——1, k3= i, k 4 = — i
We find the general solution of the homogeneous équation (see Example 4,
Sec. 1.22):
y = C1e* + C2e~* + Cs cos* + C4 sin x
We seek the particular solution of the nonhomogeneous équation in the form
y* = Aç>x?+ A xx2+ ,4a*+ A3
Differentiating y* four times and substituting the expressions obtained into
the given équation, we get
— A0x? — A xx2— A 2x — A 3~ x®+ 1
Equating the coefficients of identical powers of x , we hâve
— A 0= \ t — Ai — 0, — A 2= Q, — A3— \
Hence
y* = — x3— 1
Thejjeneral solution of the nonhomogeneous équation is found from the formula
y — y -{-y * '.
y — Ctex -j-C2e -x + C3 cos x + C4 sin x — x3— 1
Example 2. Solve the équation
y iv — y - 5 cos x
Solution. The auxiliary équation k x— 1 = 0 has the roots — 1, k2 — — 1,
k3= i , k 4= — t. Hence, the general solution of the corresponding homogeneous
équation is _
y — Ciex + C tf-* + C3 cos * + C4 sin x
Further, the right side of the given nonhomogeneous équation has the form
f (x) = M cos x + N sin x
where Af = 5 and N — 0.
Since i is a simple root of the auxiliary équation, we seek the particular
solution in the form
y* = x(A cos x + B sin x)
Putting this expression into the équation, we find
4A sin x — 4B cos x = 5 cos x
1.26 The Differential Equation of Mechanical Vibrations 97
whence
4/1 = 0, —4Æ= 5
5
or j4 = 0 , B — . Consequently, the particular solution of the differential
•
équation is
.
u — ---- 5
- x sin*
4
and the general solution is
(H
7—2082
98 Ch. î Differential Equations
where
In this case the restoring force will be equal not to — kyy but
to —k[y + ty(t)]> f°rce °f résistance will be — X[yf + q>' (/)].
and in place of équation ( 1) we will hâve the équation
Q ^ r + *-% + ky = — k<p(t) — W ( t) (2 )
or
- g - + /> -§ -+ W = /( 0 (2')
where
*.— 7 + / f 3 ?. *. = - ! • - Y T - i
p2
(1) Let Then the roots kt and k2 are real négative num-
bers. The general solution is expressed in terms of exponential
functions:
y = Cxe*‘f + C2eV (k, < 0 , k2 < 0 ) (2 )
From this formula it follows that the déviation of y for any
initial conditions approaches zéro asymptotically if t —►<». In the
given case, there will be no oscillations, since the forces of résist
ance are great compared to the coefficient of rigidity of the
spring k.
(2 ) Let y = <7; then the roots kt and k2 are equal ^and are
also equal to the négative number — ■£-) . Therefore, the general
solution will be
are
(x,y)
y = 4 sin (p/ + q> 0) \
Po) /i \
x = A cos (p/ + <Po) p0 J (7) i / / •» \ \
( y â s l A
The expressions (7) are the solutions I
of équation (4). l 0
X J X
We consider the complex quantity
z = x + iy =
= A cos (P* + <p0) + iA sin (P* + qp0)
or
Fig. 28
z = A [cos (P* + <p0) + i sin (pf + <p0)] (8 )
The imaginary and real parts of (9) are solutions of équation (4).
Expression (9) is called the complex solution of (4). Wecanrewrite (9)
as follows:
2 = A ë^éV * ( 10 )
The expression Ae‘f> is called the complex amplitude. We dénoté
it by A*. Then the complex solution (10) can be rewritten as
z = A'eW ( 11 )
ÿ , = 7 i ? = S ¥ W si" (“ ( + 'p' )
The general solution of équation ( 1) is y = y + y* or
y = ^4e“'sin (p/ 4- qp0) sin ((ùt - f qp*)
y (g—uY + P 2®2
1.28 Forced Oscillations 103
The first term of the sum on the right side (the solution of the
homogeneous équation) represents damped oscillations; it dimini-
shes with increasing t and, consequently, after some interval of
time the second term (which détermines the forced oscillations)
will dominate. The frequency co of these oscillations is equal to
the frequency of the external force /(/); the amplitude of the forced
vibrations is the greater, the smaller p and the doser w2 is to q.
Let us investigate more closely the dependence of the amplitude
of forced vibrations on the frequency co for various values of p.
For this, we dénoté the amplitude of forced vibrations by D(«):
^ ^ V (q —<û2)2-f-p2o)2
Putting 9 = (for p = 0, p, would be equal to its natural fre
quency), we hâve
a
D( o)) =
V '( P Î - û .2)2 + p2(02
t / Ï = ?
Hence, the maximum amplitude is equal to
104 Ch. 1 Differential Equations
The graphs of the function D (A.) for various values of y are shown
in Fig. 30 (in constructing the graphs we put a= 1, Px= l for
the sake of definiteness). These curves are called résonance curves.
From formula (5) it follows that for small y the maximum
value of amplitude is attained for values of A close to unity, that
is, when the frequency of the external force is close to the fre-
quency of free oscillations. If y = 0 (hence, /? = 0), that is, if there
is no résistance to motion, the amplitude of forced vibrations
increases without bound as A—>1 or as co—
lim D (A) = oo
x»—
►î
(Y=0)
At « 8 = 9 we hâve résonance.
(2) Now let us suppose that p — 0; that is, we consider the
équation of elastic oscillations without résistance but with a
periodic external force:
y" + ?ÿ = asin ait ( 6)
1.28 Forced Oscillations 105
M=~ k ' ^ = 0
Consequently,
y* = — âp t cos p t
The general solution will hâve the form
y = A sin (P* + <P0) —^ t cos p*
The second term on the right side shows that in this case the
amplitude increases without bound with the time t. This pheno-
menon, which occurs when the natural frequency of the System
coïncides with the frequency of the external force, is called
résonance.
The graph of the function y* is shown in Fig. 31.
106 Ch. 1 Differential Equations
TE* = M *. Vu h i ...... y «
)
g i = Fn(x, y v . . . . yn)
1.29 Systems of Ordinary Differential Equations___________ 107
d ny \ _f ( y u
fan ^ n \ x 9 y 1» *
Putting the expressions and ^ from équations (a) into this équation,
we get
j g = ( y + z + x ) + ( - 4 y - 3 z + 2x) + l
or
g = - 3 y - 2 2 + 3x+l (c)
(2) From the first équation of System (a) we find
or
S + 2E + » - 5‘ + ' w
The general solution of this équation is
y = (Cl + Cîx )e -* + 5x—9 (f)
and from (d) we hâve
z = (C2— 2Ci — 2C2x) e - * — 6 jc + 14 (g)
Choosing the constants Cx and C2 so that the initial conditions (b) are
satisfied,
(y)x=o= i » (z)x=o= o
we get, from équations (f) and (g),
1= C i —9, 0 = Ca—2CX+ 14
dt2 dt
Integrating this équation, we obtain its general solution:
x = C1e~t + C2e2t (a )
Whence we find
— +z = Z C ^
Integrating this équation, we find
z = C3e-<-\-C1e*t (V)
But then, from équation (P), we get
y = —(^î + Cs) e~t -{-C2e2t (6)
Equations (a), (y), and (6) give the general solution of the given System.
The differential équations of a System may contain higher-ordei
dérivatives. This then yields a System of differential équations of
higher order.
For instance, the problem of the motion of a material point under the action
of a force F reduces to a System of three second-order differential équations.
Let Fx, F», Fz be the projections of the force F on the coordinate axes. The
position o r the point at any instant of time t is determined by its coordinates x,
(/, and z. Hence, x, y , z are functions of t. The projections of the velocity veç-
dx dy dz
tor of the point on the axes will
be H t ' dt ’ dt
Suppose that the force F and, hence, its projections FXt Fy, F? dépend on
the time /, the coordinates x, y , z of the point, and on the velocity ot motion
dx du dz
of the point, that is, on .
In this problem the following three functions are the sought-for functions:
x = x(t), y = y ( t) t z = z (t)
110 Ch. 1 Differential Equations
II
Pi.1
('• Ht 9 dt 9 d tJ
d2y dx dy dz \
*■>
Xy y t Zy
y
( 8)
%
('■ dt 9 dt ’ dt J
dx dy dz \
Xy y y Zy
dt2 ~ F* 1('■ Ht 9 dt * dt J
We thus hâve a System of three second-order differential équations. In the case
of plane motion, that is, motion in which the trajectory is a plane curve (lying,
for example, in the x^-plane), we get a System of two équations for determin-
ing the fonctions x (t) and y (t):
mS = F* { i ’ x > w ) <9>
m^ Fy ( f’ x>y>-W' w ) (,0)
It is possible to solve a System of differential équations of higher order by
reducing it to a System of first-order équations. Using équations (9) and (10) as
examples, we shall show how this is done. We introduce the notation
dy
dt
Then
d2x du cPy__ dv
dt*~~ H T 9 dt2 H t
The system of two second-order équations (9) and (10) in two unknown fonc
tions *(/) and y (t) is replaced by a system of four first-order équations in four
unknown fonctions x y y t u f v:
dy_
v
dt
= x, y, u, t-)
dv
X, y f u, v)
Ht
We remark in conclusion that the general method that we hâve considered of
solving the system may, in certain spécifie cases, be replaced by some artificial
technique that gels the resuit faster.
Example 3. Find the general solution of the following system of differential
équations:
d2z
dx2 = y
1.30 Systems of Linear Differential Equations 111
Solution. Differentiate both sides of the first équation twice with respect
to *:
d4y _ d2z
lW~~dx* m
d?z
But -j-£ = y, and so we get a fourth-order équation:
d4y
dx4 = y
Integrating this équation, we obtain its general solution (see Sec. 1.22,
Example 4):
y = C1ex + C2e~ x -\-C3 cos jc+ C 4 sin x
d2u
Finding — from this équation and putting it into the first équation, we find z:
z = C}ex -\-C2e - x —C3 cos x —C4 sin x
= aux i + + • • • +alnxn
^7 = 1*1 + a , ^ t + . . . + aïnxn
at > m
satisfy the system of équations (1). Putting them into System (1),
we get:
Aaxe*f = (a11a 1+ a12a 2+ . . . + a lna „ ) e kt
k a 2ekt = (a2ia i + ü22a2+ . . . + a 2na n) ekt
-— - = 2 x i + 2 x 2 , —j j j = * i + 3*2
and
v <2)
*i : =«{««"', 4» = oi, , e“
Form the system (3) for the root k x — 1 and détermine a}1’ and c4u :
(2—1) a î l) + 2a*,) = 0 \
l«xl> + (3—1) a21) = 0 j
or
a J1’ + 2aj1>= 0
a ! l)4 - 2 a t,, ) = 0
8—2082
114 Ch, 1 Differential Equations
X t— — — C x e f - |- C 2e 4 f
II. The roots of the auxiliary équation are distinct, but indude
complex roots. Among the roots of the auxiliary équation let there
be two complex conjugate roots:
/j1= a+t'P, k î = a —ip
To these roots will correspond the solutions
x}l) = a '/1 1 ( / = 1 , 2 , . . . . n) (7)
x}2>= aj2) * (/ = 1, 2, . . . , n) (8)
The coefficients a ‘n and aJ-2) are determined from the System of
équations (3).
Just as in Sec. 1.21, it may be shown that the real and imagi-
nary parts of the complex solution are also solutions. We thus
obtain two particular solutions:
x<D — e *t (Aj.‘>cos px + Aÿ2) sin Px) |
(9)
x f — (A.J" sin px -(- A)2>cos px) j
where A)1' , Xÿ\ A.)1’, A'/* are real numbers determined in terms of
and a f ).
Appropriate combinations of functions (9) will enter iitto the
general solution of the system.
Example 2. Find the general solution of the System
dx i
~ d f= :— ^Xl *2
k 1= i, k2 = — i, k3= Ÿ " 3, k 4 = — }r 3.
J ((S ) = a (8 ) e V ~ 3 \ y W = p (3 ) g Y a t
«<*> = 1, P<2' = y
a ‘s>=l, p w = - i
« (« = 1, P“' = - y
We Write out the complex solutions:
*(1) = elt = cos / + t sin /, y(1)= y ( c o s i~\~i sin t)
*(2>= s in /, y(2)= y s i n f
} (H
conditions, the corresponding solutions differ but little for ail positive
values of /. If the System of differential équations is a System that
describes some motion, then in the case#of stability of solutions,
the nature of the motions
changes but slightly for
small changes in the initial
data.
Let us analyze an example
of a first-order équation.
Suppose we hâve the diffe-
rential équation:
t = - y +1 w
which is obtained from the system (4) by dividing the right and
left members by each other.
The origin, 0(0, 0), is a singular point of the differential équa
tion (13), since this point does not belong to the domain of exis
tence and uniqueness of the solution.
122 Ch. î Differential Equations
Let us analyze the arrangement of the curves on the phase plane in this
case. We transform expressions (A). Let
x0= M cos Ô, y0 = M sin ô
M ^ V xl + yl, tan ô = —
Xq
Then équations (A) become
* = M * -< c o s(P / —ô) \
y=jM e~( sin(p/ —ô) /
Pass to polar coordinates p and 6 in the phase plane and establish the relation-
ship p = / (0). Equations (B) become
p cos 0 = M e- * cos (0 / —ô) 1
p sin 0 = Me-* sin (p /^ ô ) / ^
Squaring the right and left members and adding, we obtain
p * = M *e-2t
or
p = Me-* (D)
How does / dépend on 0? Dividing the terms of the lower équation of (C) by
the corresponding terms of the upper équation, we get
tan 0 = tan (p /—ô)
whence
.0 +0
P
Substituting into (D), we hâve
9+6
p = Me ^
p = Me
6_
Putting Me P = M lf we finally obtain
p = M xe 3 (E)
Clearly, for arbitrary e > 0 and for ail sufficiently small |x0| and
|i/J it will be true that | jc(/) j < c, |y ( 0 |< ® for arbitrary t.
The solution is stable. Here, x and y are periodic functions of t.
In order to carry out the analysis of the intégral curves on the
phase plane, it is advisable to Write the first équation of (18) as
[see (16)]
x —Csin(p/ + Ô)
( 20)
i/ = ^ c o s (p / + 6) —-s in ( p / + 6)
& + £ ->
We hâve a System of ellipses on the phase plane, and the singular point is a
centre.
VII. Let X1= 0, X3< 0 . The solution (8) in this case becomes
x = Cj + C4eV
( 22)
y = L [ - C lC+ C2{ K - c ) e ^ ] }
Clearly, for arbitrary e > 0 and for sufficiently small |x0| and |t/0|
it will be true that |x ( f ) |< e , |# ( / ) |< e when t > 0. Hence, the
solution is stable.
Example 7. Investigate the stability of the solution of the System
dx r, du
57=°' 5 7 = -* («)
1.3Î On Lyapunov's Theory of Stability 129
— = — = £, that is, y — kx
x c_i
130 Ch. 1 Different iat Equations
This is a family of straight lines passing through the origin. The points along
the intégral curves approach the origin. The singular point O (0 , 0 ) is a nodal
point (Fig. 39).
Note that in the case of ^ = ^ > 0 the form of the solution (22)
is preserved, but when 00
I*(01 —*00 and |y (0 1 *
The solution is unstable.
X. Let ^ = 0. Then
x = Cl -\-Cit
(24)
y = — [—cCj + C2—cCtt\
S
Whence it is quite clear that x —<-oo and y —>-oo as t + 00.
The solution is unstable.
Example 9. Investigate the stability of the solution of the System of équations
dx dy
=0
di = y' dt
Solution. Find the roots of the auxiliary équation
q ^ =0» ^i = 0» A,1= X2= 0
We find the solutions to be
y — C2» x = C2t-\-C i
It is quite obvious that x —►oo as t —►+<». The solution is unstable. The
équation on the phase plane is ^“ = 0- Ttie intégral curves y — C are straight
lines parallel to the axis (Fig. 40). The singular point is called a degenerate
saddle point.
But for exceptional cases, the solution of this system is not expres-
sible in terms of elementary functions and quadratures.
To establish whether the solutions of this system are stable or
unstable, they are compared with the solutions of a linear
system. Suppose that for x —*-0 and y —► (), the functions P(x, y)
and Q (x, y) also approach zéro and approach it faster than p,
where p = l/x 2+ ÿ2; in other words,
l i m F (£1jd = 0; iif n Q Jîijd = o
p-»o P p-*-o P
Then it may be proved that, save for the exceptional case, the
solution of the system (25) will be stable when the solution of
the system
ÿ = cx + gy '
Tt= ax + by t
is stable, and unstable when the solution of the system (4) is
unstable. The exception is that case when both roots of the auxil-
iary équation lie on the axis of imaginaries; then the question of
132 Ch. 1 Differential Equations
£ ? -'(* •§ ) <26>
Put
(1)
Ay=--f(x, y) Ax (2')
When x = x0 we hâve
/
—/ (-'‘a» yà)t Aÿ0—/ (x„, y0) Ax
or
«/i—*/»= /(* ». yo)h
In this équation, x0, y0, h are known; thus we find
0 a+i = 0 a+ /(**. y * ) h
= + /(*»-!. y n -Jh
* For the proof see, for example, I. G. Petrovsky’s Lectures on the Theory
of Ordinary Differential Equations.
1.33 A Différence Method for Âpproximate Solution of Differential Equations 135
X y y’ AV
*0 yo y'»
x i = x0+ A yi y[
Ay [
•** = ■*0+2* y* y*
* *-* = * o + ( * — 2 )* y* -i y'k -t
a» ;.,
* * - i = * o + ( * — • )* y k - 1 y'k -1
& y 'k~ i
** = •*0 + ** yk y'k
and
^y'<» &2y'u • • •. bVk-z
Let us détermine the value of yk+l from Taylor’s formula setting
a = x„, x = xk+1 = xk + h:
h. hfi . /ifli
yk+ 1 — yk+ ~ { / * + ]7 2 yk + 1.2 .3 y>t + ••• + ~i~
e>i
1-
(6)
II
T
(10)
II
M
1
<
*««1«Ç
T
(11)
1
0 i = l + + - 1 + ^ . 2 + M . 2=1.1103
1 T h2 123
Similarly, for ft = 0.2 we hâve
. , 0.2 , , (0.2)2 o i (°-2)3 o ,
140 Ch. 1 Differential Equations
X y V' a y A V
*0 = 0 ÿ o = 1.0000 1/0 = 1
1 Ayô = 0.2103
A^i = 0.2324
Ay2 = 0.2568
y» = 1 .2 4 2 7 + - ^ . 1.4427+— - . 0 . 2 3 2 4 + 0 . 0 2 2 1 = 1.3995
We then find the values of y3, Ay2t A2y[. Again using formula (12) we find y4*
Hence, yx=n,4= 2e0A ^ 0.4— 1 = 1.58364. The absolute error is 0.0003; the rela-
0 0003
tive error, ~ 0.0002 = 0 .02 % (In Euler’s ipethod, the absolute error
of y 4 is 0.06, the relative error, 0.038 = 3.8%.)
Example 2 . Find approximate values of the solution of the équation
y '= y 2+ x 2
that satisfies the initial condition yo = 0 for *0 = 0. Détermine the values of the
solution for * = 0.1, 0.2, 0.3, 0.4.
Solution. We find
= O2+ O2= 0
yx-o = (2^' + 2*)x=o = 0
y'x- o = (2 y'2 + 2yy'f + 2)x=0 = 2
By formulas (4) and (4') we hâve
ÿi = — •2 = °.0°03. y, = — ^ -2 = 0.0027
X y y' Af/' AV
•*
*o = 0
il
yo — Q
o
Ayi = 0.0100
=0.0300
Ay, = 0.0501
* 4 = 0.4 y4 = 0.0204
142 Ch. 1 Differential Equations
Thusf
y3= 0 . 0 0 2 7 + • 0.0400 + 0.0300 + ^ •0 .1•0.0200 = 0.0090
Note that, to four places, y A= 0.0213. (This may be obtained by other, more
précisé, methods with error éstimation.)
% = fi(x,y,z) (2)
that satisfy the initial conditions y = y0, z —z0 when x = x0.
We détermine the values of the functions y and z for values of
the argument *0, x2, . . . , xk, xA+1, . . . . xn. Again, let
xk+i— xk — ^x = h (k —0, 1,2, ...» n — 1) (3)
We dénoté the approximate values of the functions y and z, respec-
tively, by
i/o» y i> • • • * yk< i/*+i> • • •» yn
and
2o> Z,, . . ., Zh, Zfr+j , . . .i z„
Write the récurrence formulas of type (12), Sec. 1.33:
Vk+i = y*+ y#* + y &y'k-i + j2 (4)
, 2 h , x (2/i)2 „ . (2h f ,,
y 2 —Ho+ y i/o + ~2~ yo+"3?" y°
_ . h ,, h2 h**
zo~r j i 2 ' 3! Z°
2/i (2/i)2 w -
^2 —2o+ T z° ”1” 3! Z°
' To apply these formulas, one has to know y0, y"Qi y’0", z'0t zï,
z’0" 9 which we shall now détermine. From (1) and (2) we find
y* = /i (*., y 0, z0)
Zo fî (Xo> I/o» 20)
DifFerentiating (1) and (2) and substituting the values of ■*o> ÿo> 20(
y'0 and z'0, we find
y a —(y )*=*„
2o —(2 )jc=jc0
*0 yo yo z’o
Il * 1
Ay'o AZq
Ay[ Azî
0
*2 y2 y'2 A*y[ ! z2 *2 A2z|
Aÿâ AZ2
*3 y3 yo *3 2S
1
144 Ch. 1 Difjerential Equations
From formulas (4) and (5) we find y3 and z3, and from equa-
tions (1) and (2) we find y3 and z3. Computing Ay't\ A2y[, Az2,
we find yAt and z4, etc., by applying formulas (4) and (5) once
again.
Example. Approximate the solutions of the System
y' = z, *'= y
wlth initial conditions yo = 0 and z0= l for x = 0. Compute the values of the
solutions for x = 0, 0.1, 0.2, 0.3, 0.4. f
Solution. From the given équations, we find
i/o= 2*=o=l
zo=yx=o= ®
Differentiating the given équations, we find
yo = {y")x=o = (z')x=o = 0
zo = (z )x-o = (y )x=o = f
o=( 2w)x=o==i
yô" = ly '" )x =
^ // = (2/")x =O=(^)x =O= 0
Using formulas of type (4) and (5), we find
(0.1)3•1=0.1002
3!
( 0 . 2)2 ( 0 . 2)3
y ,= o + — i- 1-2 3!
•1=0.2013
(Q.l)2 (O.l)3
*i = 1+ y 'û - 1-2 3!
•0=1.0050
( 0 . 2)2 (0 .2)3
?,= 1 + ^ - 0 - 2! 3!
•0=1.0200
0
x 0= 0 y ’o = 1 •
II Ayi = 0.0050
A ^ = 0.0150
Ayl = 0.0252
l/4 = 0.4107
H
II
TP
X z z' Az' AV
O
0H
z i= 0
II
Z 0 —1
Azi = 0.1002
Azl = 0.1014
A22 = 0.1032
10—2082
146 Ch. 1 Differential Equations
and similarly
y 4 = 0.3045+—j—-1-0452 0.0252 + —-0.1 -0.0102 = 0.4107
lt is obvious that the exact solutions of the System of équations (the solu
tions satislying the initial conditions) will be
y = sinhx, z = cosh*
And so, solutions correct to the fifth décimal place are
y A= sinh 0.4 = 0.41075, z4 = cosh 0.4 = 1.08107
Note. Since higher-order équations and Systems of équations in
many cases reduce to a System of first-order équations, the meth-
od given above is applicable to the solution of such problems.
Exercises on Chapter 1
Show that the indicated functions, which dépend on arbitrary constants,
satisfy the corresponding differential équations:
Functions Differential Equations
du 1
1. y = sin x — 1+ C e~sxnxf ■fc + y cos x = Y s in 2x.
2 . y = Cx + C — C2t
3. y2 = 2Cx + C2,
a2C
4. y* = Cx2-
1+C*
y=Cix+-£-{-c3, | Z à*y
5. dx3 ‘ x dx2
# y | 2 dy
dx2 ' x dx
Integrate the differential équations with variables separable:
9 . y dx—x d y = 0. Ans. y = Cx. 10. (1 -f u) vdu-\-(\ —u) udv = 0. Ans.
In uv + u —v = C. 11. (\ + y)dx — (1—x)dy = 0. Ans. (1 + y ) (1 —x) = C.
12. ( t * - x t t) ^ ~ - + x * + tx i = 0 . Ans. i ^ i + ln — = C. 13. ( y - a ) d x + x2dy=0.
Ans. ( y —a) = Cex . 14. z d t — (<* —at )dz = 0. Ans. zsa = C * ° . 15. —=
Solution. In two ways we calculate the volume of water that will flow out
during the time between the instants t and / + A/. Given a constant rate t»,
during 1 sec a cylinder of water with base 0.5 cm2 and altitude h flows out,
and during time At the outflow is the volume of water dv equal to
—dv= — 0 .h v d t = — 0.3 ÿ 'ï g h d f
On the other hand, due to the outflow, the height of the water receives
a négative “incrément” dh, and the differential of the volume of water out
flow is
- d v = nr'-dh= — (h + 0,7)* dft
Thus,
— (h+ 0 .7 )2dh = — 0.3 V 2gh dt
ü
whence
/^O.OSIStlO11/* —A,/‘) + 0.0732(10,/‘—hv «)+0.078( Y h)
Setting h = Q, we get the time of outflow T = 12.5sec.
37. The retarding action of friction on a disk rotating in a liquid is pro
port ional to the angular velocity of rotation cû. Find the dependence of this
angular velocity on the time if it is known that the disk begins rotating at
100 révolutions per minute and, after the elapse of one minute, rotâtes at
60 révolutions per minute. Ans. co= 100^-|-^ rpm.
38. Suppose that in a vertical column of air the pressure,at each level is
due to the pressure of the above-lying layers. Find the dependence of the pres
sure on the height if it is known that at sea level this pressure is 1 kg per cm2,
while at 500 m above sea level it is 0.92 kg per cm2.
Hint. Take advantage of the Boyle-Mariotte law, by virtue of which the
density of a gas is proportional to the pressure. The differential équation of
the problem is dp — — kpdht whence p = e~0000l7h. Ans. p = e “ 0 000l7ft.
Integrate the following homogeneous differential équations:
39. (y — x) dx + (y + x) dy = 0. Ans. y 2+ 2xy— x2 = C. 40. (x + y ) dx + x dy = 0 .
Ans. *2 + 2xy — C. 41. (x + y) d x + ( y — x) dy — Q. Ans. In (a:2 + y 2)l/*—
— a rc ta n -y = C . 42. x d y — y d x = yrx2 + y2dx. Ans. 1 + 2 C y—C2*2 = 0 .
43. (8y + 10*) dx + (5y + 7*) dy = 0 . Ans. (x + y)2 (2* + y f — C. 44. (2 }^st —s) d t+
l/jL C
+ t ds = 0. Ans. t e v * ==C or s = / l n 2 — . 45. (t — s) di + t ds = 0. Ans.
S t
— r _
t e * — C or s = / l n y . 46. xy2 dy = (x3 + y3) dx. Ans. y = x ^ /3 \n C x .
51. Détermine the curve whose subnormal is the arithmetic mean between
the abscissa and the ordinate. Ans. (x— y)2 (x + 2y) = C.
52. Détermine the curve in which the ratio of the y-intercept of the tangent
to the radins vector is equal to a constant. •
_ x <*y
Solution. By hypothesis, — m, w hence^— ) — *
53. Détermine the curve in which the ratio of the x-intercept of the normal
to the radius vector is equal to a constant.
x+ y
ày_
dx
Solution. It is given that =m, whence x2-f- y2 = m2 (x — C)2.
Y ^T y2
54. Détermine the curve in which the y-intercept of the tangent line is
equal to asec 0 , where 0 is the angle between the radius vector and the
x-axis.
Solution. Since ta n 0 = -|- and by hypothesis
y —x -r~ — o sec 0
9 dx
we obtain
dy V x i + yi
« x ïx = a x
whence
x r T +b - ( — +b
y= 2 \ e — 1e v
55. Détermine the curve for which the y-intercept of the normal drawn to
some point of the curve is equal to the distance of this point from the origin.
x
Solution. The y-intercept of the normal is y - 1— , ; therefore, by hypothesis,
y
we hâve
y+^r=
U
V x2+yt
whence
x2 = C(2y + C)
56. Find the shape of a mirror such that ail rays emerging from a single
point 0 are reflected parallel to the given direction.
Solution. For the x-axis we take the given direction, O is the origin. Let
OM be the incident ray, MP the reflected ray, and MQ the normal to the
desired curve:
a = p, OM = OQ, NM — y
NQ = NO + OQ= — x + V'x*+y* = yco tp = y-£-
whence _____
y d y = ( — x + Y x t + y !‘)dx
Intégrâting, we hâve
y2= C2+ 2Cx
Integrate the following linear differential équations:
150 Ch. 1 Differential Equations
+ C x Y 1 —x*. 60. -j^cos <+ ssin / = 1. Ans. s = sin / - f Ccosf. 61. ^ + scos / =
= -^- sin 2t. Ans. s = sin / — 1 + Ce “ Sln *. 62. y' — — y —e*xn. Ans. y —xn (e*-{-C).
& x
1
63. y '- 1---- */ = — . Ans. xny = ax~\-C. 64. y ’ + */ = — . Ans. exy — x + C.
X X / \\ &
” ln « '+ 7> - Î T ? = c - ™- ( ? + ¥ ) ^ - 2- ^ -
Ans. 7». x^ —
7 u- C . «0- * « + » * -
^ y d x _ x d y _ ' Ang je*-f.y*_ 2 arctan —= C .
x2 + i/2 * y
81. Détermine the curve that has the property that the product of the square
of the distance of any point of it from the origin into the x-intercept oï the
normal at this point is equal to the cube of the abscissa of this point. Ans.
y2 (2x2 + y2) = C.
82. Find the envelope of the following families of lines: (a)y = Cx + C2.
Ans. x2 + 4y = 0. (b) y = — -\-C2. Ans. 27x2 = 4y3. (c) —g^ = 2 . Ans. 27y — x3.
(d) C2x + C y - 1 = 0 . Ans. y2 + 4x = 0. (e) ( x - C )8 + ( y - C )2 = C2. Ans. x = 0 ,
y = 0. (f) (x—C)2 + y2 = 4C. Ans. y2 = 4x + 4. (g) (X- C ) 2+ ( y - C ) 2 = 4. Ans.
(x—y)2= 8. (h) Cx2 + C2y = 1. Ans. *4+ 4y = 0.
83. A straight line is in motion so thaï the sum of the segments it cuts off
on the axes is a constant a. Form the équation of the envelope of ail positions
of the straight line. Ans. x1/* + y,/f = a,/# (parabola).
84. Find the envelope of a family of straight lines on which the coordinate
2_ 2_ 2_
axes eut off a segment of constant length a. Ans. x 3 + y 3 = a 3 .
85. Find the envelope of a family of circles whose diameters are the doubled
ordinates of the parabola y2 = 2px. A n s.y2 — 2p
Exercises on Chapter 1 151
86. Find ihe envelope of a family of circles whose centres lie on the para-
bola y2 = 2px\ ail the circles of the family pass through the vertex of this
parabola. Ans. The cissoid x? + y2 (x + p ) = 0.
87. Find the envelope of a family of oircles whose diameters are
chords of the ellipse b2x2-\~a2y2 = a?b2 perpendicular to the x-axis. Ans.
—* + 4 = 1.
a2+ b2 ^ b 2
88. Find the evolute of the ellipse b2x2 + a2y2 = cPb2 as the envelope of its
2_
normals. Ans. (,ax)*!* + (b y = (a2— b2) 3 .
Integrate the following équations (Lagrange équations):
2C — p9
89. y = 2xy, + y ,i
Ans• * = ^ - 4 p- 3P
90. y = xy'2+ y '2. Ans. y = ( V^x+ 1 + C )2. Singular solution: y = 0.
91. y = x ( \ + y f) + (y')2. Ans. x = C e~ P -2 p + 2t y = C (p+ 1) e - P - ^ + 2
92. y = yyfZ+ 2xy'. Ans. 4Cx = 4C2— y2. 93. Find a curve with constant nor
mal. Ans. ( x - C ) 2+ y2 = a2. Singular solution: y = ± a .
Integrate the given Clairaut équations:
94. y = xy' y ’ — y ,%. Ans. y = Cx + C — C2. Singular solution: 4*/ = ( x + l) a.
95. y = xy‘ + Y 1 —y'*- Ans. y = Cx-{- V 1—C2. Singular solution: y2—x2= 1.
96. y = xy' + y'. Ans. y = C x+ C . 97. y = x y '+ — . Ans. y = C x + S i n g u l a r
1 J ^
solution: y2 = 4x. 98. y = x y '----- Ans. y = C x - r^ .. Singular solution:
y '2
, 27 a
v3= - 4 -* •
99. The area of a triangle formed by the tangent to the sought-for curve
and the coordinate axes is a constant. Find the curve. Ans. The équilatéral
hyperbola 4x y = ± a2. Also, any straight line of the family y = Cx ± aJ^C .
100. Find a curve such that the segment of its tangent between the coordi
aC
nate axes is of constant length a. Ans. y = Cx± . Singular solution:
1^1+c*
y t a = flV».
101. Find a curve the tangents to which form, on the axes, segments whose
2aC
sum is 2a. Ans. y = Cx — Singular solution: (y — x —2a)2 = Sax.
1-C*
102. Find curves for which the product of the distance of any tangent line
to two given points is constant. Ans. Ellipses and hyperbolas. (Orthogonal and
isogonal trajectories.)
103. Find the orthogonal trajectories of the family of curves y = axn. Ans.
x2+ ny2 = C.
104. Find the orthogonal trajectories of the family of parabolas y2 = 2p (x— a)
x
given straight line is the y-axis, then the équation of the trajectory will be
y = ± | - ( * + C I)T + C î .
It is required to: (1) verify that the given family of curves is indeed the
general solution; (2) find the intégral curve passing through the point ( 1, 2) if
the tangent at this point forms with the positive x-axis an angle of 45°. Ans.
m\ xm + n
Ant. y = * * l n * + l . 119.y<">=s«. Ans. y = + c ix" ~ 1+ • • • + c n - i* +
+ C „. 120. y" = a*y. Ant. a* = ln(ay + / a y ^ - C j + C , or (,=£><■* +
121. y ' = 4 - - A nt. (C1je + Ca)*= C 1yi!—a.
r
In Nos. 122-125 pick out a particular solution that satisfies the following
initial conditions: x = 0, y — — 1, y ' = 0.
122. x \f — y' = x2ex . Ans. y — e* (x — 1) + Particular solution:
y = ex (x — 1). 123. yy”— (y')2+ (y')s = 0. Ans. y + C1 ln y = x-\-C2. Particular
solution: y = — \. 124. y”--\-y' tan jt=sin 2jc. Ans. y —C2+ C l s \n x — x — ^-sin2*.
Particular solution: y = 2 s\n x — sinxcosjc—je— 1. 125. (y”)2-\-(yf)2= a2. Ans.
y= C 2—a cos (je-j-Ci). Particular solutions: y = a — 1—a cos je, y= a cos je—(û+1).
(Hint. Parametric form: y”= a c o st, y' = a s\n t.) 126. y”= — r . Ans. y =
y " ' = 0. Ans. General solution: y = C1eax + C2e” a* + C3 cos ax-{-CAsin ax. Parti-
cular solution: y« = cosax.
Integrate the following nonhomogeneous linear differential équations (find the
general solution):
148. y"—7y' + \2y = x. Ans. y = Cl(?* + Cae«* + 1 + 7- . 149. s”— a2s = t + 1.
163. y" + 6y’ + 5y = eîx . Ans. y = C1e -* + Cîe ~ * * + ^ é >*. 154. y”+ 9y = 6e3x.
Ans. y = C1 cos 3* + Ca sin 3x+-g-«s*. 156. y"—3 y '= 2 —6x. Ans. y = C 1+ C ïea* +
+ x 2. 156. yn— 2y'+3y ==e~x cos x. Ans. y = ex(A cos Ÿ 2 x + B sin y r2jc)+
e~x
-]———(5 cos x —4 sin x ). 157. y"-\-4y = 2 sin 2*. Ans. y = A sin 2x + B cos 2x—
164. Â load weighing 4 kg is suspended from a spring and increases the length
of the spring by 1 cm. Find the law of motion of the load if we assume that
the upper end of the spring performs harmonie oscillations under the law
y = sin V^lOO g t, where y is measured vertically.
Solution. Denoting by x the vertical coordinate of the load reckoned from the
position of rest, we hâve
4 d2x
where / is the length of the spring in the free state and £ = 400, as is évident
Exercises on Chapter 1 155
* = « cos( Sin ( l / — ^)
Whence (ellipse).
a mvo
166. A horizontal tube is in rotation about a vertical axis with constant
angular velocity co. A sphere inside the tube slides along it without friction.
Find the law of motion of the sphere if at the initial instant it lies on the axis
of rotation and has velocity v0 (along the tube).
dlr
Hint. The differential équation of motion is - ^ - = o ù2r. The initial data are:
d/ ‘ dt2 ~~ 1* -T V + H
dy i4/is. y = (C1+ C2^)e-^,
£ = * -y
175.
\ T x = - y - 3t-
4/is. ^ C ^ + Crf-2*
Î + - ®
176. z= —2 (C ^ 2*—C2<?-2*).
dz_
+ 4^ = 0.
dx
dAC
Ans. AC= C1e-* + Cae2*,
Tt = y + t
y = C3e~t + Cîe2tf
z — — (Cj + C3) e~ *+ C*z2*.
178. dy .
dt = x +*
dz
d i= x + y -
i4«s. z = Cgecix,
179.
y = x + c k e- c'x-
Ans. — = Clt z y * - Y *a= Cs.
180.
Exercises un Chapter / 157
j % = 2 x-3 y
191. Ans. Unstable.
\ 5 = B* + 6»-
'g = -4 x -1 0 ÿ
192. Ans. Stable.
g=12*+18y
193. Ans. Unstable.
= — Sx— \2y.
dt
194. Approximate the solution of the équation y*—y2+ x that satisfies the
initial condition y = 1 when * = 0 . Find the values of the solution for * equal
to 0.1, 0.2, 0.3, 0.4, 0.5. Ans. y x=0.6 = 2.235.
195. Approximate the value of y XsslmA of a solution of the équation y ’-f-
4 - ^ - y = ex that satisfies the initial conditions y = 1 when * = 1. Compare the
resuit obtained with the exact solution.
196. Find the approximate values of x t=1,4 and y t - lt4 of the solutions of
a System of équations -£ = y —x t ~~ = — x — 3y that satisfy the initial condi
tions * = 0, y = 1 when t = 1. Compare the values obtained with the exact
values.
CHAPTER 2
M U L T IP L E IN T E G R A L S
that is,
y ) d x d y = [ \ f ( x , y ) d x d y + \ \ f ( x , y)dxdy (3)
D D, D,
D D ,
2 f (P,) As, = 2 î (Pd As,-+ 2 / (Pi) As,-
D,
(4)
where the first sum contains ternis that correspond to the subdo
mains of Dx, the second, those corresponding to the subdomains
of D2. Indeed, since the double intégral does not dépend on the
manner of partition, we divide D so that the common bùundary
of the domains D1 and D2 is a boundary of the subdomains As,-.
Passing to the limit in (4) as As,-—>-0, we get (3). This theorem
is obviously true for any number of terms.
2.2 Calcutating Double Intégrais 161
* An interior point of a domain is one that does not lie on its boundary.
** We do not prove here that the function <D(x) is continuous.
11—2082
162 Ch. 2 Multiple Intégrais
Détermine the domain D. Here, D is tbe domain bounded by the Unes (Fig. 48)
f/ = 0, x-— 0, y = x'2, x—l
It may happen that the domain D is such that one of the func-
tions y = y l (x), */=<p2(x) cannot be represented by one analytic
W S f(x,y)dy\dx
a X cpt ( x) J
c /<fi(x) \ b /fp * ( x) \
= S( S f(x, y ) d y ) d x + U $ f(x,y)dyjdx
a ( x) y c \ ^ t (x )
C /<T* ( * ) \ b , y t ( x) \
= S( S f ( x , y ) à y ) d x - \ - K $ f(x,y)dy)dx
a \ i|> ( x) J c \ X( x) y
1d ~ I dx+ ^£>a (0
Proof. (a) Let the straight line x ~ c (a < c < b ) divide the
région D into two regular y-direction domains* Dx and D% , Then
b / c p f ( x)
c / <Pf U) \ à / ( ? t (x) • \
= $( S f(x, y ) d y ) d x + U J f(x, y ) d y ] d x = I Di + I
(*) (Pl (*) J
(b) Let the straight line y = h divide the domain D into two
regular «/-direction domains Dl and D2 as shown in Fig. 50.
Dénoté by M t and M2 the points of
intersection of the straight line y — h nlf y=<p*(x)
with the boundary L o i D. Dénoté the
abscissas of these points by a2 and bx.
The domain Dx is bounded by con
tinuons lines:
(1) «/ = <M*);
(2) the curve whose
équation we shall conditionally write
in the form
bt b ce
ÿ = <PÎ (*)
Fig. 50
having in view that qpï (*) = <p2(x)
when a a n d when and that
(f>ï( x) = h when a1^ . x ^.b^,
(3) by the straight lines x — a, x = b.
The domain D2 is bounded by the lines
* The fact that a part of the boundary of the domain Dx (and of D3) is a
portion of the vertical straight line does not stop this domain from being reg
ular in the ^-direction: for a domain to be regular, it is only necessary that
any vertical straight line passing through an interior point of the domain
should hâve no more than two common points with the boundary.
11 *
164 Ch. 2 Multiple Intégrais
since <pj(jc) = <ps (*) on [a, a j and on [&,, b], it follows that the
first and third intégrais are identically zéro. Therefore,
whence
ID= f ( P) S (5)
or
$$/(*. y ) d x d y = I D (3)
D
V=
]_ 35
3 8
= | [( Ÿ l + y f dy
=\{ V ï + Z
i + y V ~ y + £ \d y
3
2y2 , , 2ÿs
3 "l" 4 i ' 5 =S ^ + T
Note 2. Let a regular ^-direction domain D be bounded by the
Unes
*= *= y = c, y = d
and let (Fig. 55).
In this case, obviously,
<</ ’Mirt \
) \ f ( x , y ) d x d y = U J f {x, i f )dx)dy (8)
D c 'ij), (y) J
Every straight line parallel to the x-axis cuts the boundary of the domain
at no more than two points; hence, we can compute the intégral by formula (8),
setting
ti(y)=yi> ÿî(y)=y, 1
then
1/ y
Ç / (x, y) dx
0 \ÿ *
y_
Example 4. Evaluate J J e x ds if the domain D is a triangle bounded by
D
the straight lines y = x, y = 0, and x = l (Fig. 57).
Solution. Replace this double intégral by a twofold iterated intégral using
formula (4). [If we used formula (8), we would hâve to integrate the function
y_
e x with respect to x; but this intégral is not expressible in terms of elemen-
tary functions]:
over a domain D which lies between two squares with centre at the origin and
with sides parallel to the axes of coordinates, if each side of the inner square
Is equal to 2 and that of the outer square is 4 (Fig. 59).
Solution. D is irregular. However, the straight lines x — — 1 and x = \ di-
vide it into four regular subdomains Dlt Da, Da, D4. Therefore,
J J e*+y ds= JJ e*+y ds+ J J e*+y ds+ J J e * + y d s + J J ex+ yds
D D% D| D, d .
^ e x+yds= J £ J ex + y J a*+.V d# j d*
+ J £ J ex + y + J J ex + y dx
Note 4. From now on, when writing the twofold iterated intégral
s
r<Pi (X)
1d — \ f(x, y) dy \dx
a l.<Pi (x)
we will drop the brackets containing the inner intégral and will
Write
b q>, (x )
70 = $ S f(x, y)dydx
a <p, (jc)
V = $$/(*, y)ds
D
where D is (in Fig. 60) the hatched triangular région in the Xy*plane bounded
by the straight Unes jt = 0, y = 0, and x-{-y= 1. Putting the limits on the double
y= j | ( \ - x - y ) d y d K = i\) [ ( \ - x ) y - Ç y t ~Xdx = ^ ( l - x ) * d x = ^
oo o o
Thus, K= — cubic units.
o
Note 1. If a solid, the volume of which is being sought, is
bounded above by the surface z = Oi (x, y ) ~^ 0, and below by the
surface 2 = 0 , (x, y) ^ 0, and the domain D is the projection of
or
v = SS D
(*. y )—®i (*. y)] ds (i)
Further, it is easy to prove that formula (1) holds true not
only for the case where Ô, (x , y) and O, (x, y) are nonnegative, but
also where O, (x, y) and 0 2(x, y) are any continuous functions
that satisfy the relationship
« M * . y ) > ® A x , y)
Note 2. If in the domain D the function /(x, y) changes sign,
then we divide the domain into two parts: (1) the subdomain D,
174 Ch. 2 Multiple Intégrais
S = 2 1• As,-
i= 1
for any mode of partition. Passing to the limit on the right side
of the équation, we get
S=\\dxdy
If D is regular (see, for instance, Fig. 47), then the area will be
expressed by the iterated intégral
&[>,(*> 1
S= \ :=S S\ ddyy |l dx
o Lfl>i (*) J
Performing the intégration in the brackets, we obviously hâve
b
S = S [T»W—Vi(x)]dx
(cf. Sec. 12.1, Vol. I).
Example 2. Calculate the area of a région bounded by the curves
y = 2 — x2, y = x
Solution. Détermine the points of intersection of the given curves (Fig. 62).
At the point of intersection the ordinates are equal; that is,
x = 2—x2
whence
x2+ x — 2 = 0
x\ = — 2
x2= 1
We get two points of intersection: —2), Af2 (l, 1). Hence, the required
area is
2.5 The Double Intégral in Polar Coordinate* 175
* Note that in summing with respect to the index i this index will not run
thrpugh ail values from 1 to m, because not ail of the subdomains lying between
the rays 0 = 0# and 0 = 0ft + l belong to D.
2.5 The Double Intégral in Polar Coordinates 177
Vn- i [ S F ( 0 * , p / ) p ; a p,a0*J
where P(0*, p*) is a point of the subdomain As,*. Now take the
factor A0* outside the sign of the inner sum (this is permissible
since it is a common factor for ail the terms of this sum):
y „ = J ] [ Ç f ( 0 ! ,p ; ) p ; a P/] A0 *
Suppose that Ap,-—»0 and A0* remains constant. Then the ex
pression in the brackets will tend to the
intégral
®. (»*)
S F( q;, P) p dp
(eÂ)
Now, assuming that A0*—<-0, we finally
get **
P /<K,(0) \
V = U S F (0,p)prfp rf0 (3)
a Va»! (G) /
* We can consider the intégral sum in this form because the limit of the
sum does not dépend on the position of the point inside the subdomain.
** Our dérivation of formula (3) is not rigorous; in deriving this formula we
first let Api approach zéro, leaving A0* constant, and only then made A0*
approach zéro. This does not exactly correspond to the définition of a double
intégral, which we regard as the limit of an intégral sum as the diameters ofthe
subdomains approach zéro (i.e., in the simultaneous approach to zéro of A0*
and Ap|). However, though the proof lacks rigour, the resuit is true [i.e., for
mula (3) is true]. This formula could be rigorously derived by the method
uscd when considering the double intégral in rectangular coordinates. We also
note that this formula will be derived once again in Sec. 2.6 with different rea-
soning (as a particular case of the more general formula for transforming coor
dinates in a double intégral).
I2 2082
178 Ch. 2 Multiple Intégrais
Fig. 66
Conséquent 1y
2a /V 2ay-y9 \
^V = J ( J V 4a2 — x2— y2 d x j dy
0 ' 0 '
Transform the intégral obtained to the polar coordinates 0, p:
jc = p c o s 0 , y=psin0
Détermine the limits of intégration. To do so, Write the équation of the
given circle in polar coordinates; since
x2“f~y 2 = pa
y = p sin 0
it follows that
p2 — 2ap sin 0 = 0
p = 2a sin 0
Hence, in polar coordinates (Fig. 66), the boundaries of the domain are defined
by the équations
p = <D1 (0) = O, p = <D2 (0) = 2a sin 0, a = 0, P=y
= 9- o3 (3n—4)
J e-*dx
12 :
180 Ch. 2 M ultiple Intégrais
Now, if vve increase the radius R without bound (that is, if we expand
without limit the domain of intégration), we get the so-called improper iterated
intégral:
2ji / ®
e~ p* p dp dQ = lim d 0 = lim n (1 —e-R1) = n
S ( S * " p,pdp R->cc
0 \a
We shall show that the intégral J J é~x% y* dxdy approaches the limit jt if
D
a domain D' of arbitrary form expands in such manner that finally any point
of the plane is in D' and remains there (we shall conditionally indicate such
an expansion of D' by the relationship D' —►<»).
As a particular instance, let D' be a square with side 2a and centre at the
origin; then
a a
J J e~xl~yt d x d y = J J e~x*~y%dxdy
a a
Now take the factor e~y* outsidè the sign of the inner intégral (this is per-
missible since e~y% does not dépend on the variable of intégration*). Then
J ^ e ~ x*~y*dx dy — J e~y*( ^ e ~ x t d x jd y
D' -a \~ a J
a
Set J e~xi dx = Ba. This is a constant (dépendent only on a)\ therefore,
-a
a a
S 5 e~X*~ÿt dxdy = J e~y%Ba dy = Ba J e~y*dy
D' -fl -fl
= J e~yt dy j ; thus,
J J «-*•-** dxdy = BaBa = B l
We pass to the limit In this équation, by making û âpproach ifîfinity (in the
process, D' expands without limit):
r a "|2 r+ » -i2
lim C [ e~x*~y* d x d y = lim B\ = lim = f £ -* 2d * |
D’-+<a a —►co a -> « *1 I JI
D L-fl J L-® J
Æ ss D'
dxdy = n
Hence,
0b
^ e~x*dx =n
or
J e ~x2d x = V n
d<p dq>
du dv
di|? dif Au Av
du dv
Here, the outer vertical lines indicate that the absolute value of
the déterminant is taken. We introduce the notation
dtp d(p
du dv
cty d\|) = /
du dv
Thus,
As » | / | As' (4)
The déterminant / is called the futtctional déterminant of the
functions <p(u, v) and ip(u, v). It is also called the Jaçobian after
the German mathematician Jacobi.
Equation (4) is only approximate, beçause in the process
of computing the area of As we neglected infinitesimals of higher
order. However, the smaller the dimensions of the subdomains As
and As', the more exact will this équation be. And it becomes
absolutely exact in the limit, when the diameters of the subdo
mains As and As' approach zéro:
As
|/ |= lim As7
diam As'-»0
^ (y — x)dxdy
D
1 7 1
y= x + l, ÿ = x —3, y = —— x + 5
2 Ao,-
i= î
We shall call the limit o of this sum, when the greatest of the
diameters of the subdomains Ao,- approaches zéro, the area of the
surface; that is, by définition we set
n
a= lim V Ao, ( 2)
diam Aa - ->-0 t-_ j
188 Ch. 2 Multiple Intégrais
Now let us calculate the area of the surface. Dénoté by y,- the
angle between the tangent plane and the xy-plane. Using a fami-
liar formula of analytic geometry we can write (Fig. 75)
ASi = Aa, cos Y;
or
As;
A<J/ COS y; (3)
The angle y,- is at the same time the angle between the z-axis
and the perpendicular to the plane ( 1). Therefore, by équation
( 1) and the formula of analytic geometry we hâve
l
COS Y; =
V l + f f i l l . r]i ) + f y ( l i , Tl/)
Hence,
A*; = V \ T]d + fyH ln %) AS;
Putting this expression into formula (2 ), we get
Since the limit of the intégral sum on the right side of the last équation
is, by définition, the double intégral J J j / ”1
D
we finally get
a=lf D
V i + {%) + ( 0 ) dxdy (4)
This is the formula used to compute the area of the surface
z = f(x>y)-
If the équation of the surface is given in the form
x = n(y, z) or in the form y = %{x, z)
then the corresponding formulas for calculating the surfacé area
are of the form
°=JT D'
+{m)îdydz
a = if V ^ 1 + (SD + { tz) dxdz (3")
D"
where D' and D" are the domains in the xy-plane and the xz-plane
in which the givep surface is projected.
2.7 Computing the Area of a Surface 189
ÔZ _____________ X _____________
dx Y R * - x 2- y 2
dz_= _________ y
dy~ Y R i —x2—y 2
Hence,
Y l+ { ^ ) + {P y) ~ Y W - x ï - Ÿ = VT2
The domain of intégration is defined by thç condition
x2-\-y2<sR2
Thus, by formula (4) we will hâve
p [ Vr*~X*
_L o = R
2 I I
-R \ - V Td
Y R1—*1— y2
dx
Example 2. Find the area of that part of the surface of the cylinder
x2-\-y2= a2
which is eut out by the cylinder
X2+ Z 2 = ü2
Solution. Fig. 77 shows 1/ 8 th of the desired surface. The équation of the
surface has the form y = \ r a2—x2\ therefore,
ày__ x
dx y a2—x2 ’ dz
x2
The domain of intégration is a quarter of the circle, that is, it is determined
by the conditions
x2 + z2< t a2, jc^o » 2 ^ 0
Consequently,
a / V a*-x* Va* - x*
J ô
a = 8aa
2 f (Pi) As,
M = lim S f ( P ,) A s ,= S S /( P ) d s = J S /( x t y)rf*dy (2 )
Asf -*0 i= 1 D D
I=
2 tntf
i=1
Let us détermine the moment of inertia of a material plane
figure D.
Let D be located in an jn/-coordinate plane. Let us détermine
the moment of inertia of this figure relative to the origin, assuming
that the surface density is everywhere equal
to unity.
Divide the domain D into elementary
subdomains 1, 2, . . . , n) (Fig. 78).
Im each subdomain take a point P( with
coordinates rj,-. Let us call the pro-
duct of the mass of the subdomain AS,*
by the square of the distance rï = tî + T]î
an elementary moment of inertia A/, of
Fig. 78 the subdomain AS,-:
A//=(g? + »tf) AS,
and let us form the sum of such moments:
J j ( « + n » AS,-
This is the intégral sum of the function f (x, ÿ) = xiJr yi over the
domain D.
We define the moment of inertia of the figure D as the limit
of this sum when the diameter of each elementary subdomain AS,
approaches zéro:
/„ = Hm £ ( a + ttf)A S,
diam AS . - + Q i - 1
But the limit of this sum is the double intégral ^ (x2+ y2)dxdy.
D
Thus, the moment of inertia of the figure D relative to the origin is
/ o = H t xt + tf)àxdV ( 1)
D
/ # = S$ (* * + il*)dxdy
D
Solution.
fVT^x
/W = [( | y x * d y ) d x = { x^ ~ d* = Ÿ f x2 ( l - x ) d x = —
o\ o / o o 0
Ellipse of inertia. Let us détermine the moment of inertia of
the area of a plane figure D relative to some axis OL that passes
I.) -2082
194 Ch. 2 Multiple Intégrais
/= r*dxdy = (xsinq)—ycos<p)*dxdy
D D
Therefore
I = Iyy sin 2 <p— 2 Ixy sin qpcos cp+ Ixx cos2 «P (4)
here, I vv = J J x2 dxdy is the moment of inertia of the figure relative
D
to the y-axis, I * * = $ $ ÿ 2 dxdy is the moment of inertia relative
D
to the x-axis, and Ixy= \ \ xydxdy. Dividing ail terms of the last
D
équation by I, we get
___î y __
1= L / E 2/ / cos f \ _L / /sinjpX2
\ v t ) ^, x v K v r ) { r i ) + vv W t ) (5)
On the straight line OL take a point A (X , Y) such that
1
OA
V T
2
i=\
É/AS, 2 T),AS,
n * yc
2 A5<- 2 AS<
-
f=i ï =i
In the limit, as AS, —>-0, the intégral sums in the numerators
and denominators of the fractions will pass into double intégrais,
and we will obtain exact formulas for computing the coordinates
of the centre of gravity of a plane figure:
^xdxdy ^ydxdy
xc = T c-------* (2)
SSdxdy
D
SS<fc4y
D
These formulas, which hâve been derived for a plane figure with
2.11 Triple Intégrait 197
surface density 1, obviously hold true also for a figure with any
other density y constant at ail points.
If, however, the surface density is variable,
y = y (* . y) '
then the correspondis formulas will hâve the form
^ y(*> y ) x d x d y JSy(*. y )y dxdy
y __Ç___________ J D ______________
XC ff * yc
y (*. y)**dy SSyc*. v)*1* dy
D
+ fr2
assuming that the surface density at ail points is equal to 1.
Solution. By formulas (2) we hâve
or
y< *)dxdydz (2 )
V V
Let the surface bounding V below hâve the équation z = %(x, y),
and the surface bounding this domain above, the équation z = \|>(x, y)
(Fig. 84).
We introduce the concept of a threefold iterated intégral I v,
over the domain F, of a function of three variables f(x, y, z)
defined and continuous in V. Suppose that the domain D is the
projection of the domain V onto the xz/-plane bounded by the curves
y = <Pi(x)> y=--V%(x), x = a, x=b
Then a threefold iterated intégral of the function / (x, y, z) over
V is defined as follows:
>>r<p. (•*) (*• v) 'v1
r= S S S /(*. «/• z ) d z \d y \ dx (D
a L<P. (X) VX (X. y) J J
We note that as a resuit of intégration with respect to z and
substitution of limits in the braces (inner brackets) we get a func
tion of x and y. We then compute the double intégral of this
function over the domain D as has already been done.
The following is an example of the évaluation of a three
fold iterated intégral.
200 Ch. 2 Multiple Intégrais
D L 0
S **** JU
Setting up the limits in the twofold iterated in
tégral over the domain D, we obtaln
-ÎÜU ... d ,Y
= 1 { Ï - * » ( i - * - » ) i ^ | * = | è ( 1- x)*d*= 735
h = h t+ h, + •••+ Ivn
Property 2 (Theorem on the évaluation of a threefold iterated
Intégral). If m and M are, respectiwly, the smallest and largest
values of the function f (x, y, z) in the domain V, we hâve the
inequality
m V ^Iv^M V
= y )— %(x, y)] do
The proof of this property is carried out in the same way as that
for a twofold iterated intégral [see Sec. 2.2, Property 3, formula (4)].
We can now prove the theorem for evaluating a triple intégral.
Theorem. The triple intégral of a function f (x, y, z) over a
regular domain V is equal to a threefold iterated intégral over the
same domain; that is,
b (<ft (*) r 'J1(j. y)
SSS/(*. y, *)à» = \ | J J f{x, y, z)dz dy\ dx
V a 1<P, ( x) l X (*• y)
202 Ch. 2 Multiple Intégrais
/
JC2 |/2
1 —^2 —^2 ’ and above by the
surface z = c - £ - £ • T hepr°-
jection of this ellipsoid on the jct/-plane
y2 *#2
(domain D) is an ellipse, 1.
Hence, reducing the computation of vo
lume to that of a threefold iterated
intégral, we obtain 1
Fig. 86
a bV i - £ / cV
= ç
j
i - £ -
dz dy dx
- a
- 6
V \-y
a r ‘ / - 7
= 2c
I
- a
b2 dy dx
. - 6 / » - ! -
When computing the inside intégral, x is held constant. Make the substitution:
y = b ]/~ ï^ ~ sin t. dy = b ] / l - - ^ - c o s / c f t
-n
2
dx
I / ( ,--à -H ,- 5 ) sinî<b 1~S’c°s1 dt
1
JL
2
je2) djc=
- “ i o - £ ) i “■*'dt
204 Ch. 2 Multiple Intégrais
Hence,
V = -|- nabc
v=î rf
Fig. 87 Fig. 88
plane—with the plus sign if the point lies above the xy-plane,
and with the minus sign if below the xy-plane (Fig. 87).
In this case, we divide the given three-dimensional domain V
into elementary volumes by the coordinate surfaces 0 = 0,-, p = P/,
z = zk (half-planes adjoining the z-axis, circular cylinders whose
axis coincides with the z-axis, planes perpendicular to the z-axis).
The curvilinear “prism” shown in Fig. 88 is a volume element.
The base area of this prism is equal, to within infinitesimals of
higher order, to pA0Ap, the altitude is Az (to simplify notation
we drop the indices i, /, k). Thus, Ai>= pA0 ApAz. Hence, the
triple intégral of the function F ( 0 , p, z) over the domain V has
the form
/ = $ $ $ F ( 0 , p, z)pdQdpdz ( 1)
v
The limits of intégration are determined by the shape of the
domain V.
2.13 Change of Variables in a Triple Intégral 205
-S Lo
îï 0
P dp
0 L0 J
de
k_ k_R* knR*
2 2 4 4
Fig. 91
Then the problem reduces to computing the moment of inertia of the cy
linder relative to the *-axis:
Ixx —S
j SS (y2+ z 2) Yodxdydz
Changing to cylindrical coordinates, we ôbtain
2ji /R r h -i \
t= Yo S ! S S ^ + e 231"40) ^ p *p
231 (R \
= Yo j ^ j^ ^ - + 2/ip2 sin2 eJ pdp | d0 = Yo ^ j
0 10 o
[2h9R* 2hR*
= Yo[
*Yod xdydz
2c =
J J $ Yodx d y d z
2n
e\ 1 \ r cos 9 /® sin cp dr \ j dy
d9 2n ^ -l
0 _o \o / 4 2 _ 3
*c= -
" 4 D3 8
CM
-6 nR
*---- S 0
\ ( \ r2 sin 9 dr J d9 de
e
_o \o /
Obviously, by virtue of the symmetry of the hemisphere, x c —y c = 0.
r !dx = — (a > 0)
fl Lo
a Ln J a
Changing the order of intégration in the first intégral, we rewrite this équation
in the following form:
b
| [ î ‘- “ + = ,4
oP
d x = ln —
x a
Exercises on Chapter 2 211
Exercises on Chapter 2
Evaluate the intégrais: '
1 2 2 jc K 3
1 \ / y a V 2 a y - y* a a
J J f ( x t y)dxdy. 15. J J f(x, y)dxdy. Ans. J J f ( x t y)dydx.
o y• o o 0 a — Va* - j c *
N L
* If the intégral is written as J J /( * . y )dx dy then, as has already been
M K
stated, we can consider that the first intégration is performed with respect to
the variable whose differential occupies the first place; that is,
N L N /L \
$ J / (JC, y ) d x d y = J ( J /(jc , y ) tdx )dy
MK M \K
14 *
212 Ch. 2 Multiple Intégrais
1 V\-x* lV\-y•
1«. ^ ^ /(*. y) dy dx. Ans. J J f (x, y)dxdy.
-i o o -VT^ÿi
i l - y o V i - x ‘ 1 l- x
2xy
30. Compute the area of a loop of the curve F + irH
Hint. Change to new variables x = pacos0 and i/ = p6sin0. Ans.
Calculating Volumes
31. Compute the volumes of solids bounded by the following surfaces:
abc
—+ - £ + —= 1 , j c = 0, y = 0, 2 = 0. Ans
a 1b 1 c "F 32. z = 0, x2-\-y2= 1, x + y -f
+ z = 3. Ans. 3jt. 33. (x— 1)2+ (y — 1)2= 1, xy = z, 2 = 0. Ans. n. 34. x2-f- y2 —
32
— 2a* = 0, 2=0, x2+ y 2 = z2. Ans. — a3. 35. y = x 2, x = y 2, z = 0, z = \ 2 + y — x 2.
A 5 4 9
AnS' Ï44-
36. Compute the volumes of solids bounded by the coordinate planes, the
plane 2x + 3y — 12 = 0 and the cylinder z = ^ y 2. Ans. 16.
37. Compute the volumes of solids bounded by a circular cylinder of radius a,
whose axis coïncides with the z-axis, the coordinate planes and the plane
a 1 a \ 4 3J
38. Compute the volumes of solids bounded by the cylinders x2-\-y2 = a2t
x2-j-z2 = a2. Ans. j û 3. 39. y 2~l~z2 = x, x = y f z = 0. Ans. 40. x2-j-y2 -j-z2=
—a2, x2-\-y2 = R 2, a > R. Ans. [a3 — ( Y a2— R2)3]. 41. az = x2-\- y2,
ü
3
z = 0, x2-{-y2 = 2ax. Ans. — Jia3. 42. p 2 = a2 cos20, x 2+ y 2+ z2 = a2, 2 = 0.
(Compute the volume that is interior with respect to the cylinder.)
Ans. i - a 3 (3Ji + 20— 16 V~2).
Calculating Surface Areas
43. Compute the area of that part of the surface of the cone x2 + y 2~ z 2
which is eut out by the cylinder x2-\-y2 = 2ax. Ans. 2na2 V ' 2.
44. Compute the area of that part of the plane x + y + z = 2a which lies in
the first octant and is bounded by the cylinder x2-{-y2 = a2. Ans. Y 3.
45. Compute the surface area of a spherical segment (minor)if the radius of
the sphere is a and the radius of the base of the segment is b.
Ans. 2ji (û2— a Y a2— b2).
46. Find the area of that part of the surface of the sphere *2 y 2 + *a = û2
x2 y2
which is eut out by the surface of the cylinder = 1 (a > b).
50. Compute the area of that part of the surface of the paraboloid y 2 + z2 = 2ax
which lies between the parabolic cylinder y 2 = ax and the plane x = a.
Ans. y J i a 2 (3 V U — l).
64. Compute the moment of inertia of the area of a figure, bounded by the
parabola y2= ax and the straight line x = at relative to the straight line y = —a.
Ans. -=- a4,
o
Triple Intégrais
dx dy dz
65. Compute ^ ^ § if the domain of intégration is bounded
(x + y + z + \ f
by the coordinate planes and the plane x-\-y-\-z = 1. Ans.
0
~VT and J ~ Y T - Ans■ h '
0
V T-
* In Problems 68 , 69 and 71 to 73 we consider the density constant and
equal to unity.
CHAPTER 3
L I N E IN T E G R A L S A N D S U R FA C E IN T E G R A L S
Az - 0
k
The letter L under the intégral sign indicates that the intégration
is performed along the curve L.
We note two properties of a line intégral.
P ro p e rty 1 . A line intégral is determined by the element of in
tégration, the form of the curve of intégration, and the sense of
intégration.
* Here, the limit of the intégral sum is to be understood in the same sense
as ïn the case of the déftnite intégral, see Sec. 11.2 (Vol. I).
218 Ch. 3 Line Intégrais and Surface Intégrais
J Xd x + Y dy— J X dx + Yd y + J X d x + Yd y
(M ) (M) (K)
where *, and y; are the coordinates of some point lying on the arc
ASj. These limits do not dépend on the way the arc L is divided into
subarcs As(-, provided that As,- 0, and do not dépend on the choice
of the point Af,•(*,•, yf) on the subarc As,-; they are called line
220 Ch. 3 Line Intégrais and Surface Intégrais
Note. Froni this theorem it follows that the sums defined in the
preceding section, where the points yt) are the extremities
of the subarc As, and the manner of partition of the arc L into
subarcs Ast is arbitrary, approach the same lim it—the line intégral.
This theorem makes it possible to develop a method for com
puting a line intégral.
Thus, by définition, we hâve
( N) n _
5 X (x, y ) d x = lim V X (xh £/,) Axt (3)
(M) = I
where
A x i = X i— x ^ 1 = <p ( / , ) — q> (/«-_,)
Transform this latter différence by the Lagrange formula
Axt = <p(/,)—q>(ti- 1) = <p' (x,) ( t i - t i - J = <p' (x,) At,
where x(- is some value of t that lies between the values and t(.
Since the point xit y( on the subarc As,- may be chosen at plea-
sure, we shall choose it so that its coordinates correspond to the
value of the parameter x,-:
*/ = <P(*,). ÿ/ = ^(x,)
Substituting into (3) the values of xit yt and Axt that we hâve
found, we get
(ao n
S X (x, y)dx = lim £X [«p (X,.)i|>(x,)]q>' (x,) Att
(Al) 4<,--0 i= l
On the right is the limit of the intégral sum for the continuous
function of a single variable X[q> (/), i|) (*)] <P’ ( 0 on the interval [a, P].
Hence, this limit is equal to the definite intégral of the function:
(N) fi
$ X (x, y) dx r =^ X [<p (0, (0] <!>' (0 dt
( M) a
Fig. 96
Solution. To find the parametric équations of the line MN, along which the
intégration is to be performed, we Write the équation of the straight line that
passes through the given two points:
3 2 1
and dénoté ail these ratios by a single letter t\ then we get the équations of
the straight line in parametric form:
x = 3t, y = 2 t, 2 — t
Here, obviously, to the origin of the segment MM corresponds the value of tne
parameter t = 1, and to the terminus of the segment, the value / = 0 . The déri
vatives of x , y , z with respect to the parameter t (which will be needed for
cvaluating the line intégral) are easily foundf
= 3» = 2 , z^ = \
222 Ch. 3 Line Intégrais and Surface Intégrais
Now the desired line intégral may be computed by formula (4):
{N) o
$ je3 d x + 3z(/® d y - x * y d z= J [(3/)3 -3 + 3 / (2/)2 - 2 - ( 3 0 2 -2MM<
(Àf) 1
1
Example 2. Evaluate the line intégral of a pair of fonctions, 6x2y, and lOx^2,
along a plane curve y = jc3 from the point Af (1, 1) tothe point N (2, 8) (Fig. 97).
Solution. To compute the required intégral
(N)
Ç 6x2y dx+lOxy2 dy
(M )
we must hâve the parametric équations of the given curve. However, the exp-
licitly defined équation of the curve y — x 3 is a spécial case of the parametric
équation: here, the abscissa x of the point of the curve serves as the parameter,
and the parametric équations of the curve are
x = x, y = x 3
The parameter x varies from x x = \ to x2 = 2. The dérivatives with respect to
the parameter are readily evaluated:
3*s
Hence.
(AT) 2 2
^ 6x*ydx+l0xyi dy= J [6jc2jc* - 1 -f I0xxa-3x*]dx= Ç (6xs + 30x») dx
(M) 1 1
= [jc*+3 jc10]| = 3132
Formula (7) holds for each of these domains. Adding the left and
right sides, we get (on the left) the area of the given domain, on
the right, a line intégral taken over the entire boundary with
coefficient 1/ 2 since the line intégral over
the division line /* is taken twice: in the
direct and reverse senses; hence, it is equal
to zéro.
2. C o m p u tin g th e w ork of a v a ria b le
force F on sotne cu rv ed p ath L. As was
shown at the beginning of Sec. 3.1, the
work done by a force F — X(x, y, z)i +
+ Y (x, y, z ) j + Z ( x , y, z)k along a line
L = MN is equal to the line intégral
D a ÿt(x) a y t ( x)
b
=
5 [X{x, y%(x))— X (x, yl (x))] dx (1)
a
We note that the intégral
6
S * (x , »t (x))dx
a
Thus
b
J X(x, yl (x))dx= J X(x, y)dx (2)
a MPN
Substituting expressions (2) and (3) into formula (1), \ve obtain
But
J X (x, y)dx = — J X(x, y) dx
MQN NQM
(see Sec. 3.1, Property 1). And so formula (4) may be written thus:
i ï w
O
dxdy=1 I *(*» y ) d x + J
M PN NQM
x ( x > y ) dx
But the sum of the line intégrais on the right is equal to the line
intégral taken along the entire closed curve L in the clockwise
direction. Hence, the last équation can be reduced to the form
^^-dxdy= J X(x,y)dx (5)
D L (In the clockwise sense)
If part of the boundary is the segment / 3 parallel to the t/-axis,
then ^ X (x, y) dx = 0, and équation (5) holds true in this case as
u
well.
Analogously, we find
^™ -d xd y = - J Y(x,y)dy (6 )
D L (In the clockwise sense)
N { w ~ % r ) dxdy= S Xd x + Ydy
D L (in the clockwise sense)
3.4 Conditions for a Line Intégral to Be Independent of ihe Pat h 227
D L •
n l
3.4 Conditions for a Line Intégral to Be Independent of the Pat h 229
We can consider the line intégral along the arc of the space curve
* = < P (0 . = z = X (0 :
Example. Find the coordinates of the centre of gravity of one turn of the
hélix
x = ac os t , y = asint, z — bt (0^t<2:i)
if its linear density is constant.
232 Ch. 3 Line Intégrais and Surface Intégrais
s
J a cos /
0
2ji
+ b2 d/
2rt
$ Và1+ b*
Similarly, t/c = 0,
2jx _______________
^ bt Y a2 sin21 + a 2 cos2 t + b2dt
b- 4ji2 -nb
*C 2jiVra2 + *2 2-2:1
Thus, the coordinates of the centre of gravity of one turn of the hélix are
*c = 0, J/c = 0. zc = nb
J J Z cos (n, z) do
a
Let the surface o be such that any straight line parallel to the
z.-axis cuts it in one point. Then the équation of the surface may
be written in the form
z = f ( x , y)
But the différence between the intégrais on the right in the last formula
yields a volume bounded by the surface o. This means that the volume of the
solid bounded by the closed surface a is equal to the following intégral over
the surface
S S k ^ r nda^ w S S r n d a
o a
But the last intégral is equal to the area of the surface o. Indeed, b y the
définition of an intégral (noting that r n = - l ) , we obtain
« rndo=
Ao* -
lim V /v t* A o * =
"
lim
boh-»o **
VAcf* —o
ke __ ke
Hence, the flux is 4n^* = 4nA;e.
~ ÿ d x i y
or
|^-cos (n, z) = cos (n, y)
Hence,
observer looks from the end of the normal, he sees the circulation
along the curve X as being counterclockwise.
Formula (8 ) holds true for any surface if this surface can be
divided into parts whose équations hâve the form z = f(x, y).
Similarly, we can write the formulas
y, z)dy = J J [ — -^-cos(n, x) + -^ c o s (/i, z)jda (8 ')
k o
Adding the left and right sides of (8 ), (8 '), and (8 "), we get the
formula
^ X d x + Ydy + Zdz = J J [ ( | ~ — 1 cos(n, x)
k a
then the line intégral along any closed space curve X is zéro:
J X dx + Y dy + Zdz = 0 01 )
K
240 Ch. 3 Line Intégrais and Surface Intégrais
J X d x + Y d y + Zd z = J du = u(N) — u(M)
( M) ( M)
d ( ^ m v tyj = X d x - \ - Y d y + Zdz
Take the intégral along the trajectory connecting the points M i and Aft :
(M.)
I 2 1 2 f X d x + Y d y + Zdz
T m v 2 ~ m v l = j
(Mt)
where vx and v2 are the velocities at the points Afj and M a.
This last équation expresses the work-kinetic energy theorem: the increase in
kinetic energy when passing from one point to another is equal to the work of
the force acting on the mass m.
Example 2 . Détermine the work of the force of Newtonian attraction to a
fixed centre of mass m in the translation of unit mass from M l (alt bl9 cx) to
M* c,).
3.8 Ostrogradtky'i Formula 241
X=-km -L — , K = -4 m — - , Z = -k m -\--
r2 r r2 r r2 r
Then the work of the force F over the path
AfjAij is
<Mg)
x dx + y dy -|- z dz
r8
(Mx)
<Mf) (Mt)
= — km f L^L — km f d
(M.) r (Mj)
(since r2= x* + y2 + z2, it follows that Fig. 108
r dr = x d x + ydy~\~zdz). If we dénoté by r1
and r a the lengths of the radius vectors of the points and Af2» then
A- b " ( - h ~ 7 ; )
Thus, here again the line intégral does not dépend on the shape of the
curve of intégration, but only on the position of the initial and terminal
■* km
points. The fùnction u = — is called the potential of the gravitational field
generated by the mass m. In the given case,
_du v du 7 du
A = u (M2) — u (M i )
~~d~x' dy' dz '
that is, the work done in moving unit mass is equal to the différence between
the values of the potential at the terminal and initial points.
dZ{Xÿ — dxdy.dz
V
l ^ l i ï d x d y d z ^ § § Z (X ’ y ' 2) c o s ("> 2) d a
J V a
^ ^ ^ ■ d x d y d z = ^ X ( x , y, z)cos(n, x)do,
^V xJÜ
I(>
244 Ch. 3 Line Intégrais and Surface Intégrais
v u = { l Tx+J Ty + k h ) u (2')
and to consider the symbol
T M , K + 4 + *£)<,Jt+'/K+*z>
- s x + i ; r + dz
d dz
(see Sec. 3.8). Thus,
y F = d iv F (5)
(3) Form the vector product of the symbolic vector y by the vector
F = t X + j Y + kZ:
Then find
div (grad «) = ! ( ! ) + £ ( % ) + ê { l i ï )
or
dj 1•v(grad
/ i \
„ ) =d2u
_ +. ^d2*/5. +, _d2a /ftV
(9)
The right member of this expression is denoted by
* d2u , d2u , d2u
( 10)
or, symbolically,
A / a2 . a2 . d2 \
A u ~ { dx* + dy2 + dz2 ) “
The symbol
Exercises on Chapter 3
Compute the following line intégrais:
1. J y 2 dx + 2xy dy over the circle x — a c o s t %y = a s in t. Ans. 0 .
2 . J y d x —x d y over an arc of the ellipse jt = a c o s /, y = b s in t.
Ans. —2ji ab.
3' Î { x t + V*dX x2+ y2<iy) over a circle vvith centre at the origin. Ans. 0 .
Surface Intégrais
22. Prove that cos (nt z)da = 0 if o is a closed surface and n fs a
normal to it.
23. Find the moment of inertia of the surface of a segment of a sphere with
équation x2+ y2+ z2= R 2 eut off by the plane z=^H relative to the z-axis.
Ans. ^ ( 2 Æ » - 3 W /+ H9).
24. Find the moment of inertia of the surface of the paraboloid of révolution
x2+ y 2= 2cz eut off by the plane z — c relative to the z-axis. Ans. 4jic4 ^ ■.
lo
25. Compute the coordinates of the centre of gravity of a part of the surface
of the cône x2+ y2= j ^ z 2 eut off by the plane z = H . Ans. 0 , 0, - |- //.
26. Compute the coordinates of the centre of gravity of a segment of the
surface of the sphere x2+ y2+ z2= R 2 eut off by the plane z = //.
Ans. (o . 0, •
Exercise* on Chapter S 249
27. Find JJ [x cos (/fx) + y cos (ny) + z cos (nz)\ do, where o is a closed
a
surface. Ans. 3 F , where F is the volume of the solid bounded by the surface o.
28. Find J J z d x d y , where 5 is the external side of the sphere x2 + j /2 +
s
+ z2^=R2. Ans. y ji/? 3.
29. Find J J x2 dy dz + y dz dx + z dx dy,
2 2 where 5 is the external side of
s
the surface of the sphere x2-\-y2+ z2 — R2. Ans. 2jiRa.
30. Find J J Ÿ 'x 3+ y2ds, where 5 is the latéral surface of the cône
s
x* , y2 z2 n n L 2jia 2 V a2-f- b2
0 < Z< 6 . Ans. -------- 3--------- .
31. Using the Stokes formula, transform the intégral J y d x + td y + x d z .
L
IjJ (S+^+Sr)
Using the Ostrogradsky formula compute the following intégrais:
38. J J (x c o sa + f c o s P + z c o s T ) * . where S is the surface of the ellipsofd
s
Î j + I j + ^ " = ^• Ans. 4nabc. 39. (x3 cos a + y3 cos P + 23 cos y) ds, where S
• 2^
-f y 1 dz dx-\ z2 dx dy, where 5 is the surface of the cône ^ ^ =0
Jlû262 p (*
(0 «^z^<6 ). 4ns. —^— .4 1 . \ \ x dy dz + y dxdz + zd xd y, where S is the
S
surface of the cylinder jc2 + t/2 = a2, —H ^ z ^ H . Ans. 3na2H. 42. Prove the
identity J J d*dy = J g “ ^s, where C is a contour bounding the
D c
domain £>, and is the directional dérivative of the outer normal.
dn
Solution.
Setting X ; Y == , we get
IIIV l”to -u à v )d x d y d * = ^
a
(o|-« ^do
Z= Vu'z — UVg
Exercises on Chapter 3 251
Then
^ - + g - + ^ - = V (u x x + U y y + u ’z z ) — U (v x x + V yy + v"z z ) = v A ll — U AV
u A d*u , , d2u
where = ,
Solution. In Green’s formula, which was derived in the preceding problem,
put u = l . Then Au = 0, and we get the desired identity.
45. If u(x, y, 2) is a harmonie function in some domain, that is, a function
which at every point of the domain satisfies the Laplace équation
d2u . d2u . ôflu
then
ws*-
a
where o is a closed surface.
Solution. This follows directly from the formula of Problem 44.
46. Let u (x , y, z) be a harmonie function in some domain V and let there
be, in Vf a sphere. a with centre at the point M (xlt y lt zt ) and with radius R .
Prove that
« f e . »x, z1) = ^ ¥ ^ u d a
o+a
252 Ch. 3 Line Intégrais and Surface Intégrais
1 du
\T fa ~ ~ u
lz)dn • t o - j j i r Ê ï - '
lz
dn
do = 0
à
Hence,
- $“ *+jâ[ u d a - [
but
dn dr
Therefore,
+
S S a ^ da- S S u ^ d a = 0
2 ô
(O
2 Ô
Apply the mean-value theorem to the intégral on the left:
( 2)
where u (g, t|, £) is a point on the surface of a sphere of radius p with centre
at the point M (x lt y u 2 ^ .
We make p approach zéro; then w (g, tj, £) - + u ( x l t y l t ZJ:
ct
Hence, as p -►0, we get
p j j u d o - m ( x t, yu *t) 4n
SERIES
sn —u1+ wa + w, + . . . + un
If there exists a finite limit
s = Iim s„
n -►od
it is called the sum of the sériés ( 1) and we say that the sériés
converges.
If Iims„ does not exist (for example, s„—►oo as n —*■<»), then
rt—
►<*
we say that the sériés ( 1) diverges and has no sum.
Example. Consider the sériés
a+ aq + a ç * + . . . + a q " - '+ . . . (2)
This is a géométrie progression with first term a and ratio q (a 0).
Hence, in the case of | <7 1 < 1, the sériés (2) converges and its sum is
s=
1- q
d—CLQn
(2) If | q | > 1, then | qn | -►00 as n-> 00 and then -j— ^-----►± 00 as n -* 0 0 ,.
that is, lim sn does not exist. Thus, when \q \ > 1, the sériés (2) diverges.
(3) If <7=1, then the sériés (2) has the form
Cl -J - Cl -J - Q -J - . . •
In this case
sn = na, lim sn = 00
From this relation it follows that if Iim tr„_A exists, then Iim sn
/!-*• oo n -* &
exists as well; if Iim s„ exists, then lim also exists; which
«-♦■ao n -+ oo
proves the theorem.
We conclude this section with two simple properties of sériés.
Theorem 2 . I f a sériés
ai + a2 + ••• (3)
converges and its sum is s, then the sériés
ca1+ cai + . . . (4)
where c is some fixed number, also converges, and its sum is es.
Proof. Dénoté the nth partial sum of the sériés (3) by s„, and
that of the sériés (4), by o„. Then
o„ = ca1+ . . . +ca„ = c (û , + . . . +a„) = cs„
It is now clear that the limit of the nth partial sum of the
sériés (4) exists, since
Iim o„ = Iim (cs„) = c lim sn = cs
n -* c o n -* oo /i-eoo
Thus, the sériés (4) converges and its sum is equal to es.
Theorem 3. If the sériés
a1+ a2+ . . . (5 )
and
bi + ^2 + • • • (6 )
converge and their sums, respectively, are s and s, then the sériés
(ax + bj) '+ (a2+ b2) + . . . (7)
and
(a1—6 i) + (æ2—&2) + • • • (8 )
also converge and their sums are s + s and s—s, respectively.
Proof. We prove the convergence of the sériés (7). Denoting its
nth partial sum by_a„ and the nth partial sums of the sériés (5)
and (6 ) by sn and s„, respectively, we get
a n = (a i + ^i) + • • • + (a n + b n)
But
sn— s„-! = ua
Hence,
lim «„ = 0
fl-*00
which is what was to be proved.
Corollary. If the nth term of a sériés does not tend to zéro as
n —►oo, then the sériés diverges.
4.2 Necessary Condition for Convergence of a Sériés 257
T + l + f + - + 2% + -
diverges, since
Bm u„ = lim f
n -* od n -*■ ce \ “•* i ^ ^
We stress the fact that this condition is only a necessary con
dition, but not a sufficient condition; in other words, from the
fact that the nth term approaches zéro, it does not follow that the
sériés converges, for the sériés may diverge.
For example, the so-called harmonie sériés
1+ t + t + t + ---+4'+--*
diverges, although
lim u„= lim —= 0
+ ¥ + T ô + i7 + Ï2 + B + ü + T 5 + T 6 + l7 + W
V ■ 1■V
We also write the auxiliary sériés
Si.= 1 + 7 + ( t + t ) + ( t + ••• + t ) + ( îÎ + • • • + !? )
4 terms 8 terms
= 1+ 4' ÿ
16 terms
Since the sériés (2) converges, its partial sum has a limit a:
lim o„ = a
n-*-ao
From the tact that the terms of the sériés (1) and (2) are posi
tive, it follows that a„ < a, and then, by virtue of (4),
sn < o
We hâve thus proved that the partial sums s„ are bounded. We
note that as n increases, the partial sum s„ increases, and from
the fact that the sequence of partial sums is bounded and increases,
it follows that it has a limit *
lim s„ = s
n -> c o
l + ^ + p ,+ î ï + - - - + ^ + - - -
converges because its terms are smaller than the corresponding terms of the
sériés
1 l± + ± + ± + 4 -1 +
1 r 22 i 23 i 24 ' **‘ ' 2n~ *‘ ’
But the latter sériés converges because its terms, beginning with the second,
form a géométrie progression with common ratio - y . The sum of this sériés
T heorem 2. If the terms of the sériés (1) are not smaller than
the corresponding terms of the sériés (2 ); that is,
u„> vn (5)
and the sériés (2 ) diverges, then the sériés ( 1) also diverges.
* To convince ourselves that the variable s n has a limit, let us recall a con
dition for the existence of a limit of a sequence (see Theorem 7, Sec. 2.5, Vol. I):
“if a variable is bounded and increases, it has a limit.” Here, the sequence of
sums sn is bounded and increases. Hence it has a limit, i.e., the sériés con
verges.
17 :i:
260 Ch. 4 Sériés
1+ — + — +•••■
Vn
diverges because its ternis (from the second on) are greater than the correspon-
ding terms of the harmonie sériés
'-‘4 + T + - + T + -
which, as we know, diverges.
N o te . Both the conditionsthat we hâve proved (Theorems 1
and 2) hold only for sériés with positive terms. They also hold
true when some of the terms of the first or second sériés are zéro.
But these conditions do not hold if some of the terms of the sériés
are négative numbers.
the ratio of the (n-\-\)th term to the nth term, as n —<-oo, has a
(finite) limit l, that is,
lim -^ -= / (2 )
rt-+a>
then:
( 1) the sériés converges for l < 1
(2 ) the sériés diverges for l > 1 .
(For / = 1, the theorem does not yield the convergence or diver
gence of the sériés.)
P roof. (1) Let / < 1. Consider a number q ( l < . q < 1) (Fig. 109).
From the définition of a limit and relation (2) it follows that
for ail values of n after a certain integer N, that is, for N,
4.4 D*Alembert's Test 261
l üjnil 1
“a
Fig. 109 Fig. 110
sériés ( 1), after uy+1, are less than the terms of the sériés (T).
By Theorem 1, Sec. 4.3, and Theorem 1, Sec. 4.1, it follows that
the sériés ( 1) converges.
(2) Let / > 1. Then from the équation lim^i± i = / (where / > 1)
n—
►ao un
it follows that, after a certain N , that is for n ^ N , we will hâve
the inequality
(Fig. 110), or ull+1> u ,t for ail n ^ N . But this means that the
terms of the sériés increase after the term A f+ l, and for this
reason the general term of the sériés does not tend tozero. Hence,
the sériés diverges.
262 Ch. 4 Sériés
Note 1. The sériés will also diverge when lim ^ L±1 = oo. This
n -+ ao un
i j —!— |— !— L . j --------!--------- 1_ .
Solution. Here,
1 1 _ 1 _ 1
Un~ 1-2 . . . . . n ni W', + 1“ 1. 2 . . . . . / i ( n + l ) (n + l)l
un+i ni l
Un (« + !)! n + l
Hence,
lim f ^ ± i = lim — = 0 < i
a-» » un a-* » n -j- 1
The sériés converges.
Example 2 . Test for convergence the sériés
2 2a 2;J 2n
1+ T + T + - + T + -
Solution. Here,
_2n
un — n » un + 1— T + T ' ZH±1= 2 lim îîa ± l= lim 2 - ^ = 2 > 1
Un n+ 1 n-*-» /l+ 1
The sériés diverges and its general terni un approaches infinity.
l + T + ÿ + - - - + 'i ‘+---
We note that wn = — , un +\ ~ and, consequently.
1 -2 + 2 .3 + 3 -4 '* • • • + / i ( n + l ) + - “
Solution. Here,
1 1
“B— n ( n + l ) ’ “" + 1 “ l n + l ) ( n - f 2)
lim ffü ± l= lim »(«+!) lim — = 1
Un
AZ-* « n-*-« (»+!)(« + 2) «-*-» n~r2
D’Alembert’s test does not permit us to infer that the sériés converges;
but by other reasoning we can establish the fact that this sériés converges.
Noting that
1 _ 1 ___ 1_
« ( n |l ) - » n-j- 1
we can Write the given sériés in the form
The partial sum of the first n te rn is , after removing brackets and cancell-
ing, is
1
sn —
n+\
264 Ch. 4 Sériés
Hence,
lim s„ = üm ( = l
« -♦ o o n -t-o o \ n -r 1 /
t + ( t ) + ( t ) + - - - + ( 2^ n ) " + - - -
4.5 Cauchy*s Test 265
lim \ / u n — \\m 1/ —= 1
n-+<x> n -+ rc T fl
1-
lim iln 1
-S f 1
/ —= 1-lim ------
—In/i
n —> cd * ^ n -+ c c fl
S / (*) ^
converges (see Sec. 11.7, Vol. I), then the sériés (1) converges too;
(2 ) if the given intégral diverges, then the sériés ( 1) diverges
as well.
-+ -+ -+ +— +
Solution. Apply the intégral test, putting
f(x)=Tp
This function satisfies ail the conditions of the theorem. Consider the intégral
1 1
Cdx= x l ~P (N1-P — 1) when p ?= 1
1- P 1- P
J xP
ln x |? = lnyV when p = 1
Allow N to approach infinity and détermine whether the improper intégral
converges in various cases.
It will then be possible to judge about the convergence or divergence of the
sériés for various values of p.
00
For p > 1, ^ — 1— > intégral is finite and, hence, the sériés
l
converges;
ce
P dx
for p < 1, \ — = o o , the intégral is infinité, and the sériés diverges;
1
00
P dx
for p = l , \ —= o o , the intégral is infinité, and the sériés diverges.
1
We note that neither the d’Alembert test nor the Cauchy test, which were
considered earlier, décidé whether the sériés is convergent or not, since
4.7 Alternating Sériés. Leibniz* Theorem 269
then the sériés ( 1) converges, its sum is positive and does not exceed
the first term.
Proof. Consider the sum of the first n-=2m terms of the sériés (1):
Sîm = («l —«2) + («3 —«4) + • • •
From condition (2) it follows that the expression in each of the
brackets is positive. Hence, the sum s2m is positive,
21 ^3!
converges by virtue of the Leibniz theorem.
4.8 P lus-and-Minus Sériés. Abzolute and Conditional Convergence 271
By hypothesis, on has the limit a; s' and s" are positive in-
creasing quantities less than a. Consequently, they hâve the
limits s' and s". From the relationship sn = s'n— Sn it follows that
sn also has a limit and that this limit is equal to s' —s", which
means that the alternating sériés ( 1) converges.
The above-proved theorem enables one to judge about the con
vergence of certain alternating sériés. In this case, testing the
alternating sériés for convergence reduces to investigating a sériés
with positive terms.
Consider two examples.
Example 1. Test for convergence the sériés
sin a . sin 2a , sin 3a . . sin na .
i r + ^ r + - 3r + - + i r + - (3)
where a is any number.
* In this English édition we shall use the terni alternating sériés for both
types.—Tr.
272 Ch. 4 Sériés
Ul + U2 + U3 + •• • + Un + ••• (1 )
'+ T + Ÿ + T + - - '
which diverges. The sériés itself converges (this can be readily verified by
Leibniz* test).
Example 4. The alternating sériés
i-± + ± -± + ..
21^3! 41^
is absolutely convergent, since a sériés made up of the absolute values of its
terms,
We shall Drove that the résultantsériés converges, but that its sum s' is
18—2082
274 Ch. 4 Sériés
half the sum of the sériés (8 ): that is, — s. Dénoté by sn and s'n the partial
sums of the sériés (8 ) and (9). Consider the sum of 3k terms of the sériés (9):
Thus, in this case the sum of the sériés changed after its terms were
rearranged (it diminished by a factor of 2).
metric progression with ratio x). Thus, in the interval (—1, 1) the given sériés
defines the function
Let us now represent the sum of the functional sériés (1) in the
form
s (x) = sn (x) + rn (x)
where
Sn {x) = U, (x) + . . . + Un (x)
Tn W = « n + l (X) + Un + 1 (X ) + « B + s (X) + . . .
_________ 4.11 The Continuity of the Sum of a Sériés 277
From condition (3) it follows that
and therefore
I rn (a) | < e„
for ail x of the range under considération.
Thus,
| s (a) —s„( a)| < e„
for ail x of the interval [a, b], and e„—>0 as n —* oo.
Note 1 . This resuit may be represented geometrically as follows.
Consider the graph of the function y — s(x). About this curve
construct a band of width 2e„; in other words, construct the curves
t/=s(A) + e„ and i/=s(a) —en
(Fig. 114). Then for any e„ the
graph of the function s„(A)will
lie completely in the band under
considération. The graphs of ail
successive partial sums will
likewise lie within this band.
Note 2 . Not every functional
sériés convergent on the interval
[a, b] has the property indicated
in the foregoing theorem. How-
ever, there are nondominated sé
riés such that possess this pro
perty. A sériés that possesses this property is called a uniformly
convergent sériés on the interval [a, b].
Thus, the functional sériés ut (x) + u2(a) + . . . + u„ (x) + . . . is
called a uniformly convergent sériés on the interval [a, b] if for
any arbitrarily small e > 0 there is an integer N such that for
ail n ^ N the inequality
\ s { x ) — sn ( x ) \ < t
will be fulfilled for any x of the interval [a, b].
From the theorem that has been proved it follows that a domi-
nated sériés is a sériés that uniformly converges.
function. This property does not hold for the sum of a sériés con-
sisting of an infinité number of terms. Some functional sériés with
continuons terms hâve for the sum a continuous function, while
in the case of other functional sériés with continuous terms, the
sum is a discontinuous function.
Example. Consider the sériés
The terms of this sériés (each term is bracketed) are continuous functions for
ail values of x. We shall prove that this sériés converges and that its sum
is a discontinuous function.
And so the sum of the given sériés is a discontinuous function. Its graph is
shown in Fig. 115 along with the graphs of the partial sums st (x),
and s3 (x).
IM*)l<y (3)
will be fulfilled for any x of the interval [a, 6 ]. The value x + Ax
lies on the interval [a, b] and therefore the following inequality
is fulfilled:
|rjv(x + A x )|< — (3')
280 Ch. 4 Sériés
Further, for the chosen N the partial sum s,v (a) is a continuous
function (the sum of a finite number of continuous functions) and,
consequently, a positive number ô may be chosen such that for
every Ax that satisfies the condition J Ajc | < ô the following
inequality is fulfilled:
|ASa,| < ! (4)
By inequalities (2), (3), (3'), and (4), we hâve
l*l<T +T +ï = 8
that is,
| As | < e for | Ax | < ô
which means that s(x) is a continuous function at the point *
(and, consequently, at any point of the interval [a, b]).
Note. From this theorem it follows that if the sum of a sériés
is discontinuous on some interval [n, b], then the sériés is not
dominated on this interval. In particular, the sériés given in the
Example is not dominated on any interval containing the point
x — 0 , that is to say, a point of discontinuity of the sum of the
sériés.
We note, finally, that the converse statement is not true: there
are sériés, not dominated on an interval, which, however, converge
on that interval to a continuous function. For instance, every
sériés uniformly convergent on the interval [a, b] (even if it is not
dominated) has a continuous function for its sum (if, of course,
ail terms of the sériés are continuous).
Then
X X X X X
lim \ r n(t)d t= 0
n ■* « a
But from équation (2) we hâve
x x r x x n
)j u1(t)dt+ . . . + J « „ (/)d /+ . . .
< (4)
Since the partial sums of this sériés hâve a limit, this sériés
converges and its sum, by virtue of équation (3), is equal to
X
^ s (t)d t, i.e.,
a
X X X X
that the intégral [ s(t)d t of the sum of the sériés ( 1) 1s not always
a
equal to the sum of the intégrais of its terms [that is, to the sum
of the sériés (4)].
T h eo rem 2. If a sériés
«1 (*) + « , W + . . . + H „ W + . . . (5)
made up of functions having continuous dérivatives on the interval
[a, b] converges (on this interval) to the sum s (x) and the sériés
u'1(x) + iït { x ) + ...+ u 'n ( x ) + ... (6 )
made up of the dérivatives of its terms is dominated on the same
interval, then the sum of the sériés of dérivatives is equal to the
dérivative of the sum of the original sériés; that is,
s' (x) = (x) -[- ua (x) -f- u8 (jc) • + u'n (x) + . . .
P ro o f. Dénoté by F (x) the sum of the sériés (6 ):
F (x) = u[ (x) + a» (*) (at) + . . .
and prove that
F(x) = s’ (x)
Since the sériés (6 ) is dominated, it follows, by the preceding
theorem, that
X X X X
(2 ) if a sériés diverges for some value x'0t then it diverges for every
x for which
1*1 > K l
Proof. (1) Since, by assumption, the numerical sériés
an+ atx 0 + a2xl + . . . + anx%+ . . . (2 )
converges, it follows that its common term a„x{J—>-0 as n —►oo,
and this means that there exists a positive number M such that
ail the ternis of the sériés are less than M in absolute value.
Rewrite the sériés ( 1) in the form
a0+ ûi*o (^") + ) + •• • + a K ( —) + ••• (3)
and consider a sériés of the absolute values of its tcrms:
K I + K *ol + 1azxl I + •• • + | I + ••• (4 )
The terms of this sériés are less than the corresponding terms
of the sériés
X X
M+M M
*7
‘+ . . . + M
*0
+ ... (5 )
For \ x \ < \ x Q\ the latter sériés is a géométrie progression with
X
ratio < 1 and, consequently, converges. Since the terms of the
*o
sériés (4) are less than the corresponding terms of the sériés (5),
the sériés (4) also converges, and this means that the sériés (3)
or ( 1) converges absolutely.
(2) It is now easy to prove the second part of the theorem: let
the sériés ( 1) diverge at some point x'Q. Then it will diverge at
any point x that satisfies the condition |* |> |* i |. Indeed, if at
some point x that satisfies this condition the sériés converged, then,
by virtue of the first part (just proved) of the theorem, it should
converge at the point x'0 as well, since \x'0\< \x \. But this con-
tradicts the condition that at the point x'0 the sériés diverges.
Hence the sériés diverges at the point x as well. The theorem is
thus completely proved.
Abel’s theorem makes it possible to judge the position of the
points of convergence and divergence of a power sériés. Indeed,
if x0 is a point of convergence, then the entire interval (— 1*0|,
| jc0 |) is filled with points of absolute convergence. If x'0 is a point
of divergence, then the whole infinité half-line to the right of the
point |*o| and the whole half-line to the left of the point — \x'0\
consist of points of divergence.
From this it may be concluded that there exists a number R
such that for |a*| < R we hâve points of absolute convergence and
for | jc | > /?, points of divergence.
4.13 Power Sériés. Interval of Convergence 285
for points x lying outside it, the sériés diverges (Fig. 116). The
number R is called the radius of convergence of the power sériés.
At the end points of the interval (at x = R a n d a tx = — R) the
question of the convergence or divergence of a given sériés is
decided separately for each spécifie sériés.
We note that in some sériés the interval of convergence dégé
nérâtes into a point (Æ= 0 ), while in others it encompasses the
entire x-axis (R = oo).
We give a method for determining the radius of convergence of
a power sériés.
Let there be a sériés
a0+ a1x-\-a2xi + ...+ a „ x n+ . . . ( 1)
Consider a sériés made up of the absolute values of its terms:
I a01+ 1a, 1 1x | + 1a 2 11 a: |2 + 1a 3 11 x |s
+ 1a411x |4+ • • • + 1 11x |" + . . . (6)
To détermine the convergence of this sériés (with positive terms!),
apply the d’Alembert test.
Let us assume that there exists a limit:
a n + ix n + 1 I « an+1
lim 2a±±= Iitn M I ~ AX11x \ = L \x \
I—
* rr> Ptn a»xn \ n ^ , û/i
Then, by the d ’Alembert test, the sériés (6 ) converges, if L \x \ < 1 ;
that is, if |* |< -£ - . and diverges if L \x \ > 1 , that is, if |x |> - ^ - .
Consequently, sériés ( 1) converges absolutely when | * |< X - But
if | x | > - , then lim —|x | L > 1 and sériés (6 ) diverges, and
L n -+*> un
286 Ch. 4 Sériés
its general term does not tend to zéro. * But then neither does
the general term of the given power sériés ( 1) tend to zéro, and
this means that (on the basis of the necessary condition of con
vergence) this power sériés diverges ^when
From the foregoing it follows that the interval ( T ’ r ) is
the interval of convergence of the power sériés ( 1), i.e.,
R = -r-= lim
L n oo an+ l
)
Similarly, to détermine the interval of convergence we can make
use of the Cauchy test, and then
D _ _____*
lim { /K J
Tl —► CO
Thus, the sériés converges when \x \ < 1 and diverges when \x \ > 1. At
the extremities of the interval (— 1, 1) it is impossible to investigate the
sériés by means of d’Alembert’s test. However, it is immediately apparent
that when x = — 1 and when x = \ the sériés diverges.
Example 2. Détermine the interval of convergence of the sériés
2x (2x)2 . ( 2 j c ) 3
T 29 nI 3
Solution. We apply the d’Alembert test:
(2*)”* 1
n - \-1
lim
(2x)n = n lim« n + \ I2x I = I 2x I
The sériés converges if \2x\ < 1, that is, if \x \ < when ^ = y the sériés
Since the limit is independent of x and is less than unity, the sériés con
verges for ail values of x.
Example 4. The sériés 1 + x -|- (2jc)2+ (3*)3 (nx)n + . . . diverges for
ail values of x except = 0 because (nx)n —►oo as n —►oo no matter what
jc
—Q-
-a
interval of majorUation
Fig. 117
Proof. It is given that p < R (Fig. 117) and therefore the num-
ber sériés (with positive terms)
Ia Q I + Ia \ IP + Iü 2 IP2 + ** *+ Iün IP” (7 )
converges. But when | x | < p , the terms of the sériés ( 1) do not
exceed, in absolute value, the corresponding terms of sériés (7).
Hence, sériés ( 1) is dominated on the interval [— p, p].
-/? oc 0 fi R
Fig. 118
Fig. 119
Take a point £ such that p < £ < Æ (Fig. 119). The sériés (1)
converges at this point, hence lim = it is therefore possible
n -► o c
Thus the terms of the sériés (2), when ,v | ^ p, are less, in abso-
lute value, than the terms of a 'positive number sériés with con
stant terms:
—•(1 + 2q + 3q*+ . . . + n q " -'+ . . . )
But the latter sériés converges, as will be évident if we apply
the d’Alembert test:
nqn~ 1
lim • 7 <I
n -+• » (n — \)q n~ 2
4.15 Sériés in Powers of x — a 289
sériés ( 1) will diverge for |jc—a | > / ? , that is, it will diverge
outside the interval a — R < x < a + R (Fig. 120,/?).
And so the interval (a—R, a + R) with centre at the point a
will be the interval of convergence of sériés (1). Ail the properties
of a sériés in powers of x inside the interval of convergence
(— R, + R) are retained completely for a sériés in powers of
x —a inside the interval of convergence (a —R , a + /?). For example,
-R 0 R X
(a) + —I----- 1--+
a - R -- --------. a *R x 0 12 3 x
(b) * ---- 1----a1---- 1---- H--- 1--- 1---- 1------
Fig. 120 Fig. 121
f(x)= J(a) + ^ r ( a )
+ r(a)+... + —
{ ^ r( a ) + Rn(x) (1)
But P„(x) is the nth partial sum of the sériés (2); its limit is
equal to the sum of the sériés on the right side of (2). Hence,
(2 ) is true:
/ « - / « o + ^ r («)+ f e ^ 2 f w + . . . + — r : r ” ( a ) + - - -
From the foregoing it follows that the Taylor sériés is a given
function f(x) only when lim Rn(x) = 0. If lim Rn (x) 0, then the sériés
is not the given function, although it mav converge (to a different
function).
If in the Taylor sériés we put a = 0 , we get a spécial case of
this sériés known as Maclaurin's sériés:
/ w = / ( 0) + f r ( 0) + - r ( 0) + . . . + f | r ( 0) + . .. (3)
If for some function we hâve a formally written Taylor’s
sériés, then in order to prove that this sériés is a given function
it is either necessary to prove that the remainder term approaches
zéro, or to be convinced in some way that this sériés converges
to the given function.
We note that for each of the elementary functions defined in
Sec. 1.8, Vol. I, there exists an a and an R such that in the inter-
val (a—R, a + R) it may be expanded into a Taylor’s sériés or
(if a = 0) into a Maclaurin’s sériés.
19
292 Ch. 4 Seriez
Since the remainder term approaches zéro for any x, the given
sériés converges and, for its sum, has the function sinx for
any x.
Fig. 122 shows the graphs of the function sinx and of the
first three partial sums of the sériés ( 1).
This sériés is used to compute the values of sinx for different
values of x.
To illustrate, let us compute sin 10° to the fifth décimal place.
4.17 Sériés Expansion of Functions 293
since it was proved that Iim Rn (x) = 0 for any x. Hence, the
n -* oo
sériés converges for ail values of x and is the function ex.
Substituting x = — x, in (2 ), we get
y2 y3 y4
o-x _ 1 x 4 - - ____ — 4 - — -
e 1 '* ^ 2 ! 31 ‘ 41 (3)
for ail values of x the sériés converges and represents the function
COSX.
4. E x p a n sio n , in M a c la u rin ’s sériés, of th e fu n c tio n s
e* — e--*
f (x) = sinh x ^ 2
e*+ e-
/ (x) = cosh x =
These functions are readily expanded by subtracting and adding
the sériés (2) and (3) and dividing by 2.
294 Ch. 4 Sériés
Thus,
JC3 , JC5 , X1
sinh* = x + 3j + 5ï + 7 î + . . . (5)
coshx= l + £ + £ + ^ + . . . (6)
Up till now we hâve considered only sériés with real terms and
hâve not dealt with sériés with complex terms. We shall not give
the complété theory oi sériés with complex terms, for this goes
beyond the scope of this text. We shall consider only one important
example in this field.
In Chapter 7, Vol. I we defined the function e*+i* by the équa
tion
e x+ iy _ gX (C0S y _|_ (• s iu y}
(1)
s (x) = 1+ mx + « V * + •••
, m(m— —(n — 1)] „n ,
I 1 .9 m A I
* We took the absolute term equal to unity by virtue of the initial con
dition s( 0 ) = l .
296 Ch. 4 Sériés
For m — ]r we get
1 .. 1 », 1-3 , 1-3-5
V l + x —1+-K
2
-X—577+
2-4
;r;, „-V 2-4-6-8 .V4 + . . .
'2 - 4 - 6 (5)
For m = — g- we hâve
— ~ 1 ------1 ..v ,_ 1-3 , ___
| _ _ v-2 1-3-5 , , 1-3-5-7
x 2 x ^2 -4 * 2 -4-6 1 2-4-6-Sî J f — . . . (6)
1 -3*5.. .( 2n — 1) jc2" * 1
+ 2 •4 •6 . . . 2/i 2 /i+ 1 *" *
This sériés converges in the interval (—1, 1). One could prove
that the sériés converges for x ^ ± \ as well as that for these
values the sum of the sériés is likewise equal to arcsinx. Then,
setting * = 1 , we get a formula for computing j i :
• i n , , 1 1 , 1 3 1 13 -5 1
arcsinl = 2 - = l + T . -3+274. 5 2*4*6 * 7 +
i i
o o
or
y2 y3 v4 y /J
l n ( l + * ) - * - + + £ — - + . . . + (- l.)" +1- + . . . ( 1)
This équation holds true in the interval (—1, 1).
If in this formula x is replaced by —x, then we get the sériés
y2 y3 y4
ln (1 —x) = —x — —— ^ — ... (2 )
which converges in the interval (— 1, 1).
Using the sériés ( 1) and (2 ) we can compute the logarithms of
numbers lying between zéro and two. We note, without proof,
that for x = l the expansion ( 1) also holds true.
We will now dérivé a formula for computing the natural loga
rithms of ail integers.
Since in the term-by-term subtraction of two convergent sériés
we get a convergent sériés (see Sec. 4.1, Theorem 3), then by sub-
tracting équation (2 ) from équation ( 1) term by term, we find
In (l- M)—|n ( l _ , ) = l „ } ± i = 2 [* + £ + £ + . . . ]
(4)
9
If we now want to compute ln2 to, for example, nine décimal
places, we must choose p such that ^<0.000000001. This can be
done by selecting p so that the right side of inequality (4) is less
than 0.000000001. By direct choice we find that it is sufficient
to take p —8 . To nine-place accuracy we hâve
ln 2 » sg= 2
1-3 ‘ 3-33 ~ 5-3» “ 7-37 ~ 9-3» ~ 11-3U
^ T Î ^ + T S ^ ] =°-69314718°
Thus, ln 2 = 0.693147180, correct to nine places.
4.21 Sériés Evaluation of Ùefinite Intégrais 299
J e~x*dx
0
Here, the antiderivative of e~x% is not an elementary function. To
evaluate this intégral we expand the integrand in a sériés, replac-
ing x by —xi in the expansion of e* [see formula (2), Sec. 4.17]:
• y2 y4 y# y2n
°^ d x
0
Expand the integrand in a sériés: from the équation
300 Ch. 4 Sériés
we get
sin x _ . __ x*_ . x*___xs
~ “ 1 — 3! + 5! 7! ' ’ ’ *
the latter sériés converges for ail values of x. Integrating term by
term, we obtain
sin jc ___ a3 , a6 a7 .
~ a x ~ a— 5^ 3 -— 7P 7 + •••
0
The sum of the sériés is readily computed to any degree of accu-
racy for any a.
3. Evaluate the elliptic intégral
n
2
J V \ —Æ*sin 2 qpcfqs (k < 1)
0
"“ T T T * ' J s i n ^ ^ — • • •
Solution. We hâve
/ ( 0) = f t = l . / ' (0) = y'0 = 0
From the given équation we find (y")x - 0 = f" (0) = 0; further,
y '" = —y ’x* — 2xy, (y "')JC=0 = / '" (0 ) = 0 ,
yiV = - y * xt - 4 Xy ' - 2 y , (ÿ,v)x=# = —2
and, generally, di fièrent iating k times both sides of the équation by the Leib
niz formula, we find (Sec. 3.22, Vol. I)
yik + 2) = —yM)x2_ 2kxyi* - l>- k (k - 1 ) y<* -*>
Putting x = 0, we hâve
(fc+2) «, /*, i\ .,<*-2)
yo = —* ( « — i) i/o
or, setting fc+ 2 = /i,
ÿ <") = — ( „ _ 3 ) ( „ —2) y " - »
Whence
J/JV = —1-2, y ,8’ = —5-6yJv = (—1)* (1 -2) (5-6)
yÔ12’ = - 9 - l<tyj8) = (-1)3 (1 -2) (5-6) (9-10)
In addition,
yô5> = 0 , y p = 0 .............. ÿ^ +‘>= 0.........
y<*>=0, |, « » = 0 ............ yl‘k+*>= 0..........
</i7, = 0, y<"> = 0............ y(o**+3>= 0, . . .
Thus, only those dérivatives whose order is a multiple of four do not
become zéro.
Putting the values of the dérivatives that we hâve found into a Maclaurin
sériés, we get the solution of the équation
y = 1 - ^ - 1 . 2 + | i (1-2) ( 5 - 6 ) - ^ - ( 1 - 2 ) (5-6) ( 9 - 1 0 ) + . . .
Consequently,
o 1
2-1 1 2 1 1
a* 4 21 ’ a ,~ 6 3! ’ — 4! ’ ' ‘
2 1
*(fe —1)1 1
**+1 2k ~ k \ .......
o 4 = 0 ; a, = 0 ........... «** = 0 , . . .
Substituting the coefficients which we hâve found, we get the desired solution:
xtk +i
y X+ 1 2 ! +31 k\
The sériés thus obtained converges for ail values of x.
It will be noted that this particular solution may be expressed in terms of
elementary functions: taking x outside the brackets we get (inside the brackets)
an expansion of the function e*1. Hence,
y = xe**
a power sériés:
y = xr 2 akx» (2)
k=0
The coefficient a0 may be considered nonzero due to the indefi-
niteness of the exponent r.
We rewrite the expression (2) in the form
y = k2= 0 û**r+*
and find its dérivatives:
y ' - k=0 +
substituting —p for p:
yt = x~p [ 1 - 2 (— 2p + 2 )T 2 4 ( — 2p + 2) ( - 2p + 4)
xe , 1
2-4-6 ( - 2p + 2) ( - 2p + 4) ( - 2 p + 6 ) (5')
The power sériés (5) and (5') converge for ail values of x\ this
is readily found by d ’Alembert’s test. It is likewise obvious that yx
and t/2 are. linearly independent.*
The solution y x multiplied by a certain constant is called a
Bessel function of the first kind of order p and is designated by
the symbol J p. The solution yz is denoted by the symbol J „p.
Thus, for p not equal to an integer, the general solution of
équation ( 1) has the form
y = C1Jp + CiJ _p
For instance, when p = y the sériés (5) will hâve the form
\ fi x* i x* x4 i1
X L *1 2-3 + 2-4.3-5 2-4-6-3-5-7*‘ •••J
__ i r x* . x 5 *!_< 1
~ Y 7 \x ~ 3 f + 5T- ' • "J
j __l( * ) = y r- ^ c°sx
y"+^-y'+y=°
20 *
308 Ch. 4 Sériés
From (2) it follows that condition (1) will hold provided the
following conditions are valid:
lim an = a, lim bn = b (3 )
n -►oo n -♦» oo
Let us form a sériés of complex numbers
w ,+ wt + . . . + w n+ . . . (4)
4.24 Sériés with Complex Terms 309
where
w„ = un + iv„ (« = 1 . 2 , . . . )
We consider a sum of n terms of the se»ies (4), which we dénoté
by s„:
sn = wl + wî + . . . + w n (5 )
where s„ is a complex number:
*-U-)+,U*) <6>
Définition 2. If the limit
lim s„ = s = A-\-iB
n oo
exists, then sériés (4) is termed a convergent sériés and s is called
the sum of the sériés:
s = ^ » , = A + ifl (7)
From condition (6 ), on the basis of (3), follow the équations
n n
2. sinz = z - - + - [- j J+ . . . (5)
This is the sine of a complex variable. Fdt y = 0, formula (5) turns
into formula (1) of Sec. 4.17.
3. c o s z = l - ^ + 5 - S + --- (6 )
This is the cosine of a complex variable. If in formula (4) we put
(iz) in place of z on the right and on the left, then (like what
was done in Sec. 4.18) we get
eiz = cos z + i sin z (7 )
This is Euler’s formula for a complex z. If z is a real pumber,
then this formula coïncides with formula (2 ) of Sec. 4.18.
4. sinh z = z + 57 + 3 ] + ••• (8 )
5. coshz= 1 + 2 ! + - 4 ; + gj + • • • (9)
These last two formulas are similar to formulas (5) and (6 ) of
Sec. 4.17 and coincide with the latter when z = x is a real number.
By (4), (5), (6 ), (8 ), and (9) and via addition, subtraction of
the sériés and replacement of z by (iz), we get the following équa
tions:
e* = cosh z + sinh z, ( 10 )
e~z = cosh z — sinh z, (ii)
u e*+ e~z • 1 ez—e~~z
cosh z = —- — , sinh z = — ^— » ( 12)
e-* + e* . e“ z—ez
COS IZ = -------» sin iz = — 2i— (13)
Note that the sériés (4), (5), (6 ), (8 ), (9) converge for ail values
of z. This is clearly seen on the basis of Theorem 1 and of Sec.
4.24. As in the case of power sériés of a real variable, we can
consider sériés of a complex variable in powers of (z—z0), where z0
is a complex number, cn are complex or real constants
Co+ M z —z0) + c,(z—z0)*+ • • • + c„(z—z0)n+ . . . (14)
The sériés (14) is reduced to (1) by the substitution z—z„ = z*.
Ail the properties and theorems that are valid for sériés of type
(1) carry over to sériés of type (14), only the centre of thecircle
of convergence of (14) lies at point z0 instead of at the origin.
If R is the radius of convergence of sériés (14), then we Write:
the sériés converges in the domain
| z - z 0| < f l
312 Ch. 4 Sériés
!=/(*. y) (i)
that satisfies the initial condition
y = y0 for * = (2 )
Integrating the terms of équation (1) from xa to x and taking into
account that y\x=x, = y<>, we get
X
y = l f ( x , y)dx + y0 (3)
*0
In this équation, the desired function y is under the intégral sign
and so the équation is called an intégral équation.
The function y = y(x) which satisfies ( 1) and the initial condi
tions (2) satisfies équation (3). It is clear that the function y = y (x)
which satisfies (3). satisfies équation (1) and the initial conditions (2).
Let us first consider a method of obtaining approximate solutions
to équation ( 1) for the initial conditions (2 ).
Wewill take y0 for the zeroth approximation of the solution.
Substituting yQinto the integrand on the right of (3) inplace of y y
we get
X
y A x ) = [ f ( x , y0)dx + y0 (4)
y* M = $ /[ * - y A x )]d x + y a,
*o
lfx = $ (* + l)< te + l= £ -h r+ l
o 2
S=/(** y) (!)
and the initial conditions
y = y0 for x = x0 (2 )
Let f (xf y) and f'y (x, y) be continuons in a ctosed domain D:
D{x0— a ^ x ^ x 0+ a, y<>— b ï ^ y ^ y 0+ b\ (Fig. 123)(3)
Then, in some interval
x0— l < x < x 9+ l (4 )
there exists a solution to équation ( 1) which satisfies the initial
conditions (2); and the solution is unique. The number l will be
defined below.
314 Ch. 4 Sériés
P roof. Note that from the tact that f(x, y) and f'y (x, y) are
continuons in a closed domain D it follows that there exist cons
tants M > 0 and N > 0 such that for ail points of the domain
the following relations hold:
l/(*. y ) \ < M , (5)
\ry (x, y ) \ ^ N (6 )
(This property is similar to the property indicated in Sec. 2.10,
Vol. I.) The number l in (4) is the
(Xo 4 o * b)
smaller of the numbers a and -77
M
- (Xg*a,yg) that is,
l = min 7â ) (7)
* Note that if for some function F (y) the following condition is satisfied:
IF (y2)—F (yi) | < | y2 —0 i |
where y2 and y x are any points in the domain and K is a constant, then this
condition is called the Lipschitz condition. Thus, having established the rela
tion (8), we demonstrated that if a function f(x, y) has a dérivative
bounded in some domain, then it satisfies the Lipschitz condition in that
domain. The converse may not orove to be true.
4.27 P roof of the Existence of a Solution of a Différent ial Equation 315
1= 2 ut = yn (14)
i=0
Let us estimate the terms of sériés (13) in absolute value:
< ± N $ M Ix — x01dx = N — Ix — x0 \2
(the plus sign is taken if x0 < x, and the minus sign if *0 > x ).
Thus
Iy*— y x K m |x—xj! (16)
Similarly, taking into account (16),
< ± $ N \ y — y „ \ d x ^ N R \ x — x0\
4.27 Proof of the Exittence of a Solution of a Differential Equation 317
Example. Let us apply this estimate with respect to the fifth approximation
y t to the solution of the équation
y '= x + y i
given the initial conditions y0— 1 when x = 0.
Suppose the domain D is
D { — 1 /2 < x < 1 / 2 , —
that is, a = 1/2, 6 = 1. Then M = 3/2, N = 2. We then détermine
/ = min
Note that the estimate (25) is rather rough. In the case at hand, wecouldshow
by other methods that the error is tens of times less.
318 Ch. 4 Sériés
t- f ^ y ) (0
Iy — z N \ y — z\dx ( 6)
xg
Let us consider a value of x such that |a:—x0|< - ^ - . For the sake
of definiteness, we assume that xü < for the case of x < x0 the
proof is analogous.
Suppose, on the interval x —x0 < \ y —z\ assumes a maximum
value for x = x* and let it be equal to K. Then inequality (6 ) as-
Exercises on Chapter 4 319
Exercises on Chapter 4
Write the first few terms of the sériés on the basis of the given general term:
1
t• * 2 u 3 u - (- ^ l
“" “ rt+1 ' 6 n (2n)V
4 u = f_l)» +iî2
„*•
n ( » + l)
u„ = V n3+ 1— y rn2+ 1.
Test the following sériés for convergence:
17. kw= y i a—t-ô • Converges. 18. un^ 1= -r^— . Ans. Diverges. 19. Prove
n£-\-Zn-\~o n \n n
the inequality
21, î ~ 2«+ ' ' ' + ( - ') " +12»+ ••• • Ans- ti = 20.
224 - t + t - t + - (- i) 4 + - - • n= io#-
23--p- — 1)” 7jr+ --- • Ans■n = 10*.
+ 3f C 2 )8+ ... .
47. Expand x3 —2x2 + 5 x —7 in powers of (x—1). Ans. —3 + 4 ( x — !) +
+ ( x - l )2 + ( x - l ) 3.
48. Expand the polynomial x10 + 2x9 —3x7 —6x€ + 3x4 + 6xs —x — 2 in a
Taylor’s sériés in powers of (x—1); check to see that this polynomial has the
21—2082
322 Ch. 4 Sériés
- + * - ' ) 4+ . . . •
(x -i-2)n I
1 + 2mà -- J *
52. Expand cos2* in a sériés of powers of ^* — .
2/1-1
4"" 1 (* —t )
(_- n» /i------X----^-L
Ans. ( 1*1 < co).
T+ H (2n—1)1
n - 1
+ ( x- t ) + - •
Write the first four terms of the sériés expansion, in powers of *, of the
following functions:
... A . x3 . 2*6 17*7 .
55. tan x. Ans. x + T + 1 F + - ^ + . . . .
Co sx a ( * x2 i 4*4 31*° \
S6' * • ^ V -2 T + 4 r-7 2 Ô ----j-
57. éarctan * a ii i X2 X3 7x4 .
Ans. 1 + X + -2 — g— 2 4 + - -
f1 dx x^ x®
Hint. Take advantage of the formula arctan x = j ^ ^ . Ans. x — g--|—g—
1
67. Expand in a s é r ié s o f p o w e r s o f x . Ans. 1 — 2 jc + 3 jc2 — 4x3 + ...
0 + * )2
—1 < x < 1).
Using the formulas for expansion of the functions ex y sin x, cosx, ln (l+ ;c )
and (l--fjc)m into power sériés and applying various procedures, expand the
following functions in power sériés and détermine the intervals of convergence:
x? JC®
68. sinh x. Ans. —b • • • ( — « > < * < oo).
Xn
78. J ln ( l + x) dx. Ans. £ (_ l) » + i J £ ( | * | < 1)
0 n- 1
x oo
^arctanjc j ^ , tv„je2" * 1
79 dx. Ans. ^ (—1)”
(2/1+ 1)* ( - K * < 1).
n= 0
03
80. ^S 2 1 Id x . Ans. C + l n |* | + £ ( —1 ) « o o < x < 0 a n d 0 < x < o e ) .
21 *
324 Ch. 4 Sériés
A
dx * 9 /1 -8
81
I
0
1 —x9
Ans.
n= 1
9n=8'
82. Prove the équations
sin (a -f x) = sin a cos * + cos a sin x
cos (a + *) = cos a cos x — sin a sin x
by expanding the left sides in powers of x.
Utilizing appropriate sériés, compute:
83. cos10° to four décimais. Ans. 0.9848. 84. sin1° to fourdécimais.
Ans. 0.0175. 85. sin 18° to three décimais. Ans. 0.309. 86. sin ^ to four de-
4
cimals. Ans. 0.7071. 87. arctan 4- to four décimais. Ans. 0.1973. 88. In5 to
ü
three décimais. Ans. 1.609. 89. log105 to three décimais. Ans. 0.699.
90. arcsin 1 to within 0.0001. Ans. 1.5708. 91. ÿ e to within 0.0001.
Ans. 1.6487. 92. loge to within 0.00001. Ans. 0.43429. 93. cos 1 to within
0 .00001. Ans. 0.5403. _____
Using a Maclaurin sériés expansion of the function /( * ) = \ / a n ~\-xy com
pute to within 0 .001 : _
94. j/j3 0 . Ans. 3.107. Ans. 4.121. _96. j/iÔÔ. Ans. 7.937.
97. y/250. Ans. 3.017. 98. 84. Ans. 9.165. 99. f / 2 . Ans. 1.2598.
Expanding the integrand in a sériés, compute the intégrais:
9. J* —^
100. dx to five décimal places. Ans. 0*94608. 101. ^ e -* * d x to four
o o
n
4
décimais. Ans. 0.7468. 102. J s\n(x2)dx to four décimais. Ans. 0.1571.
o
1
T 05
arctan x
103. J e^x djctotwo décimais. Ans. 0.81. 104. J dx to three décimal
places. Ans. 0.487. 105. J cos Ÿlc dx to within 0.001. Ans. 0.764.
o
1
T 1
106. J ln ( \ + V x ) d x to within 0.001. i4ns. 0.071. 107. J e 4 dx to within
5
0.0001. Ans. 0.9226. 108. f dx to within
withi 0 .0001. Ans. 0.0214.
J Ÿ 1—x
o
0.5
109. J j to within 0.001. Ans. 0.494. 110. J 1
^ ± ^ d x . Ans. £ .
x 12
Exercises on Chapter 4 325
Note. When solving this exercise and the two following ones it is well to
bear in mind the équations:
*y”+ xy'-^-^x* — y= 0
Hlnt. Seek the solution in the form^ y = xP (A0+ A iX + 4*jç*4- . • •)•
1
c ‘‘ ’ [ | - # + l r - 7 f + • • • ] + c■ ' ' , [ ‘ - I r + î f — •]
sin jc cos JC
C*116.VxFind^ Vx
the solution of the équation xy”+ y ’ + x y = 0 that satisfies the
vl vl mC
initial conditions for j c = 0 , y = \ t y ' = 0 . Ans. 1 -
22 ' (1 *2)a 2 4 (l-2-3)a 2®
JC2*
îz+ ----
(*!)a 2a*
Hint. The last two differential équations are particular cases of the Bessel
équation
jc2y 0+ jcy' + (jc®— n*)y = 0
for n = ~ 2 anc* n = 0 .
117. Find the general solution of the équation
4xy”+ 2y’ + y = b
Hint. Seek the solution in the form of a sériés jc** (a0 + Ojjc + ^jc 2 + .. •)•
Ans. Cjcos V x + C t sin V~x /
118. Find the solution of the équation (1 —jc2) y*— x y ’ = 0 that satisfies the
1 jc® 1 3 jc®
initial conditions y = 0, y' = 1 when jc = 0. Ans. x 4“ “5^"^
, 1 3 5 jc7 .
326 Ch. 4 Sériés
Find the first three terms of the power-series expansion of the solutions ôf
the following differentiâl équations for the given initial conditions:
dy3
125. y '= x 2-\~y2; for x = 0, y = 1. Ans. \ + x + x 2-\—g—f - . . . .
py2 y3
126. y"= éy+ x; for jc = 0 , y = 1. ÿ ' = 0 . /1ns. 1+ ^ - + ^ + . . . .
X2 X2
127. £/' = sin y — sin *; for x = 0, y = 0 . /4/îs. — ----- g— . . . .
Find several terms of the sériés expansion of solutions of differential équa
tions under the indicated initial conditions:
y2 O y3
128. y" = y y '— xi when x = 0, y = 1 and y ' = 1. Ans. 1 + * + " 2f + “3]—H
FOURIER SERIES
cos n x d x + ÿ bn s in nx dx)
-Jl -JT / ï = l \ - JT -JT ^
i f ' * * - =jiaa
-JT
Jl îl
Consequently,
JT
J f (x) dx = jia0
-JT
whence
JT
“«= -■ § f ( x ) d x (4 )
-Jl
J cos nxsinkxdx = 0 ( 1)
- J•JT
l
Jl
JT
Fig. 126
Applying formula (5), Sec. 5.1, and integrating by parts, \ve find
Fig. 127
-n L- * o
1
ji
r? r l
\ (— x) cos kx dx + \ x cos kx dx
1-----3t 0 J
-3 71 -2 71 - 71 0 71 271 3n x
-1
F ig .'128
Conséquent!y, for the fonction at hand the Fourier sériés has the form
sinx , sin3x , sin5x . . sin ( 2p + l)x
1 + 2p-f 1
This équation holds at ail points with the exception of discontinuities.
Fig. 129 illustrâtes how the partial sums sn of the sériés represent more and
more accurately the fonction f (x) as n oo.
Example 4. A periodic fonction f (x) with period 2 j i is defined as follows:
f( x ) = x a, — (Fig. 130)
Détermine its Fourier coefficients.
JT.
n 2jt2
- ji 3
n
334 Ch. 5 Fourier Sériés
n
_2_ X cos kx n , 1 f , ■
nk k -n + T J C° SkxdX
Thus, the Fourier sériés of the given function has the form
. ji 2 . / cos x cos 2x , cos 3jc
* = 3 ~ '4H ------V - + — s-----
Since the function is piecewise monotonie, bounded and continuous, this équa
tion holds at ail points.
Putting x = n in the équation obtained, we get
n= 1
0 n
1 ji2 n
x dx
fl#=-sr I f{x)dx= T
-n
S■n 0dx+S
0
n ~2~~2
x sin kx I*
cos kxdx = —
— H * ji k |o
1 cos kx
nk k
~ w * i0Tk odd
0 for k even
jt r « n
c 1 T * u a 1 x cos kx « , 1 P . ,
b k = — \ x sir\ kxdx = — \ ------- — -\—^ \ c o s k x d x \
o L o J
for k odd
= _ _ c°s k n = i
— r- for k even
k
336 Ch. 5 Fourier Sériés
number
Putting x = 0 in the équation obtained, we get
1
(2n - l )2
n =î
it follows that, putting x = | —2n, we can Write (for ali c and d):
d d+ 2Jt d+2Jt d + 231
From the property that has been proved it follows that when
computing Fourier coefficients we can replace the interval of in
tégration (— n, n) by the interval of intégration (X, À+ 2 n), that
is, we can put •
\+2n X+2n
This follows from the fact that the function f(x) is, by hypothe-
sis, periodic with period 2 n; hence, both the functions f(x)cosnx
and f(x)s\nnx are periodic functions with period 2jt. We now
illustrate how this property simplifies the process of finding coef
ficients in certain cases.
2jt 2jt
1 T c/ v . 1 P . 1 [*sinrtx . cos nx~\
an= — \ f (x) cos nx dx ==— \ x cos nxdx = — -----------------s— =0
" îl J ' JIJ jx ^ 1 /i2 J o
o 0
2jt 2jï
. 1 P £/ x . . 1 T . , 1 T x cos nx . sin nx 12jc 2_
= — \ / ( jc) sin ha: = — \ jt sin nx dx = — --------------------- 5— =
Ji J î t j jx L / i ‘ n2 Jo n
o o
Consequently,
2 2 2 2
/ (x) = ji — 2 sin jc— —sin 2jc — r- sin 3jc — T sin 4jc — =- sin 5jc— . . .
Z o 4 o
This sériés yields the given function at ail points with the exception of points
of discontinuity (i.e., except the points = 0 , 2ji, 4jx, At these points the
jc
sum of the sériés is equal to the half sum of the limiting values of the function
f (x) on the right and on the left (to the number ji, in this case).
«0 = 4 - J f(x)dx = 0
Jt
aA= — J / (x) cos kx dx = 0 o)
-J t
Jt Jt
bk = J f(x) sin kxdx = — J / (x) sin kx dx
-Jt o
5.5 The Fourier Sériés for a Function with Period 21 339
We obtained the same coefficients as in Example 2, Sec. 5.2, but this time by
a short eut.
22 *
340 Ch. 5 Fourier Sériés
where
n n
Jt
(3)
62 = J [/ (*)—9 WJ*dx
a
the curve y = <p1 (jc) is less than that of the curve y = q>2(*), but the
root-mean-square déviation of the first curve is greater than that of
the second because the curve z/ = <p2 (x) is considerably different from
the curve y — f (x) only on a narrow section and for this reasofi
characterizes the curve y = f(x) better than the first.
Now let us return to our problem.
Let there be given a periodic function f(x) with period 2n.
From among ail the trigonométrie polynomials of order n
+ ^ (a*cos/uc + p*sin&t)
*=i
it is required to find (by ckoice of the coefficients a* and p*) that
polynomial for which the root-mean-square déviation defined by the
équation
ôî f(x) — y —^ (a* cos kx -f- p* sin kx) 1 dx
k= î
has the smallest value.
The problem reduces to finding the minimum of the function
of 2 / i + l variables a„, a lt . . . . a„, pit p„, p„.
Expanding the square under the intégral sign and integrating
termwise, we get
+ + («*cos^ + P*sin ^ ) j
7t 71 n Jl
We note that
n n
— j f(x)dx = a0, - j f (x) cqskxdx = ük
-J l -J l
ji
t + t i > * + 6î>
k=l
we will hâve
I
-J l
è ( û ï+ f c ii) + T ( a o-«o)a
k = 1
(i)
k= I
The first three terms of this sum are independent of the choice
of coefficients a 0, a 1( . . . . a„, plt . . . . p„. The remaining terms
l ( a 0 - a 0) * + | +
k= i
are nonnegative. Their sum reaches the least value (equal to zéro)
346 Ch. 5 Fourier Sériés
4 +
i t (at + w = - k ] f t (x ) d x <3')
fc=i - ji
which is called the Lyapunov-Parseval équation. We note that
this équation has been proved for a broader class of functions
than that which we here eonsider.
From what has been proved it follows that for a function which
satisfies the Lyapunov-Parseval équation (in particular, for any
bounded pieeewise monotonie function), the corresponding Fourier
sériés yields a root-mean-square déviation equal to zéro.
5.7 Mean Approximation of a Given Function 347
w here
ji n
s"(*) = i S f (0 d t
-Jt
Jt Jt
-Jt
Jt Jt "I
N ow ta k in g — o u ts id e th e b ra c k e ts an d re p la c in g th e sum of in tég
ra is by th e in té g ra l of th e su m , we o b ta in
H t)
^ [ / ( 0 cos kx cos k t + f (t) sin kx sin kt] t dt
k=i
or
th e n
2 <j„ (z ) cos z = cos 2 + 2 cos 2 cos 2 + 2 cos 2 cos 2 2 + . . .
+ 2 cos 2 cos nz = cos 2 + (1 + cos 2 2 ) + (cos 2 + cos 3 2 )
+ (c o s 2 2 + c o s 42) + . . . + [ c o s ( n — l ) z + c o s (n + l ) z ]
= 1 + 2 c o s 2 + 2 c o s 2 2 + . . . + 2 c o s ( n — l ) z + c o srtz + c o s (n + l ) z
or
2cr„ (2 ) cos 2 = 2a„ (z )— cos n z + cos ( n + 1) z
cos nz — cos ( n + \ ) z
On ( 2 ) 2 ( 1— cos z)
But
cos n z — cos (n + l ) z = 2 s i n ( 2 n + l ) - |- s i n - |-
1 — cosz = 2 3 ^ - 2
350 Ch. 5 Fourier Sériés
H ence,
sin (2n + 1)
<*«(*) = ----------—
2sin T
T h u s, é q u a tio n (1) m ay be re w ritte n as
t —x
sin (2/t + 1)
s „ (*) = -£• \ / ( O t —x
dt
2 sin
t —x
sin (2 n + 1)
M *) = - \ H t) ■dt
n2 sin
■ t——x
t-—
t — x = a, t = jc + a
T hen w e g et th e fo rm u la
sin ( 2 n + l ) y
stt(x) = - \ f ( X + a) da (2)
4S 2sinf
T he in té g ra l on th e rig h t is D i r i c h l e t ’ s i n t é g r a l .
In th is fo rm u la p u t / ( x ) = 1 ; th e n a 0 = 2, a k = 0, b k — 0 w hen
k > 0; hence, s „ ( * ) = l for an y n a n d we g et th e id e n tity
sin (2 n + l)-j
/w -4 \.f w a da
2sini
-Jt
n a
+ - f [/ (Jf + a) —/ W ] ------ -- sin na da
i 2sinT
n
+i Ç [f(x+ct) — f (x)]jCOsnada
-3 1
Put <D, (a) = flx + ay -f (*).' Since f (x) is a bounded piecewise con-
tinuous function, it follows that O, (a) is also a bounded and
piecewise continuous periodic function of a. Hence, the latter inté
gral approaches zéro as n —►oo, since it is a Fourier coefficient
of •this function. The function oc
cos T
(«) = [/ (* + « ) - / ( * ) ] -------z—
2sinT
is bounded wnen —j t < I a < — 6 and and
| ^ ( a ) | < [ M + M ]— V
2sinT
where M is the upper limit of the quantity |/(x )|. Also, the func
tion <I>2 (a) is piecewise continuous. Hence, by formulas (5) of
Sec. 5.7, the second and third intégrais approach zéro as n —*-oo.
We can thus Write
i r cosf
üm [sn(x) — f(x)] =lim - - [ [ / ( r + a ) - / ( j : ) ] ----- — sin n a d a ( 1)
«-►<» n-+ » 2 s in —
= lim - J — cqs«
—o-o « si„« 2
a
2
= lim lim lim cos-- = kx- 1 .1 = k x
a -> 0 -0 CL—►O — 0 -
g j jj ^ CL—►O — 0
* If conditions (1) and (2) are fulfilled, then we say that the function f (x)
has, at the point x, a dérivative on the right and a dérivative on the left.
Fig. 139 shows a function where /?l = tancp1, fc2 = tan<p2, kl ^=k2y H kl = k2,
that is, if the dérivatives on the right and left are equal, then the function will
be différentiable at the given point.
2 3 -2 08 2
354 Ch. 5 Fourier Sériés
bh = J / (x) sin kx dx
-Jt
These values are determined either from a table or from the graph
of the given function (by measuring the corresponding ordinates).
Then, taking advantage, for example, of the rectangular formula
[see formula ( 1'), Sec. 1 1 .8 , Vol. I], we détermine the Fourier
coefficients:
n n n
2ji
J / (*) (cos nx— i sin nx) dx = — ^ f (x) e~lnxdx
Thus
3t
C-» = Ü
f(x)einxdx (5")
-Jt
The expression for c0 and formulas (5') and (5") can be combi-
ned into a single formula
JT
cn and c_„ are termed the complex Fourier coefficients of the fun-
ction /(x).
If the function f(x) is periodic with period 21, then the Fourier
sériés of /(x) will be
00
/(x) = ^ + £ a nc o s - ^ x + b,,sin-^-x (7)
n —1
/( * ) = E v ' v ( 10 )
flS-OD
The set of wave numbers forms what is known as a spectrum.
Plotting these numbers on a number axis, we get a collection of
separate points. This collection of points is called a discrète collec
tion, and the associated spectrum is termed a discrète spectrum.
The coefficients cn defined by formulas (9) are called the complex
amplitude. In some engineering texts, the collection of moduli of
the amplitudes, | c„ |, is also called the spectrum of the function / (x).
J / ( 0 sin ^ p / d /^ sin ^ p x
_1_
2/ j/(/)rf/ + - ^ j/(/) [cos-ï / cos^x + sitiT /sinT x] ^
or
l 00 l
/(*) = i j f P + jZ f/(/)c o s Î2 i^ rf/. (4)
-/ *=* -/
Let us investigate what form expansion (4) will take when pas-
sing to the limit as /->oo,
We introduce the following notation:
2 ji kn
«1 = 7 -. «« = — . «*= T . an d A a t = • (5)
f (x) =
j f w dt + TT2 ( j* f ( 0 cosa» (*—*)d/) Aa* (6 )
*=i \_r, /
As /-►oo, the first term on the right approaches zéro. Indeed,
I l 00
-/ -Z -00
For any fixed /, the expression in the parenthèses is a function
of <xA [see formulas (5)], which takes on values from j to oo. We
will show, without proof, that if the function f (x) is piecewise
monotonie on every finite interval, is bounded on an infinité in
terval and satisfies condition ( 1), then as / -* + oo formula (6 )
takes the form
CO / O
O V
/ (*) = —■j y J f (t) cos a (t—x) d t j da (7)
- ^ j / (0cos«V — x ) dx = /(*+Q)+/(*-Q)
2
(T )
360 Ch. 5 Fourier Sériés
If /(x) is defined only in the interval (0, oo), then tor x > 0 it
may be represented by either formula (9) or (10). In the first case
we redefine it in the interval (—oo, 0 ) in even fashion; in the
latter case, in odd fashion.
Let it be noted once again that at the points of discontinuity
one should write the following expression in place of /(x) in the
5.13 Fourier Intégral 361
FW=ÿlc]f(t)<XK*tdt ( 12)
o
then formula (9) takes the form
/— *
f(x)= y j F (a) cos ax da (13)
o
The function F (a) is called the Fourier cosine transform of the
function f(x).
If in (12) we consider F (a) as given and f (t) as the unknown
function, then it is an intégral équation of the function f (t). Formula
(13) gives the solution of this équation.
On the basis of formula (10) we can write the following équations:
/— 00
<S>(a) = y J f (t) sin at dt (14)
o
/----*
f (x) = y J (a) sin ax da (15)
o
The function O (a) is called the Fourier sine transform.
362 Ch. 5 Fourler Sériés
Example. Let
f(x)=e~Q* (P > 0, jc =3:0)
From (12) we détermine the Fourier cosine transform:
oc
(1)
—oo —oo
Let us now consider the following expression, which is identically
equal to zéro:
/ (/) sin a (t—x) dt )da — 0
or
(2)
5.14 The Fourler Intégral in Cômplex Form 363
provided that each of the limits to the right exists (see Sec. 11.7,
Vol. I). But in équation (2 ) we wrote
M
[ tp (a) da = lim C <p(a) da (**)
J M-*•ce-M
«-1_
Obviously, it may happen that the limit (**) exists, while the
limits on the right side of équation (*) do not exist. The expres
sion on the right of (**) is called the principal value of the in
tégral. Thus, in équation (2) we consider the principal value of
the improper (outer) intégral. The subséquent intégrais of this sec
tion will be written in this sense.
Let us multiply the terms of (2) by—^ and add them to the
corresponding terms of ( 1); we then get
oo r » “i
/(*) = — J J / (/) (cos a (f —x) — ts in a ( f —x))dt da
or
or, compactly
f(x)= JC (a) ela* d a (5)
'•OO
where
00
c («) = •— J f ( t ) e ~ ,a‘ d t (6)
364 Ch. 5 Fourier Sériés
the équation
= --------------- (7)
l tp| (X) dx
366 Ch. 5 Fourler Seriez
The coefficients ck computed from formulas (7) are called the Fou
rier coefficients of the function f (x) relative to the system of ortho
gonal functions (1). Sériés (6 ) is called the Fourier sériés relative
to the system of functions ( 1).
Définition 2. The orthogonal system of functions
J / 2(x) d x < o o
which were considered in Sec. 4.23. Here, A,, Xît . . . , A,-, . . . are
the roots of the Bessel function, that is, numbers such that satisfy
the relation
J n(h) = 0 (t = 1, 2, ...)
We will now point out, without proof, that the system of func
tions
A = 2 A fit
(AB) = £ AiBi (2 )
\A-A\ = \ 2 ( A - A ) ! =0
r i- 1
The angle <p between two vectors is defined as
(AB)
costp = \ A\ - \ B\ (6)
5.16 The Concept of a Linear Function Space 369
Clearly, if /(*) ==(p (x), that is, / ( jc) and q>(jc) coïncide at ail
points of the interval [a, b], then | | / —qp|| = 0. But if | | / —qp|| = 0,
then / (jc) = q>(jc) at ail, except a finite number, points of [a, b].*
But in this case we also say that the éléments of the space O are
identical.
A space of piecewise monotonie bounded functions, in which are
defined the operations (7), (8 ) and the metric is given by équ
ation ( 11 ), is called a linear function space with quadratic metric.
The éléments of the space <I> are called points of the space or vectors.
Now let us consider a sequence of functions
<Pi(*). •••. <P*W. ••• (12)
lying in the space <I>.
The sequence of functions (12) is said to be orthogonal on the
interval [a, b] if for arbitrary i, / (i j) the équations
b
(<Pi. <Py) = S <P, (x) <P/ (x) d x ^O (13)
a
hold true.
On the basis of équations (I), Sec. 5.1, it follows that, for instance,
the System of functions
1, cos x, sin x, cos 2x, sin 2x, cos 3x, sin 3x, . . .
is orthogonal on the interval [ —n, «].
We will now show that expanding a function in a Fourier sériés
of orthogonal functions is similar to the décomposition of a vector
into orthogonal vectors. Suppose we hâve a vector
A = A & + A,e2+ . . . + Akek + . . . + Anen (14)
We assume that the vectors elt e2, . . . , en are orthogonal, that is,
if i /, then
(efij) = 0 (15)
To détermine the projection Ak, we form the scalar products of
each member of (14) by the vector ek. By properties (2) and (3)
we get
{Aek) = Ax (e1ek) + A2(e2ek) + . . . + A k (ekek) + . . . + A n (enek)
Taking account of (15) we obtain
(Aek) = Ak (ekek)
whence
< * = ' ’ 2 .........») 06)
* There can also be an infinité number of points, where / (jc) ^ < p (jc ).
Exercises on Chapter 5 37!
Further assume that the function f(x) has been expanded relative
to a System of orthogonal functions:
/(*) = k=î
S ûa<Pa(*)* (17)
Forming scalar products of each member of (17) by q>*(x) and
taking into account (9) and (13), we get*
( / . <Pa) = M < P a » <Pa)
w h en ce
b
\ f(x)<f (x) dx
„ _ (f •?>) __ a_________ (18)
* (f*. <P*) *
] 1<P* M l2 dx
Exercises on Chapter 5
1. Expand the following function in a Fourier sériés in the interval ( — ji, ji):
f(x) — 2x for
f( x ) = x for — ji < x C 0
- —1t- ji-------
Ans. 2 cos x , cos Sx , cos 5* lin 2x
4 ji l2 32 52
sin Sx
...)
2. Taking advantage of the expansion of the function f(x )= 1 in the interval
(0, ji) in thesines of multiple arcs, calculate the sum of the sériés 1— 5- + 4 ™
o 5
—y + ... . AnS. y .
* We présumé that the sériés obtained in the process converge and that
term-by-term intégration is legitimate.
372 Ch. 5 Fourier Sériés
/W = -2 (n - x ) for 0^ jc < ji
/ ijijc
(—l) " -1 n sin —j—
+n (<•'-«-*)X l* + n*n*
12. Expand the function f(x) = 2x in a sériés of sines in the interval (0, 1),
Exercises on Chapter 5 373
a 4 sin nnx
Ans-- z 2 *
n—1
13. Expand the function f ( x ) = x in a sériés of sines in the interval (0, /).
- . nnx
Ans. (-!)■ ♦ » —
n n - 1
' ' n
14. Expand the function
x for 0 < x 1
/«-x for 1 < x < 2
in the interval (0, 2): (a) in a sériés of sines; (b) in a sériés of cosines.
. (2/i+ 1) n*
sm----- T ~
/tns. ( a + 2 £ ( —l)n
n= 0
(2/1+ 1)»
00
... 1 4 cos (2/i + 1) nx
W (2/1+ I)'2
CHAPTER 6
^ -L ^ - 0 /3\
dx* ^ dy*~ U W
dt * ~ a Vdx2 ‘ d y 2) 0 ')
6.2 Deriving the Equation of the Vibraiing String 375
hold:
u (0 , 0 = 0 (2)
u(l, o = o • ( 2 ')
where AQ is the quantity of heat that has passed through a cross section S
during a time At.
6.4 The Equation of Heat Conduction in a Rod 383
The influx of heat AQX—AQt into the rod element during time
At will be
[-*& L H - H
k^A xSA t (4)
where cosa, cosp, cos y are the direction cosines of the vector n, or
du j
■^ = »gradn
But this is the heat that has entered the volume V during the
time At\ it is defined by formula (2). Thus, we hâve the équation
— At kn grad uds = At c p dv
where the point P (xlt ylt zx) is some point in the volume V.
2 5 — 2082
386 Ch. 6 Equations of Mathematical Physics
If k is a constant, then
div (k grad u) = k div (grad u) - k
and équation (6 ) then yields
_ co — - k (V—
P àt
4- — -4- — \
àx* ^ dy* ^ dz2 J
âT = û Au
where =^ + is the Laplacian operator. Equation (8 )
is the équation of heat conduction in space. To find its unique
solution that corresponds to the problem posed here, *it is necessary
to specify the boundary-value conditions.
Let there be a body £2 with a surface a. Iri this body we con-
sider the process of heat transfer. At the initial time the tempe-
rature of the body is specified, which means that the solution is
known for / = 0 (the initial condition):
u (x , y, z, 0) = qp (x, y y z) (9)
In addition to that we must know the température at any point M
of the surface a of the body at any time t (the boundary condi-
6.6 Solution of the First Boundary-Value Problem 387
tiùn):
u (M, t) = \p(M, t) (10)
(Other boundary conditions are possible t®o.)
If the desired function u (x , y, z, t) is independent of z, which
corresponds to the température being independent of z, we obtain
the équation
of the équation
du . d*u
-d t= a S * (4)
that satisfies the boundary-value conditions
u (x , 0) = <p(;r), (5)
u (0, 0 = f i ( 0 . o < * < r ( 6)
u (i, /)= ♦ ,(/). o < / < r (7)
that is, we hâve to find the solution u (x, t) in a rectangle bound-
ed by the straight Unes / = 0, x = 0, x = L, t — T, if the values
df the desired function are giv- .
en on thfee' of its sides: / = 0 , £i|.
J
î
1
\(x,t+0 JM î
i
(H.M) ajt>(MM
(x-h,t) (x,t) (x+h,t) 7
I
L x
F ig . 143 Fig. 144
or
* Since from the meaning of the problem T (t) must be bounded for any t
T1
!f q>(*) is bounded, it follows that y must be négative. And so we Write—A,2.
6.7 Heat Transfer in an Unboundéd Rod 391
90
00 ®
= Ç (p(a)cosX(a—x)dajdX
aX Y t = z. (14)
a Y~t
We dénoté
G
O
/C (P) = $e"z’ cos pzdz (15)
o
Differentiating,* we get
oo
K' (P) = — $ z sin pz dz
o
Intégrât ing by parts, we find
oo
K' (p) = y [e-** sin N " — ^ e~**cos Pzdz
/C (p) = Ce * . (16)
AT(0 ) = j V **dz-
And so
*(P) = 2 e (17)
Put the value (17) of the intégral (15) into (13):
®. - p*
j e-o*x,»/CosX(a—x)dX = — -^Ç—e 4
(see Secs. 3.5 and 3.6). The amount of liquid in the volume ©
at time t was
(ù
During time At the amount of liquid will change (due to changes
in density) by the amount
Q= jjjA p d » * 4 ( (7)
0 ©
Assuming that there are no sources in the volume <o, we con-
clude that this change is brought about by an influx of liquid to
an amount that is determined by équation (6 ). Equating the right
sides of (6 ) and (7) and cancelling out At, we get
(8)
S 0
We transform the iterated intégral on the left by Ostrogradsky’s
formula (Sec. 3.8). Then (8 ) will assume the form
0 0
or
I W ( - t ~ div(pv)) d£ù=0
0
Since the volume © is arbitrary and the integrand is continuous,
we obtain
div(p®) = 0 (9)
6.8 Stating Bûundary-Vatue Problems 397
or
aT- T x (P°*) •
~ T z^ = 0(Q')
This is the équation of continuity of a compressible liquid.
Note. In certain problems, for instance when consitfering the
movement of oil or gas in a subterranean porous medium to awell,
it may be taken that
®= — -j^-grad p
% + div (k grad p) = 0
or
that is, the potential function of the velocity (p must satisfy the
Laplace équation.
398 Ch. 6 Equations of Mathematical Physics
^ £ 4 _ ^ î = v0
dx*^dy* (14)
Boundary-value conditions of type (2), the Dirichlet problem, or
of type (3), the Neumann problem, are specified on the contour C.
III. The potential of a steady-state electric current. Let a ho-
mogeneous medium fill some volume V, and let an electric current
pass through it whose density at each point is given by the vector
J(x, y, z) = J J + J y J + J zk. Suppose that the current density is
independent of the time t. Further assume that there are no cur
rent sources in the volume under considération. Thus, the flux of
the vector J through any closed surface S lying inside the volume
V will be equal to zéro:
Jnds = 0
s
where » is a unit vector directed along the outer normal to the
surface. From Ostrogradsky’s formula we conclude that
div / = 0 (15)
The electric force E in the conducting medium at hand is, on the
6.9 The Laplace Equation in Cylindrical Coordinates 399
e =T . (l6)
or
J=XE
where X is the conductivity of the medium, which we shall con-
sider constant.
From the general electromagnetic-field équations it follows that
if the process is stationary, then the vector field E is irrotational,
that is, rot E = 0. Then, like the case we had when considering
the velocity field of a liquid, the vector field ispotential (see Sec. 3.9).
There is a function <p such that
£ = g ra d ç (17)
From (16) we get
J=Xgrad<p (18)
From (15) and (18) we hâve
Xdiv(grad <p) = 0
or
d x * ^ ‘ dy*~t dz2 (19)
We thus hâve the Laplace équation.
Solving this équation for appropriate boundary-value conditions,
we find the function qp, and from formulas (18) and (17) we find
the current J and the electric force E.
u = ut («2 —Ul) ~ (H )
<n
We shall seek the solution by the method of séparation of variables,
placing
u = Q(<p)R(r) (3)
Substituting into équation (T), we get
r e0 (q>) R" (r) + r 0 (9 ) R' (r) + 0 " (9 ) R(r) = 0
or
0 *(9 ) r»/?»Q-) + r / ? » ,a
0(9) /? (')
(4)
Since the left side of this équation is independent of r and the
fight is independent of 9 , it follows that they are equal to a
constant which we dénoté by —ka. Thus, équation (4) yields two
équations:
0" (9 ) + &20 (9 ) = 0 (5)
r2Rn(r) + rR, ( r ) - k iR(r) = 0 (5')
The general solution of (5) will be
0 = A cos fop + B sin £ 9 (6 )
We seek the solution of (5') in the form R ( r ) = r m. Substituting
R(r) = rm into (5'), we get
r2m (m— 1)rm~i + rrnrm~1—klrm= 0
or
m2 —£* = 0
Thus, there are two particular linearly independent solutions r* and
r~k. The general solution of équation (5') is
R = C r k + Dr~k (7)
We substitute expressions (6 ) and (7) into (3):
u„ = (Ak cos £ 9 + Bk sin £9 ) (Ckrk + Dkr~k) (8 )
Function (8 ) will be the solution of (T) for any value of k different
6.10 The Solution of Dirichlet's Problem for a Circle 403
from zéro. If k = 0, then équations (5) and (5') take the form
0 >" = 0 , rR"(r) + R'(r) = 0
and, consequently,
«o = (4» + B0V)(Co + £>olnr) (8')
The solution must be a periodic function of <p, since for one and
the same value of r for <p and <p+ 2 n we must hâve the same so*
lution, because one and the same point of the circle is considered.
It is therefore obvious that in formula (8 ') we must hâve Bg— 0.
To continue, we seek a solution that is continuous and finite in
the circle. Hence, in the centre of the circle the solution must be
finite for r = 0 , and for that reason we must hâve £>„ = 0 in (8 ')
and Dk = 0 in (8 ).
Thus, the right side of (8 ') becomes the product AgCt, which
we dénoté by A j 2 . Thus,
Thus, the sériés (9) with coefficients defined by formulas (11) will
be a solution of our problem if it admits termwise itèrated diffé
rentiation with respect to r and <p (but we hâve not proved this).
Let us transform formula (9). Putting, in place of An and B„,
their expressions ( 11) and performing the trigonométrie transfor
mations, we get
Jt ooJt
= / ( 0 | 1 + 2 2 (- --)" co s« (/—f j d / ( 12 )
-n L n=\ J
Let us transform the expression in the square brackets: *
oo ao
1 + 2 ^ (-4 -Y cosn (t — qp) = 1 + £ ( j —Y^einV-V + e-lnV-it)]
n=l * ' rt=l ^ K '
= 1
R R
+
1 __ C_e*(t-<p) i ___
R R
2
- ( t ) /?*— /■*
R2— 2Rr cos (t—<p) + r 2 (13)
l-2 -+ c o s « -< p )+ (+ -)2
by finite différences:
d*u ui + i , k ~ ^ ui,k~\~ul - \ . k
dx* x = ih , y= kh h3
(Pu «Uti — +
dy 2 x = ih , u-kh /J2
The differential équation ( 1) is replaced by a différence équation
(after cancelling out h2):
u t + 1, *— 2u/, * + «,-!, * + «/, *+i 2u(>A+ «,•, =0
or (Fig, 147)
Ui, ft = "î" (M/+J. + *+l + Ui-l, ft + U(, A-l)(4)
For each lattice point of the grid lying inside D* (and not lying
on the boundary C*), we form an équation (4). If the point (x = ih,
y --=kh) is adjacent to the point of the contour C*, then the right
side of (4) will contain known values of /*. Thus, we obtain a
nonhomogeneous System of N équations
(i.k+1) in N unknowns, where N is the number
y\
of lattice points of the grid lying in
(i-W GJc) ïm ) side D*.
We shall prove that the System (4)
has one, and only one, solution. This
is a System of N linear équations in N
0 X unknowns. It has a unique solution if
Fig. 147 the déterminant of the System is not
zéro. The déterminant of the System is
nonzero if the homogeneous System has only a trivial solution. The
System will be homogeneous if /* = 0 at the lattice points on the
boundary of the contour C*. We shall prove that in this case ail
the values uit k at ail interior lattice points of the grid are equal
to zéro. Inside the domain, let uit k be different from zéro. For the
sake of definiteness, we assume that the greatest of them is posi
tive. Let us designate it by uit k > 0 .
By (4) we write
« / . * = T (« / + ! . * + “ /.* + ! + « , - X ,* + M / , * - l ) ( 4 ')
ç(*) = ^ for 0
x (l-x )
u( 0 , t) = u (/, / ) = 0 , u(x, 0)
" /2
oo (2/i+ 1)f TPaït . (2n+ 1 ) nx
An*. u(x, “ sin ------— — .
n= 0
9. Find a solution to the équation '^ " ==aa^ that satisfies the conditions
du
= 0, u(l, t) = u0, u ( x , 0) = <p(x)
dx x = o
Point out the physical meaning of the problem.
00
Ans. u (x, 0 = ^ 0 + *cos ^ where
n =0
/
(2/t+l)n,x (—l)n4u^
^n = 7 - jW )C O S X ji(2/i + 1) *
2/
équation tan fi = — .
Hint. At the end of the rod (when x = l) a heat exchange occurs with the
environment, which has a température of zéro.
410 Ch. 6 Equations of Mathematical Physics
11. Find [by formula (10), Sec. 6.6, putting h = 0.2] an approximate solution
to the équation - ^ - = 2 that satisfies the conditions
- . . 2i4 1
sin
MIX
Ans. u(x, 0 = — > —*
7 ji n
< M Çe-t*-»,>*dt=> —
M § e ~ ates«t d t
o
In similar fashion we evaluate the second intégral. Thus, the in
tégral J e~pif ( t ) d t exist. It defines a certain function of p, which
o
we dénoté* by F (p):
oo
F (p) — \ e ~ptf (t) d t (4)
o
The function F (p) is called the L apl ac e t r a n s f o r m , or the L - t r a n s -
forrn, or simply the t r a n s f o r m of the function f ( t ) . The function
f ( t ) is known as the o r i g i n a l f un c t i o n. If F (p) is the transform
of /(/), then we Write
F(p)+f(t) (5)
f(t ) + F( p) (6)
L{ f ( t ) } = F( p) (7)
As we shall presently see, with the help of transforms it is pos
sible to simplify the solution of many problems, for instance, to
reduce the solution of difîerential équations to simple algebraic
* The function F (p), for p 7= 0, is a function of a complex variable (for
example, see V. I. Smirnov’s Course of Higher Mathematics, Vol. III, Part 2,
in Russian). The transformation (4) is similar to the Fourier transformation consi
dérée! in Sec. 5.14.
7.2 Transforms of the Functions <J0 (t), sin t, cos t 413
L { i ( a t ) ) = ^ e ~ Zf(z)dz
0
or
m m ) - ^ r ( i)
Thus, if
F( p) + Ï (t )
then
-f(f)+/(«o ai
7.4 The Linearity Property of a Transform 415
• • (*> ■ *»
or
sin at ■ (2)
• p2+ a a
Example 2. From (3), Sec. 7.2, by (1), we obtain
P_
cos at
. 1 a
' * (-5 )* + .
or
cos at + P
• p2+ a2 (3)
f(t)= ic,f,(t) (O
i= I
(C/ are constants) and
F( p)+f(t), F M + f r f )
then
F { p ) = ' k c , F l {p) (!')
i=i
Proof. Multiplying ail the terms of (1) by e~Pf and integrating
with respect to t from 0 to oo (taking the factors C (- outside the
intégral sign), we get (T).
Example 1. Find the transform of the function
/ (/) = 3 sin 4t —2 cos 51
Solution. Applying formulas (2), (3), Sec. 7.3, and (T), we hâve
L If (t)\ —3 ^ - 2 ^ ^ ______
' '' p2+ 1 6 p2+ 2 5 ~ p 2+ 1 6 p2+ 25
Example 2. Find the original function whose transform is expressed by the
formula
5 20p
F (P)
P2+ 4 P2+ 9
Solution. We represent F (p) as
5 2
416 Ch. 7 Operational Calculus and Certain of lis Applications
From the uniqueness theorem, Sec. 7.1, it follows that this is the on 1y original
function that corresponds to the given F (p).
or
+ + 2 -^ s in h a / (2 )
Similarly, by adding ( 1) and (T), we obtain
i^ c o s h a / (3)
7.7 Différentiation of Transfor ms 417
f ( p ) = p2+ 1 0 p + 41
Solution. Transform F (p) to the form of the expression on the left-hand
side of (4):
7 7 7 4
p2+ 1Op + 41 (p + 5)2+ 1 6 4 (p+ 5)*+ 4*
Thus
7 4
F(p) =
4 (p + 5)2+ 42
Hence, by formula (4) we will hâve
F (p) sin 4/
Example 2. Find the original function whose transform is given by the
formula
F (p) -- P+ 3
P2+ 2p + 10
Solution. Transform the function F (p):
p+ 3 _ ( p + l ) + 2p + 1
p2 + 2 p + 1 0 ( p + l ) 2+ 9 (p + l)* + 3*
_|_ 2 P+1 | _2_
(p + l)* + 3* ( P + l ) 2+ 32 3 ( p + l ) 2+ 32
Using formulas (4) and (5), we find the original function:
F (p) +-► cos 3/ + - |- e - f sin 3/
o
exists.
27—2082
418_______ Ch. 7 Operational Calcul us and Certain of lis Applications
^g -(a-e) (t)\dt
0
We then evaluate the intégral (2 )
oo oo
\e~P*f(t)dt
o
And so, from formula
oo
( - l ) n £-nF(P) = $ e - pttnHt)
0
which is formula ( 1).
Let us use (2) to find the transform of a power function. We
write the formula ( 1), Sec. 7.2:
( t
or
Similarly
*t cos at (5)
(P2+ a2)2
Example 3. From (1), Sec. 7.6, by (1), we hâve
1
*te- a t ( 6)
(p+a )2
Theorem. If F ^ - ^ f i t ) , then
p F (p )-H O )^f'(t) ( 1)
Proof. From the définition of a transform we can Write
ce
L { r ( t ) } = l e - r ‘f' (t )d t (2 )
o
We shall assume that ail the dérivatives f'(t), f ( t ) .........
which we encounter satisfy the condition ( 1), Sec. 7.1, and, con-
sequently, the intégral (2 ) and similar intégrais for subséquent
420 Ch. 7 OperationaL Calculas and Certain of lis Applications
p F ( p ) - ^ r (t)
p« F (p )-f-* r(o
oc
No. P(P) = $ e - ptHt) dt fit)
0
•
1
1 1
P
a
2 sin at
P%+ a2
3 P cos at
P4 +<**
1
4 e
P+ «
a
5 sinh at
p2— a 2
6
P cosh at
p2— a 2
a
7 e~at sin at
(p + a )4+ a%
P+ a e~at cos at
8 ( p + a ) 2-(-a*
ni t«
9 pn +l
2pa t sin at
10 (p2+ a2)2
p2— a2 t cos at
11
(p2+ a2)2
1 t e - at
12
(P + a )2
1 ^ (sin at— at cos at)
13 (p2 + a2)3
ao ] e~ptÿ z dt + ai ] e~pti ï £ r d t +
0 0
oo oo
+ . . . + an j e-p,x (t) dt = j e -# f (t) dt (3 )
7.10 An Auxiliary Equation for a Given Differential Equation 423
$ ■ + * -'
that satisfies the initial conditions: x b = x 'q= Q for t=Q.
Solution. Write the auxiliary équation (4')
7.10 An Auxiliary Equation for a Given Differential Equation 425
x(t) = - - ~ c o s 3 l + ± -
i £ + * w + 2x= t
that satisfies the initial conditions: xo = x'o = 0 for / = 0 .
Solution. Write the auxiliary équation (4')
x (P) (P2 + 3p + 2) =
or
* (0 = Ÿ < - | + e - ' - { « - ît
Example 4. Find the solution of the équation
d2x . _ dx , e . .
~ d F + 2 i r + 5 x = S in t
satisfying the initial conditions *0= 1 , x0 = 2 for / = 0.
Solution. Write the auxiliary équation (4'):
*(p) (p2+ 2 p + 5) = p - l + 2 + 2 1 + L { s in < }
or
x(p)(p2 + 2p + 5) = p + 4 + — -T
or
ü P+l | 29 2 1 p , 1 1
KP) 10 ' ( p + l ) 2 + 22 - r 10.2 * ( p + l ) 2+ 22 10* p * + l ’1" 5 pa+ l
Applying formulas 8, 7, 3, and 2 of the table, we get the solution
, , , 1 1 f 0 , , 29 # 1 , - 1 . ,
* (0 = î ô ^ fcos 2/+2Ôe" sin2/” Jô cos/+ “5 sin*
or, finally,
11 OQ \ 1 1
( ygcos 2/ + 2Q sin 2 t J — jq c o s i + g - s i n /
11. — ^ — ;
(p—a)*
i \ - xeat (i)
(p—a)k ' — O*
p+
_ ^ iiW !z îL -
( ',+ t ) ’+ ( Ÿ t) ( ,,+ t ) * + ( V * - + )
+(b- ^ V
w m a - )' 4
2
M^+Ae COS t y r “. - 4 -
N ( B - 4)
And, finally,
Ap+B
P2+ ai P + at
-± tr /---
A cos t sinrf Y aa ( 2)
l_
We shall not consider a type IV partial fraction, since it would
involve considérable calculations. We shall consider certain spécial
cases below. For further information see the list of books at the
beginning of this chapter.
«8 Ch. 7 Operational Calculus and Certain of tts Applications
^d“X . . = sin3jc
T+4* . „
*(<) = 2e-t+ e 2 ( - c o s ^ ^ s i n Ç f )
Example 3. Find the solution of the équation
d2x
+ j c = t cos 2 t
dt*
that satisfies the initial conditions: x = 0, xo = 0 when f = 0 .
7.13 The Convoi ut ion Theorem 429
Obviously, the. operational method may also be used to solve Systems of linear
differential équations. The following is an illustration.
Example 4. Find the solution of the set of équations
i r + 4 i n ' + 3!/=0
that satisfies the initial conditions: x = 0, y = 0 when t = 0.
Solution. We dénoté x ( t ) ^ r x( p) t y (t) -«-f- y (p) and write the System ôf
auxiliary équations:
(3p + 2) x ( p) +pÿ ( p) = j
Px(P)-h(*P + 3)ÿ(p) = 0
Solving this System, we find
—, 4p + 3 _ 1 133
X(p> p ( p + 1) (1 lp + 6) 2p 5 ( p + l ) 10(11/7 + 6)
yyfJ’ (1 lp + 6) (p + 1) 5 V P-r * Hp + 6 J
From the transforms we find the original functions, i.e., the sought-for solutions
of the System:
\ h (T) t
0
that is,
*
F1(p)F2( p ) ^ l f l (r)fAt-*)dT ( 1)
0
S f i (*)/*(t— x)dx
0
from the définition of a transform:
e - p t
{} fi (T)/* (<—T)dtj = 5 S fi W ^ (/—t) dx^dt
Pi(p)FAp)
0
7:13 The Convoi ut ivn Theorem 43»
d?x
~£+*=r«)
that satisfies the initial conditions: x0 = *i = 0 for / = 0 .
Solution. Write the auxiliary équation
X(P) (p2+ l) = F (p)
Applying the convolution formula (1) and denoting j = F 2 (p), F (p) = F1(p)t
we get
t
X (t) = 5 f (T) sin (* — T) dt (2)
0
L àt + I*i J r % = E (2)
(3)
From (3) we get
d*Q _ di
dt2 di (3')
* See, for example, Sec. 1.26, where such an équation is derived in consi-
dering the vibration of a weight on a car spring.
7.15 Solution of the Differential Equation of Oscillât ions 433
L dP + * dt + C 1 ~ dt ' 5'
And so, from (3), taking into account (4) and (5), we get
r J_V i
—t>-%-t
2 , i/ al . 2 . , i /r oî
x (t) =e je, cos/ |/ û2— Y H----- . sin t y a2— j-
lV/ aa*— 4"
The solution of this équation xIr that satisfies the initial condi
tions x = x„, x' = x„ for t = 0 is given by the formula (4), Sec. 7.15,
or by the first term of (6 ):
xfr (t) = e~ nt j^JCocos k j + x°n sin k j J (2)
the form
+ 2n Tt +■k%x = A sin (ù/ O)
To détermine the nature of the motion it is sufficient to consider
the case when xn= x'o= 0. One could obtain the solution of the
équation by formula (6 ), Sec. 7.15, but we obtain the solution by
carrying out ail the intermediate calculations.
Let us write the transform équation
x(p)(p2+ 2np + k2) = A P2+“ (t>2
from which we get
x(p) A(ù
(p* + 2np + k*) (p2+(o2) ( 2)
We consider the case when 2n=F0 (n2 < k2). Décomposé the frac
tion on the right into partial fractions:
Am Np + B Cp+ D
(p2+ 2 n p -f-k2) (p*-|-(o2) p2 2 np -\-k2
' p2+ co2 (3)
The type of oscillations for the case k^>a> is shown in Fig. 151.
Let us again return to formula (4). If 2n > 0 (which occurs in
the mechanical and electrical Systems un- xai
der considération), then the term containing
the factor e~nt, which represents. damped
natural oscillations, rapidly decreases for I I i I |'
I I I T • .
increasing t. For t sufficiently large, the I1 11I I
character of the oscillations will be deter- I I I h |!
mined by the term that does not contain 11i ! ! ! 1 M !
the factor e~nt; that is, by the term
i ! i ! ! 1 1 1 1l 1i
| 1 | I I ! I
x(0i i 1 1 1 11 M
* <*>= (*«-«,*)* + 4^ {(*»-«>«) sin o,/
— 2 niocoso)/} (7)
We introduce the notation Fig. 151
4(*»-<o2) =M cos 6 , A-2ruù ,, . c
(k1
(V — or8)2+ 4n2(û* (k*- -- rr.,
-to2)2+ 4n2co2
= M Sin 0 ( 8)
where
M-
V (k* —(o2)2+ 4n2(o2
The solution (7) may be rewritten as follows:
x(t) sin (û)f-|-6 ) (9)
Let us consider the spécial case when al = 2n=0, that is, when
there is no résistance and the frequency of the external force coïn
cides with that of the natural oscillations, Æ= co. The équation
438 Ch. 7 Operational Calculus and Certain of lis Applications
= [ * C0S ^ “ T s i n k t ] d t
o
Whence it follows directly that
Theorem. If F (p) is the iransform of the function f (t), then e~P*« F (p)
is the transform of the function f( t — /„); that is, if f (t) F (p),
then
f( t — e~ptoF (p) (1)
Proof. By the définition of a transform we hâve
ao
L {f{t-t,)}= \e-P tf { t - Q d t
0
t0 «
= \e~r ‘ f ( t - t j dt + jj e~pt f ( t —10) dt
o <•
The first intégral on the right of the équation iszerosincef (tf —10) = 0
for t < t0. In the last intégral we change the variable, putting
t —t0— z:
oc oo
L{f (t — t0)}= J e~p (*+<o)/ (z) dz = e~p1» § e~P* f (z) dz=e~P,«F (p)
o o
Thus, f( t — t ^ - ^ e - P ^ F (p).
Example. In Sec. 7.2 it was established for the Heaviside unit function that
440 Ch. 7 Operational Calculus and Certain of Its Applications
It follows, from the theorem that has just been proved, that for the function
a0 ( t — h) depicted in Fig. 153, the L-transform is
— e~Ph
that is,
(2)
60(t-h) i
h
h 0 i 1
Fig. 153 Fig. 154
that is,
( 2)
ordinary meaning of the word; physicists often call ô (t) the Dirac
function or the Dirac delta function.)
It is natural to put
oo
S 1 (4)
—oe
One also writes
o
$ ô (/)d / = 1 (5)
o
The function 6(x) finds application not only in mechanics but
in many divisions of mathematics, in particular in the solution of
numerous problems involving the équations of mathematical physics.
Let us consider the action of 6 (t) if it is represented as a force.
We find the solution to the équation
S = ô(0 (6 )
Differentiating the right and left sides of the équation with respect
to t, we get the conditional équation
o; (0 = 6(0 (17)
To get at the meaning of (17), let us examine the function
o# ( 0 h) shown in Fig. 155. It is clear that
ô ’o(t, h) = al (/, h) (18)
Exercises on Chapter 7 443
Exercises on Chapter 7
Find solutions to the following équations for the indicated initial conditions:
W
2y Wy
1. - jto + 3 tt4-2jc = 0, x = \, x' = 2 for t = 0. Ans. x = 4e~t —3e~2t.
a tL dt
H*r S x
2. — ——= 0 , x = 2, x f = 0, x" = 1 for t = 0. Ans. x = \ —
dt3 d t2
3. 2 a ^ + ( a * + t»2)x = 0, x = xc, x '= x '0 for / = 0 .
E L E M E N T S O F T H E T H E O R Y OF P R OB A B I L I TY
A N D M A TH EM A TICA L STAT1STICS
second.
re“ 0 6 0 ’
third, ^ - 0 .5 8 .
27
fourth,
® r 054-
fifth,
T O " 0-49’
sixth,
» = 0-5'-
From observations of a variety of phenomena, we can conclude
that if the number of trials in each sequence is small, then the
relative frequencies of the occurrence of event A in the different
sequences can differ substantially from one another. However, if
446 Ch. 8 The Theory of Probability
There are 4 cases in ail, of which one is favourable. Hence, the probability
of obtaining two heads is
P= 4
g
Example 5. The probability of hitting a target from one gun is , from
7
another gun, . Find the probability of destroying the target in a simultaneous
firing from both guns. The target will be destroyed if at least one of the guns
makes a hit.
Solution. This problem can be modelled as follows. Two urns contain 10
balls each, numbered from 1 to 10, with 8 red and 2 black in the first urn,
and 7 red and 3 black in the second. One bail is drawn from each urn. What
is the probability that at least one of 2 drawn balls will be red?
Since each bail of the first urn can be drawn with any bail of the second
urn, there will be a total of 100 cases: n= 100.
Let us calculate the favourable cases.
Wtien each of the 8 red balls of the first urn is drawn together with any
bail of the second urn, we will hâve at least one red bail, drawn. There will
be 10x8 = 80 such cases. Drawing each of the two black balls of the first urn
together with any one of the 7 red balls of the second urn gives us one red
bail. There will be 2 x 7 = 1 4 such cases. This makes a total of m = 80+ 14 = 94
favourable cases.
8.3 The Addition of Probabilities. Complémentary Random Events 449
The probability that there will be at least one red bail from among those
drawn is
m 94
n 100 •
♦ Note here that the "or” is inclusive and signifies the occurrence of at least
one of the events, in accord with Définition 1.
2 9 -2 0 8 2
450 Ch. 8 The Theory of Probability
Proof. Suppose
p (4 ) = - . p(A )=-
Since the events A 1 and A2 are exclusive, it follows that for a
total number n of cases, the number of cases favouring the occur
rence of Ax and A2 together is equal to 0 , and the number of ca
ses favouring the occurrence of Ax or Â2 is equal to m1+ m2.
Consequently,
P {A, or A2) = ^ ^ ^ + ^ = P( Al) + P(At)
and the proof is complété.
The proof of this theorem is the same for any number of terms:
P (A or A2 or .. .or An) = P (/!,) + P (4,) + . . . + P {An) (V)
This équation may be written thus:
p (i//)= J iP M /) (O
Zone II, P ( i42) = - j ^ , and Zone IJI, P(y43) = ~ . What is the probability of
hitting D? Event A is a hit scored in domain D. By formula (T) we hâve
p M , ) + p M * )+ p ('43 ) = T g g + î ^ + i ^ = ^
red balls being drawn from both urns will be mlm2. The probabi-
lity of the intersection of events A and B will be
P (A and 7 = nxn 2 n{ m2
V 7
If in this formula we replace and —- by their expressions
in (2 ), we get ( 1). The theorem is illustrated in Fig. 158.
ni
Fig. 158
Solution. Event A is a hit scored by the first gun. Event B is a hit scored
by the second gun. It is obvious that
p (^ ) = t = 4
The probability of event A, provided B has not taken place (a black bail
appears in the first drawing), will be
P(d/5)=-|-
We see that
P (A /£ )# P (A /B )
Theorem l. The probability of the intersection of two event s is
equal to the product (logical intersection) of the probability of one by
the conditional probability of the other computed on the condition
that the first event has taken place, that is,
P (A and B) = P(B) P(A/B) (n
8.5 Dépendent Events. Conditional Probability. Total Probability 455
------------------------V------------------------------------------------------------
Proof. We carry out the proof for events that reduce to the urn
model (that is, for the case vvhere the classical définition of pro
bability is applicable). •
An urn contains n balls, of which n 1 are white and n2 are
black. Suppose the nx white balls include / balls marked with
a star (Fig. 159).
n
L
Fig. 159
We détermine the probability of each event. One hit willoccur if: either
the first shot scores a hit and the second and third miss; or the first shot is
a miss, the second is a hit, and the third is a miss; or the first shot is a miss,
the second is a miss, and the third is a hit» Therefore, by the theorem of
multiplication arrd addition of probabilités, we get the following expression for
the probability of one hit:
P( Bl) = p1( \ - p 2) ( \ - p 3) + ( \ - p 1) p 2( l - p 3) + ( \ - p l) ( \ - p 2) p 9 = 0 M2
Reasoning in like fashion, we hâve
P (B2) = P1 P2 — Pn) ~r Pi 0 ~Pi ) P3 + 0 —Pi) P2 P3 — 0.468
P(^3) = ^lP2P3^ 0144
P ( * 4) = (1 - Pi) 0 P2 ) (1 - p a) = 0.056
Let us write down the conditional probabilities of destroying the target upon
the realization of each of these events:
P (A/ BJ = 0.4, P (AlB.2) - 0.7, P ( A/£3) - 1.0, P ( A/ BJ = 0
Substituting the resulting expressions into formula (8 ), we get the probabi
lity of target destruction:
P(i4) = P ( f l 1)-P(i4/fl1) + P(fl,)-P(i4/flt ) + P(fl3)*P(»/Ba)
P (B4) ■P (A/B4) = 0.332 •0.4 + 0.468 •0.7
+ 0.144 • 1.0 + 0.056 •0 = 0.6044
Note 2. If event A does not dépend on event B, then
P(A/B) = P(A)
and formula ( 1) takes the form
P (A and B) - P (fl) •P (4)
That is, we obtain formula (1), Sec. 8.4.
P(flsM)=^ w 1 = 0 -11
p (53M) = — 2^ i = 0-1l
p (BJA) = — = 002
Here P (/?!) = 0.25, P (Bi/A) = 0.76 became greater because event A occurred.
In this case, the probability P (A/Bi) = 0.7 is greater than the other conditional
probabilities.
Example 2. Each of two tanks fired independently at a target. The probabi
lity of the first tank destroying the target is ^ = 0.8, that of the second,
8.6 Probability of Causes. Bayes's Formula 459
P ^ ) = 0T O T ï O T 8 = m = 0-325
p^ > = ô Æ o . 8= ë =a675
460 Ch. 8 The Theory of Probability
£ p ,== i o)
<=i
must hold true for the case of a finite sequence of N values, or
= 1 (!')
<=1
for the case of an infinité sequence. The value x,- of the random
variable having the greatest probability is called the mode. The
random variable x depicted in Fig. 160 has mode x2.
* Or. briefly, the distribution law of a random variable.
8.7 A Discrète Random Variable 461
X 1 2 3 4 5 6
X 1 2 3 k
Ph p o - p ) p ( i - p ) ’ p ... (i —p ) * - * p
We also hâve
*=i
Z
*=i
p » = £ ( i - p ) * - | p = i _ '(î _ p j = i
X 1 2 3
Note. This problem can be stated in terms of the urn model, which means
it can also be applied to other problems. This remark applies to certain
other problems as well.
which is to say, A occurs in the first m trials, and does not occur
in the next n — m trials (event A occurs). Since
P(A) = p, P(Â)--=rl— p = q
then by the multiplication theorem the probability of such an
alternation of events A and À will be
p m• q n ~m
or
' s r"
p[x = ^ = CZpmqn~m ( 1)
p ( ^ i ) :=É o t -* (4)
m-k
or
k- 1
p ( x ^ 4 ) = 1~ Z
' ' m=0
Example 1. Give a graphical représentation of the binomial distribution of
the random variable x for n = 8, p = l/2 , <7 = 1/ 2 .
Solution. We déterminé ail the values of the probabilities that appear in the
table:
'•«-O»
P ( x = i ) = Cj
]_
256= 32
P (* = j ) = d f ! Y — ± —JL
\ 2J 1•2 ’ 28 64
8 8-7-6 1 7
P
'1-2-3 29 32
8 -7 -6 5 35
P
=1-2-3-4 * 2«“ 128
P
( * = 8') = C®2»=:32
8.8 Relative Frequency and the Probability of Relative Frequency 465
v 8 y *2»—64
P(* = ir ) = C* 2à= 32
Pf^ «8 )U c8 28'i _- 2561
Let us construct the distribution polygon, or frequency polygon (Fig. 161).
Example 2. What is the probability that event A will occur twice (a) in two
trials, (b) in three trials, (c) in 10 trials, if the probability of the occurrence of
the event in each trial is equal to 0.4?
p ( , = f 0) = cî.pV = - ^ ( 0 .4)M0.6)»=0.,2«
Example 3. Five independent shots are fired at a target. The probability of
a hit in each shot is 0.2. Three hits suffice to destroy the target. Détermine the
probability of target destruction.
Solution. Here, n = 5, p = 0 .2 , <7 = 0.8. It is clear that the probability of
target destruction should be computed from the formula
p<«= p ( * = 4 ) + p ( x“ t ) + p (*= t )
or from the formula
P«<*,= l - [ p ( * = y ) + p ( * = 3 - ) + p ( * = |- ) ]
30—2082
466 Ch. 8 The Theory of Probability
p ( ^ 4 ) = p ( ^ = 4 ) + ,,( ^ 4 ) + p ( * = 4 )
or
p (jc = |-^ = iC § p 3= 1 0 .1 » = 0 .0 0 1
X *1 x2 Xk */!
\
2
k=l k=\
xkPk
468 Ch. 8 The Theory of Probability
X 0 1 2 3
10
x'
x 0 1
Pk l —p P
Consequently, mx = 0 (1 — p) + l- p = p.
Note 2 . In the sequel it will be established that the expectation
M (*] of the number of occurrences of event A ity n independent
trials is equal to the product of the number of tr/als by the pro
bability p of occurrence of A in each trial:
M [*] = np (4)
8.9 The Mathematical Expectation of a Discrète Random Variable 469
x 1 1 2 3 ... k ...
Note that
mx —►1 as p —►1
rnx —» oo as p —►0
These relations can be explained by the essence of the problem.
Indeed, if the probability of occurrence of event A in each trial
is close to 1 (p « 1), then it may be expected that A will occur
in the first trial (mx « 1). But if the probability p is small (p « 0),
then one can expect that in order for event A to occur it will be
necessary to perform a large number of trials (mx » oo).
The mathematical expeatation of a random variable x is called
the centre of probability distribution of the random variable x.
470 Ch. 8 The Theory of Probability
If 2 p * = l , then
*=i n
* c =*=
2 i XkPk (5)
Formula (5) coïncides (in appearance) with formula (1) for mathe-
matical expectation.
We hâve thus established that the centre of gravity of a sériés
of masses and the expectation are computed from analogous for
mulas, whence the term “centre of probability distribution”.
\
P
p\
Pi
r f l n _ 1^1 Pi
0 x, X2 x3 Xg Xg Xg Xj X x? x \ xgo x4°x | xg Xf X*
Fig. 162 Fig. 163
X i = X i — mx x l = x2— mx / ...
1
II
&
a
£
Pk Pl p% / ... Pk
= mx— mx 2 Pk = mx —mx - 1 = 0
A=1
Thus, the expectation of a centred random variable is equal to zéro.
Note 4. It is sometimes advisable to regard a nonrandom constant
(a certain quantity) c as a random variable which assumes the
value c with probability 1 and assumes ail other values with pro-
bability 0 .
Then it is meaningful to speak of the expectation of a constant:
M [c]=c-l=c (6)
The expectation of a constant is equal to that constant.
8.10 VARIANCE.
ROOT-MEAN-SQUARE (STANDARD) DEVIATION. MOMENTS
D[.ï] =k=l
2 (x n —m x ) i Pk =fe=l
2 xlPk — 2 *=1
2 Xkmxpk +k=l
2 tti'xPk
n n n
X 1 0
I
Pk 1 P <7
q = 1 —p. Hence, /
W [ * ] = l.p + 0.? = p /
D(*) =(1 — PÏ1P + (0 — p)2q = q2p / p * q = pq
o (x) = V'pq /
Let us consider some more exam ple/in order to get a good idea
of what the concepts of variance and/standard déviation mean as
measures of dispersion of a random/variable.
Example 2. A random variable x-is s/ecifi'ed by the following distribution
law (see table and Fig. 164). /
3 4
* h 7
P
P
0 .3 0M 0 .3
0 .3 0A 0 .3
1 ■
0 2\ jr i '4 X 0 1 i? 5 x
Example 3. A random variable x has the distribution law (see table and
Fig. 165):
x 1 3 5
P1 ti
x 3
l
P l
0 3 x
Fig. 166
X Xi x2 Xk xn
Pk Pl Pi ^ Pk Pu
Pk Pl Pi Pk Pn
8.ÎJ Functions of Random Variables 475
y —A A V~2 0 AV 2 A
2 2
X0X,
i.
xi-lx i x n-ix n x
Fig. 167
(We assume that f(x) is such that the limit on the right exists.)
The limit on the right is the definite intégral of the function f (x)
from a to p. Thus
P (a < x < P) = j / (x)
a
which complétés the proof. /
478 Ch. 8 The Theory of Probability
^ f (x)dx= 1 (5)
Note that if it follows from the essence of the problem that the func
tion f (x) is defined on the finite interval (a, b), then we can regard
the function to be defined on the whole infinité interval (— oo,
-f- oo), but
ï(x) = 0
outside the interval (a, b). In this case, both (4) and (5) hold
true. The density function of the random variable fully détermines
the random variable.
8.13 The Distribution Function. Law of Uniform Distribution 479
F(x)= $ f(x)dx ( 1)
—oo
is called the distribution function.
For a discrète random variable, the distribution function is
equal to the sum of the probabilities of those values xk which are
less than x:
F(x)= 2 Pk
xk < x
9(<X<z<fl’ F(p)-Fÿ.)
F (fi)
Note tha't the density function f (x) and the corresponding dis
tribution function F (x) are connectée! by the relation
F'(x) = f(x) (3)
This follows from (1) and the theorem on differentiating a definite
intégral with respect to the upper limit.
We now consider a random variable with a uniform distribution
of probabilities. The distribution law, or the density function,
f (x) of such a random variable is de-
fined as follows:
f(x) = 0 for x < a
f(x) = c for a < x < b
f(x) — 0 for b < x
On the interval (a, b) the density func-
F»g. 175 tion f(x) has a constant value c (Fig.
175); outside this interval it is zéro.
Such a distribution is also called the law of uniform distribution.
00
From the condition J f ( x ) dx = 1 we find the value of c:
—ao
00 b
J f (x)dx — ^ cdx = c(b—a) = 1
8.13 The Distribution Function. Law of Uniform Distribution 481
hence
From the last équation it follows that the interval (a, b) of uni
form distribution is necessarily finite. Let us détermine the pro-
bability that the random variable x will assume a value lying in
the interval (a, P):
F( x) = J f(x)dx
— CD
If b < x, then
/(* ) = 0 , \f(x)dx = 0
b
hep—
M [je] — ^ xf (x)dx ( 1)
—ao
If a random variable x can assume values only on a finite inter-
val [a, b], then the expectation M [je] is expressed by the formula
_ b
M [je] = J xf (je) dx 11 ')
a
Formula (T) may be regarded as a generalization of formula (1),
Sec. 8.9.
Indeed, partition [a, b] into subintervals (je*^, je*), in each of
which we take a point £*. Consider the auxiliary discrète random
variable £ which can take on the values
^1» ^2* • *• * • • •»
Let the probabilities of thé corresponding values of the discrète
random variable be plt p2, . . . , pk, . . . , p„:
p t = f ( & A x 2, . . . .
_____________P* = /(ift) Ax*. . . . . p„ —f (i„) Axn *
* At the same time, /(g*) A** is the probability that the continuous random
variable jc wi 11 assume a value in the interval **).
8.14 Numerical Characteristics of a Continuous Random Variable 433
M [l] = £ l kpk-
k= 1
or
m[s] = y (ix) a*+ y (g.) a*, +... + y (g*) a**+...
+ U ( U a xn= t u ( i * ) Ax*
*=i
Passing to the limit as max Ax*—<-0, we get
is, for the expression (1). Formulas (1) and (T) are similar to for
mula ( 1) of Sec. 8.9 for a discrète random variable. We use the
symbol mx for the expectation.
The expectation is called the centre of probability distribution
of the random variable x (Fig. 177). If the distribution curve is
symmetric about the #-axis, that is, if f(x) is an even function,
then clearly
00
M [x] = ^ x/ (x) dx = 0
—oo
In this case the centre of the distribution will coïncide with the
coordinate origin (Fig. 178). Let us consider a centred random
variable x —mx. We will find its expectation:
oo oo oo
M [x— mx\ = J (x— mx) f ( x ) d x = J xf (x)dx— mx J f(x)dx
—oo —oo —oo
== mx—mx• 1 = 0
The expectation of a centred random variable is zéro. ,
31
484 Ch. 8 The Theory of Prohability
- '-----=r f 00
a [x] = Ÿ D [x] - \ J (x— mxy f (x) dx (3)
— 00
$ f (x) dx — 5 f ( x ) d x ^ - j (4)
- X Àf
e
P ( x < M , ) = P ( A 4 e <7) - - , i -
no. — j
distributed (the term Gaussian dis- Fig. 180
tribution is also used). The so-called
normal curve (normal distribution curve) in shown Fig. 180. A table
of the values of the function ( 1) for a = 0 , a = l is given in the
Appendix, Table 2. A similar curve was investigated in detail
in Sec. 5.9 of Vol. I.
First of ail, we will show that the density function (1) satisfies
the basic relation (5) of Sec. 8.12:
$ f{x)dx*= 1
— QD
we get
+ V 2 a/, dx = o \ f 2 d t
Hence
CO
= j (a + V 2 o t)e -‘*dt
—ce
CO CO
Th us
mx = a (3 )
or to the left (for a < 0). To save space, in most of our discussions
we will henceforth use the probability density function defined by
formula (4).
8.16 VARIANCE AND STANDARD DEVIATION
OF A NORMALLY DISTRIBUTED RANDOM VARIABLE
Suppose the probability density function (the frequency function)
of the random variable x is given by
D [* ]= — f t2 e~t,d t = f t-2t-e~1' dt
~t%
y n J yn J
—QO —QO
Integratiiig by parts, we obtain
Since
lim t-e~t* = 0 , ? e-*'dt = V n
t -►00 J
we finally hâve
D [x] = a* (2)
l {x)~ ^ 7 ü exp
will fall in the interval (a, P)
8
P (a < x < P) = ^ f(x )d x
a
or
P ( a < ï < |)) = - ^ = . j ' e x p [ — !ïj2 îi] i , ( ,)
a
(Fig. 183). Making the change of variable
r~2
we get P-a
G V~2
l
P (a < x < P) =
va
J e~<'dt ( 1')
GV 2
8.17 The Laplace Function. Normal Distribution Function 489
a>(x) = ^ r ^ e ~ ‘t dt (2 )
0
2 Ÿ~n .
3. d>( + oo) = - ^
y r ~ ~ '
o
4. O (x) is monotonie increasing on the interval (0, oo).
5. <D(x) is an odd function since
0 (_ * )= _ O (* ) (3)
6. The graph of the function O (jc) is shown in Fig. 184.
Detailed tables of the values of this function hâve been compiled
(a short table is given in the Appendix, Table 1).
1 a V2 aV2
— j e -‘'d t+ j e-i'd t
Ÿ'n
o Q
490 Ch. 8 The Theory of Probability
With the aid of the function ®(x) [see (2)], we give a final
expression for the probability of the normally distributed random
variable x falling in the interval (a, p):
P (a < x < p) = (4)
K k r )-* (r ? ï)
When a = 0 we hâve
P (a < x < P) = y (5)
Equating the right members of ( 1) for the case a = 0 and of (5),
we get
P
J a V 2n (5')
a
about the point x = a (Fig. 185). Formula (4) then takes the form
The right side does not dépend on the position of the centre of
dispersion, and so for a = 0 as well we get
P H < ï< 0 =® (^ -) (7)
Example 1. A random variable x obeys the normal distribution law with centre
of dispersion a = 0.5 and variance o2 = -^-. Détermine the probability that a
8
value of the random variable x will fall in the interval (0.4, 0.6) (Fig. 186).
= 1 {0 (0 .2)- ® ( - 0 .2)}
But <D( — 0.2)= — CD(0.2) [see formula (3)] and so we can write
Thus
= (9)
and, obviously,
Ô ( x / 2 ) =i-<t>(x) ( 10 )
Using the function <I>(x) and relation (9), weWrite (5) as follows:
P(a<x<P) = o ( | ) - ô ( £ ) (11)
and when a = 1
P ( a < x < p)=<b(p) — Ô (a)
A table of the values of the Laplace function lî>(x) is given in
the Appendix, Table 3.
Now let us define the normal distribution function. From formula
( 1), Sec. 8.13, we hâve
X
F (x) =- J / (x) dx
—»
1
a V ïïï I exP [ - T ^ l dx = P (—oo < x < x)
Using formula (4) for the case a — —oo, P = x, we get
8.18 Probable Error 493
, w = ï ? i ï exp ( “ ■»■)
will fall in the interval ( — E, E) is ÿ (Fig. 188); that is,
P (a —E < x < a + E) = Y ( 2)
The left members of ( 1) and (4) are equal, and so also are the
right members:
fM = r h e x p ( - p' p ) (l)
The probabilityof the random variable (an error, for example)
falling in the interval (a, P) will, in accordance with (5), Sec.
8.17, be
P ( a < * < P ) = l [<d ( p - | ) - < d ( P - I ) ] (2)
and, in accordance with formula (7) of Sec. 8.17,
P ( - / < x < / ) = G > (p4) (3)
B oc
The numbers and -g- in the right member of (2) are determined
by the nature of the problem; p = 0.4769.
8.19 The Reduced Laplace Function 495
Ô>(x) = -^=r [ e - p ^ d z (5 )
’ 31 o
Let us express the right member of (2) in ternis of the reduced
Laplace function-
P ( a < * < P) = ! [ è ( - | ) - Ô ( ! - ) ] (6 )
©>
A
A
II
(7)
î
d O)
° hV T (3)
d—-U
h VU
(4)
m n
2 2
/=1 /=1
p /y = l (U
P (x < x < x + Ax, y < y < y + Ay) dénotés the probability that the
two-dimensional random variable (x, y) will assume a value denoted
by a point lying in the hatched rectangle As (Fig. 193). We will
say that the “value of the random va
riable lies in the domain As”.*
The probability P (x < x: < x + Ax,
y < y < y + Ay) will also be denoted
by P [(at, i/)cAs]. Using this nota
tion, we can rewrite (2 ) as
P [(a, i/)cA s] œ f ( x , y) As (3)
We will now prove the following the-
orem, which is analogous to Theorem 1, Sec. 8.12.
Theorem I. The probability P [(x ,ÿ )c D ] that a two-dimensional
random variable (x, y) with density function f (x, y) will be in domain
D is expressed by the double intégral of the function f (x, y) over D,
that is,
P [(x, ÿ)cD] = SJ /(*• y)dxdy (4)
D
y>
9
y w m m .
0 oc p X
Fig. 194 Fig. 195
Ÿ s]
’ y 'T l - i I I (!+**) (!+*•)
V3
1 VT
_ î C dx _ r dy _ 1 arctanx VT
arctan y 1
Jl- J l+ A i+ y * Jl*
o i VT
VT
= _ L / 'ü _ o U - - - W -
Jis V4 JV3 6J 24
502 Ch. 8 The Theory of Probability
w = S j (x' v)dv
( 8)
(2)
Thus, f (x, y) may be regarded as a product of Jwo density func-
tions of normally distributed random variables x and y. As in the
In the next section we will establish the fact that the proba-
bility of a two-dimensional random variable falling in the total
ellipse of dispersion is equal to 0.97, which makes a hit a practical
certainty.
8.25 THE PROBABILITY OF A TWO-DIMENSIONAL
RANDOM VARIABLE FALLING IN A RECTANGLE WITH SIDES PARALLEL
TO THE PRINCIPAL AXES OF DISPERSION UNDER THE NORMAL
DISTRIBUTION LAW
Let
By formula (5), Sec. 8.23 (see Fig. 194), the probability of a random
variable falling in the rectangle bounded by the straight Unes
x = a, x = y —y, is expressed thus:
a
yfe:exp(-p!liM "Hh;exp(-p‘| ) *
V
( 2)
=H k ) H t y
We hâve obtained formula (4).
Example. Shots are fired at a rectangle with sides 200 m and 100 m bounded
by the Unes
x ——100, ^=100, y = —50, y = 50
The principal mean errors are equal to Ex = Blon = 50 métrés and E y — Biaf
= 10 métrés, respectively. Find the probability of hitting the rectangle in a
single shot (Fig. 198).
i 1 2 3 ... i ... n
X i * 2 * 3 ... X i ... X n
interval (ak^ v ak). The values at the end of the interval are referred
8.28 Stalisiical Sériés. Histogram 509
^ = * ( 1)
P '
~ V » ,
a 0 a, 0 aZ a3 a« a5 a6 a7 a8 X
Fig. 199
This procedure is based on the fact that ail values in the interval
(a*_lf ak) are close-lying values and are therefore considered to be
equal to the abscissa of the midvalue of the interval x\.
Example. One hundred range measurements were made yielding the following
results (frequency table):
O O
290-320
140-170
260-290
170-200
200-230
230-260
Intervals 1
1
O O
00
mk 2 5 16 24 28 18 6 1
mk 2 5 16 24 28 18 6 1
mlYPk (4)
k= 1
This formula is similar to formula (2) of Sec. 8.10.
Example. Détermine the statistical mean and the statistical variance on the
basis of the statistical data of the Example in Sec. 8.28.
Solution. By formula (2) we get
n
*>
m £= — = 2 1 *aPÂ= 95-0.02+ 125-0.05+155-0.16
k= 1
+ 185-0.24+ 215-0.28+ 245*0.18 + 275-0.06+ 305-0.01 __ 201.20
By formula (4) we obtain
n
2 (**—m i)2 x x
D * [x] = — ---- - --------- = X = X * k P k —m * x
k= 1 k= 1
= 952*0.02+ 1253*0.05+ 1552*0.16+ 1852*0.24 + 2152*0.28
+ 2452*0.18 + 2752•0.06 + 3052•0.01 — (201.20)2= 1753.56
(3)
514 Ch. 8 The Theory of Probability
a= (4 )
2 (xi— <j)a
<•=■ /K\
Note that the right member in (5) and the right member of (3),
Sec. 8.29, differ by the factor , which is close to 1 in prac-
tical problems.
Example 1. Write down the expression for the distribution law of the random
variable on the basis of the measurements given in the Example of Sec. 8.28,
and the results of computations carried out in the Example of Sec. 8.29.
Solution. On the basis of the computations carried out in the Example of
Sec. 8.29, we obtain
a = m* = 201
âî = — ^— D* = ~ . 1754=1771
n—1 99
ô = yT77l«:41
Substituting into (3), we get
r/ x 1 [ (x —201 )21
j <=i
2 I y>I
(6)
d==— h—
From (5) we détermine the standard déviation
a - (7)
1 n— 1
E E
And then we find the ratios -?■ and — .
d a
£
For a random variable that obeys the normal law, the ratios y
£•
and — are equal, respectively, to 0.8453 and 0.6745 [see formula
CT
(6), Sec. 8.22]. If the ratios ^ and — differ from 0.8453 and
CT
0.6745 by a quantity of the order of 10%, then it is agreed, con-
ditionally, that the random variable y obeys the normal law.
A corollary to the central limit theorem is an important theorem
of Laplace conceming the probability that an event will occur not
less than a times and no more than p times. We give this theorem
without proof.
Theorem 2 (Laplace). I f n in d e p e n d e n t t r i a l s a r e m o d e in each
o f w h ich th e p r o b a b i l i t y o f o cc u rre n c e of even t A is p , th e n th e
f o llo w in g r e la tio n h o ld s tr u e :
33
516 Ch. 8 The Theory of Probability
Solution. Here,
n = 1000, p = 0 .01 , (7= 0.99, a = 0 , P = 20
We then find
a — np _ 0 — 10 = _ 2 25
Ÿ"2 yltp q V"2 I / 9 T9
V>~nP _ 2 0 -1 0 _ „ 25
Y ' 2 ^n p q Y ' 2 f 9.9
By formula (8) we get
P ( 0 < m < 2 0 ) = — j^<D (2.25) — <D (- 2 .2 5 )] = <D(2.25)
to Pi —~0.1, /)2- 0 .2 , p z--0 A , pA— 0.2, ph — 0.\. It is also known that if the
CDS lies in the first point, the probability (in range) of hitting the target will
be ^ —0.15. For the othcr points \ve will hâve for this case: p2 — 0.25, p3 =
= 0.60, pA— 0.25, pb = 0.15.
In the initial adjustment of sight, a shot was fired with a miss in fange.
Déterminé the probability that the shot is made at adjustment of sight corre-
sponding to each of the five indicated points of the CDS, that is, déterminé the
probabilités of causes (hypothèses) concerning distinct errors in the position of
the CDS after a trial (shot). Ans. 0.85, 0.75, 0.40, 0.75, 0.85.
14. A die is thrown 5 times. What is the probability that a six will turn
up twice and a non-six three times? Ans. 625 3888.
15. Six shots are fired. Find the probability that not ail are plus rounds if
the probability of a plus round is p = 1/2, the probability of a minus round is
q — 1/2 (gunfire at a “narrow” target). Ans. 31/32.
16. Referring to the conditions of the preceding problem, détermine the pro
bability that there will be three plus rounds and three minus rounds. Ans. 5/16,
17. Find the mathematical expectation of the number that tums up in a
single toss of a die. ,4ns. 7/2.
18. Find the variance of a random variable x given by the following distri
bution table*
jc 2 3 5
Ans. 1.05.
19. The probability of occurrence of event A in a single trial is equal to 0.4.
Five independent trials are carried out. Find the variance of the number of
occurrences of A. Ans. 1.2.
20. In a case of gunfire at a target, the probability of a hit is 0.8. The fire
is kept up till a first hit. There are 4 shells. Find the expectation of the number
of shells used. Ans. 1.242.
21. In firing at a “narrow” target, the probability of a plus round is p — 1/4,
the probability of a minus round is q = 3,4. Find the probability of a combi
nation of two plus rounds and 4 minus rounds in six shots. Ans. 0.297.
22. The probability that an article will be defective is p = 0.01. What is the
probability that a batch of 10 articles will yield 0, 1, 2, 3 defectives? Ans. 0.9045,
Q,0904, 0.0041, 0.0011.
23. Find the probabilités of obtaining at least one hit in the case of 10 shots
if the probability of hitting the target in a single shot is p = 0.15. Ans. 1 —
— (0.85)10 « 0.803.
24. A random variable x is given by the distribution function
0 fo r jc < 0
1 fo r 0 < jc C 1,
0 fo r 1 < jc
518 Ch. 8 The Theory of Probability
25. A random variable x obeys the normal law with expectation 30 and
variance 100. Find the probability that the value of the random variable lies
in the interval (10, 50). Ans. 0.954.
26. A random variable obeys the normal distribution law with variance
a 2 = 0.16. Find the probability that a value of the random variable will differ
in absolute value from the mathematical expectation by less than0.3. Ans. 0.5468.
27. A random variable x is normally distributed with centre of dispersion
a = 0.3 and modulus of précision. h = 2. Find the probability of a hit in the
interval (0.5, 2.0). Ans. 0.262.
28. Gunfire is conducted at a strip of width 4 métrés. There is a syste-
matic error in aiming of 1 métré (undershooting). The probable error is 5 métrés.
Find the probability of hitting the strip in the case of normal dispersion.
Ans. 0.211.
29. Fire is conducted at a rectangle bounded by the straight lines x t = 10 m,
xa = 20m, y x= 15m, y2 = 35 m, in the direction of the straight line bisecting
the short side. The probable errors of normal distribution on the plane are
Ex = 5 m, £ J/= 1 0 m . Find the probability of a hit in a single shot. Ans. 0.25.
30. An error in the fabrication of an article with a given length 20 cm is
a normally distributed random variable; a = 0.2 cm. Détermine the probability
that the length of a manufactured article will differ from the given value by
less than 0.3 cm. Ans. 0.866.
31. Under the hypothèses of Example 30, détermine the fabrication error that
has probability 0.95 of not being exceeded. Ans. 0.392.
32. A random variable x is normally distributed with parameters M[*] = 5
and a = 2. Détermine the probability that the random variable will lie in the
interval (1, 10). Make a drawing. Ans. 0.971.
33. The length of an article made on an automatic machine is a normally
distributed random variable with parameters M [x |= 15, o = 0.2. Find the pro
bability of defective output if the tolérance is 15 ± 0.3. What accuracy of length
of the article can be guaranteed with probability 0.97? Make a drawing.
34. In measuring a certain quantity we get the following statistical sériés;
X 1 2 3 4
Relative 20 15 10 5
frequency
Détermine the statistical mean and the statistical variance. Ans. 2 and 1.
35. The results of measurements are tabulated as follows.
x 0.18 0.20 0.22 0.24 0.26 0.28
Relative 4 18 33 35 9 1
frequency
Find the statistical mean a and the statistical variance o2. Ans. 0.226, 0.004.
36. The probability of defective production is p = 0.02. Find the probability
that in a batch of 400 parts there will be from 7 to 10 defectives. Ans. 0.414.
37. The probability of a hit is P = y • What is the probability that in 250
shots the number of hits will lie between 100 and 150? Ans. 0.998.
38. The probability of defective production is p = 0.02. Find the probability
that among 1000 items there will be not more than 25 defectives. i4ns. 0.87.
CH A PT E.R 9
M A TRICES
01 = * 1 1*1+ 01.*. \
y 2 ~ *21*1 + *22*2 )
By (1), to each point M (xt, x2) of the .Vjjr2-plane there corresponds
a point M (yt, y2) of the y ty2-plane.
We say that équations (1) constitute linear transformations of
the coordinates. These équations map the x^-plane onto the
plane (not necessarily onto the whole plane). Since the équations
(1) are linear, the mapping is termed a linear mapping.
Ui
0 xj 0 y(
Fig. 201
x 2 1yz *2\y 2
o ■ ï 0
( 1)
A= a 2 l a 22
a 9l a 32
0 .. . 1
D é fin itio n 6. We can also consider matrices consisting of a single
column or a single row:
X-i
x2
X y = \ \ y i ÿ * • • • 0.11 (6)
Xm
where the éléments of the matrix are the projections of the vector
on the appropriate coordinate axes. Thus, we can write
*1
—Xl/ -j- x^j -J- x jt ( 10)
xt
A row matrix is also occasionally conveniently identified with
a vector.
9.3 INVERSE TRANSFORMATION
From équations (1) of Sec. 9.1,
Ui ~ #11*1 4 “ #12*2
y* ~ #21*1 4“ #22*2 i
\
(O
it follows that the mapping of x^-plane into y xy2-plane is one-to-
one, since each point of the x^-plane is associated with a unique
point of the y xy2-plane.
If the déterminant of the transformation matrix is nonzero,
#11 #12
A (A) = :0 or «11«22 «21«12 ^ ( 2)
«21 «22
or, expanded,
■y2
°2I
(3)
x* = T y i ~ T y*
The matrix of the inverse transformation îsr by formula (4),
1 1
3 3
A~l =
1 2
3 3
Example 2. The linear transformation
= * i + 2 x2
0i = 2*i + 4*a
526 C7i. 9 Matrices
The matrix (9) is called the product of the matrices (7) and (5)
and we write
b21b12 «u a,2 ^11^11 “1"bl2a2l ^liai2 4“^12^22
( 11)
bl2 b22 <*21^22 ^2iail 4“^22^21 ^21^12 4“ ^22a22
or, briefly,
B A = C ( 12)
Let us now state the rule for multiplying two matrices B and .4
if the first contains m rows and k columns, and the second, k rows
and n columns.
Schematically we hâve the équation
bu ^12 • • • bik all CLi2 . . . aij . . . avi Cl l ............. Cin
Û21 ^22 • • * ^2j • • • ^2n = . . •
bii bî2 • • * : • • • cu ■■; (13)
(1)
That is,
Y=AX (6)
Multiplication of a square matrix by a column matrix yields a
column matrix with the same number of rows.
Clearly, the transformation of a three-dimensional vector X into
the vector Y is another way of saying that three-dimensional space
is transformed into three-dimensional space.
Note that the System of équations (3) is obtained from the mat
rix équation (4) by equating the éléments of the left-hand matrix
and the right-hand matrix.
Equation (4) yields a transformation of the vector X into the
vector Y by the matrix A.
Ail the reasoning pertaining to a vector in three-dimensional
space can be earried over to a transformation of vectors in a space
of any number of dimensions.
9.6 Inverse Matrix 531
A sa, ( 1)
ai û 32 Q-à3
^11 a i2 a i3
A = A (i4) = û■*9i210*®
“ 22 “a ,2» ¥ = 0 (2)
® 81 û »8 ^ 3 »
We will prove that the inverse matrix A~l will be
4s»
A A A
4*2 4 s*
A~ l A A A (3)
4|S 4*3 4*3
A A A
A (A) = 5
Then we find the cofactors
5 —4 2
A= 0 2 —1
0 —1 3
534 Ch. 9 Matrices
(4)
120 SI
0 3 1 *2 = 9
0 12 *3 8
X - I W AD (3)
or, expanded,
An A2l A31 dt
1
*AM 2 À 22 A 3 2 d2
y
X 2
A (A)
u (4)
x3 ^ 4 1 3 A 23 A 33 d3
Multiplying together the matrices on the right, we get
Equating the éléments of the matrices on the left and on the right,
we get
« _ ^1^11 + ^2^21 + ^3^31
A
diA i2-\-d2A 22-\-d^A22
*2— A (6)
_ _d{Ai3-{-d2A 23-\-d3As3
*« — a
536 Ch. 9 Matrices
*i + 2*2 =5
3*2+ *3 = 9
*2 + 2*8= 8
by the matrix method.
Solution. Let us find the déterminant of the matrix of the System:
12 0
A(4) = 0 3 1 = 5
0 12
Let us détermine the inverse matrix by formula (3), Sec. 8.7:
2_
5
J_
5
j3
5
The matrix D will be
5
Z>= 9
8
We \vrite down the solution in matrix form, via formula (2), as lollows:
4 2 4 2
*1 I 5
5 5 l- 8 - T - 9 + T - 8
2 1
*2 = 0 9 = ° .5 + 4 . 9 - 1 . 8
T 5
î 3
*3 0 8 °-5 ~ f 9 + f 8
5 5
9.10 Orthogonal Mappings. Orthogonal Matrices 537
Equating tJie rows of tlie matrices on the left and on the right, \ve get
*, = 0 . 5 + | ~ 9 — - - 8 = 2
*a = 0 5 - - i - 9 + | - - 8 = . 3
Example 2. Solve the System of équations
xx-\-2x.î + xz —0
*2 +*3= 1
*i-|-3*2 + *3 - 2
by the matrix method.
Solution. Find the déterminant of the matrix of the System
1 2 1
A (A) = 2 1 1 = 1 9*Û
1 3 1
*2 — 1 0 1 1 — 2
*3 1 5 — 1 — 3 2 — 7
Equating the rows of the matrices on the right and on the left, we get
*i = 3, *a = 2, *a = — 7
we get
*î = « n * i+ '“ 21*2 + « 8.*s ]
X* = a 1tX1+ aitx. -f a 33x3 l (17)
^ = « 1 3 * 1 + “ 2 3 * 2 + «33*3 J
w here
X* + Xj + Xj 0
If after transforming the vector X by means of matrix .4
[see (2), Sec. 9.5] we get a vector Y
Y = AX (2)
9.11 The Eigenvector of a Linear Transformation 541
parallel to X,
Y=XX (3)
where X is a scalar, then the vector X is termed the eigenvector
of matrix A or the eigenvector of the given linear transformation;
the scalar X is termed the eigenvalue.
Let us find the eigenvector
X= X2
*3
and (5) to
(au —X) xy + a13x3+ a, 3x, = 0
Aji-Ci + (as2— X) x2 + al3x3= 0 (7)
«31*^1 4 “ « 3 2 ^ 2 4 “ (« 3 3 ^ 3 ^
a u ü 22 ^ a 23 (8)
«31 &32 û 33 *
or
A ( 4 —XE) = 0 (9)
This is a cubic équation in X. It is called the characteristic equa-
542 Ch. 9 Matrices
tion of the matrix A. From this équation we can find the eigen-
values k.
Let us consider a case where ail roots of the characteristic équa
tion are real and distinct. We dénoté them by klt k2, k3.
To each eigenvalue k there corresponds an eigenvector whose
coordinates are determined from the System (7) for an appropriate
value of k. Let us dénoté the eigenvectors by xlt t 2, t 3. It can
be shown that these vectors are linearly independent, that is to
say, none can be expressed in terms of the others. Hence, any
vector can be expressed in terms of the vectors t lf t 2, t 3, that is,
they can be taken for base vectors.
We note without proof that ail roots of the* characteristic équa
tion of a symmetric matrix are real.
Example 1. Find the eigenvectors and the corresponding eigenvalues of the
matrix
1 1
8 3
Solution. Form the characteristic équation and find the eigenvalues:
1—k 1
= 0 , .e., X2—4X — 5 = 0, Aa = — -1, k2 = 5
8 3—k
We find the eigenvector corresponding to the eigenvalue ^ = —1 from the
appropriate System of équations (7):
(1—X2) *i+ *2 —0 or 2xi -f- —0
8^i (3 — Àj) X 2 —
—0 8*i 4x2 ~ 0
Solving this System, we get x l = m, x2 = —2m, where m is an arbitrary num-
ber. The eigenvector is
x1= mi — 2mj
For the eigenvalue X2 = 5, we write the System of équations
—4xi + *2 = 0
Sxi —2*2 = 0
The eigenvector is
x2 = mi + 4m j
Example 2. Find the eigenvalues and the eigenvectors of the matrix
7 —2 0
—2 6 —2
0*—2 5
Solution. We write the characteristic équation
7—k —2 0
—2 6 -A, —2 = 0, i.e., -A ,3+18A,2-9 9 A ,+ 162 = 0
0 —2 5 —Â
The roots of this équation are: k x = 3t À2 = 6, Pi3 = 9.
9.12 The M atrix of a Linear Transformation 543
x3= — m i+ m j— ^ m k
or, a s a S y s te m o f é q u a t io n s
 .j = f l j i • 1 -j” O u * 0 - f - f lj 3 • 0 I
*3 = ^ 1 3 * 1 “ 1“ ^ 2 3 * 2 “b ^ 3 3 * 3 /
where the expressions (4) are substituted into the right-hand side.
Equating the coefficients of the vectors elt e3, e3 on the right and
left, we get the équations
x 1= bllx[ + bl3xî + biix3
x<i —b3lx3+ b33x3-b b33x3 (7)
x3 ^ b 3lx[+ b3tx3-\-b33x3
or, briefly,
X=BX' ( 8)
where
bn K 2 bi3
b3i b» bu (9)
bn ^32 b33
This is a nonsingular matrix that has an inverse, B -1, since
System (7) has a definite solution for x\, x’t, x3. If we write the
vector Y in the new basis
Y '= y \e \+ y 3ë3+ y 3ë3
then quite obviously the équation
Y = BY' (10)
35—2082
546 Ch. 9 Matrices
—1 - 1 2
B -M = -1 0 11
4 2 —4
Finally, by formula (13), we get
-1 3 01
A f — B - 1 AB = 0 10
4 — 2 2\
is called the matrix of the quadratic form ( 1). The given matrix
is symmetric.
We will regard (xlt x2, x3) as the coordinates of a point of space
or the coordinates of a vector in the orthogonal basis (elt e 2, e„),
where e t , e 2, e 3 are unit vectors.
35
548 Ch. 9 Matricés
4
X' A'j (7)
x'3
We write transformation (5) as
X' = AX ( 8)
a 2\ Û 2 Î f l23 ^24
Û 31 f l32 û 33 û 34
Third-order minors of this matrix are obtained by striking out one column and
replacing the matrix symbol || || by the sign | | of the déterminant. There are
four. Second-order minors are obtained by striking out iwo columns and one
row. There are 18. First-order minors number 12.
° 1 1 a i t a is
A= Û J I ü 22 Û 2 3 ( 2)
ûjt ûjj Ûjs
and the augmented matrix
®11 ®1* ^13 ^1
5= ûji ü22 ûs» bt (3)
atl a32 a33 ^3
The system (1) has solutions if the rank of the matrix A is
equal to the rank of the matrix B. The system does not hâve any
solutions if the rank of A is less than the rank of B. If the rank
of A and the rank of B is 3, then the system has a unique solution.
If the rank of the matrices A and B is equal to 2, then the
system has an infinitude of solutions; here, two unknowns are expres-
sed in terms of the third, which has an arbitrary value.
I f the rank of the matrices A and B is equal to 1, then the
system has an infinitude of solutions and two unknowns hâve
arbitrary values, the third being expressed in terms of these two.
The validity o! this theorem is readily established on the basis
of an analysis (familiar in algebra) of the solutions of a system
of équations. This theorem holds true for a system of any number
of équations.
Il fl/ / (011 =
a 21(0 a22 (0 a2n (t)
(l)
or || a (0 1| = (5)
f t
ûml(Z)dz ... J&mn(2 )dz
> U
552 Ch. 9 Matrices
More compacty.
(8)
or
J || a (z) || dz (9)
1o
ii*ii- x: m ( 2)
xn V )
dx„
dt
9.17 Matrix Notation for Systems of Differential Equations 553
• a m Û/X2 * * ■ a nn
Using the rule for matrix multiplication (see Sec. 9.4), we can
write the System of difïerential équations (1) in matrix form:
dxi
dt a 12 . . • a in
«U
dx 2
c i22 . . • ^2 n
©2
~dt •
(5)
W = ax (7)
where x is also called the vector solution; a is short for the
matrix ||a (/ ||.
Suppose we hâve
«i
tt2
||a || = ce = ( 8)
«„
where a,- are certain scalars.
The set of solutions of a System of difïerential équations will
be sought in the form [see formula (2), Sec. 1.30]
| x 1| = e*' | a (9)
or
X = ekta ( 10)
Substituting ( 10 ) into (7) [or (9) into (6 )], via the rule for multi
plication of a matrix by a scalar and the rule for differentiating
matrices, we get
kekta — aekla (H)
554 Ch. 9 Matrices
whence we obtain
ka = art
or
a a — ka = 0 ( 12 )
Recall that in the last équation, a is the matrix (4), k is a scalar,
and a is the column matrix (8 ). The matrix in the left member
of ( 12 ) can be written as
(a — kE) a = 0 (13)
where E is an identity (unit) matrix of order n. In expanded form,
équation (13) can be rewritten thus:
au — tl cl12 •• <*ln “i
flji ait— k . • • n a2
(14)
a . . a nn— k a n
a n(h
9.17 M atrix Notation for Systems oj Differential Equations 555
dt +
^ - = x x-{-Xi + 3xa
Solution. We Write down the matrix of the System:
10 0
A=12 0
1 13
Thus, in matrix form, the System of équations isw rittenas [see équation (5)J:
dx,
1 0 1
dt
dx2
= 1 2 0 *2
dt
dx3
1 1 3 *3
dt
Let us form the characteristic équation (16) and find its roots:
1- Jf e 0 0
We détermine ai*0, a 23\ o49\ whieh correspond to the root ks = $, from the System
- 2 a ia) =0 #
ai3,- a ? ) = 0
cxi3, + a <
28)= 0
to find
a<3) = 0 , a23) = 0 , a<3) = l
Now let us \vrite down in matrix form the solution of the System [see formula (17)J:
*i 1 0 0 Cte*
x2 = — 1 1 0 Cte**
*3 0 — 1 1
or in ordinary form
j = — C,e‘ + C ,e st
*, = - C te*t+Cse»t
Tlien, like formula (5) of Sec. 9.17, the System (2) can be written
as follows:
dxi
0 1 0 . , ,. 0
~dt
dxi
0 0 1 . , ,. 0
dt
dx„-1
0 0 0.. . 1
dt
dxn
dt a » a a • • • an
or, briefly.
a-Il-<11=Il “•INI •«il (5)
The subséquent solution is carried through as in Sec. 9.17, since
the matrix équation (5) is a particular case of équation (6 ), Sec. 9.17.
Example. Write the équation
d 2x dx
di*= p W +qx
in matrix form.
Solution. Put x = x x . Then
d x x
d t ~ ~ * 2
t + Q x i
d x 2
d t
<7 P X 2
TT “ (*) x i + ( 0 *» + • • • + a nn(t)xn
9.19 Solving a System of Linear Differential Equations 559
(3)
^ = a (t)x (6 )
or
*= (13)
matrix form:
S |M |S I M I = ^ i ^ l M l a
S ||a ||S ||a ||S ||a || = — j ^ | | a ||3 and so on
Exercises on Chapter 9
1. Find the matrix of the inverse transformation for the linear transformation
= 3 *x4 - 2 x 2, y 2= 7 xf + 5 xa
5-2
-7 3 •
2. Find the matrix of the inverse transformation
yi = *i —**. V2= xl
0 1
Ans.
-1 1
l 2 2 0
3. Find the product of the matrices
3 4 3 —1
8 —2
Ans.
18 —4
4. Given the matrices
1 2 3 5 0 7
2 0 1 and B= 12 3
3 —1 1 —1 0 2
Find the matrices A B and BA.
4 4 19 26 3 221
Ans. 9 0 16 and 14—1
8
13 —2 20 5 —4 - > l
12 3
5. Given: A — 5 7 10 and E, a third-order unit matrix. Détermine the
4 3 1
matrix A + 2E.
3 2 3
Ans. 5 9 10
4 3 3
12
6. Given the matrix A = . Find the matrix Aa.
3 4
I 7 10
Ans.
115 22
7. Suppose A =
12 20
1 3 4 1| ' F'11** * + M-
Ans.
30 42
1 1 1
8. Find the inverse of A = 5 4 3 .
10 5 1
11—4 1
Ans —25 9 —2 .
1 5 -5 1
36 *
564 Ch. 9 Matrices
W + * - 0
TABLE 1 • x
The values of the function O (x) = - J._f e~ i2dt
K » J0
and the reduced Lapi ace function <î>(je) = <D(px)
0 .0 0 0 .0 0 0 0 0 .0 0 0 0 okq 2 .2 5 0 .9 9 8 5 Q 0 .8 7 0 9 QQ
OCVt O OO
0 .0 5 0 .0 5 6 4 0 .0 2 6 9 2 .3 0 0 .9 9 8 8 0 .8 7 9 2
0 .1 0 0 .1 1 2 5 561 0 .0 5 3 8 269 2 .3 5 0 .9 9 9 1 3 0 .8 8 7 1 79
0 .1 5 0 .1 6 8 0 555 0 .0 8 0 6 268 2 .4 0 0 .9 9 9 3 2 0 .8 9 4 5 74
0 .2 0 0 .2 2 2 7 547 0 .1 0 7 3 267 2 .4 5 0 9995 2 0 .9 0 1 6 71
0 .2 5 0 .2 7 6 3 536 0 .1 3 3 9 2ô6 2 50 0 .9 9 9 6 1 0 /9 0 8 2 66
0 .3 0 0 .3 2 8 6 523 0 .1 6 0 4 265 2 .5 5 0 .9 9 9 7 1 0 .9 1 4 6 64
0 .3 5 0 .3 7 9 4 508 0 .1 8 6 6 262 2 .6 0 0 .9 9 9 8 1 0 .9 2 0 5 59
0 .4 0 0 .4 2 3 4 490 0 .2 1 2 7 261 2 .6 5 0 .9 9 9 8 0 0 .9 2 6 1 56
0 .4 5 0 .4 7 5 5 471 0 .2 3 8 5 258 2 .7 0 0 .9 9 9 9 1 0 .9 3 1 4 53
0 .5 0 0 .5 2 0 5 450 0 .2 6 4 1 256 2 .7 5 0 .9 9 9 9 0 0 .9 3 6 4 50
0 .5 5 0 .5 6 3 3 428 0 .2 8 9 3 252 2 .8 0 0 .9 9 9 9 0 0 .9 4 1 0 46
0 .6 0 0 .6 0 3 9 406 0 .3 1 4 3 250 2 .8 5 1 0 .9 4 5 4 44
0 .6 5 0 .6 4 2 0 381 0 .3 3 8 9 246 2 .9 J 0 .9 4 9 5 41
0 .7 0 0 .6 7 7 8 358 0 .3 6 3 2 243 2 .9 5 0 .9 5 3 4 39
0 .7 5 0 .7 1 1 2 334 0 .3 8 7 0 238 3 .0 0 1 .0 0 0 0 0 .9 5 7 0 36
0 .8 0 0 .7 4 2 1 309 0 .4 1 0 5 235 3 .0 5 0 .9 6 0 3 33
0 .8 5 0 .7 7 0 7 286 0 .4 3 3 6 231 3 .1 0 0 .9 6 3 5 32
0 .9 0 0 .7 9 6 9 262 0 .4 5 6 2 226 3 .1 5 0 .9 6 6 4 29
0 .9 5 0 .8 2 0 9 240 0 .4 7 8 3 221 3 .2 0 0 .9 6 9 1 27
1 .0 0 0 .8 4 2 7 218 0 .5 0 0 0 217 3 .2 5 0 9716 25
1 .0 5 0 .8 6 2 4 197 0 .5 2 1 2 212 3 .3 0 0 .9 7 4 0 24
1 .1 0 0 .8 8 0 2 178 0 .5 4 1 9 207 3 .3 5 0 .9 7 6 1 21
1 .1 5 0 .8 9 6 1 159 0 .5 6 2 0 201 3 .4 0 0 .9 7 8 2 21
1 .2 0 0 .9 1 0 3 142 0 .5 8 1 7 197 3 .5 0 0 .9 8 1 8 36
1 .2 5 0 .9 2 2 9 126 0 .6 0 0 8 191 3 .6 0 0 .9 8 4 8 30
1 .3 0 0 .9 3 4 0 111 0 .6 1 9 4 186 3 .7 0 0 .9 8 7 4 26
1 .3 5 0 .9 4 3 8 98 0 .6 3 7 5 181 3 .8 0 0 .9 8 9 6 22
1 .4 0 0 .9 5 2 3 85 0 .6 5 5 0 175 3 .9 0 0 .9 9 1 5 19
1 .4 5 0 .9 5 9 7 74 0 .6 7 1 9 169 4 .0 0 0 .9 9 3 0 15
1 .5 0 0 .9 6 6 1 64 0 .6 8 8 3 164 4 .1 0 0 .9 9 4 3 13
1 .5 5 0 .9 7 1 6 55 0 .7 0 4 2 159 4 .2 0 0 .9 9 5 4 11
1 .6 0 0 .9 7 3 6 47 0 .7 1 9 5 153 4 .3 0 0 .9 9 6 3 9
1 .6 5 0 .9 8 0 4 41 0 .7 3 4 2 147 4 .4 0 0 .9 9 7 0 7
1 .7 0 0 .9 8 3 8 34 0 .7 4 8 5 143 4 .5 0 0 .9 9 7 6 6
1 .7 5 0 .9 8 6 7 29 0 .7 6 2 1 136 4 .6 0 0 .9 9 8 1 5
1 .8 0 0 .9 8 9 1 24 0 .7 7 5 3 132 4 .7 0 0 .9 9 8 5 4
1 .8 5 0 .9 9 1 1 20 0 .7 8 7 9 126 4 .8 0 0 .9 9 8 8 3
1 .9 0 0 .9 9 2 8 17 0 .8 0 0 0 121 4 .9 0 0 .9 9 9 1 3
1 .9 5 0 .9 9 4 2 14 0 .8 1 1 6 116 5 .0 0 0 .9 9 9 3 2
2 .0 0 0 .9 9 5 3 11 0 .8 2 2 7 111 5 .1 0 0 .9 9 9 4 1
2 .0 5 0*9963 10 0 .8 3 3 2 105 5 .2 0 0 .9 9 9 6 2
2 .1 0 0 .9 9 7 0 7 0 .8 4 3 4 102 5 .3 0 0 .9 9 9 7 1
2 .1 5 0 .9 9 7 6 6 0 .8 5 3 0 96 5 .4 0 0 .9 9 9 7 0
2 .2 0 0 .9 9 8 1 5 0 .8 6 2 2 92
4 87
TABLE 2 The values of the function / ( * ) = — ___ ■■■* 2
____________________ y 2n_____
X / w X f(x) X | fM
0 .0 0 0 .3 9 8 9 1 .3 5 0 .1 6 0 4 • 2 .7 0 0 .0 1 0 4
0 .0 5 0 .3 9 8 4 1 .4 0 0 .1 4 9 7 2 .7 5 0 .0 0 9 1
0 .1 0 0 .3 9 7 0 1 .4 5 0 .1 3 9 4 2 .8 0 0 .0 0 7 9
0 .1 5 0 .3 9 4 5 1 .5 0 0 .1 2 9 5 2 .8 5 0 .0 0 6 9
0 .2 0 0 .3 9 1 0 1 .5 5 0 .1 2 0 0 2 .9 0 0 .0 0 6 0
0 .2 5 0 .3 8 6 7 1 .6 0 0 .1 1 0 9 2 .9 5 0 .0 0 5 1
0 .3 0 0 .3 8 1 4 1 .6 5 0 .1 0 2 3 3 .0 0 0 .0 0 4 4
0 .3 5 0 .3 7 5 2 1 .7 0 0 .0 9 4 0 3 .0 5 0 .0 0 3 8
0 .4 0 0 .3 6 8 3 1 .7 5 0 .0 8 6 3 3 .1 0 0 .0 0 3 3
0 .4 5 0 .3 6 0 5 1 .8 0 0 .0 7 9 0 3 .1 5 0 .0 0 2 8
0 .5 0 0 .3 5 2 1 1 .8 5 0 .0 7 2 1 3 .2 0 0 .0 0 2 4
0 .5 5 0 .3 4 2 9 1 .9 0 0 .0 6 5 6 3 .2 5 0 .0 0 2 0
0 .6 0 0 .3 3 3 2 1 .9 5 0 .0 5 9 6 3 .3 0 0 .0 0 1 7
0 .6 5 0 .3 2 3 0 2 .0 0 0 .0 5 4 0 3 .3 5 0 .0 0 1 5
0 .7 0 0 .3 1 2 3 2 .0 5 0 .0 4 8 8 3 .4 0 0 .0 0 1 2
0 .7 5 0 .3 0 1 1 2 .1 0 0 .0 4 4 0 3 .4 5 0 .0 0 1 0
0 .8 0 0 .2 8 9 7 2 .1 5 0 .0 3 9 6 3 .5 0 0 .0 0 0 9
0 .8 5 0 .2 7 8 0 2 .2 0 0 .0 3 5 5 3 .5 5 0 .0 0 0 7
0 .9 0 0 .2 6 6 1 2 .2 5 0 .0 3 1 7 3 .6 0 0 .0 0 0 6
0 .9 5 0 .2 5 4 1 2 .3 0 0 .0 2 8 3 3 .6 5 0 .0 0 0 5
1 .0 0 0 .2 4 2 0 2 .3 5 0 .0 2 5 2 3 .7 0 0 .0 0 0 4
1 .0 5 0 .2 2 9 9 2 .4 0 0 .0 2 2 4 3 .7 5 0 .0 0 0 4
1 .1 0 0 .2 1 7 9 2 .4 5 0 .0 1 9 8 3 .8 0 0 .0 0 0 3
1 .1 5 0 .2 0 5 9 2 .5 0 0 .0 1 7 5 3 .8 5 0 .0 0 0 2
1 .2 0 0 .1 9 4 2 2 .5 5 0 .0 1 5 4 3 .9 0 0 .0 0 0 2
1 .2 5 0 .1 8 2 6 2 .6 0 0 .0 1 3 6 3 .9 5 0 .0 0 0 2
1 .3 0 0 .1 7 1 4 2 .6 5 0 .0 1 1 9 4 .0 0 0 .0 0 0 1
x
_L2
TABLE 3 The values of the function <D(jt) = — --= l e 2 dz
Ÿ2* J
o
X c D (r) X 4> (X) X | <!><*)
0 .0 0 0.0000 0 .9 5 0 .3 2 8 9 1 .9 0 0 .4 7 1 3
0 .0 1 0 .0 0 4 0 1 .0 0 0 .3 4 1 3 2 .0 0 0 .4 7 7 2
0 .0 5 0 .0 1 9 9 1 .0 5 0 .3 5 3 1 2 .1 0 0 .4 8 2 1
0 .1 0 0 .0 3 9 8 1 .1 0 0 .3 6 4 3 2 .2 0 0 .4 8 6 1
0 .1 5 0 .0 5 9 6 1 .1 5 0 .3 7 4 9 2 .3 0 0 .4 8 9 3
0 .2 0 0 .0 7 9 3 1 .2 0 0 .3 8 4 9 2 .4 0 0 .4 9 1 8
0 .2 5 0 .0 9 8 7 1 .2 5 0 .3 9 4 4 2 .5 0 0 .4 9 3 8
0 .3 0 0 .1 1 7 9 1 .3 0 0 .4 0 3 2 2 .6 0 0 .4 9 5 3
0 .3 5 0 .1 3 6 8 1 .3 5 0 .4 1 1 5 2 .7 0 0 .4 9 6 5
0 .4 0 0 .1 5 5 4 1 .4 0 0 .4 1 9 2 2 .8 0 0 .4 9 7 4
0 .4 5 0 .1 7 3 6 1 .4 5 0 .4 2 6 5 2 .9 0 0 .4 9 8 1
0 .5 0 0 .1 9 1 5 1 .5 0 0 .4 3 3 2 3 .0 0 0 .4 9 8 6 5
0 .5 5 0 .2 0 8 8 1 .5 5 0 .4 3 9 4 3 .2 0 0 .4 9 9 3 1
0 .6 0 0 .2 2 5 7 1 .6 0 0 .4 4 5 2 3 .4 0 0 .4 9 9 6 6
0 .6 5 0 .2 4 2 2 1 .6 5 0 .4 5 0 5 3 .6 0 0 .4 9 9 8 4 1
0 .7 0 0 .2 5 8 0 1 .7 0 0 .4 5 5 4 3 .8 0 0 .4 9 9 9 2 7
0 .7 5 0 .2 7 3 4 1 .7 5 0 .4 5 9 9 4 .0 0 0 .4 9 9 9 6 8
0 .8 0 0 .2 8 8 1 1 .8 0 0 .4 6 4 1 4 .5 0 0 .4 9 9 9 9 7
0 .8 5 0 .3 0 2 3 1 .8 5 0 .4 6 7 8 5 .0 0 0 .5 0 0 0 0 0
0 .9 0 0 .3 1 5 9
INDEX
A BOOK OF PROBLEMS IN
ORDINARY DIFFERENTIAL
EQUATIONS
b y M . K r a s n o v a n d o th e r s
VECTOR ANALYSIS
b y M . K r a s n o v a n d o th e r s
A COURSE OF MATHEMATICAL
ANALYSIS, VOLS. M I
b y S . N ik o ls k y
M IR P U B U S H E R S