CALCULUS OF VARIATIONS MA 4311 SOLUTION MANUAL

Department of Mathematics Naval Postgraduate School Code MA/Nd Monterey, California 93943 June 11, 2001
B. Neta

c 1996 - Professor B. Neta

Contents
1 2 3 4 5

Functions of n Variables 1 Examples, Notation 9 First Results 13 Variable End-Point Problems 33 Higher Dimensional Problems and Another Proof of the Second Euler Equation 54 6 Integrals Involving More Than One Independent Variable 74 7 Examples of Numerical Techniques 80 8 The Rayleigh-Ritz Method 85 9 Hamilton's Principle 91 10 Degrees of Freedom - Generalized Coordinates 101 11 Integrals Involving Higher Derivatives 103

List of Figures
1 :::::::::::::::::::::: 2 :::::::::::::::::::::: 3 :::::::::::::::::::::: 4 :::::::::::::::::::::: 5 :::::::::::::::::::::: 6 :::::::::::::::::::::: 7 :::::::::::::::::::::: 8 :::::::::::::::::::::: 9 :::::::::::::::::::::: 10 Plot of y = ` and y = 1 tan(`) ; sec(`) 2

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

: : : : : : : : : :

5 64 64 81 82 83 84 87 90 95

ii

Credits
Thanks to Lt. William K. Cooke, USN, Lt. Thomas A. Hamrick, USN, Major Michael R. Huber, USA, Lt. Gerald N. Miranda, USN, Lt. Coley R. Myers, USN, Major Tim A. Pastva, USMC, Capt Michael L. Shenk, USA who worked out the solution to some of the problems.

iii

CHAPTER 1

1 Functions of n Variables
Problems
1. Use the method of Lagrange Multipliers to solve the problem minimize f = x2 + y2 + z2 subject to = xy + 1 ; z = 0 2. Show that where
0

is the positive root of

max cosh cosh

= cosh0

sinh = 0:

Sketch to show 0 . 3. Of all rectangular parallelepipeds which have sides parallel to the coordinate planes, and which are inscribed in the ellipsoid x2 + y 2 + z 2 = 1 a2 b2 c2 determine the dimensions of that one which has the largest volume. 4. Of all parabolas which pass through the points (0,0) and (1,1), determine that one which, when rotated about the x-axis, generates a solid of revolution with least possible volume between x = 0 and x = 1: Notice that the equation may be taken in the form y = x + cx(1 ; x), when c is to be determined. 5. a. If x = (x1 x2 xn) is a real vector, and A is a real symmetric matrix of order n, show that the requirement that

be stationary, for a prescibed A, takes the form

xT Ax ; xT x

Ax = x:
Deduce that the requirement that the quadratic form

xT Ax
1

be stationary, subject to the constraint

xT x = constant,
leads to the requirement where is a constant to be determined. Notice that the same is true of the requirement that is stationary, subject to the constraint that = constant, with a suitable de nition of .] b. Show that, if we write T Ax = xxT x the requirement that be stationary leads again to the matrix equation

Ax = x

Ax = x:
Notice that the requirement d = 0 can be written as d ; d =0
2

or

d = 0]

Deduce that stationary values of the ratio are characteristic numbers of the symmetric matrix A.

xT Ax xT x

1. f = x2 + y2 + z2

' = xy + 1 ; z = 0

F = f + ' = x2 + y2 + z2 + (xy + 1 ; z) @F = 2x + y = 0 @x @F = 2y + x = 0 @y @F = 2z ; = 0 @z @F = xy + 1 ; z = 0 @
(3) (4) (1) (2) (3) (4) (5) (6) (7)

) )

= 2z

z = xy + 1

(5) and (16) Substitute (7) in (1) - (2)

= 2(xy + 1)

) 2x + 2(xy + 1)y = 0
2y + 2(xy + 1)x = 0

(8) (9)

x + xy2 + y = 0 y + x2y + x = 0
3

9 > = >

;
(10)

xy(y ; x) = 0

)x=0

or y = 0 or x = y

x = 0 ) = 2 ) z = 1 y = 0 by(1) (7) (5) y = 0 ) = 2 ) z = 1 x = 0 by(1) (7) (5) x = y ) = 2 ) z = ;1 ) xy = ;2 (7) (5) (6)

) x = ;2
2

Not possible So the only possibility

x=y=0 z=1

=2

f =1

2. Find max cosh Di erentiate cosh ; sinh = 0 d = d cosh cosh2 Since cosh 6= 0
!

cosh

sinh = 0
0 0

The positive root is Thus the function at

0

becomes

cosh 0 No need for absolute value since

5 4 3 2 1 0 1 2 3 4 5 5

>0

Figure 1:

2 y2 + z2 = 1 3. max xyz s:t: x + a2 b2 c2 2 2 2 x y z Write F = xyz + ( a2 + b2 + c2 ; 1) , then

0 = Fx = yz + 2a2x 0 = Fy = xz + 2b2y 0 = Fz = xy + 2c2z

2 y2 + z2 ; 1 + 0=F = x a2 b2 c2

(1)

(2) (3) (4)

If any of x y or z are zero then the volume is zero and not the max. Therefore x 6= 0 y 6= 0 z 6= 0 so

x2 + 2 y 2 0 = ;xFx + yFy = ;2 a2 b2
Also

)y b

=x a2

(5)

x2 + 2 y2 ) y2 = z2 0 = ;zFz + yFy = ;2 a2 b2 b2 c2

(6)

2 2 b Then by (4) 3y2 = 1 ) y2 = b taking only the (+) square root (length) y = p b 3 3 c by (5), (6) respectfully. a z=p x=p 3 3 2 y2 + z2 = 1 has dimension The largest volume parallelepiped inside the ellipsoid x + a2 b2 c2 b c a p3 p3 p3

4. ' = ;y + x + cx(1 ; x) Volume V = min V =

Z Z

1 0

y2dx

1 0

x + cx(1 ; x)]2 dx
Z

dV (c) = dc
2
Z

1 0

2 x + cx(1 ; x)] x(1 ; x)dx = 0

Z

x (1 ; x)dx + 2 c
2

1 0

x2(1 ; x)2dx = 0
1 0

1 x4 1+2 c 1 x3 ; 1 x4 + 1 x5 3 2 1 x ; 3 4 0 3 2 5 1;1+1 = 0 1 ; 1) + 2 c 2 (3 4 3 2 5 1 +2 c 1 =0 2 12 30 5 = ; c = ; 15 6 2 y =x; 5 2 x(1 ; x)

=0

V (c) =
=

1h

x2 + 2cx2(1 ; x) + c2x2(1 ; x)2 dx

i

1 + 2e 1 + c2 1 3 12 30 3 V (c = ;5=2) = 24

5. F = xT Ax ; xT x =
X

ij

Aij xixj ;
X

x2 i
i

@F = @xk

) Ax + AT x ; 2 x = 0
Since A is symmetric min F = xT Ax + (xT x ; c) implies (by di erentiating with respect to xk k = 1 : : : n)

Akj xj +

Aik xi ; 2 xk = 0

k = 1 2 ::: n

Ax = x

Ax = x T b. = xxTAx x = To minimize we require d = d ; d =0

2

Divide by d ; d or

=0

d =0

CHAPTER 2

2 Examples, Notation
Problems
1. For the integral with

I=

x2 x1

f (x y y ) dx
0 0

f = y 1= 2 1 + y 2 write the rst and second variations I (0), and I (0).

0 00

J (y) = (1 + x)(y )2dx 0 where y is twice continuously di erentiable and y(0) = 0 and y(1) = 1: Of all functions of the form y(x) = x + c1x(1 ; x) + c2x2(1 ; x) where c1 and c2 are constants, nd the one that minimizes J:
0

2. Consider the functional

1. f = y1=2(1 + y 2) 1 y 1=2(1 + y 2) fy = 2 fy = 2y y1=2 x2 1 1=2 2 1=2 I (0) = y (1 + y ) + 2 y y dx x1 2 1 y 3=2(1 + y 2) fyy = ; 4 fyy = y 1=2y
0 ; 0
0

fy y = 2y1=2
0 0

I (0) =
00

x2

x1

1y ;4

3=2

(1 + y 2) 2 + 2y
0

1=2

+ 2 y 1 =2

dx

10

2. We are given that,after expanding, y(x) = (c1 + 1)x + (c2 ; c1)x2 ; c2x3.Then we also have that y (x) = (c1 + 1) + 2(c2 ; c1)x ; 3c2x2 and that (y (x))2 = (c1 + 1)2 + 4x(c1 + 1)(c2 ; c1) ; 6x2c2(c1 + 1) 4 +4x2(c2 ; c1)2 ; 12x3c2(c2 ; c1) + 9c2 2x Therefore, we now can integrate J (y) and get a solution in terms of c1 and c2
0 0

R1

3 (c1 + 1)2 + 10 (1 + x)(y )2dx = 2 3 (c1 + 1)(c2 ; c1 ) 7 2 ; 14 4 c2 (c1 + 1) + 3 (c2 ; c1 ) 99 2 ; 27 5 c2 (c2 ; c1 ) + 30 c2

0

To get the minimum, we want to solve Jc1 = 0 and Jc2 = 0. After taking these partial derivatives and simplifying we get 7 c ;1 =0 Jc2 = 17 c 1+ 30 15 2 6 and 17 c ; 1 = 0 Jc1 = c1 + 30 2 3 Putting this in matrix form, we want to solve
"

1 Using Cramer's rule, we have that

17 30

7 15 17 30

#"

c1 = c2

"

1 6 1 3

c1 =
and

1 6 1 3 17 30

7 15 17 30 7 15 17 30 1 6 1 3 7 15 17 30

55 = 131

:42

c2 =

1 Therefore, we have that the y(x) which minimizes J (y) is

17 30

17 30

20 =; 131

;:15

y(x) =

77 2 20 3 x + 131 x + 131 x 1:42x ; :57x2 + :15x3 186 131

;

11

Using a technique found in Chapter 3, it can be shown that the extremal of the J (y) is 1 ln(1 + x) y(x) = ln 2 which, after expanding about x = 0 is represented as

y(x) =

; ln x + ln x + R(x) 1:44x ; :72x + :48x + R(x)

ln2 x
1 1 2 2 2 1 3 3 2 3 2

So we can see that the form for y(x) given in the problem is similar to the series representation gotten using a di erent method.

12

CHAPTER 3

3 First Results
Problems
1. Find the extremals of

I=

x2

for each case a. F = (y )2 ; k2y2 (k constant) b. F = (y )2 + 2y c. F = (y )2 + 4xy d. F = (y )2 + yy + y2 e. F = x (y )2 ; yy + y f. F = a(x) (y )2 ; b(x)y2 g. F = (y )2 + k2 cos y b 2. Solve the problem minimize I = (y )2 ; y2 dx a with y(a) = ya y(b) = yb: What happens if b ; a = n ?
0 0 0 0 0 0 0 0 0 0

x1

F (x y y ) dx
0

3. Show that if F = y2 + 2xyy , then I has the same value for all curves joining the endpoints.
0

4. A geodesic on a given surface is a curve, lying on that surface, along which distance between two points is as small as possible. On a plane, a geodesic is a straight line. Determine equations of geodesics on the following surfaces: 2 dz 2 2 2 2 2 a + d d a. Right circular cylinder. Take ds = a d + dz and minimize 2 d 2 or a dz + 1 dz] b. Right circular cone. Use spherical coordinates with ds2 = dr2 + r2 sin2 d 2 :] c. Sphere. Use spherical coordinates with ds2 = a2 sin2 d 2 + a2d 2:] d. Surface of revolution. Write x = r cos y = r sin , z = f (r): Express the desired relation between r and in terms of an integral.]
v Z u u t ! v Z u u t !

13

5. Determine the stationary function associated with the integral

I=
when y(0) = 0 and y(1) = 1, where

1 0

(y )2 f (x) ds
0

f (x) =
6. Find the extremals 1 a. J (y) = y dx
Z
0

8 > > > < > > > :

;1
1

0 x< 1 4
1 4

<x 1

b. J (y) = c. J (y) =

y(0) = 0 y(1) = 1: y(0) = 0 y(1) = 1: y(0) = 0 y(1) = 1:

0 Z 1 0

yy dx
0

xyy dx
0 0

7. Find extremals for 1 2 a. J (y) = y x3 dx

Z

b. J (y) =

y2 + (y )2 + 2yex dx:
0

8. Obtain the necessary condition for a function y to be a local minimum of the functional

J (y) =

Z Z

K (s t)y(s)y(t)dsdt +

b a

y2dt ; 2

y(t)f (t)dt

where K (s t) is a given continuous function of s and t on the square R, for which a s t b K (s t) is symmetric and f (t) is continuous. Hint: the answer is a Fredholm integral equation. 9. Find the extremal for

J (y) =
10. Find the extremals

1 0

(1 + x)(y )2dx
0

y(0) = 0 y(1) = 1:
0

What is the extremal if the boundary condition at x = 1 is changed to y (1) = 0?

J (y) =

x2(y )2 + y2 dx:
0

14

1.

x2

F (x y y )dx Find the externals.

0 0 0 0

a. F (y y ) = (y )2 ; k2y2 k = constant by Euler's equation F ; y Fy = c so,

0

) (y ) = ;(ky) ; c ) y = (;(ky) ; c) = dy dy = dx ) = (;(ky) ; c) (ky) + c] = = idx p Using the fact that p du = ln ju + u + a j we get
0

(y )2 ; (ky)2 ; y (2y ) = ;(y )2 ; (ky)2 = c

0 0 0 0

1 2

1 2

1 2 2

u2 + a2

dy 2 2 ((ky) + c)1=2 = ln jky + (ky) + cj =

q q

idx = ix

ky + (ky)2 + c = e

ix
Z

Let's try another way using

Z Z

dy = dx p;c2 ; (ky)2]1=2 ky = sin( x) = sin x p ) sin 1 pky = x ) ;c ;c

;

padu ;u
2

= sin 1 u a
;

y=

p;c

k sin x

c<0

15

b. F (y y ) = (y )2 + 2y
0 0 0
0

F ; y Fy = (y )2 + 2y ; y (2y ) = ;(y )2 + 2y = c ) (y )2 = 2y ; c ) p2dy y ; c = dx

0 0 0 0 0

) (2y ; c) =

1 2

= x

2y ; c = x2

2 y=1 2 (x + c)

c. F (x y ) = 4xy + (y )2
0 0 0

use Fy = c
0

4x + 2y = c
0

1 (c ; 4x) y =2
0

y=1 2 x(c ; 2x)

16

d. F = y 2 + yy + y2
0 0 0
0

F ; y Fy = c see (21) Fy = 2y + y
0

) F ; y Fy = y + yy + y ; y (2y + y) = ;y + y ) ;y + y = c y =y ;c y = y ;c pdy y ; c = dx
0
0

2
0

y +c )=x (arc cosh p c 2

can also be written as ln j y + y2 ; c j y = x;c arc cosh p 2 c y cosh ( x ; c2) = p c

q

y = c cosh( x ; c2)

17

e. F = x y 2 ; yy + y d F =F see (12) y dx y Fy = ;y + 1
0 0
0

Fy = 2xy ; y d (2xy ; y) = ;y + 1 dx 2y + 2xy ; y = ;y + 1

0

00

2xy + 2y = 1
00 0

(2xy ) = 1
0 0

2xy = c1 c1 y =2 x dy = c21 dx x y = c21 ln j x j + c2

0 0

f. F = a(x)y 2 ; b(x)y2
0

Fy = ;2b(x)y
0

Fy = 2a(x)y d F = F ) (2a(x)y ) = ;2by y dx y a(x)y + a y + by = 0

0
0

0 0

00

Linear nonconstant coe cients. Can be Solved !

18

g. F = y 2 + k2 cos y
0

F ; y Fy = c
0 0

Fy = 2y
0 0

y 2 + k2 cos y ; 2y 2 = c1
0

y 2 = c1 ; k2 cos y dy = c ; k2 cos y 1 dx pc1 ;dy k2 cos y = dx pc1 ;dy x + c2 = k2 cos y

0

;y

+ k2 cos y = c1
q

19

2. F = y 2 ; y2
0

From problem 1a with k = 1 we have

y=

p;c sin x y(a) = ya ) y(b) = yb )

c<0 ya = yb =

yb sin x = ya sin x The solution is not unique: y = sin b sin a

If b = a + n then yb = = otherwise no solution.

sin b = yb to get a solution. sin a ya

p;c sin a p;c sin b

p;c sin(a + n p;c sin a

then yb = ya for a solution !

20

3. If F = y2 + 2xyy , show I has the same values for all curves joining the end-points.
0

Using Euler's equation (12) in Chapter 3, we need only show d F (x) = F (x) x x x : y 2 dx y
0

Fy = 2xy Fy = 2y + 2xy d F (x) = d (2xy) = 2xy + 2y dx y dx which is Fy

0

Note that

x2 x1 F

= xy2

x2 x1

d (xy2) F = dx
independent of curve.

21

4. a. Right circular cylinder min

v Z u u t

a + dz d
2
0 0

!2

F ( z z ) = a2 + z 2
0

p
0

F ; z Fz = c 2 2 Fz = 1 ( a + z ) 2
0 0

1=2

pa

2z
0

a2 + z 2 ; z 2 = c1 a2 + z 2 pa2 + z 2 = a2 c1
0 0 0 0

+ z 2 ; z 2 p 2 1 2 = c1 a +z
0 0

2 a a +z = c 1 2
0

!2

2 z2= a c1
0

!2

;a
!2

z=
0

v u u t

a2 c1 a2 c1

;a ;a

z=

v u u t

!2

+ c2

2 parameter family of helical lines.

22

4. b. Right circular cone

v Z u u t

dr 2 + r2 sin2 d d
! q
0 0

F ( r r ) = r 2 + r2 sin2
No dependence on , thus we can use (21) 1 (r 2 + r2 sin2 ) 1=2 2r Fr = 2 F ; r Fr = c1 2 r 2 2 2 p ) r + r sin ; r 2 + r2 sin2 = c1
0

r 2 + r2 sin2
0

;r
"

= c1 r 2 + r2 sin2
0

r + r sin
0

2 2 = sin c1 r

!2

r = r sin
0

r2 sin2 c2 1

;1

2 2 2 r = r sin c1 r sin ; c1 dr d = c1 r sin r2 sin2 ; c2 1

q
0

Let = r sin d = sin

Z q

;c
;

2 1

d c1

1 sec 1 r sin + c c = 2 1 sin c1 r sin = c1 sec ( ; c2c1) sin ]

23

4. c. Sphere a2 d d
v Z u u t

!2

+ a2 sin2 d
0

F ; F = c1
0
0

F(

) = a2 sin2 + a2

)
2

a2 sin2 + a2
+a
2
0

;
0

(2a2 ) a2 sin2 + a2
0 0

= c1
0

a sin
0

;a

2
!2

= c1 a2 sin2 + a2
3 5

= sin2 =
q

a sin c1 a sin2 c1

;1

sin

;1
d

sin

d 2 a c1 sin

;1 =

24

4. d. Surface is given as

~ r( )
in parametric form

x = cos y = sin z = f( )
The length L( ) =
Z

t1 t0

~ r ~ r

+ 2~ r

~ r

+~ r

~ r

dt

~ r = cos # i + sin # j + f ( ) k
0

~ r =

sin # i + cos # j
0 0

~ r ~ r = cos2 # + sin2 # + f 2( ) = 1 + f ( )]2 ~ r ~ r# = 0 ~ r# ~ r# =

2
Z

)
or

L=
Z v u u t

t1

t0

(1 + f ( )]2)
0

# 2 dt
0

L=

d (1 + f ( )]2) d#
0

!2

d#

d So F is a function of and d# F ; F = c1
0
0

1 (1 + f ( )]2) d F =2 d#
0

8 < :

!2

+
0

9;1 =

=2

2 (1 + f ( )]2)
0 0

(1 + f ( )]2)
0

; (1 +
0

f ( )]2)

2
0

(1 + f ( )]2)
0

= c1

= c1 (1 + f ( )]2)
0

25

!2

c1
0

= (1 + f ( )]2)
0

v u u u t

1 + f ( )]2
0

c1

26

5. F = f (x) y 2 dF =F Using (12) dx y y Fy = 0

0
0

Fy = 2f (x) y d (2f (x) y ) = 0 ) dx f (x) y = c y = f (cx) dy = f (cx) dx y = f (cx) dx + k using y(0) = 0 x y(x) = 0 f (c ) d
0

y(1) = 1 =

) ;

1 0

c d =1 f( ) cd +
=1
Z

Substituting for f :

1=4 0

1 1=4

cd = 1

1c = 1 2 c=2

1 + c 1 ; ;c 1 4 4

y(x) =

2 d f( )

27

6. a. J (y) = y dx y(0) = 0 y(1) = 1 0 Euler's equation in this case is d1=0 dx which is satis ed for all y. Clearly that y should also satisfy the boundary conditions, i.e. y = x: Looking at this problem from another point of view, notice that J (y) can be computed directly and we have (after using the boundary condition),
0

J (y ) = 1
Since this value is constant, the functional is minimzed by any y that satis es the boundary conditions. b. J (y) = yy dx y(0) = 0 y(1) = 1 0 Euler's equation in this case is dy=y dx which is the identity y = y which is satis ed for all y. Clearly that y should also satisfy the boundary conditions, i.e. y = x: Looking at this problem from another point of view, notice that J (y) can be computed directly and we have (after using the boundary condition), J (y) = 1 2 Since this value is constant, the functional is minimzed by any y that satis es the boundary conditions.
0 0 0 0

c. J (y) = xyy dx y(0) = 0 y(1) = 1 0 Euler's equation in this case is d xy = xy dx which is y + xy = xy or y = 0:

0 0 0

Clearly that y could NOT satisfy the boundary conditions. 28

7.

) dx a. J (y) = 0 (y 3 x 2 ( y ) F = x3 Euler equation d Fy = Fy dx 2 y Fy = x3 Fy = 0 y =c Integrate Euler's equation Fy = c =) 2 x3 2y = cx3 3 cx y= 2

Z

4 cx =) y = 8 + b
Z

b. J (y) = 0 (y2 + (y )2 + 2yex)dx F = y2 + (y )2 + 2yex Fx = 2yex Fy = 2y + 2ex Fy = 2y Euler equation d Fy = Fy dx d 2y = 2y + 2ex dx y ; 2y = ex Solve the homogeneous y ; 2y = 0 =) y = c1e 2x + c2e 2x Find a particular solution of the nonhomogeneous y ; 2y = ex =) y = 2ex Therefore the general solution of the nonhomogeneous is: y = c1e 2x + c2e 2x + 2ex
0 0
0

00

00

p ;

00

29

8. Obtain the necessary condition for a function y to be a local minimum of the functional:

J (y) =

Z Z

b b

Find the rst variation of J,

a a

K (s t)y(s)y(t)dsdt + y(t) dt ; 2 y(t)f (t)dt

Z

J (y + " ) =

Z Z

b b

a a

K (s t) y(s) + " (s)] y(t) + " (t)]dsdt + y(t) + " (t)]2dt

b a

;2
Then,

y(t) + " (t)]f (t)dt

d J (y + " ) = d"

Z Z

b b

a a
Z

fK (s t) y(s) + " (s)] (t) + K (s t) y(t) + " (t)] (s)g dsdt

b
Z

+2 y(t) + " (t)] (t)dt ; 2 Now letting " = 0 we have,

a
8 Z <Z

(t)f (t)dt
8 Z <Z

d J (y + " ) = d" "=0

+2 y(t) (t)dt ; 2 f (t) (t)dt

Z

a Zb a

K (s t)y(s)ds (t)dt +
b a

9 =

K (s t)y(t)dt (s)ds

9 =

Since the limits of s and t are constants, we can interchange s for t, and vice versa, in the second of four terms above, =
8 Z <Z

Combining the rst two terms and factoring out an (t)dt yields: =
8 Z <Z

K (s t)y(s)ds (t)dt +
b b

9 =

8 Z <Z

K (t s)y(s)ds (t)dt +2 y(t) (t)dt ; 2 f (t) (t)dt

Z Z

9 =

Setting this equal to 0 implies: 1 b K (s t) + K (t s)] y(s)ds + y(t) = f (t) 2a Which is a Fredholm equation.
Z

K (s t) + K (t s)] y(s)ds + 2y(t) ; 2f (t) (t)dt

9 =

30

9. Given F = (1 + x)(y )2. It is easy to nd that Fy = 2y (1 + x) Fy = 0 d F = 0 =) d y (1 + x) = 0 Therefore dx y dx Integrating both sides we obtain, 1 y (1 + x) = c1 =) y = (1 c + x)
0
0

Integrating again leads to y = c1 ln(1 + x) + c2 Now applying the boundary conditions, y(0) = 0 =) c1 ln(1 + 0) + c2 = 0 =) c2 = 0 1 y(1) = 1 =) c1 ln(1 + 1) = 1 =) c1 = ln 2 Therefore the nal solution is + x) y = ln(1 ln 2
0

It is easy to show that in that case the functional J (y) is 1 : ln 2 If our boundary condition at x = 1 was y (1) = 0, then c1 y = c1 ln(1 + x) and y = 1 + x c 1 Then y (1) = 1 + 1 = 0 =) c1 = 0 and we get the trivial solution.
0 0

31

10. Find the extremal:

J (y ) =
0

b 2 2 a (x y
0

+ y2) dx
0

F = x2(y )2 + y2 dF dx y

Fy = 2x2y Fy = 2y d F = 2x2y + 4xy dx y

0
0

00

Euler's Equation:

; Fy = 0
2x2y + 4xy x2y + 2xy
00 00 0 0

; 2y = 0 ; y=0
Thus (a b) must not contain the origin.

This is an Euler equation

0 ;

Let y = xr y = rxr 1 y = r(r ; 1) xr

00

Substituting, (r2 ; r) xr + 2rxr ; xr = 0 (r2 ; r + 2r ; 1) xr = 0 r2 ; r + 2r ; 1 = 0 r2 + r ; 1 = 0 2 2

p1 + 4 ;1 p5 ; 1 = r=
y= x y(x) = c1 x
;
;

1; 2

y= x
0:618

1+ 2

1:618

+ c2 x

for a

32

CHAPTER 4

4 Variable End-Point Problems

Problems
1. Solve the problem minimize I = y2 ; (y )2 dx 0 with left end point xed and y(x1) is along the curve
Z

x1

x1 = 4 :
2. Find the extremals for

I=
where end values of y are free.

1 0

1 (y )2 + yy + y + y dx 2
0 0 0

3. Solve the Euler-Lagrange equation for

I=
where

b a

y 1 + (y )2 dx
0

y(a) = A a = ;b

y(b) = B: A=B

b. Investigate the special case when

and show that depending upon the relative size of b B there may be none, one or two candidate curves that satisfy the requisite endpoints conditions. 4. Solve the Euler-Lagrange equation associated with

I=

y2 ; yy + (y )2 dx
0 0

5. What is the relevant Euler-Lagrange equation associated with

I=

1 0

y2 + 2xy + (y )2 dx
0

6. Investigate all possibilities with regard to tranversality for the problem 33

min
Z

b a

1 ; (y )2 dx
0

7. Determine the stationary functions associated with the integral

I = 0 (y )2 ; 2 yy ; 2 y dx where and are constants, in each of the following situations: a. The end conditions y(0) = 0 and y(1) = 1 are preassigned. b. Only the end conditions y(0) = 0 is preassigned. c. Only the end conditions y(1) = 1 is preassigned. d. No end conditions are preassigned.
1
h
0 0 0

8. Determine the natural boundary conditions associated with the determination of extremals in each of the cases considered in Problem 1 of Chapter 3. 9. Find the curves for which the functional

I=

x1
0

p1 + y
y

dx

with y(0) = 0 can have extrema, if a. The point (x1 y1) can vary along the line y = x ; 5: b. The point (x1 y1) can vary along the circle (x ; 9)2 + y2 = 9: 10. If F depends upon x2, show that the transversality condition must be replaced by x2 @F F + ( ; y ) @F + @y x=x2 x1 @x2 dx = 0:
" # Z
0 0 0

11. Find an extremal for

J (y) =

1 x2(y )2 ; 1 y2 dx 2 8
0

y(1) = 1 y(e) is unspeci ed.

12. Find an extremal for

J (y ) =

1 0

(y )2dx + y(1)2
0

y(0) = 1 y(1) is unspeci ed.

34

1. F = y2 ; (y )2 dF =0 Fy ; dx y d (;2y ) = 0 2y ; dx y +y =0
0
0

00

y = A cos x + B sin x
using y(0) = 0

y = B sin x
Now for the transversity condition

F + ('

slope of curve Since the curve is x = 4 (vertical line, slope is in nite) we should rewrite the condition 1 + (1 ; y ) F = 0 F ' ' y
0 0 0
0

"

; y )Fy x
0
0

=4

=0

|{z}

=0

|{z}

=0

Fy

x= =4
0

=0
=4

;2y x

=0

) ;2B cos 4

=0

)
y

B=0
0:

35

2. F = 1 y 2 + yy + y + y 2 dF =0 Fy ; dx y Fy = y + 1
0 0 0
0

Fy = y + y + 1 d F = y + 1 ; (y + y + 1) = 0 Fy ; dx y y +1;y ;y =0
0

00

00

;1=0

yH = Ax + B 2 yP = 1 2x
2 y = Ax + B + 1 2x

Free ends at x = 0 x = 1

F + ('

; y )Fy x
0
0

=0

=0 =0

F + (' Fy Fy

; y )Fy x
0
0

=1

The free ends are on vertical lines x = 0 x = 1

0

x=0

=0 =0

) )

y (0) + y(0) + 1 = 0
0

A+B+1=0 y (1) + y(1) + 1 = 0

0

x=1

5 =0 2A + B + 2

36

A+B+1 =0
2A + B + 5 2 =0 A + 3=2 = 0 1 + 1 x2 y = ;3 x + 2 2 2

9 > > > = > > >

; )

A = ;3=2 B = ;A ; 1 = 1=2

37

3. F = y 1 + y 2
0

Fy = 1 + y 2 2 1=2 (1 + y ) 2y Fy = y 1 2 Fy = p1yy +y2 F ; y Fy = c1 yy 2 = c y 1 + y 2 ; p1 +y2 1

0
0

y(1 + y 2) ; yy 2 = c1 1 + y 2
q
0 0 0

y2 = 1 + y 2 c2 1
0

y2 ; 1 c2 1 c1 dy = dx y 2 ; c2 1 c1 arc cosh cy + c2 = x 1 y=
0

arc cosh cy = xc; c2 1 1 c1 cosh xc; c2 = y 1 y(a) = A y(b) = B

OR c1 ln y + y2 ; c2 x 1 + c2 =
q

) )

c1 cosh a c; c2 = A 1 c1 cosh b c; c2 = B 1

a ; c2 = c1 arc cosh cA1 b ; c2 = c1 arc

38 cosh cB 1

9 > = >

a
This gives c1 , then

A ; arc cosh B b = c1 arc cosh c c1 1 A: c2 = a ; c1 arc cosh c 1

(1)

(2)

If a =

zero on left of (1)

;b +

A=B
zero on right of (1)

Thus we cannot specify c1 , based on that free c1 , we can get c2 using (2). Thus we have a one parameter family.

39

4. F = y2

arc cosh pyc + c2 = x 1

; y Fy = c y ; yy + y ; y (; y + 2y ) = c y ; (y ) = c (y ) = y ; c y = y ;c py dy ; c = dx
F
0
0

; yy
2

+ (y )2
0

2 2

1 1

40

5. F = y2 + 2xy + (y )2 dF =0 Fy ; dx y d (2y ) = 0 2y + 2x ; dx ;2y + 2y = ; 2x

0
0

00

y y

00

00

;y=x ;y=0 )

yh = Aex + Be yp = Cx + D

;Cx ; D = x C = ;1 D = 0 yp = ; x y = Aex + Be x ; x
;

41

6. F = 1 ; (y )2
q
0

Fy =
0

;2y ) 1 ; (y )
0 0

y = Ax + B

y =A
0

F +( F +(
q

; y ) Fy
0 0

x =a x=b

=0 =0
0 0

; y ) Fy
0

1 ; (y )2 + (
0 0

1 ; (y )2 + (y )2 1 1 ) =A =y x=a 1 1 =y ) =A x=b Therefore if both end points are free then the slopes are the same 1 ' (a) = (b) = A
0 0 0 0 0 0 0 0 0 0

y ; y) 1; ; (y ) ; y =0
q
0

x=a b

=0

42

7. F = (y )2
0

;2

yy

;2

a. y(0) = 0 y(1) = 1 dF =0 Fy ; dx y d (2y ; 2 y ; 2 ) = 0 ;2 y ; dx ;2 y ; 2y + 2 y = 0

0

00

y =0
00

y = Ax + B y(0) = 0 y(1) = 1

) )

B=0 A=1

y=x

b. If only y(0) = 0

y = Ax Fy
=0

Transversality condition at x = 1
0

imples
0

x=1

2y (1) ; 2 y(1) ; 2 = 0 Substituting for y

2A ; 2 A ; 2 = 0

Thus the solution is

A = 1; y = 1 ; x:
43

c. y(1) = 1

only

y = Ax + B y(1) = 1 Fy

y = Ax + 1 ; A
0

A+B =1

B =1;A

2y (0) ; 2 y(0) ; 2 = 0
0

x=0

=0

A ; (1 ; A) ; = 0 A (1 + ) = A=
+ +1 +

44

d. No end conditions

y = Ax + B y =A
0

2y (0) ; 2 y(0) ; 2 = 0
0

2y (1) ; 2 y(1) ; 2 = 0
0

2A ; 2 B

;2 2A ; 2 (A + B ) ; 2 ) A=0 + 2 B = 2A ; 2 = ; 2 B=;
2 A=0

=0 =0

9 > = >

45

8. Natural Boundary conditions are

Fy

x=a b

=0 (1a of chapter 3)

a. For

F = y 2 ; k2 y2
0

Fy = 2y
0

y (a) = 0 y (b) = 0
0 0

b. For F = y 2 + 2y
0

exactly the same

c. For F = y 2 + 4xy
0
0

Fy = 2y + 4x
0

y (a) + 2a = 0 y (b) + 2b = 0
0 0

46

d. F = y 2 + yy + y2
0 0

Fy = 2y + y
0

2y (a) + y(a) = 0 2y (b) + y(b) = 0

0 0

e. F = x y 2 ; yy + y
0 0

Fy = 2xy
0

2ay (a) ; y(a) = 0 2by (b) ; y(b) = 0

0 0

;y

f. F = a(x) y 2 ; b(x)y2
0

Fy = 2a(x) y
0

2a( ) y ( ) = 0
0

2a( ) y ( ) = 0
0

Divide by 2a( ) or 2a( ) to get same as part a.

g. F = y 2 + k2 cos y
0

Fy = 2y
0

same as part a

47

9. F =

p1 + y
y

Fy

dF =0 ; dx y p1 + y Fy = ; y
0

Fy = yp1y+ y 2
0
0

y y (1 + y ) ; y (1 + y ) ; yy p ; y y (1 + y ) 1 + y ; (1 + y ) ; y y ; y ; y ] = 0 ; 1 ; 2y ; y ; y y + y + y = 0 yy + y = ; 1 (yy ) = ;1 yy = ; x + c ydy = (; x + c ) dx 1y = ; x + c x + c 2 2 y = ; x + 2c x + 2c y(0) = 0 ) c = 0 Fy

yy 2 2 d F = y y 1 + y ; y y 1 + y + y p1+y 2 dx y y2 (1 + y 2)
00 0 0 0 0
0 00 0 0

p
00

dF = dx y
0

p1 + y
2
0 00

y =0
00

2 2 2

00

00

0 0

a. Transversality condition:

=1

F + (1 ; y ) Fy ]
0
0

"

p1 + y
y
0

(1p ;y )y +y 1+y2
0 0 0 0

x=x1

=0
#

x=x1

=0

(1 + y 2 + y

;y

x=x1

=0 48

Since y2 = ; x2 + 2c1x 2yy =

0

1 + y (x1) = 0
0

y (x1) = ; 1
0

; 2x + 2c
;

at x = x1
| {z
;

b.

;2(x ; 5) = ; 2x + 2c ) c =5 p10x ; x y = ; x + 10x or y = On (x ; 9) + y = 9

1 1 1 1 2 2 2 2
0

2y(x1) y (x1) = ; 2x1 + 2c1 x1 5 = 1 on the line

0

} | {z }

The slope
0

yy = ; (x ; 9)
At x = x1
0

2(x ; 9) + 2yy = 0
0

is computed

( x1 ) =

;9 ; xy(x )
1 1

Remember that at x1 y(x1) from the solution:

y(x1)2 = ; x2 1 + 2c1 x1

is the same as from the circle

) ;x ) ;x
F+(

y(x1)2 + (x1 ; 9)2 = 9

2 1 2 1

c1x1 = 9x1 ; 36
0 0

+ 2c1x1 = 9 ; x2 1 + 18x1 ()

+ 2c1x1 = 9 ; (x1

; 9)

; 81

Substituting in the transversality condition

; y ) Fy ] x = 0
0

49

we have

y (x1) = 0 ;x + c1 y x1 1 ; 9 ;x1 + c1 1;x y(x1) y(x1) = 0 x1 ; c1) = 0 1 + (x1 ;y9)( 2(x ) 1

0 0

1 + (x1)

| {z }

9;(x1 ;9)2

y2(x1) + (x1 ; 9)(x1 ; c1) = 0

| {z }

9 ; (x1 ; 9) x1 ; 9 ; x1 + c1)] = 0 9 ; (x1 ; 9) (c1 ; 9) = 0 Solve this with (*) c1 x1 = 9x1 ; 36 to get: c1 = 9 ; 36 x1

9 ; (x1 ; 9)2 + (x1 ; 9)(x1 ; c1) = 0

:36 9 + 36 = 9x 1
36 x1 = 9 45

9 ; (x1 ; 9)(; 36 x)=0

1

x1 = 36 5

c1 = 9 ; 5 = 4

y2 = ; x2 + 8x

50

10. In this case equation (8) will have another term resulting from the dependence of F on x2( ), that is x2 (0) @F dx x1 (0) @x2
Z

51

11. In this problem, one boundary is variable and the line along which this variable point moves is given by y(e) = y2 which implies that is the line x = e. First we satisfy Euler's 1 2 1 2 2 rst equation. Since F = x (y ) ; y , we have 2 8
0

and so, Therefore

d F =0 Fy ; dx y
0

0 =

d (x y ) ; y ; dx
1 4 2
0

2 = ;1 4 y ; (2xy + x y ) 1 = x2y + 2xy + 4 y

0 00 00 0

x2y + 2xy + 1 4y = 0 This is a Cauchy-Euler equation with assumed solution of the form y = xr . Plugging this in and simplifying results in the following equation for r r2 + (2 ; 1)r + 1 4 =0 1 which has two identical real roots, r1 = r2 = ; 2 and therefore the solution to the di erential equation is 2 ln x y(x) = c1x 12 + c2x 1 The initial condition y(1) = 1 implies that c1 = 1. The solution is then
00 0 ; ;

y=x

1=2

+ c2x
0

1=2

ln x:

To get the other constant, we have to consider the transversality condition. Therefore we need to solve F + ( ; y )Fy jx=e = 0 Which means we solve the following (note that is a vertical line)
0
0

; F + (1 ; y )Fy x e
0 0 0 0

= Fy jx=e = x2y jx=e = 0

0

which implies that y (e) = 0 is our natural boundary condition. 1 c x 3=2 ln x + c x 3=2 3=2 y = ;1 x ; 2 2 22 With this natural boundary condition we get that c2 = 1, and therefore the solution is
0 0 ; ; ;

1 y(x) = x 2 (1 + ln x)
;

52

12. Find an extremal for J (y) =

0

1 0

(y )2dx + y(1)2, where y(0) = 1, y(1) is unspeci ed.

0

F = (y ) + y(1) , Fy = 0 Fy = 2y . Notice that since y(1) is unspeci ed, the right hand value is on the vertical line x = 1. By the Fundamental Lemma, an extremal solution, y, must satisfy the Euler equation d F = 0: Fy ; dx y d 2y = 0 0 ; dx ;2y = 0 y = 0: Solving this ordinary di erential equation via standard integration results in the following: y = Ax + B . Given the xed left endpoint equation, y(0) = 1, this extremal solution can be further re ned to the following: y = Ax + 1. Additionally, y must satisfy a natural boundary condition at y(1). In this case where y(1) is part of the functional to minimize, we substitute the solution y = Ax + 1 into the functional to get:
2 2
0

00

00

I (A) = A2dx + (A + 1)2 = A2 + (A + 1)2 0 Di erentiating I with respect to A and setting the derivative to zero (necessary condition for a minimum), we have
Therefore and the solution is 2A + 2(A + 1) = 0

A = ;1 2 y = ;1 2 x + 1:

53

CHAPTER 5

5 Higher Dimensional Problems and Another Proof of the Second Euler Equation
Problems
1. A particle moves on the surface (x y z) = 0 from the point (x1 y1 z1) to the point (x2 y2 z2) in the time T . Show that if it moves in such a way that the integral of its kinetic energy over that time is a minimum, its coordinates must also satisfy the equations x = y = z:
x y z

2. Specialize problem 1 in the case when the particle moves on the unit sphere, from (0 0 1) to (0 0 ;1), in time T . 3. Determine the equation of the shortest arc in the rst quadrant, which passes through the points (0 0) and (1 0) and encloses a prescribed area A with the x-axis, where A 8 . 4. Finish the example on page 51. What if L = 2 ? 5. Solve the following variational problem by nding extremals satisfying the conditions

J (y1 y2) =

2 2 4y1 + y2 + y1y2 dx
0 0

y1(0) = 1 y1 4 = 0 y2(0) = 0 y2 4 = 1:
6. Solve the isoparametric problem

J (y) =
and

1 0

(y )2 + x2 dx y(0) = y(1) = 0
0

1 0

y2dx = 2:

7. Derive a necessary condition for the isoparametric problem Minimize b I (y1 y2) = L(x y1 y2 y1 y2)dx
Z

54

subject to and

b a

G(x y1 y2 y1 y2)dx = C
0 0

y1(a) = A1 y2(a) = A2 where C A1 A2 B1 and B2 are constants. I (x y) =

subject to
Z Z

y1(b) = B1

y2(b) = B2

8. Use the results of the previous problem to maximize

t1 t0

(xy _ ; yx _ )dt

t1 q t0

x _2 + y _ 2dt = 1:

Show that I represents the area enclosed by a curve with parametric equations x = x(t) y = y(y) and the contraint xes the length of the curve. 9. Find extremals of the isoparametric problem

I (y) =
subject to

(y )2dx
0

y(0) = y( ) = 0

y2dx = 1:

55

1. Kinetic energy E is given by T 1 2 (x _ +y _2 + z _ 2) dt E= 2 0 The problem is to minimize E subject to

Z

'(x y z) = 0
Let F (x y z) = 1 _2 + y _2 + z _ 2) + '(x y z) 2 (x Using (65) dF =0 j=1 2 3 Fyj ; dt yj x = dx _=0 ) ' 'x ; dt x dy y = 'y ; dt _=0 ) ' y d z_ = 0 ) z = 'z ; dt 'z ) 'xx = 'yy = 'zz
0

56

2. If ' x2 + y2 + z2 ; 1 = 0 y = z =; then 2x = x 2y 2z x+2 x=0

y+2 y =0 z+2 z=0

Solving

x = A cos 2 t + B sin 2 t p p y = C cos 2 t + D sin 2 t p p z = E cos 2 t + G sin 2 t

Use the boundary condition at t = 0 x(0) = y(0) = 0 z(0) = 1 )A = 0

C=0 E=1
Therefore the solution becomes

p p y = D sin 2 t p p z = cos 2 t + G sin 2

x = B sin 2 t x(T ) = 0 y(T ) = 0 z(T ) = ;1

The boundary condition at t = T

) B sin p2

t=0

) p2 p2

same conclusion for y ) x = B sin nT t

t=n =n T

57

y = D sin n T t n t z = cos n t + G sin T T Now use z(T ) = ; 1

) ;1 = cos n ) n is odd

+ G sin n
|

=0

{z

x = B sin n T t y = D sin n T t n t t + G sin z = cos n T T

Now substitute in the kinetic energy integral

9 > > > > > > > > > = > > > > > > > > >

n = odd

1 (x _2 + y _ 2 + z_2) dt 0 2 1 T n B 2 cos2 n t + n D 2 cos2 n t = 2 T T T T 0

Z

n t + ; sin n t + G cos T T
1 = 2 + n T
2
Z

n T

dt

T
0

n T
2

B 2 + D2 + G2 cos2 n T t
)

sin2 n T t

;2G nT

n t dt sin n t cos T T

T
0

n t dt = T cos 2n t T = 0 sin 2T 2n T 0
58

T sin2 n t dt = ; 1 T sin 2n t + 1 t = T T 2 2n T 2 0 2 0 n ; cos 2 T t + 1 2 T T dt = cos2 n t T 2 0 2n cos T t + 1 2 1 n 2 (B 2 + D2 + G2 + 1) T E=2 T 2 Clearly E increases with n, thus the minimum is for n = 1:
Z

{z

{z

Therefore the solution is

x = B sin T t y = D sin T t

z = cos sin T t + G sin T t

59

3. Min L =

1 0
Z

1 + y 2 dx
0

subject to A =
q

1 0

y dx

=8

F = 1 + y2 + y
0

; y Fy
0

= c1
0 0

y + 1 + y2 ; y
q q
0

1 + y 2 = ;1 1 1 + y 2 = (c ; y)2 1 (y

y 1 + y 2 + 1 + y 2 ; y 2 = c1 1 + y 2
q
0 0 0

p1 y+ y
0

= c1

;c)
1
0

y = i
0

(c1 ;
Z q

dy
y
)2

; (c ;1
1

y)2 + 1

(c1 ; y) dy = i (c1 ; y)2 + 1

+1

= i dx
Z

dx

Use substitution

du = 2(c1 ; y) (; ) dy (c1 ; y) dy = ; du 2 1 du = i dx ;2 u1=2 1=2 ; 21 u 1=2 = i x + c2 substitute for u

Z Z

) )

u = (c1 ; y)2 ; 1

(c1

y)2 ; 1 = ( i x + c2)
60

square both sides (c1

y)2 ; 1 = 2( i x + c2)2 ; c1 + y 2 ; 12 = (;x2 2ix c2 + c22)

2

y ; c1 y ; c1

1 = 2

;x
|

2i c2 x ; c2 2
(x+D)2
{z

+ (x + D)2 = 12

We need the curve to go thru (0, 0) and (1, 0)

x=y=0 x=1 y=0 D2 D2

; c1

+ D = 12
2 2

9 > > > > > = > > > > >

) ;c

+ (1 + D)2 = 12

; (1 + D) = 0 ; 1 ; 2D ; D = 0 2D = ; 1 D = ; 1=2 )
2 2

y ; c1

Let

+ x;1 2

= 12

k=

;c

then the equation is 2 1 2 2 x;1 + ( y + k ) = k + 2 4 To nd k1 we use the area A

A=
use:

1 0

y dx =

1 0

2 4

1 + k2 + 4

; x;
61

1 2

;k

3 5

dx

where

pa ; u du = u pa ; u 2
2 2 2

2 u +a arc sin 2 a

1 a2 = k 2 + 4 u=x;1 2

A= x; 2
s

1 2

k2 + 1 4

1 ; x; 2

k2 + 2
q

1 4

1 =2 arc sin x ; k2 + 1 4
q

; kx

1 0

1 k2 + 1 =4 4

;1 4+ ;

k2 + 2

1 4

arc sin
1 4

1=2 k2 +
q

1 4

;k
9 =

8 < :

1 2 + k ;1 4 4

1 + k2 + 4 2

=2 arc sin 1 k2 +

1 4

1 2 p A=1 k ; k + k + 1 = 4 arc sin 2 2 4k + 1 (4k2 + 1) arc sin p 1 A+1 k = 2 4 4k 2 + 1 1 4A + 2k = (4k2 + 1) arc sin p 2 = (4k2 + 1) arc cot 2k 4k + 1 So: 4A + 2k = (4k2 + 1) arc cot 2k and

x;1 2

+ (y + k)2 = k2 + 1 4

62

2 4. c2 2 + c1 =

2 2

c2 + (1 ; c1)2 =
2 c2 1 ; (1 ; c1 ) = 0

at (0 0) at (1 0)

subtract

c2 1 ; 1 + 2c1 c1 = 1=2
Now use (34): Since y = tan
0

;c

2 1

=0

L=
since

sec dx

sin = x ; c1 dx = cos d x=0 x=1

) )

sin sin
Z

; c = ; 21 = 1;c = 1
1 1

L=

2 1

sec cos d = (

Suppose we sketch the two sides as a function of 21 0 is the value such that L = arc sin 1 20 20 0 is a function of L ! 1 = 2 (L) c2 2 + 0 4 1 2 c2 2 = 0 (L) ; 4 63

L = arc sin 1 2 2

) = 2 arc sin 21

1.5

y=1.2/(2)

0.5

1/(20)

0.5

1.5 1.5

0.5

0.5

1.5

Figure 2:

L = =2 ) = 1=2 1;1=0 c2 2 = 4 4 1 )2 = 1 The curve is then y2 + (x ; 2 4

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0.2 0.2 A= /4 L= /2

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Figure 3:

64

5. Solve the following variational problem by nding the extremals satisfying the conditions:
2 2 J (y1 y2) = (4y1 + y2 + y1y2)dx
0 0

=4

y1(0) = 1 y1( =4) = 0 y2(0) = 1 y2( =4) = 1

Vary each variable independently by choosing
1 1

and

in C 2 0 =4] satisfying:

(0) = 2(0) = 1( =4) = 2( =4) = 0

Form a one parameter admissible pair of functions:

y1 + " 1 and y2 + "

Yielding two Euler equations of the form: d F = 0 and F ; d F = 0 Fy1 ; dx y2 y1 dx y2 For our problem:
0 0

2 2 F = 4y1 + y2 + y1y2
0 0

Taking the partials of F yields:

Fy1 Fy2 Fy1 Fy2

0 0

= = = =

8y1 2y2 y2 y1
0 0

Substituting the partials with respect to y1 into the Euler equation: dy = 0 8y1 ; dx 2 y2 = 8y1
0 00

Substituting the partials with respect to y2 into the Euler equation

dy = 0 2y2 ; dx 1 y1 = 2y2
0 00

65

Solving for y2 and substituting into the rst, second order equation: y2 = 1 2 y1 =) y1 = 16y1 Since this is a 4th order, homogeneous, constant coe cient, di erential equation, we can assume a solution of the form
00 0000

y1 = erx
Now substituting into y1 = 16y1 gives:
0000

r4erx r4 r2 r
This yields a homogeneous solution of:

= 16erx = 16 = 4 = 2 2i

y1 = C1e2x + C2e = C1e2x + C2e

Now using the result from above:

2x

+ C 3e2ix + C 4e 2ix 2x + C3 cos 2x + C4 sin 2x

;

y2 = 1 2 y1 d 2C e2x ; 2C e 2x ; 2C sin 2x + 2C cos 2x = 1 2 3 4 2 dx 1 2x 2x = 1 2 4C1e + 4C2e ; 4C3 cos 2x ; 4C4 sin 2x = 2C1e2x + 2C2e 2x ; 2C3 cos 2x ; 2C4 sin 2x
00 ; ; ;

Applying the initial conditions:

We now have 4 equations with 4 unknowns

y1(0) y1( 4 ) y2(0) y2( 4 )

= = = =

1 =) C1 + C2 + C3 = 1 0 =) C1e 2 + C2e 2 + C4 = 0 0 =) C1 + C2 ; C3 = 0 1 1 =) C1e 2 + C2e 2 ; C4 = 2

; ;

66

C1 C2 = C3 C4 Performing Gaussian Elimination on the augmented matrix: (;1) 1 1 1 0 1 1 1 ,! 1 1 ;1 0 0 0 0 (;1) e 2 e 2 0 1 0 = e 2 e 2 0 0 ,! e 2 e 2 0 ;1 1 2 The augmented matrix yields:
2 6 6 6 4

1 1 1 1 e2 e 2 e2 e 2
; ;

1 0 ;1 0 0 1 0 ;1

32 76 76 76 54

3 7 7 7 5

2 6 6 6 4

1 0 0
1 2

3 7 7 7 5

2 6 6 6 4

3 7 7 7 5

2 6 6 6 4

; ;

1 0 1 ;2 0 ;1 0 1 0 1 0 ;2 2

3 7 7 7 5

C1 + C2 + C3 = 1 ;2C3 = ;1 =) C3 = 1 2 1 1 ;2C4 = 2 =) C4 = ; 4 C1e 2 + C2e 2 + C4 = 0

;

Substituting C3 and C4 into the rst and fourth equations gives: 1 ;C = ) C C1 + C2 = 1 1 = 2 2 2 1 e 1 1 4 2 2 C1e + C2e = 4 and C1 = 2 ; C2 =) C2 = e Finally:
;

; ;1

1 2

= :4683 =) C1 = :0317

1 cos 2x ; 1 sin 2x +2 4 1 y2 = :0634e2x + :9366e 2x ; cos 2x + 2 sin 2x

y1 = :0317e2x + :4683e

2x

67

6. The problem is solved using the Lagrangian technique.

Z

L = ((y ) + x )dx + (y2 ; 2)dx 0 0 L = F + G = (y )2 + x2 + (y2 ; 2) where F = (y )2 + x2 and G = y2 ; 2 Ly = 2 y and Ly = 2y Now we use Euler's Equation to obtain d (2y ) = 2 y dx y = y Solving for y p p y = A cos( x) + B sin( x) Applying the initial conditions, p p y(0) = A cos( 0) + B sin( 0) =) A = 0 p y(1) = B sin( ) = 0 p If B = 0 then we get the trivial solution. Therefore, we want sin( ) = 0. p This implies that = n , n = 1 2 3 : : : Now we solve for B using our constraint. y = B sin(n x)
1
0

00

2 2 2 y dx = B sin (n x)dx = 2 0 0 sin 2 x 1 = 2 =) B 2 ( 1 ; 0) ; 0 = 2 B2 x ; 2 4 0 2 2 B = 4 or B = 2: Therfore, our nal solution is y = 2 sin(n x), n = 1 2 3 : : :

68

7. Derive a necessary condition for the isoperimetric problem. Minimize subject to

I (y1 y2) =
Z

L (x y1 y2 y1 y2) dx
0 0 0 0

b a

G (x y1 y2 y1 y2) dx = C

and y1(a) = A1 y2(a) = A2 y1(b) = B1 y2(b) = B2 where A1 A2 B1 B2 and C are constants. Assume L and G are twice continuously di erentiable functions. The fact that b G (x y1 y2 y1 y2) dx = C is called an isoperimetric constraint.
Z
0 0

Let W =

G (x y1 y2 y1 y2) dx
0 0

We must embed an assumed local minimum y(x) in a family of admissible functions with respect to which we carry out the extremization. Introduce a two-parameter family

zi = yi(x) + "i i (x)

where
1 2

i=1 2 i=1 2
(11)

C 2 (a b) and
i (a)

= i(b) = 0

and "1 "2 are real parameters ranging over intervals containing the orign. Assume W does not have an extremum at yi then for any choice of 1 and 2 there will be values of "1 and "2 in the neighborhood of (0 0) for which W (z) = C: Evaluating I and W at z gives

J ("1 "2) =

b a

L (x z1 z2 z1 z2) dx and V ("1 "2) =

0 0

b a

G (x z1 z2 z1 z2) dx
0 0

Since y is a local minimum subject to V the point ("1 "2) = (0 0) must be a local minimum for J ("1 "2) subject to the constraint V ("1 "2) = C . This is just a di erential calculus problem and so the Lagrange multiplier rule may be applied. There must exist a constant such that @ J = @ J = 0 at (" " ) = (0 0) (12) 1 2 @ "1 @ "2 where J is de ned by

J =J+ V =

b a

L (x z1 z2 z1 z2) dx
0 0

69

with

L =L+ G
We now calculate the derivatives in (12), afterward setting "1 = "2 = 0. Accordingly,

@ J (0 0) = @ "i

b a

Ly (x y1 y2 y1 y2) i + Ly (x y1 y2 y1 y2)
0 0
0

dx i = 1 2

Integrating the second term by parts (as in the notes) and applying the conditions of (11) gives

@ J (0 0) = @ "i

d L (x y y y y )] dx i = 1 2 Ly (x y1 y2 y1 y2) ; dx 1 2 y 1 2 i
0 0
0

Therefore from (12), and because of the arbitrary character of 1 or 2 the Fundamental Lemma implies d L (x y y y y ) = 0 Ly (x y1 y2 y1 y2) ; dx 1 2 y 1 2 Which is a necessary condition for an extremum.
0 0
0

70

~ be R ~ = x~ 8. Let the two dimensional position vector R i + y~ j, then the velocity vector ~ ~ ~ ~ v=x _i + y _ j . From vector calculus it is known that the triple ~ a b ~ c gives the volume of the parallelepiped whose edges are these three vectors. If one of the vectors is of length unity then the volume is the same as the area of the parallelogram whose edges are the other 2 ~ and ~ vectors. Now lets take ~ a=~ k ~ b=R c=~ v. Computing the triple, we have xy _ ; xy _ which is the integrand in I . The second integral gives the length of the curve from t0 to t1 (see de nition of arc length in any Calculus book). To use the previous problem, let L(t x y x _y _ ) = xy _ ; xy _
q

then

_2 + y _2 G(t x y x _y _) = x

= y _ Ly = ;x _ = 0 Gy = 0 = ;y Ly_ = x _ x _ = p 2 2 Gy_ = p 2y x _ +y _ x _ +y _2 Substituting in the Euler equations, we end up with the two equations: xy _;x _y = 0 y _ 2 ; (x 2 2 _ +y _ )3=2 xy _;x _y = 0 x _ ;2 + (x 2 2 _ +y _ )3=2

Lx Gx Lx_ Gx_

Case 1: y _=0 Substituting this in the second equation, yields x _ = 0. Thus the solution is x = c1 y = c2 Case 2: x _ = 0, then the rst one yields y _ = 0 and we have the same solution. Case 3: x _ 6= 0, and y _ 6= 0 In this case the term in the braces is zero, or 2 (x _2 + y _ 2)3=2 = xy _;x _y

d x _ . The right hand side can be written as y _ dt y _ x _ Now let u = y _ , we get 2 dy du = 2 3 = 2 (1 + u ) For this we use the trigonometric substitution u = tan . This gives the following:
!

71

Simplifying we get

x _ = 2y + c y _ dx =
2
q

_ )2 1 + (x y _

Substitute v = 1 ; ( 2 y + c)2 and we get

!

y+c dy 1 ; ( 2 y + c)2
!2 !2

2 c y + 2 + x + 2k which is the equation of a circle.

= 2

72

9. Let F = F + G = (y )2 + y2. Then Euler`s rst equation gives

0

d (2y ) = 0 2 y ; dx
0

) 2 y ; 2y = 0 ) y ; y=0 ) r;p =0 ) r=
00 00

p; x) + c sin(p; x) p; ) = 0. Since c = 0 would give us the The initial conditions result in c = 0 and c sin ( p trivial solution again, it must be that ; = n where n = 1 2 : : :. This implies that ; = n or eqivalently = ;n n = 1 2 : : :.
y(x) = c1cos(
1 2 2 2 2

Where we are substituting the assumed solution form of y = erx into the di erential equation to get an equation for r. Note that = 0 and > 0 both lead to trivial solutions for y(x) and there would be no way to satisfy the condition that o y2dx = 1. Therefore, assume that < 0. We then have that the solution has the form
R

We now use this solution and the requirement Therefore, we have

Z

y2dx = 1 to solve for the constant c2.

2 c2 2sin (nx)dx =

n
0 2 2

= = = = After solving for the constant we have that

s

n c n 0 c2 2 2 c2 2 2 1 for n = 1 2 : : :
!

c2 2 2 n sin udu u ; sin(2u) 2 4 sin (2 n ; 4 )

y(x) =

2 sin(nx) n = 1 2 : : :
Z
0

If we now plug this solution into the equation (y )2dx we get that I (y) = n2 which implies 0 we should choose n = 1 to minimize I (y). Therefore, our nal solution is
s

y(x) =

2 sin(x)

73

CHAPTER 6

6 Integrals Involving More Than One Independent Variable

Problem
1. Find all minimal surfaces whose equations have the form z = (x) + (y): 2. Derive the Euler equation and obtain the natural boundary conditions of the problem
Z Z h

2 2 (x y)u2 x + (x y )uy ; (x y )u dxdy = 0:

i

In particular, show that if (x y) = (x y) the natural boundary condition takes the form @u u = 0 @n @u is the normal derivative of u. where @n 3. Determine the natural boundary condition for the multiple integral problem

I (u) =

Z Z

L(x y u ux uy )dxdy

u C 2(R) u unspeci ed on the boundary of R

4. Find the Euler equations corresponding to the following functionals 2 2 a. I (u) = (x2u2 x + y uy )dxdy
Z Z

b. I (u) =

Z Z

2 2 (u2 t ; c ux )dxdt c is constant

74

1. z = (x) + (y)

S=
=

R
Z Z q

2 + z 2 dx dy 1 + zx y
0

1+
0

(x) +
!
0

(y) dx dy

@ @x
00

) p1 + (x 2 +
0

+ @@y
0 0

p1 + (2y)+
0 0

!
0

=0 + =0

Di erentiate and multiply by 1 + (x) 1 + (y) 1 +

q q q
0

2 2

+ +

2 2

00

; ;

+ 1+

2 2 2

2 2

q
00

1+

+ +

1=2

q
00

1=2

Expand and collect terms

00

(x) 1 +
00 0

+ (y ) +

00

(y ) 1 +

+ (x)] = 0

Separate the variables (x) (y) = ; 2 2 1 + + (y) 1 + + (x) One possibility is

00 0 00

(x) =

00

(y) = 0

) z = Ax + By + C
00

(x) = Ax + (y) = By + which is a plane

The other possibility is that each side is a constant (left hand side is a function of only x and the right hand side depends only on y) (y) (x) = = ; 2 2 1 + (x) 1 + + (y) Let = (x) then
00 0 0 0

1+ 2 = d = dx 1+ 2 arc tan = x + c1 75

= tan ( x + c1) Integrate again (x) = (x) =

Z

tan ( x + c1) dx cos ( x + c1) + c2

; 1 ln

e(c2

(x))

= cos ( x + c1))

(1)

Similarly for (y) (sign is di erent !) (y) = 1 ln cos( y ; D1) + D2

e(

(y) ; D2 )

= cos ( y

;D)
1

Divide equation (2) by equation (1) y ; D1 ) e ( c2 D2 + (y) + (x)) = cos( cos( x + c1) using z = (x) + (y) we have y ; D1) e ( c2 D2) e z = cos( cos( x + c1) If we let (x0 y0 z0) be on the surface, we nd y ; D1) cos( x0 + c1) e (z z0 ) = cos( cos( x + c1) cos( y0 ; D1)
; ; ; ; ;

76

2 2. F = (x y) u2 x + (x y ) uy

(x y) u2

@ F = 0 (see equation 11) + @y uy Fux = 2 (x y) ux @ F ; Fu + @x u

x

Fu = ; 2 (x y) u @ ( (x y) u ) + @ ( (x y) u ) + (x y) u = 0 ) @x x y @y The natural boundary conditions come from the boundary integral Fux cos + Fuy sin = 0
( (x y) ux cos + (x y) uy sin ) = 0 If (x y) = (x y) then (x y) (ux cos + uy sin ) = 0 ru ~ n @u = @n @u = 0 ) @n
| {z {z } | }

Fuy = 2 (x y) uy

77

3. Determine the natural boundary condition for the muliple integral problem I (u) = R L(x y u ux uy )dxdy u 2 C 2(R) u unspeci ed on the boundary of R. Let u(x y) be a minimizing function (among the admissible functions) for I (u). Consider the one-parameter family of functions u(") = u(x y) + " (x y) where 2 C 2 over R and (x y) = 0 on the boundary of R. Then if I (") = R L(x y u + " ux + " x uy + " y )dxdy a necessary condition for a minimum is I (0) = 0: Now, I (0) = ( Lu + xLux + y Luy )dxdy, where the arguments in the partial derivaR tives of L are the elements (x y u ux uy ) of the minimizing function u: Thus,
Z Z Z Z
0

Z Z

I (0) =
0

Z Z

The second integral in this equation is equal to (by Green's Theorem) (`Lux + mLuy )ds @R where ` and m are the direction cosines of the outward normal to @R and ds is the arc length of the @R . But, since (x y) = 0 on @R this integral vanishes. Thus, the condition I (0) = 0 which holds for all admissible (x y) reduces to @ L ; @ L )dxdy = 0: (Lu ; @x ux @y uy R @ L ; @ L = 0 at all points of R. This is the Euler-Lagrange equation Therefore, Lu ; @x ux @y uy (11) for the two dimensional problem. Now consider the problem
I
0

@ L ; @ L )dxdy + (Lu ; @x ux @y uy

Z Z

@ ( L ) + @ ( L ))dxdy: ( @x ux @y uy

Z Z

L(x y u ux uy )dxdy = I (u) = L(x y u ux uy )dxdy R c a where all or or a portion of the @R is unspeci ed. This condition is analogous to the single integral variable endpoint problem discussed previously. Recall the line integral presented above: (`Lux + mLuy )ds where ` and m are the direction cosines of the outward normal to @R @R and ds is the arc length of the @R . Recall that in the case where u is given on @R (analogous to xed endpoint) this integral vanishes since (x y) = 0 on @R. However, in the case where on all or a portion of @R u is unspeci ed, (x y) 6= 0. Therefore, the natural boundary condition which must hold on @R is `Lux + mLuy = 0 where ` and m are the direction cosines of the outward normal to @R.
I

Z Z

dZ b

78

4. Euler's equation

@F + @F @x ux @y uy

; Fu = 0

2 2 a. F = x2u2 x + y uy Di erentiate and substitute in Euler's equation, we have

2 2 b. F = u2 t ; c ux Di erentiate and substitute in Euler's equation, we have

2xux + x2uxx + 2yuy + y2uyy = 0

which is the wave equation.

utt ; c2uxx = 0

79

CHAPTER 7

7 Examples of Numerical Techniques

Problems
1. Find the minimal arc y(x) that solves, minimize I = y2 ; (y )2 dx 0 a. Using the indirect ( xed end point) method when x1 = 1: b. Using the indirect (variable end point) method with y(0)=1 and y(x1) = Y1 = x2 ; 4 :
Z

x1

2. Find the minimal arc y(x) that solves, minimize I = where y(0) = 1 and y(1) = 2:
Z

1 (y )2 + yy + y + y dx 2
0 0 0

3. Solve the problem, minimze I = y2 ; yy + (y )2 dx 0 a. Using the indirect ( xed end point) method when x1 = 1: b. Using the indirect (variable end point) method with y(0)=1 and y(x1) = Y1 = x2 ; 1:
x1
h
0 0

4. Solve for the minimal arc y(x) :

I= where y(0) = 0 and y(1) = 1:

1 0

y2 + 2xy + 2y dx
0

80

1. a. Here is the Matlab function de ning all the derivatives required

% odef.m function xdot=odef(t,x) % fy1fy1 - fy'y' (2nd partial wrt y' y') % fy1y - fy'y (2nd partial wrt y' y) % fy - fy (1st partial wrt y) % fy1x - fy'x (2nd partial wrt y' x) fy1y1 = -2 fy1y = 0 fy = 2*x(1) fy1x = 0 rhs2= -fy1y/fy1y1,(fy-fy1x)/fy1y1] xdot= x(2),rhs2(1)*x(2)+rhs2(2)]'

The graph of the solution is given in the following gure

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure 4:

81

2. First we give the modi ed nput.m

% function VALUE = FINPUT(x,y,yprime,num) returns the value of the % functions F(x,y,y'), Fy(x,y,y'), Fy'(x,y,y') for a given num. % num defines which function you want to evaluate: % 1 for F, 2 for Fy, 3 for Fy'. if nargin < 4, error('Four arguments are required'), break, end if (num < 1) | (num > 3) error('num must be between 1 and 3'), break end if num == 1, value = .5*yp^2+yp*y+yp+y if num == 2, value = yp+1 end if num == 3, value = yp+y+1 end end % F % Fy % Fy'

The boundary conditions are given in the main program dmethod.m (see lecture notes). The graph of the solution (using direct method) follows
Solution y(x) using the direct method 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0

0.1

0.2

0.3

0.4

0.5 x

0.6

0.7

0.8

0.9

Figure 5:

82

3. a. Here is the Matlab function de ning all the derivatives required

% odef.m function xdot=odef(t,x) % fy1fy1 - fy'y' (2nd partial wrt y' y') % fy1y - fy'y (2nd partial wrt y' y) % fy - fy (1st partial wrt y) % fy1x - fy'x (2nd partial wrt y' x) fy1y1 = 2 fy1y = -1 fy = 2*x(1)-x(2) fy1x = 0 rhs2= -fy1y/fy1y1,(fy-fy1x)/fy1y1] xdot= x(2),rhs2(1)*x(2)+rhs2(2)]'

The graph of the solution is given in the following gure

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure 6:

83

4. First we give the modi ed nput.m

% function VALUE = FINPUT(x,y,yprime,num) returns the value of the % functions F(x,y,y'), Fy(x,y,y'), Fy'(x,y,y') for a given num. % num defines which function you want to evaluate: % 1 for F, 2 for Fy, 3 for Fy'. if nargin < 4, error('Four arguments are required'), break, end if (num < 1) | (num > 3) error('num must be between 1 and 3'), break end if num == 1, value = y^2+2*x*y+2*yp if num == 2, value = 2*y+2*x end if num == 3, value = 2 end end % F % Fy % Fy'

The boundary conditions are given in the main program dmethod.m (see lecture notes). The graph of the solution (using direct method) follows
Solution y(x) using the direct method 1

0.8

0.6

0.4

0.2

-0.2 0

0.1

0.2

0.3

0.4

0.5 x

0.6

0.7

0.8

0.9

Figure 7:

84

CHAPTER 8

8 The Rayleigh-Ritz Method

Problems
1. Write a MAPLE program for the Rayleigh-Ritz approximation to minimize the integral
Z

I=

1h 0

(y )2
0

; y ; 2xy
2

dx

y(0) = 1 y(1) = 2:

Plot the graph of y0 y1 y2 and the exact solution. 2. Solve the same problem using nite di erences.

85

1.
with(plots): phi0:= 1+x: y0 :=phi0: p0:=plot(y0,x=0..1,color=yellow,style=point): phi0:= 1+x:phi1:= a1*x*(1-x): y1 :=phi0 + phi1: dy1 :=diff(y1,x): f := (dy1^2 - y1^2 - 2*x*y1): w := int(f,x=0..1): dw := diff(w,a1): a1:= fsolve(dw=0,a1): p1:=plot(y1,x=0..1,color=green,style=point): phi0:= 1+x:phi1:= b1*x*(1-x):phi2 := b2*x*x*(1-x): y2 :=phi0 + phi1 + phi2: dy2 :=diff(y2,x): f := (dy2^2 - y2^2 - 2*x*y2): w := int(f,x=0..1): dw1 := diff(w,b1): c_1:=solve(dw1=0,b1): dw2 := diff(w,b2): c_2:=solve(dw2=0,b1): b3:= c_1-c_2: b2:=solve(b3=0,b2): b1:=c_1: p2:=plot(y2,x=0..1,color=cyan,style=point): phi0:= 1+x: phi1:= c1*x*(1-x): phi2 := c2*x*x*(1-x): phi3 := c3*x*x*x*(1-x): y3 :=phi0 + phi1 + phi2 + phi3: dy3 :=diff(y3,x): f := (dy3^2 - y3^2 - 2*x*y3): w := int(f,x=0..1): dw1 := diff(w,c1): c_1:=solve(dw1=0,c1): dw2 := diff(w,c2): c_2:=solve(dw2=0,c1): dw3 := diff(w,c3): c_3:=solve(dw3=0,c1): a1:= c_1 - c_2: a_1:=solve(a1=0,c2): a2:= c_3 - c_2:

86

a_2:=solve(a2=0,c2): b1:= a_1 - a_2: c3:=solve(b1=0,c3): c2:=a_1: c1:=c_1: p3:=plot(y3,x=0..1,color=blue,style=point): y:= cos(x) +((3-cos(1))/sin(1))*sin(x) - x: p:=plot(y,x=0..1,color=red,style=line): display({p,p0,p1,p2,p3})

Note: Delete p2 or p3 (or both) if you want to make the True versus Approximations more noticable.
2

1.8

1.6

1.4

1.2

1 0 0.2 0.4 x 0.6 0.8 1

Figure 8:

87

2.
F =dy^2-y^2-2*y*x with(plots): f = (((y i+1]-y i])/delx)^2 - y i]^2 - 2*x i]*y i]) phi1 :=sum((((y1 i+1]-y1 i])/delx1)^2 - y1 i]^2 - 2*x1 i]*y1 i])*delx1,'i'=0..1): dy1 0] := diff(phi1,y1 0]): dy1 1] := diff(phi1,y1 1]): dy1 2] := diff(phi1,y1 2]): x1 0]:=0: x1 1]:=.5: x1 2]:=1: delx1 := 1/2: y1 0] := 1: y1 2]:=2: y1 1]:=solve(dy1 1]=0,y1 1]): p1:=array(1..6, x1 0],y1 0],x1 1],y1 1],x1 2],y1 2]]): p1:=plot(p1): phi2 :=sum((((y2 i+1]-y2 i])/delx2)^2 - y2 i]^2 - 2*x2 i]*y2 i])*delx2,'i'=0..2): dy2 0] := diff(phi2,y2 0]): dy2 1] := diff(phi2,y2 1]): dy2 2] := diff(phi2,y2 2]): dy2 3] := diff(phi2,y2 3]): x2 0]:=0: x2 1]:=1/3: x2 2]:=2/3: x2 3]:=1: delx2 := 1/3: y2 0] := 1: y2 3]:=2: d2 2]:=solve(dy2 2]=0,y2 2]): d2 1]:=solve(dy2 1]=0,y2 2]): d2 3] :=d2 2]-d2 1]: y2 1]:= solve(d2 3]=0,y2 1]): y2 2]:=d2 2]: p2:=array(1..8, x2 0],y2 0],x2 1],y2 1],x2 2],y2 2],x2 3],y2 3]]): p2:=plot(p2): phi3 :=sum((((y3 i+1]-y3 i])/delx3)^2 - y3 i]^2 - 2*x3 i]*y3 i])*delx3,'i'=0..3): dy3 0] := diff(phi3,y3 0]): dy3 1] := diff(phi3,y3 1]): dy3 2] := diff(phi3,y3 2]):dy3 3] := diff(phi3,y3 3]):

88

dy3 4] := diff(phi3,y3 4]): x3 0]:=0: x3 1]:=1/4: x3 2]:=1/2: x3 3]:=3/4: x3 4]:=1: delx3 := 1/4: y3 0] := 1: y3 4]:=2: d3 1]:=solve(dy3 1]=0,y3 2]): d3 2]:=solve(dy3 2]=0,y3 2]): d3 3]:=solve(dy3 3]=0,y3 2]): d3 1] :=d3 2]-d3 1]:d3 3] :=d3 2]-d3 3]: d3 1]:=solve(d3 1]=0,y3 3]): d3 3]:=solve(d3 3]=0,y3 3]): d3 1]:= d3 1]-d3 3]: y3 1]:= solve(d3 1]=0,y3 1]): y3 3]:=d3 3]: y3 2]:=d3 2]: p3:=array(1..10, x3 0],y3 0],x3 1],y3 1],x3 2],y3 2],x3 3],y3 3],x3 4],y3 4]]): p3:=plot(p3): phi4 :=sum((((y4 i+1]-y4 i])/delx4)^2 - y4 i]^2 - 2*x4 i]*y4 i])*delx4,'i'=0..4): dy4 0] := diff(phi4,y4 0]): dy4 1] := diff(phi4,y4 1]): dy4 2] := diff(phi4,y4 2]): dy4 3] := diff(phi4,y4 3]): dy4 4] := diff(phi4,y4 4]): dy4 5] := diff(phi4,y4 5]): x4 0]:=0: x4 1]:=1/5: x4 2]:=2/5: x4 3]:=3/5: x4 4]:=4/5: x4 5]:=1: delx4 := 1/5: y4 0] := 1: y4 5]:=2: d4 1]:=solve(dy4 1]=0,y4 2]): d4 2]:=solve(dy4 2]=0,y4 3]): d4 3]:=solve(dy4 3]=0,y4 4]): d4 4]:=solve(dy4 4]=0,y4 4]): d4 3]:= d4 3]-d4 4]: d4 3]:=solve(d4 3]=0,y4 3]):

89

d4 2]:=d4 2]-d4 3]: d4 2]:=solve(d4 2]=0,y4 2]): d4 1]:=d4 1]-d4 2]: y4 1]:=solve(d4 1]=0,y4 1]): y4 2]:=d4 2]: y4 3]:=d4 3]: y4 4]:=d4 4]: p4:=array(1..12, x4 0],y4 0],x4 1],y4 1],x4 2],y4 2],x4 3],y4 3],x4 4],y4 4], x4 5],y4 5]]): p4:=plot(p4): y:= cos(x) +((3-cos(1))/sin(1))*sin(x) - x: p:=plot(y,x=0..1,color=red,style=line): display({p,p1,p2,p3,p4})

1.8

1.6

1.4

1.2

1 0 0.2 0.4 0.6 0.8 1

Figure 9:

90

CHAPTER 9

9 Hamilton's Principle
Problems
1. If ` is not preassigned, show that the stationary functions corresponding to the problem
Z

y 2 dx = 0
0

y(0) = 2 y(`) = sin ` are of the form y = 2 + 2x cos `, where ` satis es the transcendental equation
Also verify that the smallest positive value of ` is between 2 and 34 : 2. If ` is not preassigned, show that the stationary functions corresponding to the problem
Z

subject to

2 + 2` cos ` ; sin ` = 0:

1 0

y 2 + 4(y ; `) dx = 0
0

y(0) = 2 y(`) = `2 are of the form y = x2 ; 2 x + 2 where ` is one of the two real roots of the quartic equation ` 2`4 ; `3 ; 1 = 0:
3. A particle of mass m is falling vertically, under the action of gravity. If y is distance measured downward and no resistive forces are present. a. Show that the Lagrangian function is 1y L=T ;V =m 2 _ 2 + gy + constant and verify that the Euler equation of the problem
Z

subject to

t2

t1

L dt = 0

is the proper equation of motion of the particle. b. Use the momentum p = my _ to write the Hamiltonian of the system. c. Show that 91

@H= =y _ @p @ H = ;p _ @y
4. A particle of mass m is moving vertically, under the action of gravity and a resistive force numerically equal to k times the displacement y from an equilibrium position. Show that the equation of Hamilton's principle is of the form t2 1 2 1 ky2 dt = 0 m y _ + mgy ; 2 t1 2 and obtain the Euler equation.
Z

5. A particle of mass m is moving vertically, under the action of gravity and a resistive force numerically equal to c times its velocity y _ . Show that the equation of Hamilton's principle is of the form t2 t2 1 2 m y _ + mgy dt ; cy _ y dt = 0: t1 t1 2
Z Z

6. Three masses are connected in series to a xed support, by linear springs. Assuming that only the spring forces are present, show that the Lagrangian function of the system is 1 mx 2 2 2 2 L=2 _2 _2 + constant 1 _ 1 + m2 x 2 + m3x 3 ; k1 x1 ; k2 (x2 ; x1) ; k3 (x3 ; x2) where the xi represent displacements from equilibrium and ki are the spring constants.
h i

92

1. If ` is not preassigned, show that the stationary functions corresponding to the problem
Z

( y )2dx = 0
0

Subject to y(0) = 2 and y(`) = sin` Are equal to,

d Ly = 0 with Using the Euler equation Ly ; dt
0

y = 2 + 2x cos `

L = (y )2 Ly = 0 Ly = 2y
0
0

We get the 2nd order ODE

;2y
Integrating twice, we have

= 0 y = 0
00 00

y = Ax + B Using our initial conditions to solve for for A and B, y(0) = 2 = A(0) + B =) B = 2 y(`) = sin ` = A` + 2 =) A = sin `` ; 2
!

Substituting A and B into our original equation gives, y = sin `` ; 2 x + 2 Now, because we have a variable right hand end point, we must satisfy the following transversality condition:

F +(

; y )Fy jx ` = 0
0
0

93

Where,

Therefore,

F = (y )2 `) ; 4 Fy = 2 sin(` = cos ` y = sin(``) ; 2

0
0

y (`)] + cos(`) ; sin(``) ; 2 2y y (`)]2 + cos(`) ; sin(``) ; 2 2 sin(``) ; 2 sin(`) ; 2 2 + cos(`) ; sin(`) ; 2 2 sin(`) ; 2 ` ` ` sin(`) ; 2 + 2 cos(`) ; sin(`) ; 2 ` ` sin(`) ; 2 + 2` cos(`) ; 2 sin(`) + 4 2 + 2` cos(`) ; sin(`)
!
0

= 0 = 0 = 0 = 0 = 0 = 0

Which is our transversality condition. Since ` satis es the transcendental equation above, sin ` ; 2 = 2 cos ` ` Substituting this back into the equation for y yields, we have,

y = 2 + 2x cos `
Which is what we wanted to show. To verify that the smallest positive value of ` is between 2 and 34 , we must rst solve the transcendental equation for `. 2 + 2` cos ` ; sin ` = 0 sin ` ; 2 2` = cos ` cos ` 1 ` = 2 tan ` ; sec ` 94

20 15 10 5 0

y
5 10 15 20 pi/2 25 0 0.5 1 1.5 l 2 2.5 3

Figure 10: Plot of y = ` and y = 1 2 tan(`) ; sec(`) Then plot the curves,

y = ` y = 1 2 tan ` ; sec ` between 0 and Pi, to see where they intersect. 1 Since they appear to intersect at approximately 2 , lets verify the limits of y = 2 tan ` ; sec ` analytically.
tan ` ; sec ` lim 1 2 2 1 sin 2 ; 1 = 2 cos 2 cos 2 1 sin 2 ; 2 = 2 cos 2 1 = ; 0 = 1
l
; !

Which agrees with the plot . Therefore, 2 is the smallest value of `

95

2. subject to
0

1h 0

(y )2 + 4(y ; `) dx = 0
0

Since L = (y )2 + 4(y ; `) we have Ly = 4 and Ly = 2y d L = 0 becomes d 2y = 4 Thus Euler's equation: Ly ; dx y dx Integrating leads to y = 2x + c21 Integrating again y = x2 + c21 x + c2 Now use the left end condition: y(0) = 2 = 0 + 0 + c2 At x = ` we have: y(`) = `2 = `2 + c21 ` + 2 =) c1 = ; 4 ` 2 2 Thus the solution is: y = x ; ` x + 2
0 0 0 0 0 0 0 0 0

y(0) = 2 y(`) = `2

Let's di erentiate y for the transversality condition: y = 2x ; 2 ` Now we apply the transversality condition L + ( ; y )Ly = 0 where = `2 and = 2`
0

Now substituting for , L, Ly , y and y and evaluating at x = `, we obtain 2 ` + 2 ; `) + (2` ; (2` ; 2 ))2(2` ; 2 ) = 0 2 2 (2` ; 2 ) + 4( ` ; ` ` ` ` 2) = 0 (2 ` ; 4`2 ; 8 + `42 + 4(`2 ; `) + 4 ` ` 4 8 4`2 ; 8 + `2 + 4`2 ; 4` + 8 ; `2 = 0 8`2 ; 4` ; `42 = 0 2`4 ; `3 ; 1 = 0 Therefore the nal solution is y = x2 ; 2 `x + 2 where ` is one of the two real roots of 2`4 ; `3 ; 1 = 0.
0

x=`

96

3. First, using Newton's Second Law of Motion, a particle with mass m with position vector y is acted on by a force of gravity. Summing the forces gives

my ; F = 0

Taking the downward direction of y to be positive, F = mgy: Thus

my + mgy = 0
1 From Eqn (9) and the de nition of T = 2 my _ 2 we obtain
Z

t2

t1

( T + F dy) dt = 0

From Eqn (10),

Z

t2

t1

(my _ y + F y) dt = 0

De ning the potential energy as gives

F y = ; V = mgy y
Z

t2 t1

(T ; V ) dt = 0

or
Z

t2 t1

If we de ne the Lagrangian L as L T ; V , we obtain the result _ 2 + gy) + constant L = m( 1 2y Note: The constant is arbitrary and dependent on the initial conditions. To show the Euler Equation holds, recall L = m( 1 _ 2 + gy) + constant 2y

2 m y _ (1 2 ; mgy) dt = 0

Ly = mg
Thus,

Ly = my _
0

d L = my dt y
0

97

Since the particle falls under gravity (no initial velocity), y = g and dL =0 Ly ; dt y The Euler Equation holds. b. Let p = my _. The Hamiltonian of the system is
0

d L = mg ; my = m(g ; y) Ly ; dt y
0

H (t x p) = ;L(t x (t x p)) + p (t x p)

_ 2 + gy) + constant + my (t x p) = ; m( 1 2y

@H= c. @p

@ H =y _ (by de nition) @p @ H = ;mg = ;my = ;p _ @y

98

4. Newton's second law: mR ; F = 0 Note that F = mg ; kR, so we have

Z

t2

t1

mR R ; mg R + kR R dt = 0
1 mR _ 2 + mgR ; 1 kR2 dt = 0 2 2

This can also be written as

Z

t2 t1

To obtain Euler's equation, we let _ 2 + mgR ; 1 kR2 L= 1 m R 2 2 Therefore LR = mg ; kR _ LR _ = mR d L = mg ; kR ; mR = 0 LR ; dt _ R 5. The rst two terms are as before (coming from ma and the gravity). The second integral gives the resistive force contribution which is proportional to y _ with a constant of proportionality c. Note that the same is negative because it acts opposite to other forces. 6. Here we notice that the rst spring moves a distance of x1 relative to rest. The second spring in the series moves a distance x2 relative to its original position, but x1 was the contribution of the rst spring therefore, the total is x2 ; x1. Similarly, the third moves x3 ; x2 units.

99

CHAPTER 10

10 Degrees of Freedom - Generalized Coordinates

Problems
1. Consider the functional

I (y) =
2. Give Hamilton's equations for

bh a

r(t)y _ 2 + q(t)y2 dt:

Find the Hamiltonian and write the canonical equations for the problem.

I (y) =

bq a

(t2 + y2)(1 + y _ 2)dt:

Solve these equations and plot the solution curves in the yp plane. 3. A particle of unit mass moves along the y axis under the in uence of a potential f (y) = ;!2y + ay2 where ! and a are positive constants. a. What is the potential energy V (y)? Determine the Lagrangian and write down the equations of motion. b. Find the Hamiltonian H (y p) and show it coincides with the total energy. Write down Hamilton's equations. Is energy conserved? Is momentum conserved? 2 c. If the total energy E is ! 10 , and y(0) = 0, what is the initial velocity? d. Sketch the possible phase trajectories in phase space when the total energy in the !6 : system is given by E = 12 a2 p Hint: Note that p = 2 E ; V (y): What is the value of E above which oscillatory solution is not possible?
q

4. A particle of mass m moves in one dimension under the in uence of the force F (y t) = ky 2et where y(t) is the position at time t, and k is a constant. Formulate Hamilton's principle for this system, and derive the equations of motion. Determine the Hamiltonian and compare it with the total energy. 5. A Lagrangian has the form a2 (y )4 + a(y )2G(y) ; G(y)2 L(x y y ) = 12
; 0 0 0

100

where G is a given di erentaible function. Find Euler's equation and a rst integral. 6. If the Lagrangian L does not depend explicitly on time t, prove that H = constant, and if L doesn't depend explicitly on a generalized coordinate y, prove that p = constant: 7. Consider the di erential equations

r2 _ = C

governing the motion of a mass in an inversely square central force eld. a. Show by the chain rule that
2 dr 2 4d r r_ = Cr d r = C r d 2 ; 2C 2r and therefore the di erential equations may be written
;

kr 2 = 0 r ; r _2 + m
;

dr d

!2

d2r ; 2r d2
b. Let r = u 1 and show that
;

dr d

!2

; r + Ckm r
2

=0

d2u + u = k : d2 C 2m

c. Solve the di erential equation in part b to obtain (1 + cos( ; 0)) u = r 1 = Ck 2m where and 0 are constants of integration. d. Show that elliptical orbits are obtained when < 1:
;

101

CHAPTER 11

11 Integrals Involving Higher Derivatives

Problems
1. Derive the Euler equation of the problem
Z

x2 x1

F (x y y y ) dx = 0
0 00

d2 @F ; d @F + @F = 0 dx2 @y dx @y @y and show that the associated natural boundary conditions are d @F ; @F y x2 = 0 dx @y @y x1 and @F y x2 = 0: @y
! !
00 0

in the form

"

00

"

00

x1

2. Derive the Euler equation of the problem

Z

x2 x1

y2

y1

F (x y u ux uy uxx uxy uyy ) dxdy = 0

! ! ! !

where x1 x2 y1 and y2 are constants, in the form @ 2 @F + @ 2 @F + @ 2 @F ; @ @F ; @ @F + @F = 0 @x2 @uxx @x@y @uxy @y2 @uyy @x @ux @y @uy @u and show that the associated natural boundary conditions are then @ @F + @ @F ; @F u x2 = 0 @x @uxx @y @uxy @ux x1
! " ! # "

and

@F u @uxx x

x2 x1

=0

"

@ @F + @ @F @y @uyy @x @uxy @F u @uyy y

"

;
y2 y1

@F @uy

y2 y1

=0

= 0:

102

3. Specialize the results of problem 2 in the case of the problem x2 y2 1 2 1 u2 + u u + (1 ; )u2 dxdy = 0 u + xx yy xx xy 2 yy x1 y1 2 where is a constant. Hint: Show that the Euler equation is r4u = 0 regardless of the value of , but the natural boundary conditions depend on :
Z Z

4. Specialize the results of problem 1 in the case

00

F = a(x)(y )2 ; b(x)(y )2 + c(x)y2:

0

5. Find the extremals 1 a. I (y) = 0 (yy + (y )2)dx

Z
0 00

y(0) = 0 y (0) = 1 y(1) = 2 y (1) = 4

0 0 00 0

b. I (y) =

(y2 + (y )2 + (y + y )2)dx
0

y(0) = 1 y (0) = 2 y(1) = 0 y (1) = 0:

0 0

6. Find the extremals for the functional

I (y) =

(y2 + 2y _ 2 + y2)dt:

7. Solve the following variational problem by nding extremals satisfying the given conditions

I (y) =

(1 + (y )2)dx
00

y(0) = 0 y (0) = 1 y(1) = 1 y (1) = 1:

0 0

103