Solutions to exercises
(c) x3/2 /[1 exp(x)] = x3/2 /[1 (1 x + x2 /2 + )] = x1/2 /[1 + O(x)] = O(x1/2 ).
= a cos2 + b sin cos + c sin2 < |a| + |b| + |c| = O(1),
f
and = O(f ). If y = kx with 2kc = b b2 4ac then = 0.
397
398 CHAPTER 15. SOLUTIONS TO EXERCISES
(c) For x > a, y(x) is strictly decreasing and for x > 2a, y < 2a3 . Set x = 2a cosh
and the equation becomes
y = 6a3 cosh 2a3 (3 cosh + cosh 3) = 2a3 cosh 3
giving
1 y
x(y) = 2a cosh cosh1 3 , y < 2a3 .
3 2a
15.1. SOLUTIONS FOR CHAPTER 1 399
dy dy (a b) sin 2x
y 2 = a sin2 x+b cos2 x to give 2y = 2(ab) sin x cos x or = p
dx dx 2 a sin2 x + b cos2 x
which can also be expressed in the form
dy (a b) sin 2x
= p .
dx 2(a + b) + 2(b a) cos 2x
d
cos x3 cos x = 3x2 sin x3 cos x cos x3 sin x.
dx
dy du
(d) If y = xx = ex ln x , putting u = x ln x the chain rule gives = eu = (1 + ln x)xx .
dx dx
d df dg
1= f (g(y)) = = f (g)g (y).
dy dg dy
dy dx
Since = f (x) and = g (y), the result follows.
dx dy
(b) Differentiate again with respect to y
1 1 2 3
d2 x d2 y d2 y
d dy d dy dx dy dx dy
= = = 2 = 2 .
dy 2 dy dx dx dx dy dx dx dy dx dx
400 CHAPTER 15. SOLUTIONS TO EXERCISES
h = f g + f g = h = (f g + f g ) + (f g + f g ).
(3) 2 2 2
The expression for h follows similarly. Since = = 1 and = 2 the
0 2 1
general result quoted is therefore true for n = 1 and 2. Suppose it to be true for n; a
further differentiation gives
n
(n+1)
X n
h = f (nk+1) g (k) + f (nk) g (k+1)
k
k=0
n n+1
X n
X n
= f (nk+1) g (k) + f (n+1s) g (s) (with s = k + 1 in second sum)
k s1
k=0 s=1
n
n + 1 (n+1) (0) n (0) (n+1)
X n n
= f g + f g + + f (nk+1) g (k) .
0 n k k1
k=1
m m
But, for all m, = = 1 and
0 m
n n n! n! n! (n + 1)! k
+ = + = +
k k1 k! (n k)! (k 1)! (n + 1 k)! k! (n k)! k! (n + 1 k)! n + 1
(n + 1)! n+1k (n + 1)! k n+1
= + = .
k! (n + 1 k)! n + 1 k! (n + 1 k)! n + 1 k
15.1. SOLUTIONS FOR CHAPTER 1 401
Thus, if the formula is true for n, it is true for n + 1: it is true for n = 2 and hence is
true for all n.
which is valid provided none of the fk (x) are zero, that is p(x) 6= 0.
Now differentiate this expression using the rule just obtained for second-order determi-
nants; then recombine the 9 terms into a third-order determinant, to obtain
a b c a b c a b c
D (x) = d e f + d e f + d e f .
g h i g h i g h i
where we have used the mean value theorem, equation 1.11 (page 22), to write
so that
F (x + h) F (x)
= f g(x) + hg g (x + h).
h
This gives the required result on taking the limit h 0.
r r
2r = 2x and 2r = 2y,
x y
p
hence the result. Alternatively, put r = x2 + y 2 to obtain
r x x r y y
= p = and = p = .
x x2 + y 2 r y x2 + y 2 r
2 2 2x 2 x2 2
2
= = + 4 2
=4 .
x y y x y y y y
15.1. SOLUTIONS FOR CHAPTER 1 403
Alternatively,
d f d dx dy
= (x 2ty) = 2y 2t .
dt y dt dt dt
Solution
for Exercise 1.24
If F = 1 + x1 x2 then the chain rule gives
dF F F x1 x + x1 x2
= x1 + x2 = 2 .
dt x1 x2 2 1 + x1 x2
dF 1 du
Alternatively, set u = x1 x2 , so = , which is a simpler method of deriving
dt 2 1 + u dt
the same result.
Differentiate this expression with respect to x1 , using the product rule,
dF 1 1
= (x1 x2 + x1 x2 ) + (x1 x2 + x1 x2 )
x1 dt x1 2 1 + x1 x2 2 1 + x1 x2 x1
1 x2 x2
= (x1 x2 + x1 x2 ) + .
4 (1 + x1 x2 )3/2 2 1 + x1 x2
F x2
Also = , and the chain rule gives
x1 2 1 + x1 x2
x
d F x2 d
= 2 3/2
(x1 x2 ),
dt x1 2 1 + x1 x2 4(1 + x1 x2 ) dt
as before.
404 CHAPTER 15. SOLUTIONS TO EXERCISES
d f du f dv
f (x, y) = + where u = x and v = y.
d u d v d
Now substitute f (x, y) = p f (x, y) into the left-hand side of give
d
f (x, y) = pp1 f (x, y)
d
and set = 1 to obtain the result.
and set = 1.
p fxk (x) = f (x) = f (x) = fxk (x)
xk (xk )
2x 2 y 2(x + y)
fx = + = 2 ,
x2 + y 2 (1 + y 2 /x2 ) x2 x + y2
2y 2 1 2(y x)
fy = = 2 .
x2 + y 2 x (1 + y 2 /x2 ) x + y2
dy fx x+y
Hence = = .
dx fy xy
But eix = cos x + i sin x, so equating real and imaginary parts gives the quoted series.
f (k) (x) = a(a 1)(a 2) (a k + 1)(1 + x)ak for all k provided a is not an integer.
406 CHAPTER 15. SOLUTIONS TO EXERCISES
and f (0) = f (0) = 0 (as expected) and f (0) = 1 and f (3) (0) = 2 giving the required
Taylor series.
(a) For the first part use the solution of exercise 1.32, with a = 1 so a(a 1) (a
k + 1) = (1)k k!, giving the quoted series. Then
x Z x
1
Z
dt 1 t + t2 + + (1)n1 tn1 +
ln(1 + x) = dt=
0 1+t 0
x2 x3 (1)n xn
= x + + + .
2 3 n
(b) The series for (1 + t)1 is valid for |t| < 1, so for |x| < 1 the integral and sum may
be interchanged.
z2 z3 z4 x2
ln(1 + z) = z + + O(z 5 ) and sinh x = x 1 + + O(x5 )
2 3 4 6
15.1. SOLUTIONS FOR CHAPTER 1 407
to give
2
x2 x2 x2 x3 x4
ln(1 + sinh x) = x 1+ + 1+ + + + O(x5 )
6 2 6 3 4
x3
2
x4 x3 x4
x
= x+ + + + O(x5 )
6 2 6 3 4
x2 x3 5x4
= x + + O(x5 ).
2 2 12
where all derivatives are evaluated at (a, b, c). For the second-order and third-order
terms we use the identities
( + + )2 = 2 + 2 + 2 + 2 + 2 + 2,
( + + )3 = 3 + 3 + 3 + 3 2 + 3 2 + 32 + 3 2 + 32 + 3 2 + 6,
(b)
3x 3x ex ln 3 ex ln 3 ln 3 ex ln 3 + ex ln 3 ln 3
(c) lim = lim = lim = .
x0 2x 2x x0 ex ln 2 ex ln 2 x0 ln 2 ex ln 2 + ex ln 2 ln 2
(b) Put F (x) = 1/f (x) and G(x) = 1/g(x) so F (a) = G(a) = 0, and
2
g (x) f (x)2 g (x)
f (x) G(x) f (x)
lim = lim = lim = lim lim .
xa g(x) xa F (x) xa f (x) g(x)2 xa f (x) xa g(x)
f (x) f (x)
Hence, provided all limits exist, lim = lim .
xa g(x) xa g (x)
15.1. SOLUTIONS FOR CHAPTER 1 409
(b) Split the integral in the same manner as in part (a), but since f (u) = f (u) the
two integrals are equal.
x sin x
Z Z Z Z
(b) dx 2x
= x tan x dx tan x but dx tan x = dx = ln | cos x| .
cos cos x
x
Z
Hence dx = x tan x + ln | cos x|.
cos2 x
1 1 1 1 1
Z Z
(c) dx x ln x = x2 ln x dx x2 = x2 ln x x2 .
2 2 x 2 4
Z Z
(d) dx x sin x = x cos x + dx cos x = sin x x cos x.
(b)
/4 /4 /4
x 1 2
Z Z
dx x tan2 x = dx x = x tan x + ln cos x x
0 0 cos2 x 2 0
1 2
= ln 2 .
4 2 32
(c)
1 1
1 1 x3
1 3 1
Z Z
dx x2 sin1 x = x sin x dx .
0 3 0 3 0 1 x2
410 CHAPTER 15. SOLUTIONS TO EXERCISES
But on putting x = sin and using the identity sin 3 = 3 sin 4 sin3 ,
1 /2 /2
x3 1 2
Z Z Z
3
dx = d sin = d (3 sin sin 3) =
0 1 x2 0 4 0 3
1
2
Z
and hence dx x2 sin1 x = .
0 6 9
aI1 = xeax I0 , aI2 = x2 eax 2I1 , aI3 = x3 eax 3I2 , , aIn = xn eax nIn1 .
Multiply the kth equation by Ak and add all the equations to obtain
n
X n
X n
X
a Ak Ik = eax Ak xk kAk Ik1 .
k=1 k=1 k=1
Now chose the Ak such that An = 1/a and for k = 1, 2, , n 1, the Ik cancel, that is
1
aAk = (k + 1)Ak+1 , k = 1, 2, , n 1, An = .
a
n! (1)nk
The solution of these equations is Ak = which gives the quoted expression.
ank+1 k!
Solution for Exercise 1.47
Z a Z 0 Z a
(a) dx f (x) = du f (a u) = dx f (a x).
0 a 0
From the first of these equations we see that y (a) = 0, so the initial conditions are
satisfied. The second equation gives y (a) = f (a), which is consistent with the original
differential equation.
we have
a(u+h)
F (u + h) F (u) 1 a(u + h) a(u)
Z
= dx f (x) = f (), where a(u), a(u + h) ,
h h a(u) h
412 CHAPTER 15. SOLUTIONS TO EXERCISES
the last result being obtained from the integral form of the Mean Value Theorem.
Taking the limit h 0 gives F (u) = a (u)f (a(u)). The same result can be derived
using the Fundamental theorem of Calculus and the chain rule.
(b) We have
b
F (u + h) F (u) f (x, u + h) f (x, u)
Z
= dx
h a h
Z b
f
Assuming that the limit h 0 exists we obtain F (u) = dx .
a u
and hence
X
1 1 1 1
Z
dx 2
= as X .
2 x(ln x) ln 2 ln X ln 2
If a > 1 the integral converges for all b because the exponential term dominates. If
a < 1 the integral diverges for all b, for the same reason. If a = 1 the integral converges
only if b > 1.
2
(d) Put x = /2 , > 0 to give ( 2x) tan x = = 2 + O().
tan
(e) Put y = x1/x , so ln y = (1/x) ln x and lim ln y = and lim x1/x = 0.
x0 x0
(f) We have
1/x
1+x 1 1+x 1
2 x + O(x3 ) = e2 .
lim = lim exp ln = lim exp
x0 1x x0 x 1x x0 x
414 CHAPTER 15. SOLUTIONS TO EXERCISES
dy
= (f g ln f + gf ) f (x)g(x)1 .
dx
(b) Since
1 1 1 1
ln y = ln(p + x) ln(p x) + ln(q + x) ln(g x)
2 2 2 2
we have
y
1 1 1 1 1 1 p q
= + + + = + 2
y 2 p+x px 2 q+x qx p2 x2 q x2
and r r
dy p q p+x q+x
= + 2 .
dx p2 x2 q x2 px qx
(c) We have
dy x yn dy y
ny n1 =1+ = therefore = .
dx 1 + x2 1 + x2 dx n 1 + x2
and then
d2 y 1 d2 y d x dy
= .
dx2 1x2 d 2 dx (1 x 2 )3/2 d
d2 y x dy
+ + y = 0,
d2 1 x2 d
p
which gives the required result since x/ x2 + y 2 = cot .
416 CHAPTER 15. SOLUTIONS TO EXERCISES
so that
2
g (x)f (g) + 3g (x)g (x)f (g) + g (x)3 f (g) 3 g (x) g (x)f (g)
Sh(x) = + .
g (x)f (g) 2 g (x) f (g)
On multiplying this out we see that Sh(x) = Sg(x) + g (x)2 Sf (g) < 0 since Sg(x) < 0
and Sf (g) < 0.
2z 2z
and hence a2 2
2 = sin(x + ay).
x y
Solution for Exercise 1.62
f f f
Differentiation gives fx = af , fy = bf and fz = cf . So the partial
x y z
derivatves are zero at ax = by = cz = 1.
0 2 4 6 8 x 10 12 14 16
1
2
Figure 15.1 Graphs of y = 1/x and y = tan x.
For the nth root, put x = n + z, and since sin x = (1)n sin z and cos x = (1)n cos z
the equation becomes
(n + z) tan z = 1 with z small.
Put = 1/n so the equation becomes (1 + z) tan z = and we require the Taylor
expansion of z() about = 0. Putting = 0 we see that z(0) = 0. Differentiation gives
1 + z
(z + z) tan z + z = 1 giving z (0) = 1,
cos2 z
418 CHAPTER 15. SOLUTIONS TO EXERCISES
1
and hence x = n + .
n
Further differentiation of the same equation allows, in principle, the calculation of
z (n) (0) for n > 2; however, such calculations are extremely tedious and error prone. A
far easier method is now outlined.
First, rewrite the equation for z in the form
tan z =
1 + z
and observe that this equation defines a function z(), with z(0) = 0, that is an odd
function of to see this note that z() satisfies the same equation. Also, for small
|z| we see that to O() the equation becomes z = + O(2 ). The power series for z()
is thus
z = + z3 3 + z5 5 + O(7 ),
where z3 and z7 are coefficients to be found. Substitute this series in to the left-hand
side of the equation and use the known series for tan z to obtain
3 3 2
tan z = + z3 3 + z5 5 + + 1 + z3 2 + + 5 +
3 15
3 1 5 2
= + z3 + + z5 + z3 + + .
3 15
Similarly the right-hand side gives
= 1 z + 2 z 2 +
1 + z
= 3 + 5 (1 z3 ) + .
Equating the coefficients of the powers of on each side of the equation gives z3 = 4/3
and z5 = 53/15 and hence
1 4 53
x = n + 2
+ + .
n 3(n) 15(n)3
(a) Since the Taylor series of ln(1 + z) is known we write, with u = x/2, and use the
identity cosh 2u = 1 + 2 sinh2 u,
ln(cosh x) = ln 1 + 2 sinh2 u
1 2 1 3
= 2 sinh2 u 2 sinh2 u + 2 sinh2 u + O(u8 ).
2 3
u2 u2
Now use sinh u = u 1 + + and sinh2 u = u2 1 + + in this expansion,
6 3
to give
u4 x2 x4
2
ln(cosh x) = 2u 1 + + 2u4 + = + O(x6 ).
12 2 12
(b) Similarly
1 1 1
ln(1 + sin x) = sin x sin2 x + sin3 x sin4 x + O(x5 ).
2 3 4
x2
sin x = x 1 +
6
giving
x2 x2 x2 x3 x4
ln(1 + sin x) = x 1 + 1 + + + O(x5 ),
6 2 3 3 4
x2 x3 x4
= x + + O(x5 ).
2 6 12
(c) Similarly
(2x)2 (2x)4 x4
1 1
sin2 x = 1 + + O(x6 ) = x2 + O(x6 ).
2 2 2 24 3
x2
f1 (x) = f1 (0) + xf1 (x) = < 0.
1 + x
Hence, f1 (x) < 0 for x > 0. Similary f2 (x) > 0 for x > 0.
sin x cos3 x
ln cos x 1
lim = lim = , hence lim y = e1/2 .
x0 tan2 x x0 cos x 2 sin x 2 x0
1
1 1
Z
I = dy sin y +
0 y y
Z
1 sin y 1 sin( z) 2 sin y
Z Z
= dy + dz = dy , (z = y),
0 y 0 z 0 y
2n 2 2n 2
1X 1 1 1 1
Z X Z
lim = dx hence lim = dx = ln 3.
n n 1 + k/n 0 1+x n n+k 0 1+x
k=1 k=1
422 CHAPTER 15. SOLUTIONS TO EXERCISES
Hence
1 iy sin y 2i 2 y
S= e 1 = + sin
iy y y 2
and hence
1 y 2y 2 y
lim sin + sin + + sin y = sin2 .
n n n n y 2
h i1/n
1
(d) If Pn = n (n + 1)(n + 2) . . . (2n) then
n n
1X 1X
ln Pn = ln n + ln(k + n) = ln(1 + k/n).
n n
k=1 k=1
3A 1 2 y 2 3A 1
y = + x but also x = x = y .
3x2 3 x 3 3x2
This solution is singular at x = 0, which is independent of the initial condition, y(1) =
A. The equation is linear.
suggesting that we set z = x/ + F (x), which gives the required coupled equations.
Note we could also set z = x + F (x), to obtain a different pair of equations, but the
former are more useful when 1.
One solution is p = 0, that is, p = c, a constant, so the general solution is y = cx+f (c).
A particular solution is given by the other solution of this equation, x = f (p),
which gives p in terms of x, and by substituting this into the original equation we obtain
one, or more, particular solutions. The geometric interpretation of this is discussed in
section 2.6.
15.2. SOLUTIONS FOR CHAPTER 2 425
1 1 1
y1 = 1 + x, y2 = 1 + x + x2 , y3 = 1 + x + x2 + x3 .
2 2 3!
The kth iterate adds the term xk /k!, so the nth iterate should be
1 1
yn (x) = 1 + x + x2 + + xn .
2 n!
This expression is correct for n = 1 and 2 and by substituting into the iterative formula
we see that yn+1 has the same form, hence it is true for all n.
The series for y(x) is a power series with coefficients un = 1/n!. The radius of
un
convergence is given by, see equation 1.32 (page 32), lim = , so the series is
n un+1
valid for all x.
(b) From the equation y (0) = 1. Differentiating the equation gives y (x) = y 2 + 2xyy ,
so y (0) = A2 . Another differentiation gives y (x) = 4yy + 2xy 2 + 2xyy , so y (0) =
4A. Thus the third-order Taylor series is
1 2
y(x) = A + x + A2 x2 + Ax3 .
2 3
This example shows that the iterative scheme, to a given order, usually produces more
terms of the series solution than the equivalent order Taylor series.
(c) Since 1 + xy 2 for x 0, y > 0 and hence y(x) is monotonic increasing.
For sufficiently large x, xy 2 1, so y z where z = xz 2 and this equation has the
solution 2/z = c2 x2 , where c is a constant. As x c, z(x) . But y(x) > z(x)
for all x > 0, thus y(x) also tends to infinity at some finite x: the actual position of
this singularity depends upon A, but is difficult to determine, but see exercise 2.21
(page 68).
A(1 + x2 ) 1
which rearranges to y = .
A(1 + x2 ) + 1
(b) Write the equation in the form (1 + x)y = x(1 + y) which separates to
dy x Aex
Z Z
= dx giving ln(1 + y) = C + x ln(1 + x) that is y = 1.
1+y 1+x 1+x
(c) This equation separates directly and, using the initial conditions, gives
Z y Z x 2
dy x 1 1
= dx hence y = 1 + x ln(1 + x) 1.
0 1+y 0 1+x 2 2
This is a quadratic equation for z, and hence y, with the two solutions
1
y= 1 2x 1 4C 8x .
2
15.2. SOLUTIONS FOR CHAPTER 2 427
dv dv F (v) v
x + v = F (v) which rearranges to = .
dx dx x
dv ev
= which gives ev = C + ln x that is y = x ln (C + ln x) .
dx x
dv 1 + 3v dv 1 v2
x +v = , that is x = .
dx 3+v dx v+3
One solution is v 2 = 1, that is, y = x. The equation for v separates to
v+3 dx 2 1
Z Z Z
dv = or dv + = C + ln x.
1 v2 x 1v 1+v
1+v
Hence = Ax and this becomes
(1 v)2
x+y 1
2
= A with solutions y = 1 + Bx 1 + 4Bx , (B = 2A).
(x y) B
The initial condition, y(1) = 0, gives B + 1 4B + 1 = 0; only the minus sign gives
real roots and these are B = 0 and 2. The solution B = 2 gives
1
y= 1 + 2x 1 + 8x .
2
If B = 0 we need to take the limit as B 0, or expand in B for small B: using
LHospitals rule we obtain y = x, which is one of the particular solutions found
above.
1 1 + 4Bx
The other solution, y = x + + x as B , and gives the other
B B
particular solution found above.
(d) If y = xv the equation becomes
dv 1 + v2 dv 1v
x +v = which rearranges to x = .
dx 1+v dx 1+v
Hence
2 dx
Z Z
dv 1 = that is ev = A|x|(1 v)2 .
1v x
This gives |x|ey/x = A(x y)2 ; here y cannot be expressed as a simple formula in x.
428 CHAPTER 15. SOLUTIONS TO EXERCISES
dv 3 v + 3v 2 dv 3(1 v)
x +v = which rearranges to x = .
dx 2 + 3v dx 2 + 3v
Hence
5 dx
Z Z
dv 3 =3 that is e3v |1v|5 = Ax3 or x2 e3y/x = B|xy|5 .
1v x
As in the previous example y cannot be expressed as a simple formula in x.
(f) If the two straight lines intersect at (a, b) we set = x a and = y b, that is we
move the point of intersection to the origin. The equation becomes
d 4 3 dv 2v 2 + 3v 2
= and with = v(), = 2
d 3 + 4 d 3 + 4v
that is
3 + 4v d
Z Z
dv = 2 .
2v 2 + 3v 2
Integration gives ln(2v 2 + 3v 2) = C 2 ln or (2v 1)(v + 2) = A/ 2 , where A is an
arbitrary constant. The lines intersect at (1, 1) hence the solution is (2y x 1)(y +
2x 3) = A.
If we set A = 0 in this general solution we obtain the two particular solutions 2y x = 1
and y + 2x = 3, which can also be derived from the equation for () by setting = k
and solving the resultant quadratic to give k = 1/2 and k 2.
e1/x 1
Z
(yp) = 4 which gives yp = A dz z 2 ez , z= .
x x
Hence the general solution is y = Ax2 e1/x 1 2x 2x2 . This linear equation has an
essential singularity at x = 0, where the coefficient if y has a repeated zero.
(d) On substituting x = 0 into the equation we see that y (0) = 0, so the solution is
stationary at x = 0.
The equation can be written in the form
d
cos x (y cos x) = (1 + cos2 x) tan x, y(0) = 2, and hence
dx
Z x Z c
1 + cos2 x
1
y cos x 2 = dx sin x = dc 1 + , (c = cos x)
0 cos2 x 1 c2
1 2 1
= cos x , hence y = 1 + .
cos x cos x cos2 x
This solution can be written as y = 2 (1/ cos x 1)2 , which, by inspection, has
a local maximum at x = 0 (because 1/ cos x has a minimum here). In fact since
1/ cos x = 1 + x2 /2 + O(x4 ), y = 2 x4 /4 + , so y (0) = y (0) = 0, and the solution
has a very flat maximum at x = 0.
The solution is singular where cos x = 0, which is where the coefficients of y are zero.
Q
v f + v (f + f P )) = Q, but f + f P = 0 and hence v = .
f
430 CHAPTER 15. SOLUTIONS TO EXERCISES
Rx
Because f (a) = 1, this gives v = A + adt Q(t)/f (t), and hence the general solution is
Z x
Q(t)
y(x) = f (x) A + dt .
a f (t)
Since f (x) = 1/p(x), where p(x) is defined in equation 2.17 (page 63), we see that this
solution is identical to that in equation 2.19.
dz 2(2x2 1) 2x
z= .
dx x(1 x2 ) 1 x2
(d) By writing the equation in the form y y/x = y 2 /x3 we see that n = 2. Putting
z = 1/y gives xz + z = x2 , that is (xz) = x2 , which integrates to give
y = x2 /(1 + Ax).
(b) If
u = a0 + a1 x + a2 x2 + + an xn + ,
then
Substituting these series into the equation for u and collecting the powers of x gives
x0 : a1 = 0,
x1 : 0 = 0,
x2 : 3a3 + a0 = 0,
xn : (n2 1)an+1 + an2 = 0, n 2.
Thus the values of a0 and a2 are undetermined and a1 = 0. We obtain two independent
solutions by setting (a0 , a2 ) = (1, 0) and (0, 1). The first of these gives a5 = a8 =
a3k+2 = 0 and the second gives a3 = a6 = a3k = 0, k = 1, 2, , so the two solutions
are
1 1 (1)k
a3 = 2
, a6 = suggesting that a3k = .
2 1 (2 1)(52 1)
2 (22 1)(52 1) ((3k 1)2 1)
The latter expression satisfies the recurrence relation, and is true for k = 1 and 2, so
by induction it is true for all k 1. Similarly
(1)k
b3k+2 = .
(42 1)(72 1) ((3k + 1)2 1)
The series for u1 and u2 are power series in x3 and using equation 1.32 (page 32) the
radii of convergence are given by
a3k
r13 = lim = and r3 = lim b3k+2 = .
2
k a3k3 k b3k1
15.2. SOLUTIONS FOR CHAPTER 2 433
(c) The general solution is u(x) = c1 u1 (x+ c2 u2 (x), for some constants c1 and c2 , so
c1 u1 (x) + c2 u2 (x)
y =
x (c1 u1 (x) + c2 u2 (x))
c1 (3a3 x + ) + c2 (2 + 5b5 x3 + )
= .
c1 (1 + a3 x3 + ) + c2 (x2 + )
Thus the initial condition gives
2c2 2u1 (x) Au2 (x)
A= and hence y(x) = .
c1 x (2u1 (x) Au2 (x))
The denominator of y is
x6
A 1 A
d(x) = x 1 x2 x3 + x5 + + .
2 3 30 72
p
If A 1 then the first two terms show that d(x) is zero
close to x = 2/A. At this
point the 3 rd and 4 th terms of the bracketed term are 2 3 2 A3/2 and 2152 A3/2 , so we
p
deduce that y(x) has a pole near x = 2/A.
2A
Note that the similar equation y = xy 2 , y(0) = A, has the solution y = , which
p 2 Ax2
has a pole at x = 2/A.
x + 1 A(x 1) exp(x2 )
yx1
Z
ln = dx 2x that is y = .
yx+1 1 A exp(x2 )
If y(1) = 1, A = e so
x + 1 + (x 1) exp(1 x2 )
y= .
1 + exp(1 x2 )
434 CHAPTER 15. SOLUTIONS TO EXERCISES
(c) Inspection shows that v = x is a particular solution. This satisfies the condition
v(1) = 1, so is the required solution.
solution for y is
p(x)
y(x) = a + R .
C dx p(x)
exp(x2 /2) 1
y =x+ Rx gives y(0) = = a,
C 0 dt exp(t2 /2) C
and hence
a exp(x2 /2)
y =x+ Rx .
1 a 0 dt exp(t2 /2)
Rx
This solution is singular when a 0 dt exp(t2 /2) = 1.
(b) In this example v = x and v = 1/x are particular solutions, and with y1 = x,
y2 = 1/x and R = 1/(1 + x), equation 2.28 becomes
yx x 1/x 1
Z Z
ln = dx = dx 1 .
y 1/x x+1 x
x Bex
y= .
1 Bxex
dz x
= (b2 + z 2 )
dx 1 + x2
which is separable, and gives
b 2 b b 2
z = b tan A + ln(1 + x ) that is y = 2 tan A + ln(1 + x ) ,
2 x 2
d2 w
+ abxn w = 0.
dx2
436 CHAPTER 15. SOLUTIONS TO EXERCISES
Since
z u = z u + z 1 u and z u = z u + 2z 1 u + ( 1)z 2 u,
A2 1 xd 1d
y= , 1, 2 = , d= 1 4ab.
ax(xd A) 2
p 2 p
q
u p + + 3/2 u = 0.
p 4p 2 p
Dividing by p gives the quoted result.
and hence g + qg = 0.
f + p1 f + p0 f = 0 and g + p1 g + p0 g = 0.
2a + b 1 a(a + b)
W = bx2a+b1 , p1 = and p0 =
x x2
15.2. SOLUTIONS FOR CHAPTER 2 439
giving the equation x2 y (2a + b 1)xy + a(a + b)y = 0 which has a singular point at
x = 0. The solution of this linear equation can be found in a variety of ways: one is to
put y = Ax to form a quadratic in , with solutions = a, a + b, giving the functions
f and g, as expected, and the general solution y = (C + Dxb )xa .
If b = 0, f = g and W (f, g) = 0. Nevertheless, the equation for y(x) exists and becomes
x2 y (2a1)xy +a2 y = 0; putting y = Ax gives (a)2 = 0, and hence the solution
y = xa . Now substitute y = xa v(x) into the equation to obtain x2 v + xv = 0, having
the solution v = A + B ln x. Thus when b = 0 the general solution is y = (A + B ln x)xa ,
which cannot be obtained from the general solution valid when b 6= 0.
xa2 a2
p1 = and p0 =
1 ax 1 ax
giving the equation (1 ax)y + xa2 y a2 y = 0 which has a singular point at x = 1/a.
dW g f
= f g f g = p1 f + p0 f p1 g + p0 g
dx p2 p2
p1 p 1
= gf g f = W.
p2 p2
x Aa2
v= +B ex(2+a1 /a2 ) ,
2a2 + a1 2a2 + a1
1
y = A cosh x + B sinh x sin x.
2
The initial conditions give a = A and b = B 1/2, hence the solution is
1 1
y = a cosh x + b + sinh x sin x.
2 2
(c) The general solution of the homogeneous equation is A cosh 2x+B sinh 2x, so we ex-
pect a particular solution of the inhomogeneous equation to be y = C, and substituting
this into the equation gives 4C = 6. Thus the general solution is
3
y = A cosh 2x + B sinh 2x .
2
The initial conditions give A 3/2 = a and 2B = b, hence the solution is
3 1 3
y = a+ cosh 2x + b sinh 2x .
2 2 2
442 CHAPTER 15. SOLUTIONS TO EXERCISES
(d) The general solution of the homogeneous equation is A cos 3x + B sin 3x, so we
expect a particular solution of the inhomogeneous equation to be y = C + Dx, and
substituting this into the equation gives 9C + 9Dx 1 + 2x. Thus the general solution
is
1
y = A cos 3x + B sin 3x + (1 + 2x).
9
The initial conditions give A + 1/9 = a and 2/9 + 3B = b, hence the solution is
1 1 2 1
y= a cos 3x + b sin 3x + (1 + 2x).
9 3 9 9
(e) To find the general solution of the homogeneous equation we set y = ex to obtain
2 6 = 0, that is = 3 and 2: thus the general solution is Ae3x + Be2x . We
expect the particular integral to have the form y = C cos 2x + D sin 2x and substituting
this into the equation and equating the coefficients of cos 2x and sin 2x to zero gives
the linear equations D + 5C = 9 and 5D C = 7, with solutions D = 22/13,
C = 19/13. Thus the general solution is
19 22
y = Ae3x + Be2x cos 2x sin 2x.
13 13
The initial conditions give the equations
19 44
A+B =a+ , 3A 2B = b +
13 19
1 82 1
with solution A = + 2a + b and B = (1 + 3a b); hence the solution is
5 13 5
1 82 1 19 22
y= + 2a + b e3x + (1 + 3a b) e2x cos 2x sin 2x.
5 13 5 13 13
1
c1 = tan x sin x = + cos x and c2 = tan x cos x = sin x,
cos x
with solutions
x
c1 (x) = sin x ln tan + and c2 (x) = cos x.
2 4
x
Thus y = c1 (x)f + c2 (x)g = cos x ln tan + . The general solution is
2 4
x
y = A cos x + B sin x ln tan + cos x.
2 4
15.2. SOLUTIONS FOR CHAPTER 2 443
y = c1 e2t + c2 e3t = c1 x2 + c2 x3 ,
where c1 and c2 are arbitrary constants. The initial conditions give 1 = c1 + c2 and
0 = 2c1 3c2 , hence c1 = 3/5 and c2 = 2/5.
d2 y d2 y
dy dy d dy dy
x = , x x = 2 = x2 2 + x
dx dt dx dx dt dx dx
and
d3 y d2 y d3 y d2 y
d d dy d dy dy
x x x = 3 =x x2 2 + x = 3 3 2 +2 .
dx dx dx dt dx dx dx dt dt dy
In terms of t the given equation becomes
d3 y d2 y dy
3
6 2
+ 11 6y = et/2 .
dt dt dx
The general solution of the homogeneous equation is found by substituting y = et into
the equation, to obtain 3 62 + 11 6 = ( 1)( 2)( 3) = 0, so the general
solution is y = A exp(t) + B exp(2t) + C exp(3t). For the inhomogeneous equation, put
y = det/2 and we find that d = 8/15, so the general solution is
8 1/2
y = Ax + Bx2 + Cx3 x .
15
r r
y y y 2
x 1 p 1 = d, that is x y = d,
x y/x x
which
gives the four equations x y = d, x y = d, x + y = d and
x + y = d. The last of these cannot be satisfied for real x and y, thus there are
three real solutions:
y = x + d, x 0, y d, (A)
y = x d, x d, y 0, (B)
y+ x = d, 0 x, y d (C)
1 0.1 0.2
0 0 0
0 2 2
0.5 1
x
1.5 1 1.5 2.5
x 3 0 0.2 0.4 0.6
x
0.8 1
x Cy x y
+ 2 1 = 0 giving fC (x, y, C) = 2 + 2 .
f (x, y, C) =
C d C d
x
r
Hence fC = 0 gives C = d , so the envelope is given by
y
x 1 x 1 2
r
p + y = 1, that is xy = d , a hyperbola.
d x/y d y 4
1 1 x2
y= ax2 and the initial condition gives y = A .
2a 2 4A
But y(1) = 1, so
! ! !
3 5 3 5 7 3 5
a = ( 5 1) ln + ( 5 + 1) ln + = ln + .
2 2 2 2 2 2
(h) This is a linear, first-order equation which can be written in the form
d
exp( cosh x) (y exp(cosh x)) = sinh 2x,
dx
which can be integrated,
Z
y exp(cosh x) = a+2 dx sinh x cosh x exp(cosh x)
Z
= a+2 dc ce2 = a + 2(c 1)ec , c = cosh x,
(i) This is a linear, first-order equation which can be written in the form
1 d 1
yx2 = x2 x2 y = a + x5 .
2
hence
x dx 5
1 1
The initial condition gives a = 1/5, and y = x3 2 .
5 x
15.2. SOLUTIONS FOR CHAPTER 2 447
dz d z sin3 x
cos x +2z sin x = 2 cos x tan3 x which can be written in the form = 2
dx dx cos2 x cos5 x
which has the general solution
z sin3 x 1 c2
Z Z
= a2 dx =a=2 dc , c = cos x
cos2 x cos5 x c5
1 1
= a 4 + 2.
2c c
If y(0) = 2, then z(0) = 1/2 and hence a = 0, so that
1 2 cos x
z =1 that is y = , 0x< .
2 cos2 x cos 2x 4
dz z 1 d 1
+ = 3 which can be written in the form (xz) = 2
dx x x dx x
which has the general solution
a 1 x2
z= + 2 and hence y = .
x x 1 + ax
(b) For the homogeneous equation, set y = xa , so (a 1)2 = 0. This gives one solution
y = x. Now set y = xv(x) to obtain xv + v = 0, so the second solution is y = x ln x,
and the general solution of the homogeneous equation is y = x(A + B ln x).
For the inhomogeneous equation, set x = et to write it in the form y 2y + y = te3t ,
which suggests y = (at + b)e3t . Substituting this into the equation gives
1 1
4ate3t + 4(a + b)e3t = te3t hence a = , b= ,
4 4
1
so the general solution is y = x(A + B ln x) x3 (1 ln x).
4
(c) Write the equation in the form xy (x) = yy (x), and set x = et to express this as
y (t) = (1 + y)y (t). Now put p = y (t) and express p in terms of y, so p (t) = pp (y)
and the equation becomes pp (t) = (1 + y)p.
448 CHAPTER 15. SOLUTIONS TO EXERCISES
dy
Z
b+t=2 .
a + (1 + y)2
The form of this integral depends upon the sign of a. For a = 2 > 0, we obtain
1 1 1 1+y 1 1
(b + t) = tan hence y = 1 + tan b + ln x .
2 2 2
Substituting these into the equation shows that v = ex cos x is a particular solution.
Now put y = f (x)v(x), to give
and the equation becomes f + 2f = 0. This has the general solution f= c1 + c2 e2x ,
hence the general solution of the original equation is y = c1 ex + c2 ex cos x.
(b) With these values of a, b and c the values of or , and the solutions are as shown
in the table.
a b c b2 4ac general solution
i) 2 3 1 1 3/2 tanh(x/2 + )/2
ii) 9 0 4 122 3 tanh(6x + )/2
iii) 1 2 1 0 1/( x) + 1
iv) 1 4 5 22 2/5 + tan(x + )/5
y x y y
z = 2 = (c + dz) (ax + by) 2 = c + (d a)z bz 2 .
x x x
(b) Put x = Aet and y = Bet in the equations for x and y, to obtain
(a )A + bB = 0 and cA + (d )B = 0.
(c) Differentiation gives Z (x) = E (x + y) = E(x + y) = Z(x) and the initial condition
is Z(0) = E(y). But the equation for Z(x) is linear, so the the solution of this equation
with the initial condition Z(0) = A, for any constant A, is Z(x) = AE(x). Now put
A = E(y) to obtain E(x + y) = E(x)E(y).
dx 1
L (y) = = .
dy dy/dx
(e) Differentiate the original equation n 1 times to give E (n) (x) = E (n1) (x), so
E (n) (0) = 1 for all n. This gives the Taylor series for E(x).
Differentiating the equation for L(y) n times gives L(n+1) (y) = (1)n y n1 , so L(n+1) (1) = (1)n1
which gives the Taylor series about y = 1. The relation
1+z 1
L = L(1 + z) + L = L(1 + z) L(1 z)
1z 1z
(c) We have
so that
f
f
=A with f (0) = C(a), g(0) = S(a).
g g
15.2. SOLUTIONS FOR CHAPTER 2 453
The functions C(x + a) and S(x + a) satisfy the same linear equations as C(x) and
S(x), but with different initial conditions. Thus C(x + a) and S(x + a) must be linear
combinations of C(x) and S(x):
C(x + a) = AC(x) + BS(x), S(x + a) = CS(x) + DC(x).
Putting x = 0 gives C(a) = A and S(a) = D. Differentiation gives
C (x + a) = S(x + a) = AC (x) + BS (x)
= AS(x) + BC(x).
Putting x = 0 gives S(a) = B and similarly, by considering S (x + a) we obtain
B = S(a), and hence
C(x + a) = C(x)C(a) S(x)S(a) and S(x + a) = S(x)C(a) + C(x)S(a).
(d) Represent the solution as a point in the Cartesian space with coordinates (C(x), S(x)).
At x = 0 the coordinates are (1, 0) and as x increases the point traces out the cir-
cle C(x)2 + S(x)2 = 1. From the initial point the point moves anticlockwise, be-
cause C (0) = 0 and S (0) = 1. At no value of x is C = S = 0: if this were
the case a solution would be C = constant and S = constant, which would contradict
the uniqueness theorem. Thus as x increases the point moves continuously, anticlock-
wise round the circle. At some value of x = X > 0 it must reach the initial point
(C(X), S(X)) = (1, 0). Because the equations are autonomous we may use this as an
initial condition to see that (C(2X), S(2X)) = (1, 0), and hence using induction we see
that (C(nX), S(nX)) = (1, 0), n = 0, 1, 2, .
Alternatively we can use the result found in part (c) with x = a = X, to obtain
C(2X) = C(X)2 S(X)2 = 1 and S(2X) = 2S(X)C(X) = 0, and use induction to
prove the general result.
It follows, using the result found in part (c), that C(x) and S(x) are periodic functions
with period X, C(x + X) = C(x) and S(x + X) = S(x) for all x.
(e) Consider the vales of x for which the points on the circle (0, 1), (1, 0) and (0, 1)
are first reached. If these are X1 , X2 and X3 we have
Z X1 Z 0 Z 1 Z 1
dC dC dC p
X1 = dx =
= = , (C = 1 C 2 < 0).
0 1 C 0 S 0 1 C2
For the second point
Z X2 1 0 1
dC dC dz
Z Z Z
X2 = X1 + dx = X1 + = X1 + = X1 + = 2X1 ,
X1 0 C 1 1 C2 0 1 z2
where C = 1 C 2 < 0. For the third point
Z X3 Z 1 Z 0
dC dC p
X3 = X2 + dx = X3 +
= X2 + = 3X1 , (C = 1 C 2 < 0).
X2 0 C 1 1 C2
Finally, for the last quarter of the circle
Z XX 1 1
dC dC
Z Z
X = X3 + dx = X3 + = X3 + = 4X1 ,
X3 0 C 0 1 C2
which proves the required result.
454 CHAPTER 15. SOLUTIONS TO EXERCISES
(sinh x cos x cosh x sin x) y + 2 sinh x sin x y (cosh x sin x + sinh x cos x )y = 0
cos 2x
p0 = 4/ sin2 2x, p1 = 2 giving the equation sin2 2x y + sin 4x y 4y = 0.
sin 2x
and
1 1 f u
p1 W = f f =2 , hence p1 = .
f f f u
u f
Thus the differential equation with solutions f and 1/f is y y u2 y = 0, u= .
u f
Multiply the first and second rows by p0 and p1 , respectively, and add to the third row
to obtain
f g h
dW
= f g h = p2 (x)W (x).
dx
p2 f p2 g p2 h
Hence Z x
W (x) = W (a) exp dx p2 (x) .
a
f f g 2f g 2 2f g
v = 2 + , but f = qf, g = qg,
g g g3 g2
2g 2g
= (f g f g) = v.
g3 g
456 CHAPTER 15. SOLUTIONS TO EXERCISES
Hence
2g g 2
g
v = v 2 2 v ,
g g g
2 2
v g v 3 v
= 6 2q, hence = 2q.
v g v 2 v
af + bg av + b
v= = and S(v) = 2q = S(v).
cf + dg cv + d
x
y =
m( + )
m ()
1 2
= (x ) 1 + O( )
m() m()
m
1 2
= x 1 + (x ) + O( ) .
m() m
But = f () + O( 2 ) and using the equation of the normal through the adjacent
point (, ) we obtain an equation for the x-coordinate of the point of intersection of
these two normals,
m
x = (1 + mf ), m() = f ().
m
The y-coordinate is given by
1 1
1 f ()2 .
y = (x ) =
m () m ()
s ds
(s) = lim = .
s0 d
Figure 15.3
1 1
E= ALh2 + Agh2 .
8 4
460 CHAPTER 15. SOLUTIONS TO EXERCISES
A O x P B
Consider a hole AB as shown in the figure. The force on a
r
particle P in the direction BA is F = kr cos , and if OP = x,
so cos = x/r, then F = kx, k > 0.
Newtons equation of motion is mx = kx and hence for a
particle initially at B,
with x(0) = X and x(0)= 0 the so-
lution is x = X cos( k t) and the period is 2/ k, which is
independent of X, that is the position of the hole.
Figure 15.4
15.3. SOLUTIONS FOR CHAPTER 3 461
The functional is stationary if the first-order term is zero for all h(x), otherwise S
would change sign with . Using the result quoted inpthe text (after equation 3.5)
and proved in exercise 4.4 (page 128) this gives 1 + y (x) =constant, that is
y (x) =constant and y(x) = x + . The boundary conditions then give y = Bx for
the stationary path. With this value for y(x), the integrand is real if B > 1 and has
the value S = 1 + B.
Hence
p p 2 2
1 + ( + )2 = 1 + 2 + + + O(3 ).
1 + 2 2(1 + 2 )3/2
462 CHAPTER 15. SOLUTIONS TO EXERCISES
(b) With = y (x) and = h (x) we see, using the argument described in the text,
that the term O() in the expansion of S[y + h] S[y] is zero if y (x) =constant, hence
the straight line defined by equation 3.6 makes the functional stationary. With this
choice of y(x), = m and the second term in the above expansion gives the result
quoted. The second-order term is positive for 6= 0 and all h(x), so the functional has
a minimum along this line.
and the first resultfollows. There is equality only if a = b, that is u = 0. Divide the
first inequality by 1 + z 2 to derive the second result.
This depends only upon C, D and the boundaries a and b: the value of the functional is
therefore independent of the chosen path, and hence this functional has no stationary
paths.
Alternatively, consider the difference
Z b h i
S = S[y + h] S[y] = dx C(y + h ) + D (Cy + D)
a
Z b h i
= C dx h (x) = C h(b) h(a) .
a
15.3. SOLUTIONS FOR CHAPTER 3 463
Since h(a) = h(b) = 0, S = 0 for any y(x). That is, there is no stationary path.
If C and D depend upon x then
Z b
S = dx C(x)h (x).
a
If C is a constant, because h(a) = h(b) = 0, S = 0 for all y(x), as expected from the
previous analysis. Otherwise, for any C(x) we can chose h(x) so that C(x)h (x) > 0
for a < x < b: since S[y] is stationary only if S = O(2 ), this shows that S[y] has no
stationary path.
2 BA (B A) 3/2
a = 3/2 and hence y(x) = A + (1 + x) 1 .
3 2 1 (23/2 1)
F F x F y
= 0, =p and
= p
y x x + y 2
2 y x2 + y 2
giving
dF F F F x + y y
= + y + y = p .
dx x y y x2 + y 2
Since F does not depend explicitly upon y, we have
2 F 2F
d F
= 2
y +
dx y y xy
and
2F xy 2F 1 y 2 x2
= , = =
xy (x2 + y 2 )3/2 y 2 (x2 + y 2 )1/2 (x2 + y 2 )3/2 (x2 + y 2 )3/2
464 CHAPTER 15. SOLUTIONS TO EXERCISES
which gives
x2 y xy x(xy y ) x3 (y /x)
d F
= = = .
dx y (x2 + y 2 )3/2 (x2 + y 2 )3/2 (x2 + y 2 )3/2 (x2 + y 2 )3/2
Also
y (x + y y )y x(xy y )
dF
= p 2 2 3/2
= 2 ,
y dx x2 + y 2 (x + y ) (x + y 2 )3/2
d F dF
so, in this case, = .
dx y y dx
F dy
d F F dy F
= + +
dx y y y dx y y dx x y
2 F 2F 2F
= 2
y + y +
y y y xy
which gives the required expression and is the left-hand side of the inequality.
The right-hand side of the inequality is
dF F F F
= + y + y
y dx y x y y
2F F 2 F 2 F
=
+ + y + 2y
xy y yy y
which differs from the left-hand side by the term F/y. Thus, only if F is independent
of y are the derivatives equal.
we obtain
yy y 2 1/2
z = 2 3/2
+ 2 1/2
1 + y 2 ,
(1 + y ) (1 + y )
1
2
2
2 2 1
yy y 2 1 ,
= yy + 1 + y y 1 + y =
(1 + y 2 )3/2 (1 + y 2 )3/2
hence the equation z = 0 becomes yy 1 y 2 = 0. But
y 2
y
d y 2 2 d y
= 2 giving yy y = y , if y 6= 0,
dx y y y dx y
and hence
d F F 1 2 d y
= y 1 .
dx y y (1 + y 2 )3/2 dx y
(b) If the left-hand side is zero we have
d y d y
y2 = 1 or y 2 y = 1.
dx y dy y
Now define z = y /y and consider z to be a function of y, so in the following z = dz/dy
note this is possible because x may be considered a function of y so y /y can be
expressed in terms of y. Now put the second equation in the form y 3 z z (y) = 1, which
can be integrated directly to give z 2Z = C 2 y 2 , for some constant C. Hence, since
dy p dy
z = y /y, = (Cy)2 1 giving p = x + D. Finally, set Cy = cosh
dx (Cy)2 1
to give = C(x + D), that is y = (1/C) cosh(Cx + CD), which is the required solution,
if C = A and CD = B.
1 1 1 1
1 4 3 1 3 4
Z Z Z
S[x2 ] = dt s2 + st s2 t2 =
ds ds s t + s t
0 0 0 3 4 t=0
1
1 4 1 3 31
Z
= ds s + s = .
0 3 4 240
AR x RB dx
sin 1 = = p , and sin 2 = = p ,
SR x2 + h21 RO (d x)2 + h22
where the distances are defined in figure 3.10 (page 113), we see that the distance
travelled by the light is stationary when sin 1 = sin 2 , that is 1 = 2 . Further since
h21 h22
f (x) = 2
+ > 0,
(x2 + h1 ) 3/2 ((d x)2 + h22 )3/2
2 2
p 3
1 + + = 1 + 1 + + O( ) ,
2(1 + ) 8(1 + )2
15.3. SOLUTIONS FOR CHAPTER 3 467
and so
2 2
p
( + ) 1 + + = 1+ 1+ +
2(1 + ) 8(1 + )2
+ 1 + 1 + + ,
2(1 + )
(2 + 3) 2 2 (4 + 3)
= 1++ + + .
2 1+ 8(1 + )3/2
If y(x) is a stationary path of S then the term O() is zero. Since h(0) = h(1) = 0 it
follows, as in the text, that y (x) =constant
is a possible solution. Since y(0) = 0 and
y(1) = B this gives y(x) = Bx and S[y] = B 1 + B.
Alternatively, using equation 3.12 (page 100), with F (y ) = y 1 + y , we see that
the stationary path is given by F (y ) = constant and hence y = constant, that is
y = mx + c: since y(0) = 0 and y(1) = B this gives y(x) = Bx.
(b) On substituting Bx for y(x) we see that S takes the value,
1
2 (4 + 3B)
Z
S = dx h (x)2 + O(3 ).
8(1 + B)3/2 0
Then, provided B > 1, S is positive and the functional is a minumum on the sta-
tionary path.
ln(x/a)
A = d and B = d + c ln(b/a) and hence y(x) = A + (B A) .
ln(b/a)
468 CHAPTER 15. SOLUTIONS TO EXERCISES
then, provided F (z) is not a constant or a linear function of z, y (x) is also a constant.
(c) On the stationary path y (x) is a constant and hence d2 F/dy 2 is constant and
1 d2 F b
Z
S = 2 2 dx h (x)2 + O(3 ).
2 dy a
(b) The length along a curve is just the sum of the small elements which in the limit
R p
0 becomes the integral L[z] = 12 d 2 + z ()2 .
(c) The functional L[z] is the same type as that considered in section 3.3.1 hence its
minimum value is given when z() is a linear function of . The boundary conditions
give the result quoted.
Hence p the distance between the points 1 and 2 along the curve () is L[] =
R 2
1
d 2 + 2 sin2 .
Hence
a Z a
p p
(vy )2 + (c2 v 2 )(1 + y 2 ) vy (1 + y 2 )c2 v 2 vy
Z
T [y] = dx = dx .
0 c2 v 2 0 c2 v 2
M V 2 + mv 2 = M V 2 + mv 2 Energy conservation
M V mv = M V + mv Linear momentum in the direction of the block motion
and hence, since F depends only upon y and not y, the stationary points are given by
the equations,
S yk yk1 yk+1 yk
= F F = 0, k = 1, 2, , N.
yk
/2 /2
d
Z Z
dx (y + h )h (y + h)h dx (y h yh) .
S[y+h] = 2 and S[y, h] = 2
d 0 0
Rb
(b) We have S[y + h] = a dx (y + h )2 x3 . Hence
b b
d (y + h ) y h
Z Z
S[y + h] = 2 dx h and S[y, h] = 2 dx .
d a x3 a x3
Rb
dx (y + h )2 + (y + h)2 + 2ex(y + h) . Hence
(c) We have S[y + h] = a
b b
d
Z Z
dx (y + h )h + (y + h)h + ex h and S[y, h] = 2 dx [y h + (y + ex ) h] .
S[y+h] = 2
d a a
R1 p p
(d) We have S[y + h] = 0 dx x2 + (y + h)2 1 + (y + h )2 . Hence
" #
1
p
d (y + h)h x2 + (y + h)2 (y + h )h
Z p
S[y+h] = dx p 1 + (y + h )2 + p
d 0 x2 + (y + h)2 1 + (y + h )2
and " p #
1
p
y 1 + y 2 x2 + y 2 y
Z
S[y, h] = dx p h+ p h .
0 x2 + y 2 1 + y 2
472 CHAPTER 15. SOLUTIONS TO EXERCISES
so that
b b
d
Z Z
S[y + h] = ds dt K(s, t) y(s)h(t) + h(s)y(t) + O() .
d a a
where, in the second integral we have put t = s and s = t and then changed the
integration order of the first integral to obtain the final result.
Unless z(x) = C, the integrand is almost everywhere positve and hence the integrand
is zero only if z(x) = C.
dy
Z
= x or 2 c y = A x.
cy
Putting x = 0 gives 2 c = A and hence y = Ax/2 x2 /4; putting x = 1 gives
y(1) = 1 = A/2 1/4, and hence y = x(5 x)/4.
2 G 2 G 2 G
d G G
y = 2
yy + y y.
dx y y y yy y
2 G 2 G 2 G
d G G G
y = y y + y y = y + y + 2y y ,
dx y y y y y yy y y
2 F 2F 2F F
2
y +
y + = 0.
y yy xy y
If F (x, y, y ) = G(y, y ) the third term is zero and if y = this equation becomes
Gy (, 0) = 0, assuming that Gy y (, 0) and Gyy (, 0) exist.
Let g(y) = G(y, 0) be a function of y. The equation Gy (, 0) = 0 shows that must
be at a stationary point of g(y) whereas the equation G(, 0) = c, found in part (a),
imposes the weaker restriction that c lies in the domain of g(y).
Thus, in general the constant solution y = of the first-integral, is not a solution of
the Euler-Lagrange equation.
474 CHAPTER 15. SOLUTIONS TO EXERCISES
so the functional is
1 2
Z
J[y] = dx x2 = .
2 3
The function is continuous provided > 0 and hence on this class of continuous functions
J[y] can be made arbitrarily small, but not zero.
15.4. SOLUTIONS FOR CHAPTER 4 475
(c) The given functions behave similarly to the piecewise continuous function defined
in part (b), as seen in figure 15.5 which depicts graphs for = 0.1 and 0.01.
0.5
-0.5
-1
Figure 15.5 Graphs of the functions y(x) for = 0.1 (solid
line) and 0.01 (dashed line).
22 1 x2
Z
J[y] = dx
2 0 (x2 + 2 )2
Z 1
2 1
= d sin2 where tan 1 =
2 0
and the second integral is obtained by putting x = tan . Integration gives
1
J[y] = (21 sin 21 ) = 2 tan
2 2 2 1 + 2
2 1
2 1 tan + 1+2
= 2 .
1 2 tan1
Since tan = + O( ) we see that J[y] = 2/ + O(2 ). Since 0 < < 1, J[y] > 0,
1 3
The boundary term is zero, because h(a) = h(b) = 0, so equation 4.9 becomes
Z b
F
S[y, h] = dx (x) h (x).
a y
476 CHAPTER 15. SOLUTIONS TO EXERCISES
On a stationary path S = 0 for all admissible h(x), so the result proved in exercise 4.4
shows that F/y = C for some constant C.
1
Z b
2 2
S[y + h] = G y(b) + h(b) + dx (y + h ) + (y + h)
2 a
diiferentiation with respect to and then setting = 0 gives the Gateaux differential
Z b
S[y, h] = Gy (y(b))h(b) + dx (h y + hy) .
a
Now integrate by parts and use the fact that h(a) = 0 to cast this in the form
Z b
S[y, h] = y (b) Gy (y(b)) h(b) dx y y h.
a
(b) On the variations with h(b) = 0 the boundary term of S is zero. For S[y] to be
stationary it is necessary that S[y, h] = 0 and it follows from the fundamental lemma
that y y = 0 with y(a) = A.
On the path defined by this equation
S[y, h] = y (b) Gy (y(b)) h(b).
Since we require S[y, h] to be zero for all allowed h, which includes those variations
for which h(b) 6= 0, we must have
y (b) = Gy (y(b)).
On the variations with h(b) = 0 the boundary term of S is zero. For S[y] to be
stationary it is necessary that S[y, h] = 0 and it follows from the fundamental lemma
that
d F F
= 0, y(a) = A.
dx y y
15.4. SOLUTIONS FOR CHAPTER 4 477
Since we require S[y, h] to be zero for all allowed h, which includes those variations
for which h(b) 6= 0, we must have
Solution
p for Exercise 4.16
If F = 1 + y 2 / y we have
p
F 1 + y 2 F y
= and = p
y 2y 3/2 y
y 1 + y 2
which expands to
p
y y 2 y 2 y 1 + y 2 1 + y 2
p + = 0 that is y = .
y(1 + y 2 )3/2 2y 3/2
p
y 1 + y 2 2y 3/2 1 + y 2 2y
(b) A second division gives 22 similar triangles of height 22 h and a line of length 2l.
After n divisions there are therefore 2n similar triangles of height 2n h and a continuous
line of length 2l. Since this is true for any l, the length of the line is unbounded.
This function is stationary at the root of S (y1 ) = 22y1 /3 1/2, that is y1 = 3/44
0.0682.
dy 1d 1 A
y 2 = (1 + x)2 + ,
y =
dx 2 dx 4 2
and integrating again, y(x)2 = B + Ax 16 (1 + x)3 . The boundary conditions then give
y(0)2 = B 61 = 1, so B = 76 , and y(1)2 = 67 + A 86 = 4, so A = 25 6 . Hence the
solution is
1 1
y(x)2 = (1 + x) 25 (1 + x)2 3 = 3 + (1 + x)(6 + x)(4 x).
6 6
The solution is written in this way because it is easier to understand. The cubic
f = (1 + x)(6 + x)(4 x) is zero at x = 6, 1 and 4; f is positive for x < 6 and
negative for x > 4. It follows that y is real only for x < x1 , for some x1 < 6, and
possibly for some x in the interval 1 < x < 4, depending upon the magnitude of f
in this interval. Numerical calculations, which you are not expected to do, show that
x1 6.33 and that y is real in the interval (0.264, 3.59).
(c) The Gateaux differential is
Z 2
1 y h
S[y, h] = Bh(2) + 2 dx 2 , y(1) = A,
2 1 x
Z 2
1 1 d y
= B + y (2) h(2) 2 dx h ,
2 2 1 dx x2
the second result being obtained using integration by parts and the fact that h(1) = 0.
Using the subset of variations with h(2) = 0 and using the fundamental lemma shows
that the stationary paths must satisfy the Euler-Lagrange equation,
d y dy
2
= 0 that is = x2 with y(1) = A,
dx x dx
for some constant . On the paths that satisfy this equation
1
S[y, h] = B y (2) h(2),
2
and since h(2) need not be zero, S[y] is stationary only on those paths that satisfy
y (2) = B, because it is necessary that S[y, h] = 0 for all allowed h. The general
solution of y = x2 is y(x) = x3 /3 + and the boundary conditions give
1 1 1
A= + , B = 4 so = B and = A B.
3 4 12
Hence y(x) = B x3 1 /12 + A.
480 CHAPTER 15. SOLUTIONS TO EXERCISES
where we have integrated by parts and used the fact that h(b) = 0. Using the subset
of variations with h(0) = 0 and the fundamental lemma shows that S[y] is stationary
only on those paths that satisfy the Euler-Lagrange equation with F = yy 2 and with
the single boundary condition y(b) = B 2 . Since F is independent of x, so we may use
the first-integral, equation 4.13 (page 130), to give y y 2 = c2 , y(b) = B 2 , where c is a
positive constant (since y(b) > 0 the constant must be positive).
On the paths that satisfy this equation
1 1
S[y, h] = 2h(0)y(0) 3 ,
y (0)3 A
so S[y] is stationary only if y (0) = A > 0. The general solution is given by (since
y(0) = (Ac)2 ),
Z y
dy y dy x
= that is = .
dx c (Ac)2 y c
Hence y = Ac + x/2c and the boundary condition at x = b gives 2Ac2 2Bc + b = 0,
that is
1 p
c= B B 2 2Ab
2A
giving the two solutions
2
x 1 p
y (x) = Ac + , c = B B 2 2Ab .
2c 2A
The fact that the sum is zero for all k is the equivalent of the fundamental lemma of
the Calculus of Variations.
The Euler-Lagrange equation for the functional on the right-hand side of this equation
is again y = 0.
2G 2G
d F F d F F d G
= + y.
dx y y dx y y dx y xy y 2
But,
2G 2G
d G
+ 2
y = ,
xy y dx y
so the Euler-Lagrange equations for F and F are identical, as expected.
which is satisfied by the functions y(x) = 0 and y(x) = x2 . Thus the given function
satisfies the Euler-Lagrange equation except at x = 0 where y (x) is not defined.
15.4. SOLUTIONS FOR CHAPTER 4 483
which reduces to
Fy y 1
3
x Fy y Fx y + Fy = 0. (15.1)
x x
(b) The Euler-Lagrange equation for F is Fy y y + Fy y y + Fx y Fy = 0. But
d2 y x
d 1 dy
2
=
= 3,
dx dy x dx x
where z(x) is any function and the set of equally spaced points xk = k/(n + 1) defined
in the question. Hence the functional becomes
n n n
1X X X 1
S = (yk+1 yk )2 h yk2 2h xk yk , h = ,
h n+1
k=0 k=0 k=0
n
1X 2 2 2 2k
= (yk+1 yk ) h yk + yk .
h n+1
k=0
15.4. SOLUTIONS FOR CHAPTER 4 485
(a) If n = 1 there are two terms in the sum; the first is y12 /h, since y0 = 0, and the
second is (1/h h)y12 2hy1 , and since h = 1/2 this gives
7 2 1
S(y1 ) = y y1 .
2 1 2
This function is stationary where S/y1 = 7y1 1/2 = 0, that is y1 = 1/14 = 0.0714,
compared to the exact value of y(1/2) = 0.0697.
The difference between this approximation to S and that obtained in exercise 4.21 is
because the approximations to the functional are different. In both cases we approxi-
mate the solution by the same type of polygon; but in the first case we evaluated the
integrals exactly; in the second case we made an additional approximation to evaluate
the integrals. For the approximation used in exercise 4.21 we have
"Z #
Z 1 Z 1/2 1
dx y (x)2 = 4y12 dx + dx = 4y12 ,
0 0 1/2
"Z #
1 1/2 1
1 2
Z Z
2
dx y(x) = 4y12 dx x +2
dx (1 x)2
= y ,
0 0 1/2 3 1
"Z #
1 1/2 1
1
Z Z
dx 2xy(x) = 4y1 dx x2 + dx x(1 x) = y1 .
0 0 1/2 2
which simplify to the given equations. These have the solutions y1 = 35/624 0.0561
and y2 = 43/624 0.0689, which are the approximate values of the solution at x = 1/3
and 2/3 respectively.
486 CHAPTER 15. SOLUTIONS TO EXERCISES
case F (z) = z 2 /2, F (z) = 1 and hence the integral is positive and the stationary path
is a minimum.
15.4. SOLUTIONS FOR CHAPTER 4 487
Thus
b
F F F
Z
S[y, h] = dx h + h + h .
a y y y
Integration by parts gives
b b Z b
F F d F
Z
dx h = h dx h .
a y y a a dx y
Using the fundamental theorem of the Calculus of Variations we see that a necessary
condition for the functional to be stationary on a function y(x) is that it satisfies the
equation
d2
F d F F
+ = 0,
dx2 y dx y y
with the given boundary conditions.
The general solution of this is y(x) = A cos x + B sin x + D cosh x + E sinh x. The
boundary conditions at x = 0 give
For the length of a curve we use a variant of equation 1.5 (page 15). Suppose that
increases from to + , then to O(), x and y increase by x () and y ()
respectively. Hence the length of the small element of the curve is, using Pythagoras
theorem p
s = x ()2 + y ()2 + O(2 ),
and the length of the curve between 1 and 2 is
Z 2 p
s= d x ()2 + y ()2 .
1
For the cycloid, x () = a(1 cos ), y () = a sin and the length of the arc OP is
Z q Z p
s = a d (1 cos )2 + sin2 = a d 2 2 cos ,
0 0
Z
= 2a d sin(/2) = 4a (1 cos(/2)) = 8a sin2 (/4),
0
(b) The initial point (X, Y ), where Y = AX, satisfies the equation X 2 +(Y R)2 = R2 ,
which becomes (1 + A2 )X = 2AR. Substituting this into the above equation for T gives
the required, rather surprising, result.
so that 3 2
x 2A y 2A x 2/3
= , =1 =1 .
b 3b A 3b b
But z = 2c2 sin cos and, since x = 21 c2 (2 sin 2) + d, x = 2c2 sin2 , so that
x 2 + z 2 = 4c4 sin2 and
Z b
2c 2cb
T = d = .
2g 0 2g
But, from the analysis preceeding the exercise, c = A/ sin b and so
s
2A b
T = .
g sin b
A2 + b2
(b R)2 + A2 = R2 giving R = .
2b
The time of passage is given by equation 5.6, with z = A y. The parametric equations
x = R(1 cos ) and y = A R sin satisfy the equation of the circle and as the particle
moves downwards from (0, A), increases from = 0 to = b where y = 0, that is
A 2Ab
sin b = = 2 ,
R A + b2
so b depends only on the ratio = b/A. Since z = R sin , using the relation dz/dx =
z ()/x (), equation 5.6 becomes
Z b s 2 s Z
1 x () + z ()2 R b 1
T = d = d .
2g 0 z() 2g 0 sin
and hence
a2 a4 a6 a2 a4
= + +O , + = A +O .
2A 4A3 A5 2A A3
A
Hence y (x) x and y+ A. With = + , y(x) A and the solution approxi-
a
mates a right circular cylinder. With = , y(x) Ax/a, so the solution increases as
|x| increases. We shall see later that both these solution behave like the exact solutions.
where we have used the relations 2 cosh2 u = 1 + cosh 2u to evaluate the integral and
sinh 2u = 2 sinh u cosh u to cast the result in this form. Dividing this by a2 we see that
S[f ] 2
the dimensionless area S[f ]/a2 depends only upon , 2 = 2 ( + sinh cosh ) .
a
(b) Since 2 sinh cosh = sinh 2 we define
1 sinh 2 1 sinh 2
F () = + giving F () = (cosh 2 1) .
2 2 2 3
Hence F () = 0 if
(cosh 2 1)
1= = tanh .
sinh 2
A 1 e
e + e = 1 + e2 .
=
a 2 2
494 CHAPTER 15. SOLUTIONS TO EXERCISES
A 1
= e , ( 1).
a 2
For large ,
1 1 1 1
cosh = e 1 + e2 e and sinh = e 1 e2 e
2 2 2 2
so 2
S[f ] 2 1 2 e 2
2 + e = 2 + .
a2 4 2
Since 1, e2 , that is e2 / 2 1/, so the first term dominates.
1.25 y
y=tanh
1
0.75
0.5 y=1/
0.25
0 1 2 3 4
Figure 15.6 Graph of y = tanh and y = 1/.
(b) At the stationary point the area is, using the result obtained in exercise 5.11
2 1 1
S = 2a + 2 sinh m cosh m
m m
2a2 2a2
= 1 + sinh2 m = cosh2 m since m sinh m = cosh m .
m m
But, by definition,
A 1
= g(m ) = cosh m hence S = 2a2 m g(m )2 = 2A2 m .
a m
15.5. SOLUTIONS FOR CHAPTER 5 495
A=1.2 A=3
1.25 3
y+(x) y+(x)
1
2
0.75
y(x)
0.5
1 y(x)
0.25
x x
-1 -0.5 0 0.5 -1 1 -0.5 0 0.5 1
Figure 15.7 Graphs of y (x) for A = 1.2, on the left, and A = 3 on the right.
(b) Substituting the general solution (for any c) into the functional gives
r
1 1
Z 1
x2 1
Z p
dx 4c4 + x2 ,
S[y] = 4 2
dx 4c + x 1 + 4 = 3
2c 1 4c 4c 1
1
= 2c + 3 . (15.2)
6c
In order to determine which path gives the largest value of S[y] we consider the difference
1 1 1
S[y ] S[y+ ] = 2(c c+ ) + 3 ,
6 c3 c+
2
c + c+ c + c2
= (c+ c ) + 2 ,
6(c+ c )3
4
= (c+ c )(A 1) > 0 if A > 1,
3
where we have used the relations c+ c = 21 and c2+ + c2 = A, which follow directly
from the original quadratic equation for c2 . This relation shows that S[y ] > S[y+ ] for
A > 1.
If A = 1, c+ = c = 1/ 2 and S = 4 2/3. Further if A 1 we have
r !
2 A 1 A 1 1
c = 1 1 2 = 1 1 +
2 A 2 2A2 8A4
where we have used the binomial expansion 1 x = 1 21 x 18 x2 + . Hence
2 1 1
c+ = A 1 +
4A2 16A4
and on taking the square root
1/2
1 1 1
c+ = A 1 1 + + = A 1 + .
4A2 2A2 8A2
Similarly
1 1 1 1
c2 = 1+ + giving c = 1+ + .
4A 4A2 2 A 8A2
15.5. SOLUTIONS FOR CHAPTER 5 497
Putting c+ = A and c = 1/2 A we obtain the following approximations
x2 1
y+ A + A and y + Ax2 Ax2 , A 1.
2A 4A
Substituting these approximations for c into the integral 15.2 for S we obtain
4
S[y+ ] 2 A and S[y ] A3/2 .
3
12 S[y]
10
S[y ]
8
6
S[y+]
4
2
A
0
1 1.5 2 2.5 3 3.5 4
Figure 15.8 Graphs of S[y ].
so y (x) is not defined at |x| = 1. Hence we define a function that approaches yG (x) as
0 for some parameter . We need only consider positive values of x:
0, 0 x < 1 , 0 < 1,
y (x) = A
A (1 x), 1 x 1.
Then
r
1
r
A A2
Z
S[y ] = 2 dx A (1 x) 1 + 2 ,
1
Z r
2 p v 4p 4
= A(A2 + 2 ) dv 1 = A(A2 + 2 ) A3/2 as 0.
0 3 3
498 CHAPTER 15. SOLUTIONS TO EXERCISES
C=(c,d)
3
l3 X 2
3
X
l1
l2 1
1 2
O A
Figure 15.9
The point X has the coordinates (x, y) and we need to find these coordinates so that
the length L = l1 + l2 + l3 is stationary. With the geometry shown
p p p
l1 = x2 + y 2 , l2 = (a x)2 + y 2 , l3 = (c x)2 + (d y)2 ,
y y dy
sin 1 = , sin 2 = , sin 3 = ,
l1 l2 l3
x ax cx
cos 1 = , cos 2 = , cos 3 = .
l1 l2 l3
The derivatives are
L x ax cx
= = cos 1 cos 2 cos 3 = 0,
x l1 l2 l3
L y y dy
= + = sin 1 + sin 2 sin 3 = 0.
y l1 l2 l3
Now let k , k = 1, 2, 3, be the angles between the intersecting lines, as shown on the
right of the figure, so 1 + 2 + 3 = 2. Also 1 = 1 2 , 2 = 3 + 2 and
3 = + 1 3 , so that
The first of these equations has the solutions 1 = 2n+2 and 1 = (2n+1)+2 , but
only the first of these also solves the second equation, and then only if cos 2 = 1/2,
that is 2 = 1 = /3, and hence 1 = 2 = 3 = 120.
In order to classify this stationary point we need the second derivatives: these are
2L x2 (a x)2 (c x)2
1 1 1
= 3 + +
x2 l1 l1 l2 l23 l3 l33
y2 y2 (d y)2
= 3 + 3 + > 0.
l1 l2 l33
15.5. SOLUTIONS FOR CHAPTER 5 499
Similarly,
2L x2 (a x)2 (c x)2
= 3 + 3 + >0
y 2 l1 l2 l33
2L xy (a x)y (c x)(d y)
= 3 + + .
xy l1 l23 l33
For a minimum we need Lxx > 0, Lyy > 0 and = Lxx Lyy L2xy > 0. Using the above
expressions we find that
a2 y 2 1 2 1 2
= + yc + xd 2xy + y(c x) (d y)(a x) > 0.
(l1 l2 )3 (l1 l3 )3 (l2 l3 )3
Hence the stationary point is a minimum.
On squaring and adding these we see that the cross-terms cancel and that
2 ( )
2 2
ds 2 d d 2
=r + sin .
dt dt dt
and !
d sin cos
p p =0
d 2 + sin2 2 + sin2
Expanding this gives the equation quoted.
(a) Using as the dependent variable, the initial condition = 0 gives c = 0 and hence
() = constant, which is the equation of the great circles through the poles.
(b) If a = b = /2, the origin may be chosen to give (a ) = 0. The equation for ()
can be simplified by noting that, for any f ()
d
( f ()) = f () + 2 f (),
d
Since there are three lengths, b, A and B, we expect the solution to depend upon only
two ratios, which we take to be A = A/b and B = B/b. Defining a third dimensionless
ratio, = c/b, gives
q
1 2 1
B = f () where f () = A cosh A 2 sinh .
x2 1/x
= e + O(x4 ), if x A.
4A
Now suppose that x A and set x = A u, where u is small and positive, and the
Taylor expansions are, to first-order in u
1 1 u 1 1 1 u 1
cosh = cosh + 2 sinh , sinh = sinh + 2 cosh
x A A A x A A A
and also
1/2 p
p
q
2 2
p u
A x = u 2A u = 2Au 1 = 2Au + O(u).
2A
These expansions give
1 p 1
f (x) = A cosh 2Au sinh + O(u), u = A x.
A A
Thus as x 0, f (x) and as x A (from below) f (x) A cosh(1/A) from below.
Also for 0 < x A, f (x) is continuous and positive. It follows that f (x) has at least
one minimum for 0 < x A and that if B > min(f ) the equation for has at least two
real roots; if B < min(f ) there are no real roots.
502 CHAPTER 15. SOLUTIONS TO EXERCISES
It is difficult to prove that there is only one minimum, but numerical results suggest
this to be the case. In the following figure we plot graphs of the scaled function
p
f (Ay) 1 1
g= = cosh 1 y 2 sinh , 0 < y 1,
A Ay Ay
10
A=0.35
8
A=5
A=1 A=0.5
6
4 A=10
0
0 0.2 0.4 0.6 0.8 1
Figure 15.10 Graphs of the function g(y) for A = 10, 5, 1, 0.5
and 0.35.
(c) If A 1, x is necessarily small and for f (x) we may use the approximation derived
in part (b),
q
q 2 2
A x
1 1/x 2 1 A x
f (x) e A A x2 so that f (x) e1/x 2+q .
2 2 x2 x 2 2
A x
1 v02
x = d + c2 (2 sin 2) , y =A+ c2 sin2
2 2g
15.5. SOLUTIONS FOR CHAPTER 5 503
where c and d are constants and the path starts at (x, y) = (0, A), where = 0 , and
ends at (b, 0), where = b . We need equations for the four unknowns c, d, 0 and b ,
in terms of A, b and v0 . The initial conditions give
1 v02
d = c2 (20 sin 20 ) and c2 sin2 0 = .
2 2g
The final end point conditions give
1 v02
b = d + c2 (2b sin 2b ) and c2 sin2 b = A + .
2 2g
From these equations we see that 0 and b are related by the equation
sin2 0 v02
2 = k2 = that is sin2 0 = k 2 sin2 b .
sin b 2Ag + v02
Then b is determined by
1 2n o
b = c (2b sin 2b ) (20 sin 20 )
2
v02 n o
= 2 (2b sin 2b ) (20 sin 20 ) , 0 = sin1 (k sin b ).
4gk 2 sin b
This gives b , which then allows 0 to be determined and from these c and d are found.
(b) In the limiting cases v02 2Ag, we expect the solution to be close to the v0 = 0
solution found in the text. In this cases k 2 v02 /(2Ag) 1, so 0 is small and, to a
first approximation is given by 0 = k sin b . Thus the above equation for b becomes
A n
3
o
b= (2b sin 2b ) + O(k ) .
2 sin2 b
The function on the right-hand side of this equation is monotonic increasing for 0
b < : for small b it behaves as 2Ab /3 and it is infinite at b = . Hence, for all
b 0 there is a unique real solution. In the limit v0 = 0 this is the same equation
determined in the text the equation immediately preceding 5.9. With this value of
b we have
A + v02 /2g
c2 = and 0 = k sin b + O(k 3 ).
sin2 b
If v02 2Ag we should expect gravity to have little effect because the initial kinetic
energy (mv02 /2) greatly exceeds the initial potential energy (mgA), so the motion will
be close to the straight line joining (0, A) to (b, 0).
In this case k 1 and we can write
1 2Ag
k2 = , = 1,
1+ v02
so is the ratio of the potential and kinetic energies. Then
sin b
sin 0 = or 0 = sin1 (sin b sin b )
1+
504 CHAPTER 15. SOLUTIONS TO EXERCISES
where
1
=1 = .
1+ 1++ 1+ 2
Now expand the equation for 0 as a Taylor series in ,
1
0 = b tan b + 2 tan3 b + O( 3 ).
2
This equation already shows that the path is approximately a straight line, because
b 0 is O(), and this short segment of the ellipse is approximated, to this order, by
a straight line. However, we shall continue with the analysis.
The equation relating b to b is obtained using the following expansion, correct to O(),
20 sin 20 = 2b 2 tan b sin (2b 2 tan b ) + O(2 )
= 2b sin 2b 4 tan b sin2 b + O(2 )
so that the equation for b becomes
v02 p
b= 1 + tan b = A tan b .
g
Thus b is the angle between the downward vertical and the straight line between the
end points.
Now put = b tan b , where is a parameter such that = 0 and b when
= 1 and 0, respectively. The x-coordinate is
1 2
x= c {(2 sin 2) (20 sin 20 )}
2
and since, to first-order,
2 sin 2 = 2b 2 tan b sin (2b 2 tan b )
= 2b sin 2b 2 tan b sin2 b
we find that x = 2c2 (1 ) tan b sin2 b . But c2 sin2 b = A(1 + )/, tan b = b/A
and = /2 so x = (1 )b. For the y-coordinate, since sin = (1 ) sin b
A
y = (1 + ) c2 (1 )2 sin2 b
A A
= (1 + ) (1 + )(1 )2 A.
As expected this gives the parametric equation of a straight line between the initial and
final points.
v(1 + v 2 ) 1 v2
Z
dv 1 2v dx
Z Z
x = or dv = dv = .
dx 1 v2 v(1 + v 2 ) v 1 + v2 x
This integrates to
v f
= Ax and since v = this gives x2 + (f )2 = 2 .
1 + v2 x
This equation represents a circle of radius with centre at (0, ).
y3
The integrand is independent of x, so the first-integral is p = c3 . Symmetry
1 + y 2
about x = 0 suggests that y(x) is even, so y (0) = 0 and then y(0) = c, where c is
positive. Rearranging this gives
Z y Z y
y 6 c3 du 1
y 2 = 1 or x = = c3 du p .
c c u 6 c6
c (u 2 c2 )(c4 + c2 u2 + u4 )
If (a) = a then A, c and a are related by A = c cosh a and, from the above integral
a
a 1 a
Z
cosh a = dv p . that is = f (a )
A 0 1 + cosh v + cosh4 v
2 A
where z
1 1
Z
f (z) = dv p .
cosh z 0 1 + cosh v + cosh4 v
2
506 CHAPTER 15. SOLUTIONS TO EXERCISES
we have f (z) / cosh z < . Thus the equation a/A = f (a ) has real solutions only
if a < A: for large separations of the ends, a > A, there are no solutions of the
Euler-Lagrange equation. Numerical evaluation of the integral gives = 0.701
Now we show, by approximating f (z), that for small z, f (z) is increasing and for large
z it is decreasing, so f (z) has at least one maximum and the equation a/A = f (a ) has
at least two real roots for small a/A.
For small v
3 1 1 1
p =q = 1 v2 + v4 + ,
2 4
1 + cosh v + cosh v 1 + sinh2 v + 1 4 2 24
3 sinh v
z 2
f (z) = z 3 + .
3 3 3
Provided cosh2 v + cosh4 v > 1, that is v > 0.722, we may expand the square root to
give Z
dv 1 1
g(z) = 1 + .
z cosh2 v 2 cosh2 v 8 cosh4
But
dv 22n 2nz
Z Z
2n
= 22n dv e2nv 1 + e2v 1 + O(e2z ) .
= e
z cosh2n v z 2n
Using the expressions for y and s we find that the surface area is
Z 2 Z 2
S = 4a 2
d (1 cos ) sin(/2) = 8a 2
d sin3 (/2)
0 0
/2
64 2
Z
= 32a2 d sin3 = a .
0 3
Similarly the volume is
Z 2 Z 2
3
V = a3 d (1 cos ) = a3 d 1 + 3 cos2 = 5 2 a3
0 0
where we have used the fact that the mean of odd powers of the cosine function is zero,
Z 2+a
dx cos2n+1 x = 0 for any real a.
a
and hence
2 2/3 1/3 5/3
S(x) = 6 a x + O(x7/3 ).
5
Similarly the volume for small is
Z 3 2 !
3 9 a3 8
V () = a d + O( ) = 2
+O(10 ) = 62/3 a1/3 x8/3 +O(x10/3 ).
0 6 8.6 8
and hence
1 32 2
S() = 4a2 3 sin(/2) 2 cos(/2) + sin(3/2) with S() = a .
3 3
But
1
1 1
Z Z
3
d sin = d (3 sin sin 3) = 3 cos + cos 3
0 4 0 4 3 0
2 3 1
= cos + cos 3
3 4 12
and
1
Z Z
d sin2 = d (1 cos 2)
0 2 0
( )
1 2 1 1 1
Z
= sin 2 d sin 2
4 2 2 0 2 0
i
1 2 1 1 1h
= sin 2 cos 2
4 2 2 4 0
1 2 1 1
= sin 2 + (1 cos 2).
4 4 8
and finally
Z h i Z
d 2 sin = 2 cos + 2 d cos
0 0 0
( )
h i Z
2
= cos + 2 sin d sin
0 0
2
= cos + 2 sin 2(1 cos ).
If = , V = a3 ( 2 /2 8/3).
15.5. SOLUTIONS FOR CHAPTER 5 509
dy sin 1
tan = = = ,
dx 1 cos tan(/2)
where we use the identities sin 2x = 2 sin x cos x, cos 2x = 12 sin2 x. Hence tan tan(/2) = 1,
so cos( + /2) = 0 which means that + /2 is an odd integer multiple of /2. But
when = 0, = /2 and when = , = 0, so + /2 = /2.
(b) If s() is the length OQ the straight line QR is of length l s() and the horizontal
and vertical distances from Q to R are (l s()) cos and (l s()) sin , respectively.
Since = /2 /2 we see that the coordinates of R are
(c) Since s() = 8a sin2 (/4), see exercise 5.3 (page 149), the length OQC is given by
putting = , LOCD = 4a. Then if l = LOCD
and
(b) The second derivative is obtained using the chain rule again,
d2 y
d 1 dY 1 1 d 1 dY
= =
dx2 dz 2z dz dx/dz 2z dz 2z dz
2
2
1 1 d Y 1 dY 1 d Y dY
= = z .
2z 2z dz 2 2z 2 dz 4z 3 dz 2 dz
2
1 d Y dY
2
Hence y + y = 0 becomes z 2 + 2 Y = 0, giving the same Euler-
4z 3 dz dz
Lagrange equation for Y (z), as before.
(b) If y = G(z) the chain rule gives y (x) = G (z)z (x) and the functional becomes
Z b
S[z] = dx G (z)2 z 2 .
a
d 2
G (z) z G (z)G (z)z 2 = 0.
dx
15.6. SOLUTIONS FOR CHAPTER 6 511
d 2
But G (z) z = G (z)2 z + 2G (z)G (z)z 2 , and hence the Euler-Lagrange equa-
dx
tion becomes z G (z)2 + G (z)G (z)z 2 = 0. But G(z) is monotonic, so G (z) 6= 0, and
the Euler-Lagrange equation is
G (z)z + G (z)z 2 = 0.
Now make the same change of variables in the original Euler-Lagrange equation y = 0.
The chain rule gives
2
d2 y d2 z
dy dz dz
= G (z) and = G
(z) + G (z) ,
dx dx dx2 dx2 dx
Since
F F y F F y F y
= = Fy G (z) and = + = Fy G (z) + Fy G (z)z
z y z z y z y z
the Euler-Lagrange equation is
d
(Fy G (z)) Fy G (z) Fy G (z)z = 0,
dx
where F and its derivatives, Fy and Fy , are evaluated at y = G(z) and y = G (z)z .
and since dx/dr = cos r sin , this gives the required result.
512 CHAPTER 15. SOLUTIONS TO EXERCISES
(b) Put w = 1/r, and this equation becomes w + w = 0, with the general solution
w = A cos + B sin . Hence the quoted result.
(c) Differentiate the definitions x = r cos , y = r sin with respect to the new indepen-
dent variable x to obtain
1 = r cos r sin and y = r sin + r cos ,
where the prime denotes differentiation with respect to x. Eliminate r by multiplying
the first equation by sin , the second by cos and subtracting, to give r2 = xy y.
Now eliminate by multiplying the first equation by cos , the second by sin and
adding, to give rr = x + yy , and hence
dr dr d (x + yy )r
= / = .
d dx dx xy y
Hence the functional becomes
Z b s Z b p
d (x + yy )2 r2 1 + y 2
S= dx r2 + 2
= dx (xy y) .
a dx (xy y) a |xy y|
Assuming that xy y > 0, that is > 0 this gives the result quoted.
(d) The stationary path in Cartesian coordinates is the straight line y = mx + c. The
boundary conditions give m = 1 + /(b a) and c = a/(b a), so
a
y = 1+ x .
ba ba
In polar coordinates the point x = y = a is r = a 2, = /4. At the other end
cos = b/r and sin = (b + )/r, so B = (b a)/a and A = (b a + )/a giving
a
r= .
(b a + ) cos (b a) sin
When = 0 this gives r = 0 unless cos = sin when r = a/ cos , that is x = a, which
is clearly incorrect. In this limit the straight line passes through the origin and the
problem is that polar coordinates are not defined here (r = 0, but is undefined).
15.6. SOLUTIONS FOR CHAPTER 6 513
r 2 + r2
1 + y (x)2 = 2.
(r cos r sin )
Since we are assuming that y (x) is bounded, from the definition of admissible functions,
r cos r sin 6= 0 on the curve. Here, for simplicity, we assume r cos r sin > 0:
in the opposite case the analysis changes slightly, but the final result is the same. The
functional becomes Z b
dx r r 2 + r2
S[r] = d ,
a d r cos r sin
where tan a = y(a)/a and tan b = y(b)/b. Using equation 6.12 this gives
Z b p
S[r] = d r r2 + r 2 .
a
F r 2 + 2r2 F rr
The derivatives of F = r r2 + r 2 are = and
= , giving
r r2 + r 2 r r2 + r 2
the Euler-Lagrange equation,
rr r 2 + 2r2
d
= 0.
d r2 + r 2 r2 + r 2
Expanding this gives
rr + r 2 rr (rr + r r ) 2r2 + r 2
= 0,
r2 + r 2 (r2 + r 2 )3/2 r2 + r 2
and this reduces to r3 r 3r2 r 2 2r4 = 0; division by r3 gives the quoted equation.
In order to simplify this, consider the first two derivatives of 1/r ,
r () d2 r () r ()2
d 1 1
= +1
and 2
= +1 + ( + 1) +2
d r r d r r r
and hence
r ()2 r+1 d2
1
r () = ( + 1) .
r d2 r
Thus if we set = 2 and substitute for r (), our equation becomes
r3 d2 d2 z
1 1
2r = 0 or + 4z = 0 where z = 2 .
2 d2 r2 d2 r
The general solution of the equation for z is z = r2 = A cos 2 + B sin 2, but since
2xy x2 y 2
sin 2 = 2 sin cos = and cos 2 = cos2 sin2 = ,
r2 r2
this becomes A(x2 y 2 ) + 2Bxy = 1.
514 CHAPTER 15. SOLUTIONS TO EXERCISES
dx dx
Z
By choosing = A(x), that is u(x) = , this functional becomes that quoted.
du A(x)
(b) The Euler-Lagrange equation for the original functional can be written in the form
y +A y /ABy /A = 0. Comparison with the given equation shows that A /A = 1/x,
hence A = 1/x, and also By /A = 4x2 y +8x2 , so that B(x, y) = 2xy 2 +8xy +g(x), where
g(x) is an arbitrary function of x; since g(x) does not contributeto the Euler-Lagrange
equation we set g(x) = 0. Then, from part (a) u = x2 /2, x = 2u and the functional
becomes
d2 y
1 2
Z
2
S[y] = du y (u) + 2y + 8y with Euler-Lagrange equation 4y = 8.
2 du2
F 1 F 2
= and = 3 ,
x (xx )2 x x x
2x
d 1 2 d 1 2
2
3 = 0, but = 3 ,
dy (xx ) x x dy (xx )2 2
x x 3 x x
n n
F X dyj F X
=2 Akj and = 2 Bkj yj , 1 k n,
yk j=1
dx yk j=1
(b) Using the standard rules for matrix multiplication we see that
n X
X n n X
X n
y Ay = yi Aij yj , y By = yi Bij yj ,
i=1 j=1 i=1 j=1
516 CHAPTER 15. SOLUTIONS TO EXERCISES
But A1 Bzk = k2 zk , so on multiplying by z j for each j and using the fact that
2
zj zk = jk we obtain aj + j aj = 0, for j = 1, 2, , n.
The second of these equations gives = L2 , where L is a constant, and hence the
first becomes L2 3 + V () = 0. In this coordinate system the Euler-Lagrange
equations for and z are uncoupled.
Hence the Euler-Lagrange equation for F1 is y = f (x, y). For F2 the equation is
Z x Z x
d dy
du f (u, y) + y du fy (u, y) = 0
dx dx c2 c2
which simplifies to
Z x Z x
d2 y
f (x, y) + du y (x)fy (u, y) + y (x) du fy (u, y) = 0,
dx2 c2 c2
so that
y x
dg g g dg
Z Z
= + y = dv f (x, v) + y du f (u, y) and hence F1 F2 = .
dx x y c1 c2 dx
Fy = y Ay + By , Fyy = Ay , Fxy = cy ex + Ax ,
cy + cy + (Ax By ) ex = 0.
Setting c = 1 and (Ax By ) = yex yields the required equation. As in the case
treated in the text, there are several possible solutions of this equation. We choose
A = 0, B = 21 y 2 ex to give the functional
1
Z
dx y 2 y 2 ex .
S[y] =
2
so that
Z b d
S2 [y1 , y2 ] = dx F x, y1 , y2 , y1 , y2 + ,
a dx
h ib
= S1 [y1 , y2 ] + (x, y1 , y2 ) .
x=a
The boundary term is independent of the path so the stationary paths of the two
functions, S1 and S2 are identical.
d
gk (x)yk = gk (x)yk gk yk , k = 1, 2,
dx
to obtain S2 .
F ex F 1
= , = .
y 2 y ex y y 2 y ex y
ex
d 1
+ = 0.
dx 2 y ex y 2 y ex y
ex y + ex y y ex
+ = 0.
4(y ex y )3/2 2 y ex y
dy dY du dY dy dY
= = ex hence ex = = Y ,
dx du dx du dx du
and hence Z 1 p
S= du Y (u) + Y (u).
0
In this case F = Y + Y and F /Y = F /Y = (Y + Y )1/2 /2, and the Euler-
Lagrange equation is
d 1 1
= 0 or Y + 3Y + 2Y = 0.
du 2 Y + Y 2 Y +Y
15.6. SOLUTIONS FOR CHAPTER 6 521
(c) If x = ln u then
dy dY dx dY d2 y dY d2 Y
= = ex and 2
= ex + e2x 2 .
dx du du du dx du du
Substituting these relations into the equation y (3ex 1)y + 2ex y = 0 gives
e2x Y (u) + ex Y (u) + 3ex 1 ex Y (u) + 2e2x Y = 0,
which reduces to Y + 3Y + 2Y = 0.
The Euler-Lagrange equation for this functional is w (u) + e2u n2 w = 0, with the
(b) There is one parameter, t, and at t = 0, y = z, because this is the initial condition.
so that 2 2 2 " 2 2 #
dy 1 dy 2 dx dy1 dy2
= .
dx dx dx dx dx
Hence
d 2 " 2 2 #
dx dx dy1 dy2
Z
G = dx
c dx dx dx dx
" 2 2 #
d
1 dy1 dy2
Z
= dx .
c 1 + g (x) dx dx
F F
y2 + y1 = 2y1 y2 2y2 y1 = constant.
y1 y2
so that
dy 1 dy1 dx dy 2 dy2 dx
= + g1 (x) , = + g2 (x) ,
dx dx dx dx dx dx
and the functional 2 2 !
d
dy 1 dy 2
Z
G= dx +
c dx dx
becomes, to first-order in ,
" 2 2 #
d
1 dy1 dy2 dy1 dy2
Z
G= dx + + 2 g1 (x) + g2 (x) .
c 1 + g3 (x) dx dx dx dx
In all cases G = G, for all admissible y1 and y2 , only if g (x) = 0, that is if g(x) is a
constant, which we set to unity in the following. The resulting first-integrals are:
F
(i) 1 = 1, 2 = = 0, hence =c giving y1 = c,
y1
F
(ii) 2 = 1, 1 = = 0, hence =c giving y2 = c,
y2
F F
(iii) = 1, 1 = 2 = 0, hence F y1 y2 = c giving y1 2 + y2 2 = c.
y1 y2
where c is a constant.
In this example the first-integral arising from the invariance with respect to trans-
lations in x can also be derived from the two first-integrals due to the invariance under
the translations in y1 and y2 . Thus not all symmetries lead to new first-integrals.
15.7. SOLUTIONS FOR CHAPTER 7 525
d2 y
d 2 dy 2 dy
x + x2 y 5 = 0 which expands to 2
+ + y 5 = 0.
dx dx dx x dx
+2yy 1 + x2 y 4
= 0.
That is
2 5
2x2 y + xy = 0.
y + y +y
x
The solutions of this equation are the solutions of 2x2 y + xy = 0, that is y = A/ x, A
being a constant, and any solution of the original Euler-Lagrange equation.
and
n
" n
#d n
d d
F dk F d F
Z X X Z X
dx = k dx k
c yk dx yk c dx yk
k=1 k=1 c k=1
" n
#d n
d
F F
X Z X
= k dx k
yk c yk
k=1 c k=1
where we have used the Euler-Lagrange equation to derive the last line.
Hence, equation 15.3, becomes
" n
! n
#d
X F X F
0 = F y + k
yk k yk
k=1 k=1 c
n
!!
d
F d F
Z X
+ dx F y .
c x dx yk k
k=1
The term in brackets inside the integral is, on expanding the total derivative
n
F F X F F F F
yk + yk yk yk = 0,
x x yk yk yk yk
k=1
(b) The integrand is homogeneous of degree zero in x, so the transformation to the new
independent variable x, where x = x, with a constant leaves the integral invariant.
To use Noethers theorem we set = 1 + , so (x) = x and = 0, see equation 7.13.
Then, equation 7.15 becomes
F
F y x = c where F = xyy 2 ,
y
which gives the required result.
528 CHAPTER 15. SOLUTIONS TO EXERCISES
1
yFy + x (F y Fy ) = constant where F = x3 y 2 y 2 .
2
Substituting for F gives
x3 y 3 y + x4 y 2 y 2 = c = constant.
A4 x4+2 + A4 2 x4+2 = c,
d
x3 y 2 y 2x3 yy 2 = 0.
2
dx
Differentiation gives
1 3
x3 y 2 y + x3 yy 2 + 3x2 y 2 y = 0 or y + y 2 + y = 0.
y x
d 3
(yy ) + (yy ) = 0,
dx x
which integrates directly to yy = Ax3 , where A is a constant. A further integration
gives
1 2 A B A
y = 2 + that is y 2 + 2 = B,
2 2x 2 x
where B is a constant. This gives the general solution of the Euler-Lagrange equation.
Further, if yy = Ax3 the first-integral reduces to Ay 2 + A2 x2 = c.
15.7. SOLUTIONS FOR CHAPTER 7 529
dy dy 1
(d) If x = ua then = and the functional becomes
dx du aua1
2
1 dy
Z
S[y] = du u2a+1 y 2 .
a du
Z
2
With a = 1/2 this simplifies to the equivalent functional S[y] = du (yy ) . The
integrand, F = (yy )2 , is independent of u, so the first-integral is y Fy F = (yy )2 = c2 ,
where c is a constant. Hence yy = c, which integrates to 12 y 2 = cu+b for some constant
b: because u = x2 this is the same solution as derived in part (c). On the other hand,
the function y = Ax1/2 from part (b) becomes Au1/4 which is not a solution of the
equation yy (u) = c.
But the integrand is zero and the end points c and d are arbitrary and hence
F d F F
y y F = constant.
y dx y y
15.8. SOLUTIONS FOR CHAPTER 8 531
Hence 2 is positive definite and the stationary point is a minimum. The eigenvalues
of H are (0.689, 3.58, 9.73), which are all positive.
and D1 = fxx . If D2 > 0 then fxx fyy > 0 and fxx and fyy must have the same sign.
For a minimum the quadratic form must be positive definite, which gives the first
set of conditions. For a maximum the quadratic form must be negative definite, which
gives the second set of conditions.
If D2 < 0 we proceed as follows. Suppose the stationary point is at (a, b), define
u = x a, v = y b, so the quadratic form is 2 = u2 fxx + 2uvfxy + v 2 fyy and 2 = 0
when
u fxy D2
=
v fxx
Hence when D2 < 0 there are two lines through the stationary point, given by u =
fxy D2 v on which the quadratic form is zero and it has different signs either
side of these lines. Near the stationary point the cubic terms in the expansion of (x, y)
can be neglected, so it behaves like 2 and the stationary point is a saddle.
15.8. SOLUTIONS FOR CHAPTER 8 533
The origin. Near here f 2(x + 8y)2 so f (0, 0) > f (x, y) except along the line
x + 8y = 0, on which f (8y, y) = 64 65y 4 , so this stationary point is a saddle. Also,
H(0, 0) = 0, and the stationary point is degenerate.
The point ( 5, 5/2). The second derivatives have the value fxx = 56, fxy = 32
and fyy = 704 and the Hessian determinant is
56 32
H= giving det(H) = D2 = 38400, D1 = 56,
32 704
so this stationary is a minimum.
The point ( 5, 5/2). The second derivatives and the Hessian have the same values
as at the point ( 5, 5/2) so this stationary is also a minimum.
Now use the Schwarz inequality, page 41, with bi = 1 for all i, to see that det(H) 0
(with equality only if all the xi values are the same) and since aa = 2N > 0 we see
that the stationary point is a minimum.
If P (x) 6= 0 for a x b, 2 has the same sign as P for all h, and the stationary
path is either a minimum (P (x) > 0)or a maximum (P (x) < 0), that is there are no
conjugate points.
Alternatively, Jacobis equation is (P u ) = 0, with solution
Z x
dv
u(x) = P (a) 6= 0 for x > a,
a P (v)
provided P (x) does not change sign, at which point the integral may cease to exist.
But, by definition the left-hand side is zero and since u(a) = u(b) = 0, 2 [y, u] = 0.
Hence, from equation 8.7, we see that for h = u, S = O(3 ) and the nature of the
stationary point cannot be determined by this order of expansion.
dw u u 2 u
= P + 2 P P ,
dx u u u
which gives
u u
d du
P + P Q = 0 and hence P Qu = 0.
u u dx dx
which expands to
d dh d
Fy y Fyy Fyy h = O().
dx dx dx
Take the limit 0 to obtain the required result.
and Jacobis equation becomes u + u = 0 with = Q(a)/P (a). The Taylor expan-
sion of the solution is
1
u(x) = u(a) + (x a)u (a) + (x a)2 u (a) + O((x a)3 ).
2
But by definition, u(a) = 0 and u (a) = 1, so from the Jacobis equation u (a) = 0
and the approximate solution is u(x) = (x a) + O((x a)3 ), which is positive for
sufficiently small b a.
and
z cos
Fzz = = 3 3 .
2z 3/2 1+z 2 2c sin
These expressions can be used to find Q,
z
3 p 2
1d
Q = 1+z +
4z 5/2 2 dx z 3/2 1 + z 2
3 1 d cos 1
= +
4c5 sin6 2c3 d sin3 2c2 sin2
3 cos2
3 1 1 1
= 5 6 + 5 2 2 4 = 5 4 .
4c sin 4c sin sin sin 2c sin
this becomes
d2 u
sin2 2u = 0.
d2
Substituting the given function into this equation shows that the general solution is
A cos
u() = +B 1 ,
tan sin
cos
But, for > 0, 2 > sin 2 giving > sin cos and > , showing that for
sin2 sin
0 < < , u() is an increasing function and there are no conjugate points.
so that
b
y f (x)
Z
S[y + h] S[y] dx p h.
a 1 + y 2
538 CHAPTER 15. SOLUTIONS TO EXERCISES
and hence Z b h i
S[y + h] S[y] d dx h (x) = d h(b) h(a) = 0.
a
Z 1
dx P (x)h 2 Q(x)h2 .
2 [y, h] =
0
Z 1
(d) Since S[y + h] = exp dx F (x, y + h ) on differentiation with respect to
0
we obtain
1
d
Z
S[y + h] = S[y + h] dx h Fy (x, y + h ).
d 0
Putting = 0 gives the Gateaux differential,
Z 1
S[y, h] = S[y] dx h Fy (x, y ).
0
Putting = 0 gives
(Z
1 2 Z 1
)
2
2 S[y, h] = S[y] dx h Fy (x, y ) + dx h Fy y (x, y )
0 0
Hence Z 1
2 S[y, h] = S[y] dx h 2 Fy y (x, y ).
0
(b) The integrand of the second variation is given by the second-order term in the
Taylor expansion of F (x, y + h, y + h , y + h ). This is
1
Fyy h2 + Fy y h 2 + Fy y h 2 + Fyy hh + Fy y h h + Fyy hh .
F2 =
2
Hence
Z b
dx Fyy h2 + Fy y h 2 + Fy y h 2
2 [y, h] =
a
Z b
+2 dx (Fyy hh + Fy y h h + Fyy hh ) .
a
where
dFyy d2 Fyy
P0 (x) = Fyy +
dx dx2
dFy y
P1 (x) = Fy y 2Fyy and P2 (x) = Fy y .
dx
where c is a constant. If c = 0 then either y(x) = 0 (which does not satisfy the boundary
condition at x = 1) or y = x.
Thus the stationary path is y = x and on this path S[y] = 0, so it gives a global
minimum.
Further
d
P (x) = Fy y = 4y 2 (3y 2 1) = 8x2 and Q(x) = 2(y 2 1)2 8 yy (y 2 1) = 0
dx
Z 1
giving 2 [y, h] = 8 dx x2 h 2 > 0.
0
(b) If y(1) = A 6= 1 this becomes a much more difficult problem, since y = x is not a
solution and near x = 0 the stationary path is quite different from this, no matter how
small A 1.
The first-integral, y 2 (y 2 1)(3y 2 + 1) = c, shows that if y(0) = 0 then either,
(a) c = 0, with y(x) = 0 or y(x) = x, or
(b) c 6= 0 and limx0 y 2 y 4 = c/3.
The first possibility gives solutions that do not fit the boundary conditions.
Consider the second case. Near x = 0 the equation is approximated by y 2 y 4 = c1 , and
this has the solution y = c2 x2/3 , with c1 = c62 (2/3)4 .
Now write the first-integral as the quadratic in y 2 ,
c 1 p
4 2
3y 2y 1 + 2 = 0 with solution y 2 = y 4y 2 + 3c .
y 3y
If c = 0 this reduces to y 2 = (y 2y)/3y, and the upper sign gives the previous
equation, y 2 = 1.
15.8. SOLUTIONS FOR CHAPTER 8 541
If c 6= 0, then since y(0) = 0, we must have c > 0 and only the upper sign gives a real
solution; hence the required equation for y is
Z y s
2 1 p
2
3y
y = y + 4y + 3c giving x(y) = dy p . (15.5)
3y 0 y + 4y 2 + 3c
1
c=0
0.8 c=1
c=0.3
0.6
c=0.1
0.4 c=0.01
0.2
00 0.5 1 1.5 2
Figure 15.11 Typical graphs of f (y), defined in equa-
tion 15.6, for c = 0.1, 0.3 and 1.
Equation 15.6 shows that the value of A needs to be such that the area under f (y) for
0 < y < A is unity. It follows from the figure that if c > 0 we must have A > 1 and
that no solutions exists if A < 1.
Then from theorem 8.6 (page 223), with 2 S[y, h] = 2S[h], we see that S[h] > 0 if the
interval (0, a) does not contain a conjugate point, that is a < . In this case, put
= k/a to give
a a
k2
Z Z
dx f (x)2 > dx f (x)2 for any k < .
0 a2 0
It follows that S[y + h] < S[y] for all h, proving the result.
Hence the second variation is positive if b > 1 or b < 0 and the stationary path is a
minimum of S[y]. If 0 < b < 1, P < 0 and the stationary path is a maximum of S[y].
Consider the variation h(x) = sin(nx/a), so h (x) = O(n) and choose n > 1, to
see that this not a strong minimum.
15.8. SOLUTIONS FOR CHAPTER 8 543
u 2b2 u d2 u
d du
+ =0 which expands to (1bx)2 +2b(1bx) +2b2 u = 0.
dx (1 bx)2 (1 bx)4 dx 2 dx
Now use the identity xy = d(xy)/dt yx to obtain the alternative expression for the
area Z tb h itb
A= dt xy + xy .
ta ta
15.9. SOLUTIONS FOR CHAPTER 9 545
R tb
Adding these two expressions gives 2A = ta
dt xy xy + bB aA.
y
The areas of the triangles OA A and OB B are, respectively,
1 1
2 aA and 2 bB. Since the area of OBB is the sum of the
areas of OAA , OAB and A ABB , we obtain B
1 tB
Z xB
1 1
Z
dt (xy xy) + aA + dx y(x) = bB, A
2 tA 2 xA 2
x
which is the relation derived in the question. O A B
b
1 b
1 dt dy d dx d
Z Z
S= d x y = d x()y () x ()y() .
2 a d d dt d dt 2 a
2 2
1 a2
Z Z
Acy = d (x y y x) = d (1 cos )2 ( sin ) sin
2 0 2 0
2
a2
Z
= d (2 2 cos sin ) = 3a2 .
2 0
Note that we are free to choose a convenient interval for t. In this case = y 2 /x and
the equations for x and y are, respectively
d y 2
d y
= 0, 2 = 0.
dt x2 dt x
Integrating these gives z = y/x =constant, in both cases.
= xx + yy + constant.
so
s2 = r2 sin2 2 + r2 cos2 2 + z 2 = r2 2 + z 2
giving E = r2 , F = 0, G = 1 and 2 = r2 2 + z 2 .
(b) Equations 9.18 and 9.19, with u = and v = z, are respectively
2
d r = 0 and d q z = 0.
q
dt dt
r2 2 + z 2 r2 2 + z 2
s s
z(s) = z1 + (z2 z1 ) , (s) = 1 + (2 1 ) , 0 s S,
S S
where S is the length of the curve. Setting t = s/S gives the previous solution, but any
other parameter will suffice, for instance s = S sin(t/2) with 0 t 1.
dx y x + y x y x(xx + y y)
= p 2 .
dt x 2
x + y 2 (x + y 2 )3/2
dy y 2 + y y y y(xx + y y) p 2
= p 2 x + y 2 ,
dt y x2 + y 2 (x + y 2 )3/2
y y=x y=x a
o
45 x
A
O B
Figure 15.12 Diagram showing the first right angled triangle, ABC.
Since the angle CAB is /4 and the length of the diagonal is a the
triangle is isosceles and the length of the shorter sides is a/ 2.
It follows that the apex of the N th triangle has coordinates x = y = N a/2; if this is at
(1, 1) then N a = 2.
On the segments parallel to the x-axis the functional has value 0 (because y = 0),
but on each of the other segments its value is S1 = a/2. So the whole integral is
S = N S1 = 1.
On making a arbitrarily small, and increasing N so that N a = 2, we see that all
points on the zig-zag path are at most a distance a/2 from the straight line y = x.
h i
But sinh1 z = ln(z + 1 + z 2 ), so S = 2a 2 + ln(1 + 2) .
y
O
O P x
Figure 15.13
(b) Consider the loop surrounding the origin. The values of when x = 0 are given by
= k sin and if (0, ) is the real root of this equation the required values of are
0, : but the points corresponding to are identical. Since y() = a(1 k cos ) >
y(0) = a(1 k), = 0 corresponds to the bottom of the loop and = the top. Hence
the area of the loop is given by
Z Z
2
d 1 + k 2 2k cos k sin
A = d (yx y x) = a
0 0
= a2 (1 + k 2 ) 2k sin + k( cos sin ) .
A = a2 k 2 2 + k cos .
(c) The loop surrounding the origin intersects the positive x-axis when y = 0, that is
cos = 1/k: the root required is the negative root in the interval (, 0) which we
denote by , so 0 < < . At this point x () = 0 (since x () = y()) so the curve
is perpendicular to the x-axis. It follows, because of symmetry, that the maximum
width of the loop is 2x0 where x0 = a(k sin ). Hence adjacent loops intersect when
2x0 = 2a, that is when k satisfies the equation
p
= k 2 1 cos1 (1/k).
= a ln(cosh v).
The shape of the curve defined by these equations is shown in figure 15.14.
y/a
x/a
Figure 15.14 Graph showing the curve defined by the parametric equations
x = a(utanh u), y = a/ cosh u. The cusp on the y-axis, at y = a, corresponds
to u = 0 and near here y/a = (1 (3x/a)2/3 .
552 CHAPTER 15. SOLUTIONS TO EXERCISES
(b) This equation can be rearranged to give y =constant, with general solution y = mx + c.
The boundary conditions give c = 0 and m = Y /X.
(c) Equation 15.8, involving x and y, can be rearranged to given an equation involving
only one derivative, y (x) because it is a homogeneous function of degree 1 in x and y,
because x is homogeneous with degree zero, see exercise 1.25(c) (page 28). This is in
contrast to the example treated in the following exercise.
d d
(2x + y) = 0 and (x + 2y) = 0,
dt dt
which can both be integrated once to give 2x + y = A and x + 2y = C. Notice that
neither of these can, alone, be rearranged to given an equation for y (x), as was possible
in the previous exercise. Integrate again to obtain 2x + y = At + B, x + 2y = Ct + D.
(b) The boundary condition at t = 0 gives B = D = 0 and that at t = 1 gives x = Xt,
y = Y t.
(c) In this case both equations are needed to find a solution because the equations are
not homogeneous functions of x and y.
and the Euler-Lagrange equation has the solution y = 0. The solution passing through
(0, A) is therefore y = mx + A, for some m to be determined. The natural boundary
condition, equation 10.7, is
y
Fy = p = 0,
1 + y 2
that is, y = m = 0, so that y = A, which defines a straight line parallel to the x-axis.
The same reasoning as used in the text gives the required result.
dz dz . dx 1 dx 4b
z = = = and = sin2 ,
dx d d tan d
15.10. SOLUTIONS FOR CHAPTER 10 555
and the time is given by the functional, see equation 5.6 (page 150),
Z /2 r s Z /2 s
1 dx 1 + z 2 b b
T = d =2 d = .
2g 0 d z g 0 g
Integrating this and assuming that v(x) 0 and that y (x) 0 gives
1 x
Z
y(x) = du v(u).
c 0
Using the subset of variations with h(a) = h(b) = 0 and the fundamental lemma of the
Calculus of Variations we see that S[y] is stationary only on those paths satisfying the
equations y + y = 0. On these paths the Gateaux differential is
S[y, h] = 2h(b) y (b) + gb y(b) 2h(a) y (a) + ga y(a)
and this is zero for all variations only if ga y(a) + y (a) = 0 and gb y(b) + y (b) = 0.
g 2 g gL gL2
L + 6AL + 2B = 0 and L + 6A = 0 = A= , B= ,
2 6 4
g 2 2
x x 4Lx + 6L2 .
giving the solution y(x) =
24
Solution for Exercise 10.11
(a) The Gateaux differential is given in equation 10.14 and since h(a) = h(b) = 0 this
reduces to
b Z b 2
F d F d F F
S[y, h] = h + dx + h.
y a a dx2 y dx y y
Using the subset of varied paths for which h (a) = h (b) = 0, we see that y(x) satisfies
the Euler-Lagrange equation
d2
F d F F
2
+ = 0, y(a) = A, y(b) = B.
dx y dx y y
The other boundary conditions are obtained by considering those paths for which
h (a) = 0 and those for which h (b) = 0, which gives Fy |a = Fy |b = 0.
p
Integration gives x = c c2 y 2 , or y 2 + (x c)2 = c2 . This is the equation of a
circle, radius c with centre at (c, 0), and is also the solution for y < 0.
The transversality conditions shows that the stationary path must be perpendicular
to the line y = x a. But the only lines that are perpendicular to a circle are its
diameters, and hence this line must pass through the centre of the circle, that is c = a,
giving the stationary path y 2 + (x a)2 = a2 .
The same result follows algebraically, but this derivation is more difficult. With
= y x + a, the boundary condition 10.26 becomes
y 1
Fy + (y Fy F ) = 0 that is p =0
y 1 + y 2 c
p
but y 1 + y 2 = c therefore y = 1.
If the intersection is at x = v, the equation
for y , with y = 1 gives y = c/ 2and
from the equation for the circle v = c(1 1/ 2); the required root is v = c(1 + 1/ 2),
the other root corresponding to y = 1. Now substitute these coordinates into the
straight line equation, y = x a, to see that c = a.
1 2
x= c (2 sin 2) , y = A c2 sin2 , 0 b .
2
p
The integrand of the functional may be taken to be F = 1 + y 2 / A y, so the
transversality condition 10.26 gives, since = x/a + y/b 1,
y
1 1 dy a
p =0 that is = .
A y 1 + y 2 a b dx b
dy 1 a
Hence the equation for b is = = . Finally, at = b the end of the
dx tan b b
cycloid is on the line x/a + y/b = 1 that is
c2 1
2b sin 2b + A c2 sin2 b = 1.
2a b
558 CHAPTER 15. SOLUTIONS TO EXERCISES
The equation for can be written in the form cos( ) = AB/, and this has real
roots only if |AB| . If |AB| > the equation has only complex roots and the
ellipse and the line do not intersect.
Z p
(b) The functional is S[y] = dx 1 + y 2 where the pairs of coordinates (u, v), on
u
the ellipse, and (, ), on the line, satisfy the equations
u2 v2
e = + 1 and l = + 1
a2 b2 A B
and also v = y(u) and = y().
The general solution of the Euler-Lagrange equation is y = mx + c, where m and c are
constants, which are chosen to satisfy the boundary conditions.
For the boundary conditions, equation 10.29, we require
y m 1
Fy = p = and y Fy F = .
1+y 2 1 + m2 1 + m2
The boundary conditions on the ellipse give
u m v 1 mu v
2 = 0 and hence = 2. (15.9)
a 2 1 + m2 b 1 + m2 a2 b
The boundary conditions on the straight line give
1 m 1 1 A
=0 and hence m= , (15.10)
A 1 + m2 B 1 + m2 B
which is the condition for the stationary path to be perpendicular to the straight line.
Also these points lie on the boundary curves,
u2 v2
2
+ 2 =1 and v = mu + c (15.11)
a b
15.10. SOLUTIONS FOR CHAPTER 10 559
and
+ = 1 and = m + c. (15.12)
A B
Thus we have six equations for the six parameters (u, v), (, ) and (m, c) that we need
to find.
The distance along the stationary path is
Z p p
S[y] = dx 1 + m2 = ( u) 1 + m2 . (15.13)
u
Now m = A/B is given directly, equation 15.10, and subtracting equations 15.11
from 15.12 gives v = ( u)m: rearranging this and substituting for v from 15.9
gives 2
b
= m + mu 1 .
a2
Substitute this into the first of equation 15.12 gives
2
2 u
2 2 2 b 2 2 2
+ 2 = AB 2 .
B +A +A u 1 = AB or ( u) B + A
a2 a
Using
equations 15.11 and 15.9 we obtain (since u > 0) u = aB/ and since 1 + m2 =
A2 + B 2 /B we have
A2 + B 2 AB
S[y] = ( u) = .
B A2 + B 2
The solution of the associated Euler-Lagrange equation satisfing the boundary con-
dition at x = 0 is y = mx for some constant m. The boundary condition on the
parabola gives, on using equation 10.26, a2 y = 2y and hence v = a2 /2. At this point
y = ma2 /2, and since this point lies on the parabolawe obtain m2 = 4/a2 2. Thus
there are stationary paths if a < 2 and none if a > 2. Note that the stationary path
through (1, 0) is not given by this method.
Using the same arguments as in the text we see that y1 and y2 satisfy the Euler-Lagrange
equations
y1 = 0, y1 (0) = A, 0 x , and y2 = 0, x L,
the first to make the path stationary and the second to ensure that the path is continous.
In addition, the natural boundary condition at x = L, see equation 10.7, gives Fy =
y (L) = 0, that is = 0.
The Euler-Lagrange equations have the following solutions
where , and are constants, and the natural boundary condition at x = L, y (L) = 0,
gives = 0.
Continuity at x = gives = + A and the other condition at x = gives
+ 2C = 0, and these two equations give
2CA A
= , and = .
1 + 2C 1 + 2C
Hence the solution is
1 + 2C( x)
A
, 0 x ,
y(x) = 1 + 2C
A
, x L.
1 + 2C
Note that if A = 0, y(x) = 0.
15.10. SOLUTIONS FOR CHAPTER 10 561
Using the subset of variations for which = 0 gives equations 10.50 and 10.51, for y1 (x)
and y2 (x). On these paths S reduces to equation 10.49.
The terms O() is, by definition, zero on a stationary path and since
Fy y = 2 1 6y + 6y 2
Hence for all allowed h and 0 < || 1, S[y + h] S[y] < 0, so this stationary path is
a local maximum of S[y].
interesting solutions; denote these by x1 (y) and x2 (y). These must also be solutions of
the quartic, f (x) = f (y). But the root structure of a quartic and a cubic is different; for
instance roots of the quartic will coalesce are different values of y than for the cubic, so
it is unlikely that there is a range of y for which x1 (y) and x2 (y) satisfy both equations.
There may, however, be accidental coincidences; we now show that there are none.
Consider the differences,
f (x) f (y) = (x y)F (x, y) and g(x) g(y) = (x y)G(x, y)
where F (x, y) and G(x, y) are respectively symmetric cubic and quadratic functions of
x and y. The solution x = y is of no interest, so we require the solutions of the equations
F (x, y) = 3(x3 + y 3 ) + 3x2 y + 3xy 2 4(x2 + y 2 + xy) + x + y = 0,
G(x, y) = 2(x2 + y 2 ) + 2xy 3(x + y) + 1 = 0.
These equations are more conveniently expressed in terms of the variables u = x + y
and v = xy, (so when x = y, u2 = 4v)
F (u, v) = 3u3 4u2 6uv + u + 4v = 0,
G(u, v) = 1 3u + 2u2 2v = 0,
which gives
1
1 3u + 2u2
v=
2
and
F = (1 u)(3u2 6u + 2) = 0.
If u = 1 then v = 0 and (x, y) = (0, 1) and (1, 0), which are the solutions found in the
text.
If 3u2 6u + 2 = 0, u = 1 3/3 and v = 1/3 3/6, so u2 = 4v giving x = y.
Hence there are no real solutions other than those found in the text.
For small |2 |
1
tan 2 = 2 + 23 + O(25 )
3
1 1 2
= d 1 + d2 + d3 + O(d5 ) = d 1 + d2 + O(d5 ),
3 3 3
and hence
1 1 1 2 2 4
p2 = = = 1 d + O(d ) .
tan 2 d(1 + 32 d2 + O(d4 )) d 3
564 CHAPTER 15. SOLUTIONS TO EXERCISES
Integrating by parts and using the fact that h(b) = 0 then gives the required result,
b
dFy
Z
S2 [y, h] = ak kF a, y (a) h(a)Fy a, y (a)
dx h .
a dx
Since this equation is true for all in a neighbourhood of the origin it follows that
ky (a) + h(a) = 0, as in equation 10.76, and k 2 y (a) + 2kh (a) = 0. Thus the second
derivative becomes the simple expression
b
d2 S2 h Z
2
i
= k 1 Fx a, y
(a) + dx h 2 Fy y (x, y ).
d2 a
But,
x 1 2x(3y 2 1)
F = so that Fx = and Fy y = ,
1 + y 2 1 + y 2 (1 + y 2 )3
and since y (a) = 1, see equation 10.81 the second derivative becomes
b
d2 S2 1 x(3y 2 1) 2
Z
= k2 + 2 dx h .
d2 2 a (1 + y 2 )3
It follows that provided 3y 2 > 1 the second variation is positive for all nonzero k and
h(x), and that the stationary path is a weak local minimum.
In each integral of the sum put (n+1/2)u = p1+w, 0 w 1, and (n+1/2)u = n+w
in the last integral to write this in the form
n Z 1 Z 1/2
b2 X p1+w b2 n+w
S1 [z] = dw 2 + dw .
(n + 1/2)2 p=1 0 2
1 + B sin w (n + 1/2)2
0 1 + B 2 sin2 w
But
1 1
p1+w 1 p
Z Z
dw p dw = , p = 1, 2, , n
0 1 + B 2 sin2 w 0 1+ B2 2
sin w 1 + B2
and
1/2 1
n+w 1 n+1
Z Z
dw (n + 1) dw =
0 1 + B 2 sin2 w 0 1+ B2 2
sin w 1 + B2
so that
n+1
b2 X b2
S1 [z] p= (1 + O(1/n)) .
(n + 1/2)2 1 + B 2 p=1 2 1 + B2
But B = O(n) so S1 [z] = O(1/n) for large n. Hence, given any number > 0, an n can
be found such that S1 [z] < .
(b) Since max(z (x) = O(n) the derivative is not bounded and z(x) satisfies the D0
norm.
15.10. SOLUTIONS FOR CHAPTER 10 567
f (x, y)y
p = 0 at x = b.
1 + y 2
Since f (x, y) 6= 0, this means that y (b) = 0 and that the stationary path is perpendic-
ular to the line x = b.
Using the subset of variations for which h(v) = 0, we see that y(x) satisfies the Euler-
Lagrange equation. Then the boundary terms shows that the boundary condition at
x = v is
Fy (v, y(v), y (v)) + G (y(v)) = 0.
(b) If the right end of the path satisfies (v, y(v)) = 0 and the varied path is y + h,
and ends at v + for some , the same analysis that leads to equation 10.19 gives
x + y (v)y + h(v)y = 0.
It is convenient to write
y(v + ) + h(v + ) = y(v) + (y (v) + h(v)) + O(2 )
then differentiate with respect to , and then set = 0 to obtain the Gateaux differential
Z v
S = G (y(v)) y (v) + h(v) + F (v, y(v), v (v)) + dx (hFy + h Fy ) .
a
y Boundary curve
Stationary path
r
r
c
x
c
Figure 15.15
so that
f (x, y)
y Fy F = p exp( tan1 y )(y 1).
1 + y 2
The boundary condition 10.26 then gives
f (x, y)
p {( + y )x + (y 1)y } = 0 at x = v.
1 + y 2
If the gradient of and the stationary path at x = v are respectively tan C and tan ,
so x = y tan C and y = tan , this boundary condition becomes
which rearranges to (tan tan c ) = 1 + tan tan c , giving the required result.
and since on an admissible path h (1) can take any value we must have y (1) = 0. The
Euler-Lagrange equation is therefore
The general solution of this equation is y(x) = x5 /240 + Ax3 + Bx2 + Cx + D. The
boundary conditions y(0) = y (0) = 0 give D = C = 0, and the other two conditions
give the equations
1 1
0= +A+B and 0 = + 6A + 2B
240 12
with solution A = 3/160 and B = 7/480 giving y(x) = x2 (1 x)(2x2 + 2x 7)/480.
so that
L
dE
Z
= M gh(L) + dx h (y + h ) gh
d 0
d4 y Mg
= (x)g, y(0) = y (0) = y (L) = 0, y (3) (L) = .
dx4
If is independent of x the general solution of this equation that satisfies the boundary
conditions at x = 0 is
g 4
y(x) = x + Ax3 + Bx2
24
and the constants A and B are determined by the two conditions at x = L; since
g 2 g
y (2) (x) = x + 6Ax + 2B and y (3) (x) = x + 6A
2
these give the equations
g Lg L
A= (M + L) and B = M+ .
6 2 2
Hence
g 4 x3 Lx2
L
y(x) = x (M + L)g + M+ g,
24 6 2 2
g 2 2 Mg 2
= x (x 4xL + 6L2 ) + x (3L x).
24 6
Since Z h i Z
dx y h = y h y h + dx y (4) h
On collecting relevant terms together and using the fact that h(x) is continuous at x =
and that h1 (0) = h2 (L) = 0 this becomes
y2 y1 M g h(x) + (y1 h1 y2 h2 )
E =
x= x=
h i h i
y1 h1 + y 2 h2
x=0 x=L
Z Z L
(4) (4)
+ dx y1 g h1 + dx y2 g h2 .
0
Now choose the subset of variations that make all the boundary terms zero to see that
y1 and y2 satisfy the Euler-Lagrange equations
(4)
y1 = g, y1 (0) = 0, 0 x ,
(4)
y2 = g, y2 (L) = 0, x L.
Also, since h1 (0) 6= 0 and h2 (L) 6= 0 the natural boundary conditions at x = 0 and L
are
y1 (0) = y2 (L) = 0.
Finally choose the subset of variations for which h (x) is continuous to see that y1 () =
y2 (), that is y (x) is continuous at x = .
For the sake of completeness we now show how these conditions can be used to
find the solution when is independent of x. This analysis was not requested in the
question.
The two solutions of the Euler-Lagrange equation that fit the boundary conditions
at x = 0 and L are
g 4
y1 (x) = x + a1 x3 + b1 x,
24
g
y2 (x) = (L x)4 + a2 (L x)3 + b2 (L x),
24
so there are four further constants to be determined by the conditions just derived. The
conditions of the second and third derivatives at x = give
gL
M g = gL 6(a1 + a2 ) and 6a2 L = 6(a1 + a2 ) + (2 L).
2
572 CHAPTER 15. SOLUTIONS TO EXERCISES
Hence
Mg Lg M g Lg
a1 = ( L) and a2 =
6L 12 6L 12
and
gL3 Mg gL3 Mg
b1 = + (L )(2L ) and b2 = + (L2 2 ).
24 6L 24 6L
The second equation gives m2 = m1 and the first shows that the only solution is
m1 = m2 . Hence the functional has no corners.
Fy = 4y y 2 3 and F y Fy = 3y 2 2 y 2
The first equation gives m2 = m1 (we ignore the solution m2 = m1 ), and then the
second equation gives
m1 = m, m2 = m with m = 3.
1.5
Cycloid
1
Boundary circle
0.5
r
0
L
0 0.5 1 1.5 2 2.5 3 R
x2 4b a2 2x2
dA 4b p 2
= a x2 = ,
dx a a2 x2 a a2 x2
so that A (x) = 0 when x = a/ 2 (since x > 0). If x = a/ 2 r2 , A 2 , so that
4b a a2
A(x) has a local maximum at this point, with value A = a2 = 2ab.
a 2 2
This is a quadratic equation in x and since the coefficient of x is positive it has a single
minimum, which is most easily seen by writing it in the form
2
a2 + b 2 (Aa + b(b B))2
a
D2 = 2
x 2 2
(Aa + b(b B)) + A2 + (b B)2 .
a a +b a2 + b 2
Since sin A 6= 0 and sin B 6= 0 (because 0 < A, B < ) we have sin(2A + B) = 0 and
sin(A+2B) = 0, that is 2A+B = n and A+2B = m, with n and m positive integers,
both smaller than 3. Hence 3A = (2nm) and 3B = (2mn), and 3C = (3nm).
The bounds on A and B give n = m = 1 and hence A = B = C = /3.
which gives
2 (ab Ab Ba)2
2 1 1
D = + 2 = .
4 a2 b a2 + b 2
d2 |d|
D2 = 2 (a2 + b2 + c2 ) = that is D= .
a2 + b 2 + c2 a2 + b 2 + c2
and have the solution x = z and y = 1 2z. Hence the expression for f (x, y, z) becomes
If a + 4b + c < 0 this stationary point is a maximum. Using the expressions for x(z)
and y(z) the results quoted in equation 11.19 are obtained.
Fx = yz (y + 2z) = 0,
Fy = xz (x + 4z) = 0,
Fz = xy (4y + 2x) = 0.
The second two equations can be rearranged to give t(x 4) = and s(x 4) = 2,
so that s = 2t. Also t = /(x 4), so the first equation gives x = 8; hence t = 1/4
and s = 1/2, giving 2y = x and 4z = x.
578 CHAPTER 15. SOLUTIONS TO EXERCISES
2r2 2r3
2
F = 2rh + r h + .
sin 3 tan
2r3 2r2 r
cos
F = 2r2 2 + = cos =0
sin 3 sin2 sin2 3
where we have used the identity tan x = 2t/(1 t2 ). Adding unity to each side of the
last equation gives the second of equations 11.5.
15.11. SOLUTIONS FOR CHAPTER 11 579
Fy = 2y 1 (xzw)2 = y = 0 or (xzw)2 = 1,
Fz = 2z 1 (xyw)2 = z = 0 or (xyw)2 = 1,
Fw = 2w 1 (xyz)2 = w = 0 or (xyz)2 = 1.
If x = 0 then the three remaining equations for have no solution: hence we discard
the solution x = y = z = w = 0. The equation (yzw)2 = 1 gives = x2 (on
multiplying by x2 and using the constraint equation). Similarly, = y 2 = z 2 = w2 ,
hence (xyzw)2 = 4 = 1 and = 1 ( = 1 is not allowed); so there are 16 stationary
points, x = 1, y = 1, z = 1 and w = 1, all of which give f = 4.
Fx = 2(x A) 2a = x = a + A,
Fy = 2(y B) 2b = y = b + B,
Fz = 2(z C) 2c = z = c + C.
Aa + Bb + Cc d
(a2 + b2 + c2 ) + Aa + Bb + Cc = d = = .
a2 + b 2 + c2
But at the stationary point
F q2 F p2
= = 0 and = = 0.
p (p + q)2 q (p + q)2
Clearly since both p and q are positive, p = q and then = 1/4. The constraint
equation then gives p = q = 2c.
F = ax2 + by 2 + cz 2 1 (x2 + y 2 1) 2 (x + y + z 1)
2
Fx = 2x(a 1 ) 2 = 0 = x =
2(a 1 )
2
Fy = 2y(b 1 ) 2 = 0 = y =
2(b 1 )
2
Fz = 2cz 2 = 0 = x = .
2c
The constraints now give the following equations for 1 and 2 : on the plane,
2 1 1 1
+ + = 1. (15.15)
2 a 1 b 1 c
1 1 4c
+ = .
a 1 b 1 2c(3c )
2V 2 p
S(r, h) = rh + 2r r2 + h2 .
r 3
Thus
S 2 2rh p
= r + = 0 and hence r2 + h2 = 3h or r = 2 2 h,
h 3 r 2 + h2
582 CHAPTER 15. SOLUTIONS TO EXERCISES
and hence
4 3 2 3
V = 2r = r 2 .
3 3
15.12. SOLUTIONS FOR CHAPTER 12 583
1 1
1 2 1 3
Z Z
2 2 2 2
dy f (+y) y = dy y f () + yf () + y f () + y f ( + )
2 2 2 6
for some in (, ), where we have used Taylors series, (section 1.3.8). Since
4 4 5
Z Z Z
2 2
= 3, 2 2
dy 2 y 2 y 2 =
dy y dy y y = 0, ,
3 15
we see that
b
4
Z
dx f (x)g(x ) = f () + O( 3 ), = .
a 3
and the associated Euler-Lagrange equation is 2y + = 0. This has the general solution
y(x) = x2 + ax + b,
4
where a and b are constants. The boundary condition at x = 0 gives b = 0: that at
x = 1 gives a = /4, so the solution is y(x) = x(1 x). The constraint gives
4
Z 1
A= dx x(1 x) = giving y(x) = 6Ax(1 x).
4 0 24
and the associated Euler-Lagrange equation is 2(xy ) + = 0, which has the general
solution
y(x) = x + a ln x + b,
2
where a and b are constants. The boundary condition at x = 1 gives b = /2: that at
x = 2 gives 0 = a ln 2 /2, so the solution is
ln x
y(x) = (1 x) + .
2 2 ln 2
584 CHAPTER 15. SOLUTIONS TO EXERCISES
and hence
2 ln 2 ln x
y(x) = 1x+ .
3 ln 2 2 ln 2
Hence
24 46 + 232
c1 c2 = .
48(1 )2
Also
1 1
1 1
Z Z
2
c3 = dx z 2 = dx (3 + 2 2x) = 1 + .
0 4(1 + )2 0 12(1 + )2
Thus the constraint becomes
24 46 + 232 1
c= 1 .
48(1 )2 12(1 + )2
Define = E[y ] E[y+ ], so we need to show that > 0. Write in the form
d d+
= ac0 + Lc cosh + cosh
c0 c0
1 2 a + 2d a + 2d+ a
c0 cosh + cosh sinh
2 c0 c0 c0
d + d d+ d
= ac0 + 2Lc0 cosh cosh
2c0 2c0
2 a a + d+ + d d+ d
c0 sinh cosh cosh .
c0 c0 c0
y
y F y F = p y 2 = c,
1+y 2
where c is a constant.
The boundary condition at x = v is given by equation 12.31 with (x, y) = y: hence
as in exercise 12.10, c = 0 and,the equation for the stationary path is exactly the same
as in exercise 12.10, that is 2 = (x B)2 + y 2 , for some constant B. The boundary
condition at x = 0 gives = B, so the stationary path is a semicircle of radius with
centre at (, 0) and hence v = 2. Since the shape created is a sphere of radius , its
area and volume are A = 42 and V = 43 3 = A3/2 /(6 ).
The left-hand end of the cable is constrained to the curve = x = 0, so the boundary
condition at x = 0 is, from equation 12.31,
(y )y
0 = F y = p , (x = 0), which gives d = 0.
1 + y 2
There
p is a natural boundary condition at x = 0, so here F y = 0, where F = (y
2
) 1 + y , so
y (y )
p = 0, that is y(0) = or y (0) = 0.
1 + y 2
The first equation gives cosh(d/c) = 0, which cannot be satisfied, and the second gives
sinh(d/c) = 0, which gives d = 0.
The transversality condition, equation 12.31, with = x/a + y/b 1, gives
y y 1 v a
p = 0 at x = v, and hence sinh = .
1+y 2 a b c b
This is one equation relating v and c. The other is given by the length constraint
Z v p
v ac bL
L= dx 1 + y 2 = c sinh = hence c = .
0 c b a
Thus the required solution is
Lb ax Lb a
y = + cosh , 0x sinh1 .
a bL a b
Finally, at x = v we have v/a + y(v)/b = 1 and since
Lb 1 b Lp 2
y(v) = + cosh sinh =+ a + b2 ,
a a a
this gives
Lp 2 vb Lp 2 Lb2 a
=b a + b2 =b a + b2 2 sinh1 .
a a a a b
sin 1 + sin 1
sin 1 cos 2 + sin 2 cos 1 2
2
2
2
= .
cos 1 + sin 1
2 cos 1 cos 2 2
2
2
2
590 CHAPTER 15. SOLUTIONS TO EXERCISES
Assuming that 1 6= 2 , because we have already dealt with this solution, gives
1 + 2
cos2 = cos 1 cos 2
2
which simplifies to 1 = cos 1 cos 2 + sin 1 sin 2 = cos(1 2 ), the only solution of
which is 1 = 2 + 2n.
Since the total energy is the sum of these three components we obtain the given result.
The constrains are just the lengths along each portion of the cable.
where
y1 (x), 0 x ,
y(x) = with y1 () = y2 ().
y2 (x), x a,
Now evaluate the functional on the varied path y + h, using the method described in
section 10.5.2. The corner moves to the point ( + u, y() + v), where u and v are
independent variables. Thus
Z +u Z a
E[y+h] = M g y() + v +g dx F (y1 +h1 , y1 +h1 )+g dx F (y2 +h2 , y2 +h2 )
0 +u
p (15.16)
where F (y, y ) = (y ) 1 + y 2 .
We have, as in section 10.5.2 (but with notation changes)
Differentiate equation 15.16 with respect to and then set = 0 to obtain the Gateaux
differential,
h i
E[y] = M gv + gu F (y1 , y1 ) F (y2 , y2 )
x=
Z Z a
dx h1 Fy1 + h1 Fy1 + g dx h2 Fy2 + h2 Fy2 .
+g
0
15.12. SOLUTIONS FOR CHAPTER 12 591
First consider the subset of variations for which u = h() = 0, to obtain the Euler-
Lagrange equations satisfied by y1 and y2 :
d
Fyk Fyk = 0, y1 (0) = B y2 (a) = A.
dx
p
Since F = (y ) 1 + y 2 is independent of x we the first integrals
yk k
p = ck = constant, k = 1 and 2. (15.17)
1 + yk 2
This expression must be zero for all u and v and hence we have the conditions
lim (F y Fy ) = lim (F y Fy ) .
M = Fy2 Fy1 and (15.19)
x+ x
The first of these equations represents the resolution of forces in the vertical direction
at x = : the second equation is the resolution of forces in the horizontal direction.
In addition the first integral, equation 15.18, represents the fact that the horizontal
component of the tension in the cable is constant. Since F y Fy is c1 or c2 we see
that the second of these conditions gives c1 = c2 = c. Using the actual expression for
F the first condition becomes
( )
(y2 2 )y2 (y1 1 )y1
M = p p = c (y2 y1 ) . (15.20)
1 + y2 2 1 + y1 2
Now we have sufficient conditions to solve the problem, as may be seen by substituting
the solutions 15.19 into these equations.
592 CHAPTER 15. SOLUTIONS TO EXERCISES
d2 x d2 y
s 2
+ sy (s) = 0 and s 2 sx (s) = 0.
ds ds
On putting t = s, so s = 1 these integrate to
x + y = and y x = .
Differentiate the second with respect to s and use the first to substitute for x to obtain
2 y + y = which has the general solution y = + a cos(s/ + ), where a and are
constants. From this we obtain x = y = a sin(s/ + ).
(b) The curve is closed and has length L, that is x(0) = x(L) and y(0) = y(L), so
that L/ = 2. Further (x + )2 + (y a)2 = a2 , so a is the radius of the circle of
circumference L, that is 2a = L.
Integrating these and using s, the arc length, for the independent variables gives as in
exercise 12.16
dx dy
= y and = + x,
ds ds
with solutions
where is a constant.
d d
(y + ) + y = 0 and (z ) + = 0,
dx dx
and the natural boundary condition is z (b) = 0. These equations simplify to
y + y + = 0 and = z ,
d4 y d2 y
y = 0, y(a) = A1 , z(a) = A2 , y(b) = B1 , z (b) = y (b) = 0.
dx4 dx2
(b) The Euler-Lagrange equation for the functional J[y] is given using the general result
given in section 10.2.1, but see also exercise 4.33 (page 143),
d2
F d F F
+ = 0,
dx2 y dx y y
dv
v = v 2 + g, v(0) = 0.
ds
594 CHAPTER 15. SOLUTIONS TO EXERCISES
Integration gives
v iv
v
Z h
dv =s that is ln(g v 2 = 2s
0 g v 2 0
1 v 1 2 1 2
Z
y =A dv v that is g(A y) = v v0 .
g v0 2 2
Multiplying by the mass gives the energy equation: the left-hand side is the loss in
potential energy as the particle falls through a distance A y; the right-hand side is
the gain in kinetic energy.
(b) If R = 0
2 g2
1
f (v)2 = + 2 2
g 2 2
v2 v2
p
2 + 2
2 2 2
1 2 v
= g + hence =
v2 f (v) g 1 2 v 2
v v2
Z
x(v) = dv p .
g v0 1 (v)2
1 1
y =A 2
sin2 = A 2 (1 cos 2) .
2 g 4 g
At the terminus = 0, so = / tan b . Since tan > 0 for (0, /2) and tan < 0
for (/2, ) the result follows.
If b = /2 the cycloid is tangent to the x-axis at the terminus. If b > /2 it crosses
the x-axis and reaches a point lower than the end point. Thus type A motion has
b < /2 and type B motion has b > /2.
and using the condition v = Vt when = 0 we see that the lower sign gives the required
solution. Hence
R f (v)
R2 g 2
g = p
g 2 + 2
where f (v) is defined in the question. Then, as in the text,
2
2 2 2 1
H = + (g ) = R = 2 + (g )2
v
and
dH d d d
H = g(g ) and H Hv R = ( g)
dv dv dv dv
596 CHAPTER 15. SOLUTIONS TO EXERCISES
so that
d 1 d f (v)
HHv = R R g( g) = p
dv v dv 2 + 2
As in the text, equation 12.77
dx v d dy ( g)v d
= and =
dv HHv dv d HHv dv
and hence
dx p v dy p v
= 2 + 2 and = ( g) 2 + 2 .
dv f (v) dv f (v)
With / = A/b this gives the straight line y = A(1x/b) through the terminal points.
If A < b the above expression for s(t) is valid only t < t0 where v(t0 )= 0: for t t0
the bead is stationary. The equation for the time to reach the end, s = A2 + b2 is the
15.12. SOLUTIONS FOR CHAPTER 12 597
same but now both roots are positive and only one satisfies t < t0 and this gives the
above expression for t. If v02 < 2g(b A) this time is complex and the bead does not
reach the point x = b.
But
1
Z
dx eax+ibx = eax+ibx
a + ib
eax
= (a cos bx + b sin bx + i(a sin bx b cos bx))
a + b2
2
so that
e2
Z
d e2 (cos sin ) = 3 cos + (1 22 ) sin )
1 + 42
598 CHAPTER 15. SOLUTIONS TO EXERCISES
and hence
1 2 gR h 2 2
i
v = 3 cos + (1 2 ) sin 3e
2 1 + 42
so that v(/2) = 0 if 22 + 3e = 1.
The speed is zero when and satisfy
This equation defines a function (), the angle at which the bead stops. If > 1 ,
where 1 is the solution of this equation when = /2, the implicit function theorem
gives the rate of change of (1 ), d/d = g /g with the derivatives evaluated at
= /2 and = 1 . Thus
4 + 3(1 )e
(1 ) = = 1.36
3(1 2e )
Putting = /2 + gives
1 1
Z
x() = d (1 cos 2) = (2 sin 2)
2gB 2 0 4gB 2
1 1
Z
y() = A d sin 2 = A (1 cos 2).
2gB 2 0 4gB 2
which is a linear inhomogeneuos equation. Otherwise f (y) is not linear and the Euler-
Lagrange equation is a nonlinear equation.
and the Euler-Lagrange equation gives y = /2. The constraint then gives = 2a, so
y = a.
600 CHAPTER 15. SOLUTIONS TO EXERCISES
1 + ln y + 1 + 2 x2 = 0 y = exp 1 1 2 x2 .
that is
x2
1
Hence y(x) = exp 2 .
2 2
dy a cos a
tan = = =
dx b sin b tan
hence
1 d a d ab
= that is = 2 .
cos2 d b sin2 d a cos + b2 sin2
2
dv abv 2
v + 2 = g(a cos b sin ).
d a cos2 + b2 sin2
15.12. SOLUTIONS FOR CHAPTER 12 601
If z = v 2 /2 this gives
dz 2abz
+ 2 = g(a cos b sin ).
d a cos + b2 sin2
2
d(zf )
= g(a cos b sin )f (), z(0) = 0,
d
and integration gives
g
Z
z() = dw (a cos w b sin w)f (w)
f () 0
y () + p(q + w)y = 0.
where
A 2 (A2 )
41d
F () = (q + w)A p 2 + .
A 2 d A2
But
A 2 (A2 ) A 2 d A A 2A 2 d2
1d 1
2
+ = + = = A ,
A 2 d A2 A2 d A A A2 d 2 A
and hence
d2
1
F () = (q + w)pA4 A .
d 2 A
(b) The coefficient of is made unity by defining wpA4 = 1 and then
d2
q 1
F () = + A 2 , where A = (wp)1/4
w d A
and the Euler-Lagrange equation is
s
x
d2 v v() w(t)
Z
+ F ()v = 0, y= , (x) = dt .
d 2 (wp)1/4 a p(t)
604 CHAPTER 15. SOLUTIONS TO EXERCISES
Now multiply this equation by p (x) , integrate and use the orthogonality relation 13.23
to obtain Z b
du p (u) F (u) = p yp hp ,
a
606 CHAPTER 15. SOLUTIONS TO EXERCISES
which gives a value for yp . Substituting this value for yk into the original sum for y(x)
gives a solution of the inhomogeneous equation in the form
Z b Z b
X 1
y(x) = du F (u)k (u) k (x) = du G(x, u)F (u)
k hk a a
k=1
X k (u) k (x)
where G(x, u) = .
hk k
k=1
yn = cos nx, n = n2 , n = 0, 1, 2, .
The orthogonality condition is more difficult to establish in this case. First consider I0n ,
Z Z
I0n = dx sinh 0 x sin n x = i dx sin i0 x sin n x
0
Z 0
sin(n i0 )
= dx cos(n i0 )x =
0 n i0
1
= ( n + i 0 ) sin n cos i 0 cos n sin i 0 ,
n2 + 02
where (z) is the imaginary part of z. Using the definitions of k we see that the term
in the outer brackets is real, and hence I0n = 0.
For n, m 6= 0 we have
Z
1
Z
Inm = dx sin n x sin m x = dx (cos(n m )x cos(n + m )x)
0 2 0
1 sin(n m ) sin(n + m )
=
2 n m n + m
m sin n cos m n cos n sin m
=
n2 m2
n m
= cos n cos m cos n cos m .
n2 m
2
If n 6= m this is zero. The case n = m can be obtained from this using LHospitals
rule. Alternatively,
Z
2 1 2
1 1
Inn = dx sin n x = cos n =
0 2 2 1 + n2
so Inn /2 as n .
Finally
1 1 1
Z
I00 = dx sinh2 0 x = cosh2 0 = .
0 2 2 1 02
608 CHAPTER 15. SOLUTIONS TO EXERCISES
since
Z
1 1
Z
dx sinh2 0 x = cosh2 0 ,
dx x sinh 0 x = ( 1) cosh 0 ,
0 0 0 2
where the definition sinh 0 = 0 cosh 0 has been used. For n 1, we use the
results
Z Z
1 1
dx sin2 n x = cos2 n
dx x sin n x = ( 1) cos n ,
0 n 0 2
to obtain
R
dx x sin n x 2( 1) cos n
an = R0 2 = n = 1, 2, 3 ,
0 dx sin n x n ( cos2 n )
giving
2( 1) cosh 0 X cos k sin k x
x= sinh 0 x 2( 1) .
0 ( cosh2 0 ) k=1
k ( cos2 k )
15.13. SOLUTIONS FOR CHAPTER 13 609
(a) If < 0 the nontrivial solution cannot be made to satisfy the boundary conditions.
If = 0 the solution y = 1 satisfies the boundary condition.
If > 0, put = 2 , ( > 0) giving the general solution (note it is easier to use the
complex form here)
y = Aeix + Beix .
which is zero only if A1 A2 +B1 B2 = 0, that is the vectors a = (A1 , A2 ) and b = (B1 , B2 )
are orthogonal.
1 a=1/5
a=1/10
0.5
0
1 2 3 4 5 6 7 8 9 10 11
-0.5
-1
Figure 15.17 Graphs of the functions u = sin and u = a for a = 1/5
and 1/10.
For > 0 these curves intersect if < c 1/a, giving real zeros; for > c the zeros
are complex. There are about N 1/a zeros because there is one zero every time
passes through an integer. Hence there are a finite number of real zeros.
Consider the inner product of two distinct eigenfunctions, yi and yj with j > i.
1
Z Z
Iij = dx sin i x sin j x = dx (cos(j i )x cos(j + i )x)
0 2 0
1 sin(j i ) sin(j + i )
= sin k = ak , k = i, j
2 j i j + i
a j cos i i cos j j cos i + i cos j
=
2 j i j + i
aj i
= (cos i cos j ) .
j2 i2
It is obvious that cos j + cos i 6= 0, but this is easily proved. We note that
and since cos j cos i = 2 sin(j i ) sin(j + i ) , it follows that Iji 6= 0.
2 2
Solution for Exercise 13.11
Z
dx
(a) Comparing with the equation in exercise 2.31 (page 74) we see that p = exp =x
x
and that the self-adjoint form is
2
d dy
x + x y = 0.
dx dx x
(b) In this example p = x, q = x 2 /x so v = 1/ p = 1/ x and
2 2 41
1
I(x) = 1 + 4x x =1 .
4x2 x x2
15.13. SOLUTIONS FOR CHAPTER 13 611
d2 v
+ e2 2 v = 0,
2
y(x()) = v().
d
Compare the nth coefficient of this and the original series to obtain the first result.
P
(ii) Put z = x in equation 13.29 eix sin t = n= Jn (x)eint , and now set t = +s,
so sin t = sin s to obtain eix sin t = in int
P
n= Jn (x)e e . Compare the nth
coefficient of this and the original series to obtain the second result.
(iii) Put t = 0
X
X
X
1= Jn (z) = J0 (z) + Jn (z) + Jn (z) = J0 (z) + 2 J2n (z),
n= n=1 n=1
(b) With the boundary conditions y(a) = y(1) = 0 the system is regular and we have:
d2
q 1
f () = A 2
w d A
is continuous and hence has a minimum value Qm . If < Qm then f () + < 0 for
all and the result proved in the text shows that any solution v() has at most one
zero, so cannot satisfy the boundary conditions. Hence there are no eigenvalues smaller
than Qm .
1 d2 y
+ ay = 0 that is v () + a3 v() = 0.
a2 d 2
Thus if y(x) is a solution of y + xy = 0, v() = y(a) is a solution of v + a3 v = 0.
(c) Suppose the solution of y + xy = 0 with the condition y(0) = 0 is v(x); then
v(x) = y(1/3 x) where y(x) is a solution of y + xy = 0.
If = rn3 then v(1) = y(rn ) = 0, so y(rn x) is an eigenfunction with eigenvalue n = rn3 ,
and there are infinitely many of these.
There are no negative eigenvalues because y(0) = 0 and there can be no other zeros.
614 CHAPTER 15. SOLUTIONS TO EXERCISES
where
1 1 d
Z Z
ak = d (u() ) sin k = du sin u sin k (u sin u)
du
1
Z
= du sin u(1 cos u) sin k (u sin u)
Z
1 d
= du sin u cos k (u sin u)
k du
h i Z
1
= sin u cos k (u sin u) du cos u cos k (u sin u)
k
Z
= du cos u cos k (u sin u)
k
Z
h i
= du cos (k + 1)u k sin u + cos (k 1)u k sin u
2k
2
= (Jk+1 (k) + Jk1 (k)) = Jk (k).
k k
By expanding this and comparing with the previous integrand we obtain the differential
equation,
n2
2 i
g + i(n x cos t)g = sin t sin t + 1 2 .
x x
Consider a solution g = A + B cos t; by substituting this in the left hand side we see
that if B = i/x and A = in/x2 a solution is obtained. This solution is periodic, hence
the result.
<0
If < 0, put = 2 , ( > 0), the general solution is y = A cosh x + B sinh x: the
boundary condition at x = 0 gives A = 0 and the boundary condition at x = gives
tanh = , > 0.
If > there are no real solutions of this equation the gradient of the left and right
hand sides at = 0 are, respectively, and , so if > , > tanh for > 0.
If < , the same reasoning shows that there is one real positive solution which we
denote by 0 .
>0
If > 0, put = 2 , ( > 0), the general solution is y = A cos x + B sin x: the
boundary condition at x = 0 gives A = 0 and the boundary condition at x = gives
tan = , > 0.
If > , the first positive solution, 0 is in (0, /2) and the solution k , is in the
interval (k, (k + 1/2)), k = 0, 1, .
If < , the first positive solution, 1 is in (/2, 3/2) with kk < (k + 1/2),
k = 1, 2, .
Thus we have the following,
which has infinitely many zeros for x > 1. If 4B < 1 the general solutions is
y = x Axq + Bxq , q = 1 4B,
du du d2 u d2 u du
x = and x2 2
= 2 +
dx dt dx dt dt
and the equation for u becomes u (t) + u (t) + ( + 1/4)u = 0. Putting u = ept gives
p2 + p + ( + 1/4) = 0 and hence the general solution is
u = et/2 Ae t + Be t ,
= 2 , > 0,
(Ax + Bx )
y = (A B ln x) , = 0,
i ln x i ln x
= 2 , > 0.
Ae + Be
(i) < 0: the solution is bound at the origin only if B = 0, so y = Ax giving y(0) = 0
and y(1) = A. Hence there are no nontrivial solutions.
(ii) = 0: In the case the bound solutions are y = A: if c 6= 0, the solution is y = c,
with eigenvalue = 0.
(iii) > 0: the solution is not defined at the origin for any A or B, except A = B = 0.
Aa + Ba = 0
= A = B = 0.
Ab + Bb = 0
A B ln a = 0, A B ln b = 0 = A = B = 0.
Aei ln a + Aei ln a = 0
= e2i ln(b/a) = 1,
Aei ln b + Aei ln b = 0
n
hence n = n2 , n = , and yn = c sin (n ln(x/a)), for some constant c.
ln(b/a)
Thus this simple variational estimate provides a fairly good approximation to the period.
since y(0) = y(1) = 0 the boundary term vanishes and C[y] = S[y]. Putting y = yn
gives the result.
The functional is
Z 1 Z 1 Z 1
2 2 2 2 2 2
dx x2 sin2 x.
S(a) = dx pz qz = a dx cos x a
0 0 0
R1 R1
But 0 dx x2 sin2 x = 1
2 0 dx x2 (1 cos 2x) and
1 1
x2 1 1
Z Z
2
dx x cos 2x = sin 2x dx x sin 2x
0 2 0 0
Z 1
1 h x i1 1 1
= cos 2x + dx cos 2x = .
2 0 2 0 2 2
1 1
Hence 1 S(a) = 2 3 + 2 2 9.587.
(b) For the trial function z = ax(1 x), the constraint gives
Z 1
a2
2 2 2 2 1 2 1
C[z] = a dx x (1 x) = a + = = 1.
0 3 4 5 30
Hence 1 < 9.71. The first bound is smaller, so is the better approximation.
Hence
2 1 2
1 S(a1 ) =
2 = 1.76476.
4 2
Note that to 10 significant figures the value of the first eigenvalue is 1.762682254 it
can be shown to be the first zero of Ai(u)Bi (1 u) Bi(u)Ai (1 u), where Ai
and Bi are Airy functions.
With the two parameter trial function z = a1 sin x/2 + a2 sin 3x/2 the constraint
gives
Z 1 Z 1
2 2 1 2 3 1 1
a1 dx sin x + a2 dx sin2 x = a21 + a22 = 1.
0 2 0 2 2 2
15.14. SOLUTIONS FOR CHAPTER 14 625
The functional is
Z 1 2 Z 1 2
x 3 3x x 3x
S(a) = dx a1 cos + a2 cos dx x a1 sin + a2 sin
0 2 2 2 2 0 2 2
2 2
1 1 1 1 2
a1 + 9a22 a21 + a22 2 a1 a2
= + +
8 4 2 4 9 2
2 2
1 1 9 1 1 2
= 2 a21 + 2 a22 + 2 a1 a2 ,
8 4 8 4 9
where we have used the integrals quoted in the question. Thus the equation is
2
1 2 2
4
2 2 2
a = a
2
2 9 1 2
2
2 4 2 9
and the eigevalues are given by the quadratic equation 2 23.4489 + 38.2261 = 0
and the smallest root is 1.7627.
These n linear equations for a can be written in the matrix form M a = a where Mij
is defined in the question.
(b) If n = 1 we have
Z 1
2
M11 = 2 dx x sin2 x
0
Z 1 Z 1
2
= dx x(1 cos 2x), but since dx x cos 2x = 0,
0 0
and since
1
1 (1)k 16
Z
dx x cos kx = , M12 =
0 (k)2 9 2
626 CHAPTER 15. SOLUTIONS TO EXERCISES
and
1
1
Z
M22 = 4 2 dx x (1 cos 4x) = 4 2 ,
0 2
giving the eigenvalue problem
1 16
2
2 9 2
1 a = a,
16
2
4
9 2 2
which is just equation 14.21.
(c) If p = 1, q = x and k = sin kx we have
Z 1 Z 1
(H 1 S)ij = 2 2 ij dx cos ix cos jx 2 dx x sin ix sin jx
0 0
Z 1
= 2 ijij 2 dx x sin ix sin jx = Mij .
0
Z b
2 dx (py ) + (q + w)y h.
a
Using the class of variations with h(a) = h(b) = 0, we see that the Euler-Lagrange
equation,
d dy
p + (q + w)y = 0,
dx dx
must be satisfied by a stationary path. Further, since S = 0 for all admissible paths
the given boundary conditions must also be satisfied.
(b) Since wy = qy + (py ) we have
Z b Z b Z b
2
k dx wyk = dx qyk2 + dx yk (pyk )
a a a
h ib Z b
= yk pyk dx pyk 2 qyk2 .
a a
Using the constraint condition and the boundary conditions to replace y (a) with y(a)
and y (b) with y(b), this becomes
Z b
k = dx pyk 2 qyk2 + p(b)y(b)2 p(a)y(a)2 = S[yk ].
a
15.14. SOLUTIONS FOR CHAPTER 14 627
The functional is
Z 1 Z 1
S[a] = dx (2ax)2 a2 dx x2p (1 x2 )2
1 1
Z 1 Z 1
= 8a2 dx x2 2a2 dx x2p (1 2x2 + x4 )
0 0
2 8 1 2 1
= a 2 + .
3 2p + 1 2p + 3 2p + 5
5 6
Hence 1 S(a) = 1 .
2 (2p + 1)(2p + 3)(2p + 5)
Taking the lowest eigenfunction of the simpler problem, z = a cos(x/2), to be the trial
function the constraint gives
Z 1 x a2
a2 dx cos2 = = 1,
0 2 2
and the functional becomes
Z 1 x
S(a) = dx y 2 by 2 sin
0 2
1 2 2 1
Z x Z 1 x x
= a dx sin2 a2 b dx sin cos2
4 0 2 0 2 2
2 2 2b 2 2 4b
= a a and hence 1 .
8 3 4 3
(c) With the trial function z = a cos(n 1/2)x, the constraint gives a2 = 2 and the
functional becomes
Z 1 Z 1
S(a) = a2 2 (n 1/2)2 dx sin2 (n 1/2)x a2 b dx sin(x/2) cos2 (n 1/2)x.
0 0
But
1 1
sin(x/2) cos2 (n 1/2)x = sin(x/2) + sin(2n 1/2)x sin(2n 3/2)x ,
2 4
so
1
1 1 1 1
Z
dx sin(x/2) cos2 (n 1/2)x = +
0 4 2n 1/2 2n 3/2
1 1
=
(4n 1)(4n 3)
and hence
2
a2 b
1 2 2 1 1
S(a) = a n 1 , and since a2 = 2
2 2 (4n 1)(4n 3)
2b 1 1
n n2 2 1 2 , n=n .
16n 1 2
and for the lowest eigenvalue 1 , 1 S[y], where y is any admissible function.
Alternatively this Sturm-Liouville equation can be written in the form
Z 2 Z 2
d 1 dyk 2 d 1 dyk
k xyk = and hence k dx xyk = dx yk .
dx x dx 1 1 dx x dx
Z 2
Since, by definition, hk = dx xyk2 , putting hk = 1 gives
1
2 2 2 2
d 1 dyk 1 1 dyk
Z Z
k = dx yk = yk yk + dx .
1 dx x dx x 1 1 x dx
With k = 1 the boundary conditions y1 (1) = 0 and y1 (2) y1 (2) = 0 then give,
2 2
1 1 dy1
Z
1 = y1 (2)2 + dx .
2 1 x dx
Hence, since the functional S[y] has a minimum the Rayleigh-Ritz method shows that
6 12
1 S[z] = + ln 2 = 0.331.
7 7
This is an upper bound because S[y] is a positive quadratic functional having a minimum
value which is attained in the allowed space of functions.
d2 y
+ y=0
dw2 4
630 CHAPTER 15. SOLUTIONS TO EXERCISES
3
The boundary condition at x = 2 gives the equation tan = 2 . The smallest
2
positive solution of this is = 0.317.