You are on page 1of 6

Math2221

Test 2B

S2, 2014

Time allowed: 50 minutes


1.

Solve the boundary-value problem

2x2 u 00 + 3xu 0 3u = 14x2


2.

for 1 < x < 4, with u(1) = 9 and u(4) = 29.

Find the general solution of


u 00 4u 0 + 4u =

e2x
.
x

You may assume the following formulae for the method of variation of
parameters applied to the ODE u 00 + p(x)u 0 + q(x)u = f(x):
v10 =

u = u1 v1 + u2 v2 ,
3.

fu2
,
W

v20 =

fu1
.
W

For the differential operator


2

Lu = xu 00 ex u 0 + 4x2 u
find the coefficients b0 (x), b1 (x), b2 (x) of the adjoint operator
L v = b2 v 00 + b1 v 0 + b0 v.
4.

The Bessel function of order n {0, 1, 2, . . .} is defined by the series


Jn (z) =

X
(1)k (z/2)2k+n

k!(n + k)!

k=0

Find the coefficient Cn in the identity


zJn1 (z) + zJn+1 (z) = Cn Jn (z)
for z C and n {1, 2, 3, . . .}.

Questions continue over the page.

5. Let hf, gi be an inner product with associated norm kfk =


hf, gi when f and g are such that
kfk = 1,
6.

kgk = 3,

p
hf, fi. Find

k3f + gk = 4.

Consider the ODE


3z2 u 00 + z(1 z)u 0 u = 0.

(i) Find the indicial equation for r and the recurrence relation for Ak if
u=

Ak zk+r .

k=0

(ii) Which root of the indicial equation yields a solution u that is bounded
at z = 0?
(iii) For this root, find A1 and A2 given that A0 = 1.
7.

Show that if the boundary-value problem


u 00 + u = f(x) for 0 < x < 3/4,
u=0
u0 + u = 0

at x = 0,
at x = 3/4,

has a solution u, then the function f must satisfy


Z 3/4
f(x) sin x dx = 0.
0

Solutions
1.

[8 marks] Let Lu = 2x2 u 00 + 3xu 0 3u and observe that




Lxr = 2r(r 1) + 3r 3 xr = (2r2 + r 3)xr = (2r + 3)(r 1)xr ,

so the general solution of the homogeneous equation Lu = 0 is


uH = Ax3/2 + Bx.
Also, Lx2 = 7x2 so a particular solution of Lu = 14x2 is uP = 2x2 . Hence,
the general solution of Lu = 14x2 is
u = uH + uP = Ax3/2 + Bx + 2x2 .
Since u(1) = A + B + 2 and u(4) = A/8 + 4B + 32, the boundary conditions
give
A + B + 2 = 9,
A + B = 7,
and thus
A/8 + 4B + 32 = 29,
A/8 + 4B = 3.
Therefore A = 8 and B = 1 and the solution is
u = 8x3/2 x + 2x2 .
2. [8 marks] Since 2 4 + 4 = ( 2)2 the functions u1 = e2x and
u2 = xe2x are linearly independent solutions of the homogeneous equation. Their Wronskian is
2x

2x
e

xe
4x

W = 2x 2x
2x = e
2e
e + 2xe
so applying the formulae with f(x) = e2x /x we have
v10

(e2x /x) xe2x


=
= 1
e4x

and

v20

(e2x /x) e2x


=
= x1 .
4x
e

Thus, v1 = x and v2 = log |x|, implying that


uP = u1 v1 + u2 v2 = e2x (x) + xe2x log |x| = e2x (x + log |x|)
is a particular solution, and hence the general solution (with C = B 1) is

u = uH + uP = Ae2x + Bxe2x + uP = e2x A + Cx + x log |x| .
3

[3 marks] The definition of L gives

3.

L v = (xv) 00 + (ex v) 0 + 4x2 v = (2v 0 + xv 00 ) + (2xex v + ex v 0 ) + 4x2 v


2
2
= xv 00 + 2 + ex v 0 + 4x2 2xex v,
showing that
2

b1 = 2 + ex ,

b2 = x,

b0 = 2x(2x ex ).

[6 marks] Since

4.

zJn1 (z) = z
=

X
(1)k (z/2)2k+n1

k=0

X
k=0

k!(n 1 + k)!

X
(1)k z2k+n 22kn+1
k=0

k!(n 1 + k)!

(1)k (z/2)2k+n
2(n + k)
k!(n + k)!

and
zJn+1 (z) = z
=

X
(1)k (z/2)2k+n+1

k=0

X
k=1

k!(n + 1 + k)!

X
(1)k z2k+n+2 22kn1

k!(n + 1 + k)!

k=0

(1)k1 z2k+n 22kn+1 X (1)k (z/2)2k+n


=
(2k),
(k 1)!(n + k)!
k!(n + k)!
k=1

we have
zJn+1 (z) + zJn1 (z) = 2n
= 2n


(z/2)n X (1)k (z/2)2k+n 
+
2(n + k) 2k
n!
k!(n + k)!
k=1

X
(1)k (z/2)2k+n
k=0

k!(n + k)!

= 2nJn (z),

so C = 2n. The identity holds for any integer n, not just n > 1, if we agree
that 1/k! = 0 for k {1, 2, 3, . . .}.
5.

[4 marks] Since

k3f + gk2 = h3f + g, 3f + gi = k3fk2 + 2h3f, gi + kgk2 = 9kf2 k + 6hf, gi + kgk2


we have
16 = 9 + 6hf, gi + 9
so 6hf, gi = 2 and thus hf, gi = 1/3.
4

6.
(i) [7 marks]
3z2 u 00 + z(1 z)u 0 u = (3z2 u 00 + zu 0 u) z2 u 0

X
X


k+r
=
3(k + r)(k + r 1) + (k + r) 1 Ak z

(k + r)Ak zk+r+1
=

k=0

k=0

[3(k + r) + 1](k + r 1)Ak zk+r

k=0

(k 1 + r)Ak1 zk+r

k=1
r

= (3r + 1)(r 1)A0 z

X


+
(3k + 3r + 1)(k + r 1)Ak (k + r 1)Ak1 zk+r ,
k=1

so the indicial equation is (3r + 1)(r 1) = 0 and the recurrence


relation is
Ak1
Ak =
for all k > 1.
3k + 3r + 1
(ii) [1 mark] The roots of the indicial equation are r1 = 1 and r2 = 1/3.
Since r1 > 0 but r2 < 0, we need to choose r = r1 to obtain a solution
that is bounded at z = 0.
(iii) [3 marks] With r = 1,
Ak =

Ak1
3k + 4

for all k > 1.

Thus, putting A0 = 1 gives


A1 =
so

7.

A0
1
=
7
7

and A2 =

A1
1
= ,
10
70



z z2
u=z 1+ +
+ .
7 70

[5 marks] First consider the homogeneous problem


v 00 + v = 0 for 0 < x < 3/4,
v = 0 at x = 0,
0

v + v = 0 at x = 3/4.
5

(1)

The general solution of v 00 +v = 0 is v = A cos x+B sin x and since v(0) = A


we see that A = 0. Thus, v = B sin x and v 0 = B cos x, implying that when
x = 3/4,

3
3
2
2
v 0 + v = B cos
+ B sin
= B
+B
=0
4
4
2
2
for every B. Thus, v = B sin x is the general solution of the homogeneous
problem (1). The operator Lu = u 00 + u is formally self-adjoint so
for all u, v D,

hLu, vi = hu, Lvi


where

Z 3/4
hf, gi =

f(x)g(x) dx
0

and D, the domain of L, is defined to be the set of all C2 functions satisfying


the homogeneous boundary conditions. Hence, if a solution u exists then
for any constant B,
hf, vi = hLu, vi = hu, Lvi = hu, 0i = 0

when v = B sin x,

and in particular, with B = 1,


Z 3/4
f(x) sin x dx = 0.
0

You might also like