Second Semester, 2016

School of Mathematics and Statistics

MAST20030: Differential Equations

Writing time: 3 hours

Student Number:

Authorised materials:

• Mobile phones, smart watches and internet or communications devices are forbidden.
• Calculators, tablet devices or computers must not be used.
• One A4 sheet, handwritten only on both sides, is permitted.
• No books, printed/photocopied materials or notes are permitted.

Instructions to students:
• You should attempt all questions. There are 6 questions whose marks are indicated.
• The total number of marks available is 100.

Instructions to invigilators:
• Initially, students are to receive a 14 page script book.
• Some students may require an additional script book to complete the examination.

Page 1 of 3
 Question 1. (15 marks)

(a) Write down the definition of the wronskian of two functions f and д. Explain two ways in which evaluating a
wronskian can help solve a homogeneous second-order linear ordinary differential equation. (5 marks)
(b) Check that y(x ) = x −1 is a complementary solution of

1
x 2y 00 (x ) + 3xy 0 (x ) + y(x ) = .
x
Now use reduction of order to find the general solution. (10 marks)

Question 2. (16 marks)

This question concerns the Bessel equation of order 1:

d2y dy
x2 2
+x + (x 2 − 1)y = 0.
dx dx
(a) Substitute the Ansatz y(x ) = ∞ j=0 c j x into this ordinary differential equation, assuming that you can bring the
j
P
differential operators inside the infinite sum, to derive a recurrence relation for the c j . For which j ∈ Z>0 are
the c j free parameters? (8 marks)
(b) Compute the c j with j 6 8 and verify explicitly that your results are consistent with the following formal
solution:
∞
X (−1)m x 2m+1
y(x ) = .
m=0
22mm!(m + 1)!
What is the radius of convergence of this formal solution? (5 marks)

(c) For this formal solution to be a genuine solution of the order 1 Bessel equation, we need to know that
∞ ∞
(−1)m x 2m+1  X d (−1)m x 2m+1
" #
dy d
 X 
=  =
dx dx m=0 22mm!(m + 1)!  m=0 dx 22mm!(m + 1)!
 ∞ ∞
d2y d2  X (−1)m x 2m+1  X d2 (−1)m x 2m+1
 " #
and = = .
dx 2 dx 2 m=0 22mm!(m + 1)!  m=0 dx 2 22mm!(m + 1)!

What should we check in order to guarantee that these relations hold? [You do not have to actually perform this
check, just state what should be done.] (3 marks)

Question 3. (18 marks)

(a) Write down the definition of the Laplace transform of a function f (t ) defined for positive real t. Give an
example of a function f (t ), defined for all 0 < t < ∞, for which the Laplace transform does not exist.
(4 marks)
(b) Find the solution of the ordinary differential equation

1 1
 if x 6 2,
y 00 (x ) + y 0 (x ) + y(x ) =  y(0) = y 0 (0) = 0. (10 marks)
4 0
 if x > 2,

(c) Explain, without referring to your explicit solution, why the ordinary differential equation of the previous part
does not have any genuine solutions. What are the best differentiablity properties that a solution can have?
(4 marks)

Page 2 of 3
 Question 4. (12 marks)

(a) Discuss briefly a situation in which it might be efficient to solve a homogeneous vector-valued linear ordinary
differential equation with constant coefficients using a matrix exponential. (2 marks)
(b) Consider the following system of ordinary differential equations:

x 0 (t ) + 2x (t ) − y(t ) = 6, x (0) = 0,
y (t ) + 2y(t ) − x (t ) = 0,
0
y(0) = 0.

Convert this into a vector-valued ordinary differential equation and find the solution by diagonalisation. [You
should not compute a matrix exponential.] (10 marks)

Question 5. (25 marks)

2d
The differential operator − dx 2 was shown in lectures to be self-adjoint (hermitian) on an appropriate vector space
of functions f on [0, 1] satisfying f 0 (0) = f 0 (1) = 0.
(a) Give one important property of the eigenvalues of this operator and two important properties of its eigenfunc-
tions. [You do not need to prove these properties.] (3 marks)

(b) Derive the eigenvalues λn and eigenfunctions X n (x ) for these boundary conditions. [Merely stating the answers
will not earn any marks.] (10 marks)
(c) Show that the Fourier series of the even extension of f (x ) = x, x ∈ [0, 1], is given by
∞
1 4 X 1
− 2 cos(π (2m + 1)x ). (8 marks)
2 π m=0 (2m + 1) 2

(d) When does this Fourier series, defined for all x ∈ R, converge to x? Would you expect the partial sums to
exhibit Gibbs phenomena? (4 marks)

Question 6. (14 marks)

A string of length 1 is collared at both ends frictionlessly so that its vibrations, described by a function u (x,t ),
satisfy Neumann boundary conditions. We model the striking of the string at one end with the initial conditions

u (x, 0) = 0, ∂t u (x, 0) = αx,

where α is a small positive constant. Find the formal solution of the wave equation that satisfies these initial and
boundary conditions.

Page 3 of 3