CHEN30101

ONE AND A HALF HOURS

THE UNIVERSITY OF MANCHESTER

MATHEMATICAL METHODS 3

xx January 2016
[Time of Examination]

Answer ALL parts of the question in Section A (worth 20 marks).

Answer any TWO of the four questions in Section B (each worth 10 marks). If more than two
questions from Section B are attempted, then credit will be given for the best two answers.

Calculators may be used in accordance with the university regulations.

Tables of mathematical formulae are available and may be used without restriction.

1 of 5 P.T.O.
 CHEN30101
SECTION A

Answer ALL parts of this question

A1.
(a) Verify that the problem of finding the eigenpair {λ, u} such that
−u00 (x) = λ u(x), x ∈ (0, 1); u(0) = 0, u(1) = 0, (E)
has solution(s)
λj = j 2 π 2 ; uj = sin(jπx), j = 1, 2, . . .
[2 marks]
(b) The deflection u(x) in an elastic string of unit length can be shown to satisfy the boundary
value problem,
−u00 (x) = f (x), x ∈ (0, 1); u(0) = 0, u(1) = 0. (D)
Using (E), show that u(x) can be written as a Fourier series
∞
X cj
u(x) = sin(jπx),
λ
j=1 j

where the cj ’s are the Fourier coefficients of f (x).

[4 marks]
(c) We would like to construct approximations to the deflection u(x) in the case that the load
function f (x) in part (b) satisfies
f (x) = x(1 − x).
(i) Compute the first three Fourier coefficients {c1 , c2 , c3 } of the function f (x).
Hint: integrate by parts and use the fact that
Z 1
2
(1 − 2x) cos(jπx) dx = 2 2 (1 − cos(jπ)) .
0 j π

[4 marks]
(ii) Construct the truncated Fourier series approximation uj (x) with j = 1 and j = 3 terms.
Compare the approximate solutions at the point x = 0.5 with the exact value u(0.5) =
0.0260 (working to 4 decimal places).
[2 marks]
(d) Describe the construction of a centered difference approximation to the deflection u(x) based
on dividing the string into n subintervals of equal length h.
[4 marks]
Setting f (x) = x(1 − x), compute the centered difference approximation to u(x) for n = 2 and
compare the exact and approximate solutions at the point x = 0.5. Give one reason why the
accuracy is guaranteed to improve if the number of subintervals is increased.
[4 marks]
End of Question 1.

2 of 5 P.T.O.
 CHEN30101
SECTION B

Answer TWO of the four questions

B2. A storage tank is discovered to have suffered a leak and a plan to redress the leak is implemented.
The height of liquid in the tank is then given by

z(t) = 10 − 5e−t/10 + 4e−t/4

where the three terms represent the usual height, the leak and the rescue plan. The variables t and
z are measured in hours and metres respectively.

(a) Based on values of z at t = 0, t = 7 and one other between t = 0 and t = 7 carry out two
iterations of a bisection search to estimate the time that the height of liquid in the tank is at
a minimum. [5 marks]

(b) The whole plant must be evacuated if the level of liquid falls below z = 8.15. Does this scenario
happen? [1 mark]

(c) At what time, close to t = 10 does the tank regain a height of 8.5? Give your answer to three
decimal digits. [Hint: construct a Newton iteration.] [4 marks]

End of Question.

B3. Consider the modified Bessel equation of order j = 1/2:

 
2 00 0 2 1
x y + xy − x + y = 0,
4

together with initial conditions, y(1) = 0.9377 and y 0 (1) = 0.7624.

(a) Write this second-order ordinary differential equation as a system of two first-order equations
each with a condition at x = 1.
[2 marks]

(b) Carry out one step of the four-stage Runge–Kutta method (rk4) with step size h = 0.1 in order
to estimate y(1.1). Record your final answer to an accuracy of four decimal places. [7 marks]

(c) Compare your values at x = 1.0 and x = 1.1 with the exact values of y(x) that are given by
r
2
I1/2 (x) = sinh x.
πx
[1 mark]

End of Question.

3 of 5 P.T.O.
 CHEN30101

B4. A slab of a radioactive material is heated internally by radioactive decay at an exponentially

decreasing rate. The temperature u(x, t) is known to satisfy the partial differential equation

ut = uxx + e−t sin x (pde)

and to satisfy the boundary conditions at the cooled sides of the slab

u(0, t) = u(π, t) = 0 (bcs)

for all t > 0.

(a) Show that up = te−t sin x is a particular solution of the boundary value problem. [1 mark]

(b) Explaining all steps, use separation of variables in the form u = T (t)X(x) to reduce the
homogeneous problem
ut = uxx with u(0, t) = u(π, t) = 0
to the eigenvalue problem for the function X(x)

X 00 − µX = 0 with X(0) = X(π) = 0. (E 0 )

[3 marks]

(c) Using the result from question A1(a) write down expressions for the eigenvalues and eigen-
functions satisfying (E 0 ). What are the corresponding solutions for T (t)?
Hence, write down the general solution of the boundary value problem (pde) and (bcs).
[4 marks]

(d) If it is also known that the initial temperature profile is given by

u(x, 0) = g(x),

where g(x) is a given function, show that the temperature at later times is given by the infinite
series expansion
X∞
−t
u(x, t) = te sin x + ck exp(−k 2 t) sin(kx),
k=1

where {ck }∞
k=1 are the Fourier coefficients of g(x).

[2 marks]

End of Question.

4 of 5 P.T.O.
 CHEN30101

B5. An iron bar of length 2 units is differentially heated (one side only) so that the initial temperature
is the unit step function u(z), which is zero when z < 0 and unity when z ≥ 0. Both ends of the bar
(z = ±1) are then insulated and the bar is allowed to cool. The resulting temperature T (z, t) inside
the bar satisfies the heat equation
∂T ∂ 2T
= (H)
∂t ∂z 2
together with the end conditions ∂T∂z
= 0 at z = ±1.

(a) The boundary value problem is to be solved using the FTCS (forward time centred space)
approximation method with a step size in z of h = 0.4.

(i) For what values of z will the temperature T be computed? [1 mark]

(ii) What is the maximum allowable time step that can be used? [1 mark]
(iii) For a time step k = 0.01, carry out one step of the FTCS approximation method. Record
all your answers to an accuracy of four decimal places. [5 marks]

(b) Describe an implicit finite difference scheme that could be used instead of FTCS. Explain how
this method differs from the explicit scheme and state one advantage of implicit approximation
methods.
[3 marks]

End of Question.

END OF EXAMINATION PAPER

5 of 5