lOMoARcPSD|3449400

Exam 2016, questions

Mathematical Methods 4 (University College London)

StuDocu is not sponsored or endorsed by any college or university

Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

All questions may be attempted but only marks obtained on the best four solutions will
count.
The use of an electronic calculator is not permitted in this examination.

1. One-dimensional linear waves with surface displacement h(x, t) evolve in an lake of

variable depth according to

htt + (c2 hx )x = 0, in − 1 ≤ x ≤ 1, t ≥ 0.

In this lake the local wave speed c is given by c(x) = (1 − x2 )1/2 . At t = 0 the lake is
disturbed, so that the initial surface displacement h(x, 0) = f (x) and initial surface
velocity ht (x) = g(x). You may assume that h(x, t) remains bounded throughout
the domain.

(a) Show that for t > 0 the surface displacement is given by

∞
X
h(x, t) = Re Ak Pk (x) exp (iωk t),
k=0

where the {Ak } are complex coefficients. Find the angular frequencies {ωk }.
(b) Obtain the formula for the coefficients {Ak } in terms of integrals involving f (x)
and g(x). Briefly explain, in terms of Sturm-Liouville theory, any orthogonality
result used.
(c) Find h(x, t) for the particular case with f (x) = 0 and g(x) = 6x2 + 3x − 2 and
verify that in this case
2 √
h(0, t) = − √ sin ( 6t).
6

You are given that the Legendre polynomials Pk (x) are solutions of Legendre’s
equation
(1 − x2 )y ′′ − 2xy ′ + ν(ν + 1)y = 0, (ν ∈ R)
when ν = k (k ≥ 0 integer), and that they are the only solutions which are non-
singular at both x = ±1.
The following formulae may prove useful
Z 1
1 dk  2 k 2
Pk (x)2 dx =

Pk (x) = k k
(x − 1) , .
2 k! dx −1 2k + 1

MATH7402 PLEASE TURN OVER

1
Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

2. The Laplace transform L[f ](s) = f¯(s) of a function f (t) defined on [0, ∞) is defined
to be Z ∞
¯
f (s) = f (t) e−st dt.
0

(a) Given that g(t) is another function defined on [0, ∞), find
Z t 
n
(i) L[ t ] (ii) L f (u)g(t − u) du ,
0

where n ∈ N is an integer.
(b) The outputs x(t) and y(t) of an electronic component are related to the input
F (t) by the equations

ẋ(t) + 2x(t) + 3y(t) = F (t), x(0) = y(0) = 0,

ẏ(t) + 4y(t) + x(t) = 0.

where dots denote time derivatives. Show that

Z t


y(t) = β F (t − u) e−u − e−5u du,
0

where β > 0 is a constant to be found.

(c) If the input has the form of a short positive pulse at t = 0 (lasting for a time
∆t ≪ 1), find the time at which the output y(t) is a maximum.

3. The steady temperature distribution u(r, θ) in a circular wedge-shaped plate satisfies

Laplace’s equation
1 ∂ 2u
 
2 1 ∂ ∂u
∇ u≡ r + 2 2 = 0,
r ∂r ∂r r ∂θ
where (r, θ) are the usual polar coordinates, and the domain is (0 ≤ r ≤ a, 0 ≤ θ ≤
β). The straight boundaries of the plate are insulated so that uθ (r, 0) = uθ (r, β) = 0
and the curved outer edge is held at a fixed temperature so that u(a, θ) = h(θ). Show
that ∞
X
u(r, θ) = Ak rαk cos (αk θ) ,
k=1
Z β
where Ak = γk h(θ) cos (αk θ) dθ.
0

The sequences of constants {αk } and {γk } should be found as part of your solution.

MATH7402 CONTINUED

2
Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

4. Consider the second-order linear ordinary differential equation

x2 y ′′ + (1 − α − β)x − x2 y ′ + (αβ + (γ − 1)x) y = 0.

(†)

where α, β, γ ∈ R are constants.

(a) Seek a Frobenius-type power series solution of (†) of the form

∞
X
y(x) = ak xk+c (a0 6= 0),
k=0

and show that the coefficients {ak } satisfy the recurrence relation

k+c−γ
ak = ak−1 .
(k + c)2 − (α + β)(k + c) + αβ

Find the possible values taken by c.

(b) Give a brief outline of the method for finding the two linearly independent
solutions in the following cases (do not try to calculate the solutions explicitly):
(i) The constants α and β are distinct and |α − β| ∈
/ N.
(ii) The constants α and β are equal.
(c) Explain the difficulty with calculating one of the solutions in the case where
the constants α and β are distinct and |α − β| ∈ N. Indicate which of the two
solutions is problematic.
(d) For case (i) above, describe the simplified form taken by one of the two solutions
when γ = n + α for n ∈ N. Which other values of γ give similar behaviour?

MATH7402 PLEASE TURN OVER

3
Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

5. Consider the following eigenvalue problem, defined on the interval 0 ≤ x ≤ a,

x2 y ′′ + xy ′ + (λx2 − n2 )y = 0, y(0) finite, y(a) = 0. (†)

Here n ≥ 0 is an integer.

(a) (i) Write the equation (†) in Sturm-Liouville form.

(ii) Briefly explain the difference between regular and singular Sturm-Liouville
eigenvalue problems and classify (†).
(iii) Use the results given at the end of the question to find the eigenvalues {λk }
and eigenfunctions {yk (x)} of (†). Write down the orthogonality condition
satisfied by the eigenfunctions.
(b) The generating function formula for Bessel functions is
   ∞
x 1 X
exp t− = tm Jm (x).
2 t m=−∞

Use this to obtain the results

(i)
x

mJm (x) = Jm−1 (x) + Jm+1 (x) .
2
(ii)
π
1
Z
Jm (x) = cos (x sin θ − mθ) dθ.
2π −π

Hint for (ii): use the substitution t = exp (iθ).

You may use the fact that the only solution of Bessel’s equation

z 2 w′′ + zw′ + (z 2 − n2 )w = 0, (n ≥ 0, integer)

that is not singular at z = 0 is the Bessel function of the first kind Jn (z). This
function has infinitely many real zeroes {jnm , m ∈ N} satisfying Jn (jnm ) = 0.

MATH7402 CONTINUED

4
Downloaded by Roy Vesey (rdv0044@googlemail.com)
 lOMoARcPSD|3449400

6. A real function f (x) and its Fourier transform fˆ(k) are related through
Z ∞ Z ∞
ˆ 1 1
f (k) = √ f (x)e −ikx
dx and f (x) = √ fˆ(k)eikx dk.
2π −∞ 2π −∞

(a) Find the Fourier transform of


 1 0≤x<1
F (x) = −1 −1 < x < 0 .
0 |x| ≥ 1


(b) Find the Fourier transform of the convolution

Z ∞
(f ∗ g)(x) = f (x − y)g(y) dy.
−∞

(c) Show that the solution u(x, y) to the boundary value problem in the half-plane

∇2 u = 0 in − ∞ < x < ∞, y > 0,

u(x, 0) = f (x) with f (x) ∈ L1 (R)
u(x, y) → 0 as x2 + y 2 → ∞,

is given by
∞
y f (x − t)
Z
u(x, y) = dt.
π −∞ t2 + y 2
(d) Evaluate u(x, y) for the function f (x) = F (x) given in part (a).
(e) Verify that the solution you have found in (d) satisfies

lim u(x, y) = F (x).

y→0

MATH7402 END OF PAPER

5
Downloaded by Roy Vesey (rdv0044@googlemail.com)