MATHEMATICAL TRIPOS Part II

Monday, 3 June, 2019 1:30 pm to 4:30 pm MAT2

PAPER 1

Before you begin read these instructions carefully.

The examination paper is divided into two sections. Each question in Section II
carries twice the number of marks of each question in Section I. Section II questions
also carry an alpha or beta quality mark and Section I questions carry a beta quality
mark.
Candidates may obtain credit from attempts on at most six questions from Section
I and from any number of questions from Section II.
Write on one side of the paper only and begin each answer on a separate sheet.
Write legibly; otherwise you place yourself at a grave disadvantage.

At the end of the examination:

Tie up your answers in bundles, marked A, B, C, . . ., J according to the code letter
affixed to each question. Include in the same bundle all questions from Sections I
and II with the same code letter.
Attach a completed gold cover sheet to each bundle.
You must also complete a green master cover sheet listing all the questions you have
attempted.
Every cover sheet must bear your examination number and desk number.

STATIONERY REQUIREMENTS
Gold cover sheet
Green master cover sheet
Script paper
Rough paper

You may not start to read the questions

printed on the subsequent pages until
instructed to do so by the Invigilator.
 2

SECTION I
1I Number Theory
(a) State and prove the Chinese remainder theorem.
(b) Let N be an odd positive composite integer, and b a positive integer with
(b, N ) = 1. What does it mean to say that N is a Fermat pseudoprime to base b? Show
that 35 is a Fermat pseudoprime to base b if and only if b is congruent to one of 1, 6, 29
or 34 (mod 35).

2H Topics in Analysis
P∞ Let Tn be the nth Chebychev polynomial.
P∞ Suppose that γi > 0 for all i and that
i=1 γi converges. Explain why f = i=1 γi T3i is a well defined continuous function on
[−1, 1].
P
Show that, if we take Pn = ni=1 γi T3i , we can find points xk with

−1 6 x0 < x1 < . . . < x3n+1 6 1

P
such that f (xk ) − Pn (xk ) = (−1)k+1 ∞i=n+1 γi for each k = 0, 1, . . . , 3
n+1 .

Suppose that δn is a decreasing sequence of positive numbers and that δn → 0 as

n → ∞. Stating clearly any theorem that you use, show that there exists a continuous
function f with
sup |f (t) − P (t)| > δn
t∈[−1,1]

for all polynomials P of degree at most n and all n > 1.

3G Coding and Cryptography

Let X and Y be discrete random variables taking finitely many values. Define the
conditional entropy H(X|Y ). Suppose Z is another discrete random variable taking values
in a finite alphabet, and prove that

H(X|Y ) 6 H(X|Y, Z) + H(Z).

[You may use the equality H(X, Y ) = H(X|Y ) + H(Y ) and the inequality H(X|Y ) 6
H(X).]
State and prove Fano’s inequality.

Part II, Paper 1

 3

4H Automata and Formal Languages

(a) State the pumping lemma for context-free languages (CFLs).
(b) Which of the following are CFLs? Justify your answers.

(i) {wwR | w ∈ {a, b}∗ }, where wR is the reverse of the word w.

(ii) {0p 1p | p is a prime}.

(iii) {am bn ck dl | 3m = 4l and 2n = 5k}.

(c) Let L and M be CFLs. Show that the concatenation LM is also a CFL.

5J Statistical Modelling
The Gamma distribution with shape parameter α > 0 and scale parameter λ > 0
has probability density function
λα α−1 −λy
f (y; α, λ) = y e for y > 0.
Γ(α)

Give the definition of an exponential dispersion family and show that the set of Gamma
distributions forms one such family. Find the cumulant generating function and derive
the mean and variance of the Gamma distribution as a function of α and λ.

Part II, Paper 1 [TURN OVER]

 4

6C Mathematical Biology
An animal population has annual dynamics, breeding in the summer and hibernating
through the winter. At year t, the number of individuals alive who were born a years ago
is given by na,t . Each individual of age a gives birth to ba offspring, and after the summer
has a probability µa of dying during the winter. [You may assume that individuals do not
give birth during the year in which they are born.]
Explain carefully why the following equations, together with initial conditions, are
appropriate to describe the system:
∞
X
n0,t = na,t ba
a=1
na+1,t+1 = (1 − µa )na,t ,

Seek a solution of the form na,t = ra γ t where γ and ra , for a = 1, 2, 3 . . ., are

constants. Show γ must satisfy φ(γ) = 1 where

∞ a−1
!
X Y
φ(γ) = (1 − µi ) γ −a ba .
a=1 i=0

Explain why, for any reasonable set of parameters µi and bi , the equation φ(γ) = 1
has a unique solution. Explain also how φ(1) can be used to determine if the population will
grow or shrink.

Part II, Paper 1

 5

7A Further Complex Methods

The Beta function is defined by
Z 1
Γ(p)Γ(q)
B(p, q) := tp−1 (1 − t)q−1 dt = ,
0 Γ(p + q)

where Re p > 0, Re q > 0, and Γ is the Gamma function.

(a) By using a suitable substitution and properties of Beta and Gamma functions,
show that Z 1
dx [Γ(1/4)]2
√ = √ .
0 1 − x4 32π

(b) Deduce that

 √  4 [Γ(5/4)] 2
K 1/ 2 = √ ,
π
where K(k) is the complete elliptic integral, defined as
Z 1
dt
K(k) := p .
0 (1 − t2 )(1 − k 2 t2 )

[Hint: You might find the change of variable x = t(2 − t2 )−1/2 helpful in part (b).]

8E Classical Dynamics
(a) A mechanical system with n degrees of freedom has the Lagrangian L(q, q̇),
where q = (q1 , . . . , qn ) are the generalized coordinates and q̇ = dq/dt.
Suppose that L is invariant under the continuous symmetry transformation q(t) 7→
Q(s, t), where s is a real parameter and Q(0, t) = q(t). State and prove Noether’s theorem
for this system.
(b) A particle of mass m moves in a conservative force field with potential energy
V (r), where r is the position vector in three-dimensional space.
Let (r, φ, z) be cylindrical polar coordinates. V (r) is said to have helical symmetry
if it is of the form
V (r) = f (r, φ − kz) ,
for some constant k. Show that a particle moving in a potential with helical symmetry
has a conserved quantity that is a linear combination of angular and linear momenta.

Part II, Paper 1 [TURN OVER]

 6

9B Cosmology
[You may work in units of the speed of light, so that c = 1.]
By considering a spherical distribution of matter with total mass M and radius R
and an infinitesimal mass δm located somewhere on its surface, derive the Friedmann
equation describing the evolution of the scale factor a(t) appearing in the relation
R(t) = R0 a(t)/a(t0 ) for a spatially-flat FLRW spacetime.
Consider now a spatially-flat, contracting universe filled by a single component with
energy density ρ, which evolves with time as ρ(t) = ρ0 [a(t)/a(t0 )]−4 . Solve the Friedmann
equation for a(t) with a(t0 ) = 1.

10D Quantum Information and Computation

Introduce the 2-qubit states
 
z x |00i + |11i
|βxz i = (Z X ) ⊗ I √ ,
2

where X and Z are the standard qubit Pauli operations and x, z ∈ {0, 1}.
(a) For any 1-qubit state |αi show that the 3-qubit state |αiC |β00 iAB of system
CAB can be expressed as
1
1 X
|αiC |β00 iAB = |βxz iCA |µxz iB ,
2
x,z=0

where the 1-qubit states |µxz i are uniquely determined. Show that |µ10 i = X |αi.
(b) In addition to |µ10 i = X |αi you may now assume that |µxz i = X x Z z |αi. Alice
and Bob are separated distantly in space and share a |β00 iAB state with A and B labelling
qubits held by Alice and Bob respectively. Alice also has a qubit C in state |αi whose
identity is unknown to her. Using the results of part (a) show how she can transfer the
state of C to Bob using only local operations and classical communication, i.e. the sending
of quantum states across space is not allowed.
(c) Suppose that in part (b), while sharing the |β00 iAB state, Alice and Bob are
also unable to engage in any classical communication, i.e. they are able only to perform
local operations. Can Alice now, perhaps by a modified process, transfer the state of C
to Bob? Give a reason for your answer.

Part II, Paper 1

 7

SECTION II
11G Coding and Cryptography
What does it mean to say that C is a binary linear code of length n, rank k and
minimum distance d? Let C be such a code.
(a) Prove that n > d + k − 1.
Let x = (x1 , . . . , xn ) ∈ C be a codeword with exactly d non-zero digits.
(b) Prove that puncturing C on the non-zero digits of x produces a code C ′ of length
n − d, rank k − 1 and minimum distance d′ for some d′ > ⌈ d2 ⌉.
P
(c) Deduce that n > d + 16l6k−1 ⌈ 2dl ⌉.

12H Automata and Formal Languages

Let D = (Q, Σ, δ, q0 , F ) be a deterministic finite-state automaton (DFA). Define
what it means for two states of D to be equivalent. Define the minimal DFA D/ ∼ for D.
Let D be a DFA with no inaccessible states, and suppose that A is another DFA on
the same alphabet as D and for which L(D) = L(A). Show that A has at least as many
states as D/ ∼. [You may use results from the course as long as you state them clearly.]
Construct a minimal DFA (that is, one with the smallest possible number of states)
over the alphabet {0, 1} which accepts precisely the set of binary numbers which are
multiples of 7. You may have leading zeros in your inputs (e.g.: 00101). Prove that your
DFA is minimal by finding a distinguishing word for each pair of states.

Part II, Paper 1 [TURN OVER]

 8

13J Statistical Modelling

The ice cream data frame contains the result of a blind tasting of 90 ice creams,
each of which is rated as poor, good, or excellent. It also contains the price of each ice
cream classified into three categories. Consider the R code below and its output.

> table(ice_cream)
score
price excellent good poor
high 12 8 10
low 7 9 14
medium 12 11 7
>
> ice_cream.counts = as.data.frame(xtabs(Freq ~ price + score-1, data=table(ice_cream)))
> glm.fit = glm(Freq ~ price + score,data=ice_cream.counts,family="poisson")
> summary(glm.fit)

Call:
glm(formula = Freq ~ price + score - 1, family = "poisson", data = ice_cream.counts)

Deviance Residuals:
1 2 3 4 5 6 7 8 9
0.5054 -1.1019 0.5054 -0.4475 -0.1098 0.5304 -0.1043 1.0816 -1.1019

Coefficients:
Estimate Std. Error z value Pr(>|z|)
pricehigh 2.335e+00 2.334e-01 10.01 <2e-16 ***
pricelow 2.335e+00 2.334e-01 10.01 <2e-16 ***
pricemedium 2.335e+00 2.334e-01 10.01 <2e-16 ***
scoregood -1.018e-01 2.607e-01 -0.39 0.696
scorepoor 3.892e-14 2.540e-01 0.00 1.000
---
Signif. codes: 0 ’***’ 0.001 ’**’ 0.01 ’*’ 0.05 ’.’ 0.1 ’ ’ 1

(Dispersion parameter for poisson family taken to be 1)

Null deviance: 257.2811 on 9 degrees of freedom

Residual deviance: 4.6135 on 4 degrees of freedom
AIC: 51.791

(a) Write down the generalised linear model fitted by the code above.
(b) Prove that the fitted values resulting from the maximum likelihood estimator of
the coefficients in this model are identical to those resulting from the maximum likelihood
estimator when fitting a Multinomial model which assumes the number of ice creams at
each price level is fixed.
(c) Using the output above, perform a goodness-of-fit test at the 1% level, specifying
the null hypothesis, the test statistic, its asymptotic null distribution, any assumptions of
the test and the decision from your test.

Part II, Paper 1

 9

(d) If we believe that better ice creams are more expensive, what could be a more
powerful test against the model fitted above and why?

14A Further Complex Methods

(a) Consider the Papperitz symbol (or P-symbol):
 
 a b c 
P α β γ z . (†)
 ′ ′ ′ 
α β γ

Explain in general terms what this P -symbol represents.

[You need not write down any differential equations explicitly, but should provide an
explanation of the meaning of a, b, c, α, β, γ, α′ , β ′ and γ ′ .]
(b) Prove that the action of [(z−a)/(z−b)]δ on (†) results in the exponential shifting,
 
 a b c 
P α+δ β−δ γ z . (‡)
 ′ 
α + δ β′ − δ γ′
[Hint: It may prove useful to start by considering the relationship between two solutions, ω
and ω1 , which satisfy the P -equations described by the respective P -symbols (†) and (‡).]
(c) Explain what is meant by a Möbius transformation of a second order differential
equation. By using suitable transformations acting on (†), show how to obtain the P -
symbol
 
 0 1 ∞ 
P 0 0 a z , (⋆)
 
1−c c−a−b b
which corresponds to the hypergeometric equation.
(d) The hypergeometric function F (a, b, c; z) is defined to be the solution of the
differential equation corresponding to (⋆) that is analytic at z = 0 with F (a, b, c; 0) = 1,
which corresponds to the exponent zero. Use exponential shifting to show that the second
solution, which corresponds to the exponent 1 − c, is

z 1−c F (a − c + 1, b − c + 1, 2 − c; z).

Part II, Paper 1 [TURN OVER]

 10

15B Cosmology
[You may work in units of the speed of light, so that c = 1.]
Consider a spatially-flat FLRW universe with a single, canonical, homogeneous
scalar field φ(t) with a potential V (φ). Recall the Friedmann equation and the Ray-
chaudhuri equation (also known as the acceleration equation)
 2  
ȧ 2 8πG 1 2
=H = φ̇ + V ,
a 3 2
ä 8πG  2 
=− φ̇ − V .
a 3

(a) Assuming φ̇ 6= 0, derive the equations of motion for φ, i.e.

φ̈ + 3H φ̇ + ∂φ V = 0 .

(b) Assuming the special case V (φ) = λφ4 , find φ(t), for some initial value φ(t0 ) = φ0
in the slow-roll approximation, i.e. assuming that φ̇2 ≪ 2V and φ̈ ≪ 3H φ̇.
(c) The number N of efoldings is defined by dN = d ln a. Using the chain rule,
express dN first in terms of dt and then in terms of dφ. Write the resulting relation
between dN and dφ in terms of V and ∂φ V only, using the slow-roll approximation.
(d) Compute the number N of efoldings of expansion between some initial value
φi < 0 and a final value φf < 0 (so that φ̇ > 0 throughout).
(e) Discuss qualitatively the horizon and flatness problems in the old hot big bang
model (i.e. without inflation) and how inflation addresses them.

16I Logic and Set Theory

State the completeness theorem for propositional logic. Explain briefly how the
proof of this theorem changes from the usual proof in the case when the set of primitive
propositions may be uncountable.
State the compactness theorem and the decidability theorem, and deduce them from
the completeness theorem.
A poset (X, <) is called two-dimensional if there exist total orders <1 and <2 on X
such that x < y if and only if x <1 y and x <2 y. By applying the compactness theorem
for propositional logic, show that if every finite subset of a poset is two-dimensional then
so is the poset itself.
[Hint: Take primitive propositions px,y and qx,y , for each distinct x, y ∈ X, with the
intended interpretation that px,y is true if and only if x <1 y and qx,y is true if and only
if x <2 y.]

Part II, Paper 1

 11

17G Graph Theory

Let G be a connected d-regular graph.
(a) Show that d is an eigenvalue of G with multiplicity 1 and eigenvector

e = (1 1 . . . 1)T .

(b) Suppose that G is strongly regular. Show that G has at most three distinct
eigenvalues.
(c) Conversely, suppose that G has precisely three distinct eigenvalues d, λ and µ.
Let A be the adjacency matrix of G and let

B = A2 − (λ + µ)A + λµI.

Show that if v is an eigenvector of G that is not a scalar multiple of e then Bv = 0. Deduce

that B is a scalar multiple of the matrix J whose entries are all equal to one. Hence show
that, for i 6= j, (A2 )ij depends only on whether or not vertices i and j are adjacent, and
so G is strongly regular.
(d) Which connected d-regular graphs have precisely two eigenvalues? Justify your
answer.

18F Galois Theory

(a) Suppose K, L are fields and σ1 , . . . , σm are distinct embeddings of K into L.
Prove that there do not exist elements λ1 , . . . , λm of L (not all zero) such that

λ1 σ1 (x) + · · · + λm σm (x) = 0 for all x ∈ K.

(b) For a finite field extension K of a field k and for σ1 , . . . , σm distinct k-

automorphisms of K, show that m 6 [K : k]. In particular, if G is a finite group of
field automorphisms of a field K with K G the fixed field, deduce that |G| 6 [K : K G ].
(c) If K = Q(x, y) with x, y independent transcendentals over Q, consider the group
G generated by automorphisms σ and τ of K, where

σ(x) = y, σ(y) = −x and τ (x) = x, τ (y) = −y.

Prove that |G| = 8 and that K G = Q(x2 + y 2 , x2 y 2 ).

Part II, Paper 1 [TURN OVER]

 12

19I Representation Theory

(a) State and prove Schur’s lemma over C.
In the remainder of this question we work over R.
(b) Let G be the cyclic group of order 3.

(i) Write the regular RG-module as a direct sum of irreducible submodules.

(ii) Find all the intertwining homomorphisms between the irreducible RG-modules.
Deduce that the conclusion of Schur’s lemma is false if we replace C by R.

(c) Henceforth let G be a cyclic group of order n. Show that

(i) if n is even, the regular RG-module is a direct sum of two (non-isomorphic) 1-

dimensional irreducible submodules and (n−2)/2 (non-isomorphic) 2-dimensional
irreducible submodules;

(ii) if n is odd, the regular RG-module is a direct sum of one 1-dimensional irre-
ducible submodule and (n − 1)/2 (non-isomorphic) 2-dimensional irreducible
submodules.

20G Number Fields

√
Let K = Q( 2).
(a) Write down the ring of integers OK .
(b) State Dirichlet’s unit theorem, and use it to determine all elements of the group
×
of units OK .
√
(c) Let P ⊂ OK denote the ideal generated by 3 + 2. Show that the group
×
G = {α ∈ OK | α ≡ 1 mod P }

is cyclic, and find a generator.

Part II, Paper 1

 13

21F Algebraic Topology

In this question, X and Y are path-connected, locally simply connected spaces.
(a) Let f : Y → X be a continuous map, and X b a path-connected covering space of
b
X. State and prove a uniqueness statement for lifts of f to X.
(b) Let p : X b → X be a covering map. A covering transformation of p is a
homeomorphism φ : X b → X b such that p ◦ φ = p. For each integer n > 3, give an
example of a space X and an n-sheeted covering map pn : X bn → X such that the only
covering transformation of pn is the identity map. Justify your answer. [Hint: Take X to
be a wedge of two circles.]
(c) Is there a space X and a 2-sheeted covering map p2 : X b2 → X for which the
only covering transformation of p2 is the identity? Justify your answer briefly.

22H Linear Analysis

Let F be the space of real-valued sequences with only finitely many nonzero terms.
(a) For any p ∈ [1, ∞), show that F is dense in ℓp . Is F dense in ℓ∞ ? Justify your
answer.
(b) Let p ∈ [1, ∞), and let T : F → F be an operator that is bounded in the
k · kp -norm, i.e., there exists a C such that kT xkp 6 Ckxkp for all x ∈ F . Show that there
is a unique bounded operator Te : ℓp → ℓp satisfying Te|F = T , and that kTekp 6 C.
(c) For each p ∈ [1, ∞] and for each i = 1, . . . , 5 determine if there is a bounded
operator from ℓp to ℓp (in the k · kp norm) whose restriction to F is given by Ti :
xn
(T1 x)n = nxn , (T2 x)n = n(xn − xn+1 ), (T3 x)n = ,
Pn n
x1 j=1 xj
(T4 x)n = 1/2 , (T5 x)n = .
n 2n

(d) Let X be a normed vector space such that the closed unit ball B1 (0) is compact.
Prove that X is finite dimensional.

Part II, Paper 1 [TURN OVER]

 14

23H Analysis of Functions

(a) Consider the topology T on the natural numbers N ⊂ R induced by the standard
topology on R. Prove it is the discrete topology; i.e. T = P(N) is the power set of N.
(b) Describe the corresponding Borel sets on N and prove that any function
f : N → R or f : N → [0, +∞] is measurable.
P
(c) Using Lebesgue integration
P theory, define n>1 f (n) ∈ [0, +∞] for a function
f : N → [0, +∞] and then n>1 f (n) ∈ C for f : N → C. State any condition needed for
the sum of the latter series to be defined. What is a simple function in this setting, and
which simple functions have finite sum?
(d) State and prove the Beppo Levi theorem (also known as the monotone conver-
gence theorem).
(e) Consider f : R × N → [0, +∞] such that for any n ∈ N, the function t 7→ f (t, n)
is non-decreasing. Prove that
X X
lim f (t, n) = lim f (t, n).
t→∞ t→∞
n>1 n>1

Show that this need not be the case if we drop the hypothesis that t 7→ f (t, n) is non-
decreasing, even if all the relevant limits exist.

24F Riemann Surfaces

Define X ′ := {(x, y) ∈ C2 : x3 y + y 3 + x = 0}.
(a) Prove by defining an atlas that X ′ is a Riemann surface.
(b) Now assume that by adding finitely many points, it is possible to compactify X ′
to a Riemann surface X so that the coordinate projections extend to holomorphic maps
πx and πy from X to C∞ . Compute the genus of X.
(c) Assume that any holomorphic automorphism of X ′ extends to a holomorphic
automorphism of X. Prove that the group Aut(X) of holomorphic automorphisms of X
contains an element φ of order 7. Prove further that there exists a holomorphic map
π : X → C∞ which satisfies π ◦ φ = π.

Part II, Paper 1

 15

25F Algebraic Geometry

(a) Let k be an algebraically closed field of characteristic 0. Consider the algebraic
variety V ⊂ A3 defined over k by the polynomials

xy, y 2 − z 3 + xz, and x(x + y + 2z + 1).

Determine
(i) the irreducible components of V ,

(ii) the tangent space at each point of V ,

(iii) for each irreducible component, the smooth points of that component, and

(iv) the dimensions of the irreducible components.

(b) Let L ⊇ K be a finite extension of fields, and dimK L = n. Identify L with An

over K and show that
U = {α ∈ L | K[α] = L}
is the complement in An of the vanishing set of some polynomial. [You need not show
that U is non-empty. You may assume that K[α] = L if and only if 1, α, . . . , αn−1 form a
basis of L over K.]

26H Differential Geometry

Let n > 1 be an integer.
(a) Show that Sn = {x ∈ Rn+1 : x21 + · · · + x2n+1 = 1} defines a submanifold of Rn+1
and identify explicitly its tangent space Tx Sn for any x ∈ Sn .
2
(b) Show that the matrix group SO(n) ⊂ Rn defines a submanifold. Identify
explicitly the tangent space TR SO(n) for any R ∈ SO(n).
(c) Given v ∈ Sn , show that the set Sv = {R ∈ SO(n + 1) : Rv = v} defines a
submanifold Sv ⊂ SO(n + 1) and compute its dimension. For v 6= w, is it ever the case
that Sv and Sw are transversal?
[You may use standard theorems from the course concerning regular values and
transversality.]

Part II, Paper 1 [TURN OVER]

 16

27K Probability and Measure

Let X = (X1 , . . . , Xd ) be an Rd -valued random variable. Given u = (u1 , . . . , ud ) ∈
Rd we let
φX (u) = E(eihu,Xi )
be its characteristic function, where h·, ·i is the usual inner product on Rd .
(a) Suppose X is a Gaussian vector with mean 0 and covariance matrix σ 2 Id , where
σ > 0 and Id is the d × d identity matrix. What is the formula for the characteristic
function φX in the case d = 1? Derive from it a formula for φX in the case d > 2.
(b) We now no longer assume that X is necessarily a Gaussian vector. Instead we
assume that the Xi ’s are independent random variables and that the random vector AX
has the same law as X for every orthogonal matrix A. Furthermore we assume that d > 2.

(i) Show that there exists a continuous function f : [0, +∞) → R such that

φX (u) = f (u21 + . . . + u2d ).

[ You may use the fact that for every two vectors u, v ∈ Rd such that
hu, ui = hv, vi there is an orthogonal matrix A such that Au = v. ]

(ii) Show that for all r1 , r2 > 0

f (r1 + r2 ) = f (r1 )f (r2 ).

(iii) Deduce that f takes values in (0, 1], and furthermore that there exists α > 0
such that f (r) = e−rα , for all r > 0.

(iv) What must be the law of X?

[Standard properties of characteristic functions from the course may be used without
proof if clearly stated.]

Part II, Paper 1

 17

28K Applied Probability

Let S be a countable set, and let P = (pi,j : i, j ∈ S) be a Markov transition matrix
with pi,i = 0 for all i. Let Y = (Yn : n = 0, 1, 2, . . . ) be a discrete-time Markov chain on
the state space S with transition matrix P .
The continuous-time process X = (Xt : t > 0) is constructed as follows. Let
(Um : m = 0, 1, 2, . . .) be independent, identically distributed random variables having the
exponential distribution with mean 1. Let g be a function on S such that ε < g(i) < 1ε
for allPi ∈ S and some constant ε > 0. Let Vm = Um /g(Ym ) for m > 0. Let T0 = 0 and
n−1
Tn = m=0 Vm for n > 1. Finally, let Xt = Yn for Tn 6 t < Tn+1 .
(a) Explain briefly why X is a continuous-time Markov chain on S, and write down
its generator in terms of P and the vector g = (g(i) : i ∈ S).
(b) What does it mean to say that the chain X is irreducible? What does it mean
to say a state i ∈ S is (i) recurrent and (ii) positive recurrent?
(c) Show that

(i) X is irreducible if and only if Y is irreducible;

(ii) X is recurrent if and only if Y is recurrent.

(d) Suppose Y is irreducible and positive recurrent with invariant distribution π.

Express the invariant distribution of X in terms of π and g.

Part II, Paper 1 [TURN OVER]

 18

29J Principles of Statistics

In a regression problem, for a given X ∈ Rn×p fixed, we observe Y ∈ Rn such that

Y = Xθ0 + ε

for an unknown θ0 ∈ Rp and ε random such that ε ∼ N (0, σ 2 In ) for some known σ 2 > 0.
(a) When p 6 n and X has rank p, compute the maximum likelihood estimator
θ̂M LE for θ0 . When p > n, what issue is there with the likelihood maximisation approach
and how many maximisers of the likelihood are there (if any)?
(b) For any λ > 0 fixed, we consider θ̂λ minimising

kY − Xθk22 + λkθk22

over Rp . Derive an expression for θ̂λ and show it is well defined, i.e., there is a unique
minimiser for every X, Y and λ.
Assume p 6 n and that X has rank p. Let Σ = X ⊤ X and note that Σ = V ΛV ⊤
for some orthogonal matrix V and some diagonal matrix Λ whose diagonal entries satisfy
Λ1,1 > Λ2,2 > . . . > Λp,p . Assume that the columns of X have mean zero.
(c) Denote the columns of U = XV by u1 , . . . , up . Show that they are sample
principal components, i.e., that their pairwise sample correlations are zero and that they
have sample variances n−1 Λ1,1 , . . . , n−1 Λp,p , respectively. [Hint: the sample covariance
between ui and uj is n−1 u⊤i uj .]

(d) Show that

ŶM LE = X θ̂M LE = U Λ−1 U ⊤ Y.
Conclude that prediction ŶM LE is the closest point to Y within the subspace spanned by
the normalised sample principal components of part (c).
(e) Show that
Ŷλ = X θ̂λ = U (Λ + λIp )−1 U ⊤ Y.
Assume Λ1,1 , Λ2,2 , . . . , Λq,q >> λ >> Λq+1,q+1 , . . . , Λp,p for some 1 6 q < p. Conclude
that prediction Ŷλ is approximately the closest point to Y within the subspace spanned
by the q normalised sample principal components of part (c) with the greatest variance.

Part II, Paper 1

 19

30K Stochastic Financial Models

(a) What does it mean to say that (Mn , Fn )n>0 is a martingale?
(b) Let (Xn )n>0 be a Markov chain defined by X0 = 0 and
    1
P Xn = 1 | Xn−1 = 0 = P Xn = −1 | Xn−1 = 0 = ,
2n
  1
P Xn = 0 | Xn−1 = 0 = 1 −
n
and
  1   1
P Xn = nXn−1 | Xn−1 6= 0 = , P Xn = 0 | Xn−1 6= 0 = 1 −
n n
for n > 1. Show that (Xn )n>0 is a martingale with respect to the filtration (Fn )n>0 where
F0 is trivial and Fn = σ(X1 , . . . , Xn ) for n > 1.
(c) Let M = (Mn )n>0 be adapted with respect to a filtration (Fn )n>0 with
E[|Mn |] < ∞ for all n. Show that the following are equivalent:

(i) M is a martingale.

(ii) For every stopping time τ , the stopped process M τ defined by Mnτ := Mn∧τ ,
n > 0, is a martingale.

(iii) E[Mn∧τ ] = E[M0 ] for all n > 0 and every stopping time τ .

[Hint: To show that (iii) implies (i) you might find it useful to consider the stopping time
(
n if ω ∈ A,
T (ω) :=
n + 1 if ω 6∈ A,

for any A ∈ Fn .]

Part II, Paper 1 [TURN OVER]

 20

31E Dynamical Systems

For a dynamical system of the form ẋ = f (x), give the definition of the alpha-limit
set α(x) and the omega-limit set ω(x) of a point x.
Consider the dynamical system

ẋ = x2 − 1 ,
ẏ = kxy ,

where x = (x, y) ∈ R2 and k is a real constant. Answer the following for all values of k,
taking care over boundary cases (both in k and in x).

(i) What symmetries does this system have?

(ii) Find and classify the fixed points of this system.

(iii) Does this system have any periodic orbits?

(iv) Give α(x) and ω(x) (considering all x ∈ R2 ).

(v) For x0 = (0, y0 ), give the orbit of x0 (considering all y0 ∈ R). You should give
your answer in the form y = y(x, y0 , k), and specify the range of x.

Part II, Paper 1

 21

32C Integrable Systems

Let M = R2n = {(q, p)|q, p ∈ Rn } be equipped with its standard Poisson bracket.
(a) Given a Hamiltonian function H = H(q, p), write down Hamilton’s equations
for (M, H). Define a first integral of the system and state what it means that the system
is integrable.
(b) Show that if n = 1 then every Hamiltonian system is integrable whenever
 
∂H ∂H
, 6= 0.
∂q ∂p

Let M̃ = R2m = {(q̃, p̃)|q̃, p̃ ∈ Rm } be another phase space, equipped with its
standard Poisson bracket. Suppose that H̃ = H̃(q̃, p̃) is a Hamiltonian function for M̃ .
Define Q = (q1 , . . . , qn , q̃1 , . . . , q̃m ), P = (p1 , . . . , pn , p̃1 , . . . , p̃m ) and let the combined
phase space M = R2(n+m) = {(Q, P)} be equipped with the standard Poisson bracket.
(c) Show that if (M, H) and (M̃ , H̃) are both integrable, then so is (M, H), where
the combined Hamiltonian is given by:

H(Q, P) = H(q, p) + H̃(q̃, p̃).

(d) Consider the n−dimensional simple harmonic oscillator with phase space M and
Hamiltonian H given by:
1 1 1 1
H = p21 + . . . + p2n + ω12 q12 + . . . + ωn2 qn2 ,
2 2 2 2
where ωi > 0. Using the results above, or otherwise, show that (M, H) is integrable for
(q, p) 6= 0.
(e) Is it true that every bounded orbit of an integrable system is necessarily periodic?
You should justify your answer.

Part II, Paper 1 [TURN OVER]

 22

33B Principles of Quantum Mechanics

A d = 3 isotropic harmonic oscillator of mass µ and frequency ω has lowering
operators
1
A= √ (µωX + iP) ,
2µ~ω
where X and P are the position and momentum operators. Assuming the standard
commutation relations for X and P, evaluate the commutators [A†i , A†j ], [Ai , Aj ] and
[Ai , A†j ], for i, j = 1, 2, 3, among the components of the raising and lowering operators.
How is the ground state |0i of the oscillator defined? How are normalised higher
excited states obtained from |0i? [You should determine the appropriate normalisation
constant for each energy eigenstate.]
By expressing the orbital angular momentum operator L in terms of the raising and
lowering operators, show that each first excited state of the isotropic oscillator has total
orbital angular momentum quantum number ℓ = 1, and find a linear combination |ψi of
these first excited states obeying Lz |ψi = +~|ψi and k|ψik = 1.

34B Applications of Quantum Mechanics

A particle of mass m and charge q moving in a uniform magnetic field B = ∇ × A =
(0, 0, B) and electric field E = −∇φ is described by the Hamiltonian
1
H= |p − qA|2 + qφ ,
2m
where p is the canonical momentum.
[ In the following you may use without proof any results concerning the spectrum
of the harmonic oscillator as long as they are stated clearly.]
(a) Let E = 0. Choose a gauge which preserves translational symmetry in the y-
direction. Determine the spectrum of the system, restricted to states with pz = 0. The
system is further restricted to lie in a rectangle of area A = Lx Ly , with sides of length Lx
and Ly parallel to the x- and y-axes respectively. Assuming periodic boundary conditions
in the y-direction, estimate the degeneracy of each Landau level.
(b) Consider the introduction of an additional electric field E = (E, 0, 0). Choosing
a suitable gauge (with the same choice of vector potential A as in part (a)), write down
the resulting Hamiltonian. Find the energy spectrum for a particle on R3 again restricted
to states with pz = 0.
Define the group velocity of the electron and show that its y-component is given by
vy = −E/B.
When the system is further restricted to a rectangle of area A as above, show that
the previous degeneracy of the Landau levels is lifted and determine the resulting energy
gap ∆E between the ground-state and the first excited state.

Part II, Paper 1

 23

35D Statistical Physics

(a) Explain, from a macroscopic and microscopic point of view, what is meant by
an adiabatic change. A system has access to heat baths at temperatures T1 and T2 , with
T2 > T1 . Show that the most effective method for repeatedly converting heat to work,
using this system, is by combining isothermal and adiabatic changes. Define the efficiency
and calculate it in terms of T1 and T2 .
(b) A thermal system (of constant volume) undergoes a phase transition at temper-
ature Tc . The heat capacity of the system is measured to be
(
αT for T < Tc
C=
β for T > Tc ,

where α, β are constants. A theoretical calculation of the entropy S for T > Tc leads to

S = β log T + γ.

How can the value of the theoretically-obtained constant γ be verified using macroscopi-
cally measurable quantities?

Part II, Paper 1 [TURN OVER]

 24

36E Electrodynamics
A relativistic particle of charge q and mass m moves in a background electromagnetic
field. The four-velocity uµ (τ ) of the particle at proper time τ is determined by the equation
of motion,
duµ
m = qF µν uν .
dτ
Here F µν = ηνρ F µρ , where Fµν is the electromagnetic field strength tensor and Lorentz
indices are raised and lowered with the metric tensor η = diag{−1, +1, +1, +1}. In the
case of a constant, homogeneous field, write down the solution of this equation giving
uµ (τ ) in terms of its initial value uµ (0) .
[In the following you may use the relation, given below, between the components of
the field strength tensor Fµν , for µ, ν = 0, 1, 2, 3, and those of the electric and magnetic
fields E = (E1 , E2 , E3 ) and B = (B1 , B2 , B3 ),
1
Fi0 = −F0i = Ei , Fij = εijk Bk
c
for i, j = 1, 2, 3.]
Suppose that, in some inertial frame with spacetime coordinates x = (x, y, z) and
t, the electric and magnetic fields are parallel to the x-axis with magnitudes E and B
respectively. At time t = τ = 0 the 3-velocity v = dx/dt of the particle has initial value
v(0) = (0, v0 , 0). Find the subsequent trajectory of the particle in this frame, giving
coordinates x, y, z and t as functions of the proper time τ .
Find the motion in the x-direction explicitly, giving x as a function of coordinate
time t. Comment on the form of the solution at early and late times. Show that, when
projected onto the y-z plane, the particle undergoes circular motion which is periodic in
proper time. Find the radius R of the circle and proper time period of the motion ∆τ in
terms of q, m, E, B and v0 . The resulting trajectory therefore has the form of a helix
with varying pitch Pn := ∆xn /R where ∆xn is the distance in the x-direction travelled by
the particle during the n’th period of its motion in the y-z plane. Show that, for n ≫ 1,
 
2πEn
Pn ∼ A exp ,
cB

where A is a constant which you should determine.

Part II, Paper 1

 25

37D General Relativity

Let (M, g) be a spacetime and Γ the Levi-Civita connection of the metric g. The Riemann
tensor of this spacetime is given in terms of the connection by

Rγ ραβ = ∂α Γγρβ − ∂β Γγρα + Γµρβ Γγµα − Γµρα Γγµβ .

The contracted Bianchi identities ensure that the Einstein tensor satisfies

∇µ Gµν = 0 .

(a) Show that the Riemann tensor obeys the symmetry

Rµ ραβ + Rµ βρα + Rµ αβρ = 0 .

(b) Show that a vector field V α satisfies the Ricci identity

2∇[α ∇β] V γ = ∇α ∇β V γ − ∇β ∇α V γ = Rγ ραβ V ρ .

2

Calculate the analogous expression for a rank 0 tensor T µν , i.e. calculate ∇[α ∇β] T µν in
terms of the Riemann tensor.
(c) Let K α be a vector that satisfies the Killing equation

∇α Kβ + ∇β Kα = 0 .

Use the symmetry relation of part (a) to show that

∇ν ∇µ K α = Rα µνβ K β ,
∇µ ∇µ K α = −Rα β K β ,

where Rαβ is the Ricci tensor.

(d) Show that
K α ∇α R = 2∇[µ ∇λ] ∇[µ Kλ] ,
and use the result of part (b) to show that the right hand side evaluates to zero, hence
showing that K α ∇α R = 0.

Part II, Paper 1 [TURN OVER]

 26

38A Fluid Dynamics

A disc of radius R and weight W hovers at a height h on a cushion of air above
a horizontal air table - a fine porous plate through which air of density ρ and dynamic
viscosity µ is pumped upward at constant speed V . You may assume that the air flow is
axisymmetric with no flow in the azimuthal direction, and that the effect of gravity on the
air may be ignored.

(a) Write down the relevant components of the Navier-Stokes equations. By esti-
mating the size of the individual terms, simplify these equations when ε := h/R ≪ 1 and
Re := ρV h/µ ≪ 1.

(b) Explain briefly why it is reasonable to expect that the vertical velocity of the
air below the disc is a function of distance above the air table alone, and thus find the
steady pressure distribution below the disc. Hence show that
3πµV R
W = .
2ε3

Part II, Paper 1

 27

39A Waves
The equation of state relating pressure p to density ρ for a perfect gas is given by
 γ
p ρ
= ,
p0 ρ0

where p0 and ρ0 are constants, and γ > 1 is the specific heat ratio.
(a) Starting from the equations for one-dimensional unsteady flow of a perfect gas
of uniform entropy, show that the Riemann invariants,
2
R± = u ± (c − c0 )
γ−1
are constant on characteristics C± given by

dx
= u ± c,
dt
where u(x, t) is the velocity of the gas, c(x, t) is the local speed of sound, and c0 is a
constant.
(b) Such an ideal gas initially occupies the region x > 0 to the right of a piston in
an infinitely long tube. The gas and the piston are initially at rest. At time t = 0 the
piston starts moving to the left with path given by

x = Xp (t) , with Xp (0) = 0 .

(i) Solve for u(x, t) and ρ(x, t) in the region x > Xp (t) under the assumptions that
2c0
− γ−1 < Ẋp < 0 and that |Ẋp | is monotonically increasing, where dot indicates
a time derivative.
[It is sufficient to leave the solution in implicit form, i.e. for given x, t you
should not attempt to solve the C+ characteristic equation explicitly.]

(ii) Briefly outline the behaviour of u and ρ for times t > tc , where tc is the solution
2c0
to Ẋp (tc ) = − γ−1 .

(iii) Now suppose,

t1+α
Xp (t) = − ,
1+α
where α > 0. For 0 < α ≪ 1, find a leading-order approximation to the
solution of the C+ characteristic equation when x = c0 t − at, 0 < a < 12 (γ + 1)
and t = O(1).
[Hint: You may find it useful to consider the structure of the characteristics
in the limiting case when α = 0.]

Part II, Paper 1 [TURN OVER]

 28

40C Numerical Analysis

(a) Describe the Jacobi method for solving a system of linear equations Ax = b as
a particular case of splitting, and state the criterion for its convergence in terms of the
iteration matrix.
(b) For the case when  
1 α α
A =  α 1 α ,
α α 1
find the exact range of the parameter α for which the Jacobi method converges.
(c) State the Householder-John theorem and deduce that the Jacobi method con-
verges if A is a symmetric positive-definite tridiagonal matrix.

END OF PAPER

Part II, Paper 1