MATHEMATICAL TRIPOS

2024-25

GUIDE TO COURSES
IN PART II

This booklet contains informal and non-technical descriptions of courses to be examined

in Part II in the academic year 2024-25, as well as a summary of the overall structure of
Part II.

This Guide to Courses is intended to supplement the more formal descriptions contained
in the booklet Schedules of Lecture Courses and Form of Examinations.

These and other Faculty documents for students taking the Mathematical Tripos are
available from the undergraduate pages on the Faculty’s website at https://www.maths.
cam.ac.uk/undergrad/

Revised 7 October 2024

1 Introduction
Each lecture course in the Mathematical Tripos has an official syllabus, or schedule, that sets out formally,
and in technical terms, the material to be covered. The schedules are listed in the booklet Schedules
of Lecture Courses and Form of Examinations that is available for download at https://www.maths.
cam.ac.uk/undergrad/course/schedules.pdf. The Schedules booklet is the definitive reference for
matters of course content and assessment, for students, lecturers and examiners.
The present guide, by contrast, provides an informal description of each lecture course in Part II. These
descriptions are intended to be comprehensible without much prior knowledge, and to convey something
of the flavour of each course, and to suggest some preparatory reading, if appropriate.
A summary of the overall structure of Part II is also given below, including the distribution of questions
on examination papers.

Lectures and examinations post COVID-19

On 5 May 2023 the WHO Director-General declared, with great hope, the end to COVID-19 as a global
health emergency. Therefore, it is the working assumption of the Faculty that lectures and examinations
in 2024-25 will all be held ‘normally’ and in person. Students are expected to attend lectures in order
to take full advantage of the benefits of in-person teaching.
While arrangements for supervisions are made by Colleges and Directors of Studies, the Faculty antici-
pates that supervisions will also be held ‘normally’ and in person.

Changes to lecture courses since last year

ˆ There are minor changes to the wording of the schedules of Algebraic Topology, Number Fields,
Asymptotic Methods and Fluid Dynamics in 2024-25 as compared to 2023-24.
ˆ The 24-lecture Algebraic Topology will be given in the Michaelmas (rather than the Lent) Term.
ˆ The 16-lecture Riemann Surfaces will be given in the Lent (rather than the Michaelmas) Term.

2 Structure of Part II
The structure of Part II may be summarised as follows:

ˆ There are two types of lecture courses, labelled C and D. C-courses are all 24 lectures, D-courses
may be 16 or 24 lectures. This year there are 10 C-courses and 27 D-courses. There is in addition
a Computational Projects course (CATAM).
ˆ C-courses are intended to be straightforward, whereas D-courses are intended to be more challeng-
ing.
ˆ There is no restriction on the number or type of courses you may present for examination.
ˆ The examination consists of four papers, with questions on the courses spread as evenly as possible
over the four papers subject to:
◦ each C-course having four Section I (‘short’) questions and two Section II (‘long’) questions;
◦ each 24-lecture D-course having no Section I questions and four Section II questions;
◦ each 16-lecture D-course having no Section I questions and three Section II questions.
ˆ Only six questions from Section I may be attempted on each paper.
ˆ Each Section I question is marked out of 10 with one beta quality mark, while each Section II
question is marked out of 20 with one quality mark, alpha or beta. Thus each C-course and 24-
lecture D-course carries 80 marks and a number of quality marks, while each 16-lecture D-course
carries 60 marks and a number of quality marks. The Computational Projects course carries 150
marks and no quality marks.

1
3 Distribution of Questions on the Examination Papers
The distribution of Section II (‘long’) questions on the examination papers is as follows:

C-Courses Paper 1 Paper 2 Paper 3 Paper 4

Number Theory * *
Topics in Analysis * *
Coding and Cryptography * *
Automata and Formal Languages * *
Statistical Modelling * *
Mathematical Biology * *
Further Complex Methods * *
Classical Dynamics * *
Cosmology * *
Quantum Information and Computation * *

D-Courses Paper 1 Paper 2 Paper 3 Paper 4

Logic and Set Theory * * * *

Graph Theory * * * *
Galois Theory * * * *
Representation Theory * * * *
Number Fields * * *
Algebraic Topology * * * *
Linear Analysis * * * *
Analysis of Functions * * * *
Riemann Surfaces * * *
Algebraic Geometry * * * *
Differential Geometry * * * *
Probability and Measure * * * *
Applied Probability * * * *
Principles of Statistics * * * *
Stochastic Financial Models * * * *
Mathematics of Machine Learning * * *
Asymptotic Methods * * *
Dynamical Systems * * * *
Integrable Systems * * *
Principles of Quantum Mechanics * * * *
Applications of Quantum Mechanics * * * *
Statistical Physics * * * *
Electrodynamics * * *
General Relativity * * * *
Waves * * * *
Fluid Dynamics * * * *
Numerical Analysis * * * *

2
4 Informal Description of Courses

C-Courses
Number Theory Michaelmas, 24 lectures
Number Theory is one of the oldest subjects in mathematics and contains some of the most beautiful
results. This course introduces some of these beautiful results, such as a proof of Gauss’s Law of
Quadratic Reciprocity, and a proof that continued fractions give rise to excellent approximations by
rational numbers. The new RSA public codes familiar from Part IA Numbers and Sets have created new
interest in the subject of factorisation and primality testing. This course contains results old and new
on the problems.
On the whole, the methods used are developed from scratch. You can get a better idea of the flavour of
the course by browsing Davenport The Higher Arithmetic CUP, Hardy and Wright An introduction to
the theory of numbers (OUP, 1979) or the excellent Elementary Number Theory by G A and J M Jones.
(Springer 1998).

Automata and Formal Languages Michaelmas, 24 lectures

The course deals with three basic ideas: the idea of computability for a function; what it means for a
function to be non-computable; and how computational power depends on programming constructs. Such
questions appear straightforward, but they are not. In seeking answers students will meet fundamental
and recurring concepts such as state, transition, non-determinism, reduction and undecidability. They
remain important in current theoretical computer science even as technology changes from day to day.
Three classes of models are used to illustrate some fundamental aspect of computation, most of which
were developed long before computers existed. In increasing order of power these are:

1. finite memory, finite automata and regular expressions;

2. finite memory with stack: pushdown automata;
3. unrestricted: Register machines, Post systems, µ-recursive functions, λ-calculus and combinatory
logic.

Prerequisites are in IA Numbers and Sets, including the notions of set, function, relation, product,
partial order, and equivalence relation. Elementary notions of graph and tree will also be introduced.
The recommended text is Automata and Computability by D. C. Kozen (Springer, 1997).

Coding and Cryptography Lent, 24 lectures

When we transmit any sort of message errors will occur. Coding theory provides mathematical techniques
for ensuring that the message can still be read correctly. Since World War II it has been realised that the
theory is closely linked to cryptography – that is to techniques intended to keep messages secret. This
course will be a gently paced introduction to these two commercially important subjects concentrating
mainly on coding theory.
Discrete probability theory enters the course as a way of modelling both message sources and (noisy)
communication channels. It is also used to prove the existence of good codes. In contrast the construction
of explicit codes and cryptosystems relies on techniques from algebra. Some of the algebra should already
be familiar – Euclid’s Algorithm, modular arithmetic, polynomials and so on – but there are no essential
prerequisites. IB Linear Algebra would be useful. IB Groups, Rings and Modules is very useful.
The book by Welsh recommended in the schedules (Codes and Cryptography, OUP), although it contains
more than is in the course, is a good read.

3
Topics in Analysis Lent, 24 lectures
Some students find the basic courses in Analysis in the first two years difficult and unattractive. This is
a pity because there are some delightful ideas and beautiful results to be found in relatively elementary
Analysis. This course represents an opportunity to learn about some of these. There are no formal
prerequisites: concepts from earlier courses will be explained again in detail when and where they are
needed. Those who have not hitherto enjoyed Analysis should find this course an agreeable revelation.

Statistical Modelling Michaelmas, 24 lectures

This course is complementary to Part II Principles of Statistics, but takes a more applied perspective.
There will be approximately 16 hours of lectures and eight hours of practical classes. The lectures will
cover linear and generalised linear models, which provide a powerful and flexible framework for the study
of the relationship between a response (e.g. alcohol consumption) and one or more explanatory variables
(age, sex etc.).
In the practical classes, we will learn how to implement the techniques and ideas covered in the lectures by
analysing several real data sets. We will be making extensive use of the statistical computer programming
language R, which can be downloaded free of charge and for a variety of platforms from https://www.
stats.bris.ac.uk/R/. An excellent editor for R can be downloaded from https://www.rstudio.com/
products/RStudio/.
This course should appeal to a broad range of students, including those considering further research in
any aspect of Statistics and those considering careers in data-intensive industries (investment banking,
insurance, etc.). Those interested might like to try downloading R and experimenting with one of several
excellent tutorials available by following the links at https://www.statslab.cam.ac.uk/~pat/.

Cosmology Michaelmas, 24 lectures

This course presents a mathematically rigorous description of 13.8 billion years of history, from the Big
Bang to the present day and beyond. The course starts by deriving the equations which describe an
expanding universe. The need to include a number of surprising and mysterious ingredients, such as
dark energy, dark matter, and inflation, will be discussed. Subsequently, the course will introduce the
mathematics necessary to understand the first few minutes after the Big Bang, when the universe was
very hot and the elements were forged. The course ends by explaining how small perturbations in the
early universe subsequently grew into the glorious galaxies and structures that we see today. You will
need to be comfortable with Newtonian dynamics, special relativity and some basic facts about quantum
mechanics. No knowledge of astrophysics or general relativity is needed.

Classical Dynamics Michaelmas, 24 lectures

This course follows on from the dynamics sections of Part IA Dynamics and Relativity and also uses
the Euler–Lagrange equations from Part IB Variational Principles. The laws of motion for systems of
particles and for rigid bodies are derived from a Lagrangian (giving Lagrange’s equations) and from a
Hamiltonian (giving Hamilton’s equations) and are applied, for example, to the axisymmetric top.
One advantage of the formalism is the use of generalised coordinates; it is much easier to find the kinetic
and potential energy in coordinates adapted to the problem and then use Lagrange’s equations than to
work out the equations of motion directly in the new coordinates. At a deeper level, the formalism gives
rise to conserved quantities (generalisations of energy and angular momentum), and leads (via Poisson
brackets) to a system which can be used as a basis for quantization.
The material in this course will be of interest to anyone planning to specialise in the applied courses. It is
not used directly in any of the courses but an understanding of the subject is fundamental to Theoretical
Physics.

4
Mathematical Biology Lent, 24 lectures
The aim of the course is to explain from a mathematical point of view some underlying principles of
biology, ranging from biochemistry and gene regulation to population dynamics and spread of infectious
disease. In particular we examine mechanisms for feedback control, sensitivity amplification, oscillations,
developmental instabilities, pattern-formation, competitive growth, and predator-pray interactions.
The material should be of interest to anyone who is fascinated by the richness of biological dynamics,
but has been discouraged by too detail-oriented biological explanations. Mathematical methods include
basic stochastic theory, nonlinear dynamics, differential equations, and numerical analysis. The concepts
and techniques are not very difficult, and intuitive guiding principles and illustrative examples will be
favoured over rigorous proofs. This is an exciting field with large unexplored territories for applied
mathematicians.
While this course fits well with Dynamical Systems and there are places where understanding in one
course will help the other, it is not essential and this course does not rely on any other Part II course,
assuming concepts only from Part IA (Differential Equations and Probability) and IB (Methods).

Quantum Information and Computation Lent, 24 lectures

Quantum processes can provide enormous benefits for information processing, communication and se-
curity, offering novel features beyond the possibilities of standard (classical) paradigms. These benefits
include (i) new kinds of algorithms (so-called quantum algorithms) providing an exponentially faster
method for some computational tasks, (ii) new modes of communication such as quantum teleportation,
and (iii) the possibility of unconditionally secure communication in quantum cryptography. Most of these
exciting developments have occurred in just the past few decades and they underpin striking applications
of quantum technologies that are currently being developed.
This course will provide an introduction to these topics. No previous contact with the theory of com-
putation or information will be assumed. IB Quantum Mechanics is essential, but only to provide prior
exposure to basic ideas. This course rests on quantum theory in just a finite-dimensional setting, so
the principal mathematical ingredients (from finite-dimensional linear algebra) will be readily accessible.
We will begin by expounding the postulates of quantum mechanics in this setting (using Dirac nota-
tion) and then immediately make connections to information (quantum states viewed as information
carriers, quantum teleportation) and computation (notion of qubits and quantum gates). Then we will
discuss quantum cryptography (quantum key distribution), and quantum computing, culminating in an
exposition of principal quantum algorithms, including the Deutsch–Jozsa algorithm, Grover’s searching
algorithm and an overview of Shor’s quantum factoring algorithm. The course is cross-disciplinary in its
conceptual ingredients and will be of interest to pure and applied mathematicians alike.

Further Complex Methods Lent, 24 lectures

This course is a continuation in both style and content of Part IB Complex Methods, which is the only
prerequisite. It will appeal to anyone who enjoyed that course. The material is classical — much of it
can be found in Whittaker and Watson’s ‘Modern Analysis’, written in 1912. The passage of time has
not diminished the beauty of material, though the Faculty Board decided against naming the course
‘Modern Analysis’.
The course starts with revision of Complex Methods and continues with a discussion of of the process of
analytic continuation, which is at the heart of all modern treatments of complex variable theory. There
follows a section on special functions, including the Gamma function (which is basically the factorial
function when looked at on the real line, but on the complex plane it really blossoms) and the Riemann
zeta and its connection with number theory. Then the theory of series solutions of differential questions
in the complex plane is developed, and suddenly the treatment given in Part IA Differential Equations
makes sense. Naturally, the messy business of actually solving specific equations by series is not in the
style of the course. Particularly important are those equations that have exactly three singular points,
all regular. This leads to a study of the properties of the delightful hypergeometric function, of which
almost every other function you know can be thought of as a special case. This is the high point of the
course, involving nearly all the theory that has preceded it.

5
 D-Courses
Graph Theory Michaelmas, 24 lectures
Discrete mathematics is commonplace in modern mathematics, both in theory and in practice. This
course provides an introduction to working with discrete structures by concentrating on the most ac-
cessible examples, namely graphs. After a discussion of basic notions such as connectivity (Menger’s
theorem) and matchings (Hall’s marriage theorem), the course develops in more detail the theory of
extremal graphs, ideas of graph colouring, and the beautiful theorem of Ramsey. A significant feature is
the introduction of probabilistic methods for tackling discrete problems, an approach which is of great
importance in the modern theory.
There are no formal prerequisites but it will be helpful to recall some of the elementary definitions from
the Part IA Probability course. The attractions and drawbacks of Graph Theory are similar to those
of that course and of the Part IA Numbers and Sets course; whilst the notions are not conceptually
difficult, the problems might on occasion require you to think a little.
The text Modern Graph Theory by Bollobás is an excellent source and contains more than is needed for
the course. For a lighter introduction try Wilson’s Introduction to Graph Theory, or for a little more
look at Bondy and Murty’s old but now online Graph Theory with Applications.

Representation theory Michaelmas, 24 lectures

This course, suitable for pure and applied mathematicians, is an introduction to the basic theory of linear
(matrix) actions of finite groups on vector spaces. The key notion we define is the character of a linear
representation: this is a function on conjugacy classes of the group which determines the representation
uniquely. Orthogonality relations between characters lead to a convenient and efficient calculus with
representations, once the basic character table of the group has been computed. Later in the course
‘finite’ is replaced by ‘compact’ generalising the results with little extra effort.
The Linear Algebra course is essential and Groups, Rings and Modules is helpful.

Algebraic Topology Michaelmas, 24 lectures

Topology is the abstract study of continuity: the basic objects of study are metric and topological
spaces, and the continuous maps between them (concepts which were introduced in Part IB Analysis
and Topology). One important difference between topology and algebra is that in constructing continuous
maps one has vastly more freedom than in constructing algebraic homomorphisms; thus problems which
involve proving the non-existence of continuous maps with particular properties (e.g. the problem of
showing that Rm and Rn are not homeomorphic unless m = n) are hard to solve using purely topological
methods. The technique that has proved most successful in tackling such problems is that of developing
algebraic invariants, which assign to every topological space (in a suitable class) an algebraic structure
such as a group or vector space, and to every continuous map a homomorphism of the appropriate kind.
Thus questions of the non-existence of continuous maps are reduced to questions of non-existence of
homomorphisms, which are easier to solve.
Two particular algebraic invariants are studied in this course: the fundamental group, and the simplicial
homology groups. Of these, the former is easier to define, but hard to calculate except in a few particular
cases; the latter requires the erection of a considerable amount of machinery before it can even be
defined, but once this is done it becomes relatively easy to calculate. The course concludes with a classic
example of the application of simplicial homology: the classification of all compact 2-manifolds up to
homeomorphism.
Apart from Part IB Analysis and Topology, the only prerequisite is a modicum of geometrical intuition.
Concepts and techniques of algebraic topology are used almost everywhere in mathematics where topo-
logical spaces occur; they also arise in many research areas in mathematical physics, as well as pure
mathematics.
For introductory reading, browse Basic Topology by M.A. Armstrong (Springer-Verlag).

6
Galois Theory Michaelmas, 24 lectures
The most famous application of Galois theory – discussed at the end of this course – is the proof that
the general quintic equation with rational coefficients cannot be solved by radicals. Apart from this,
Galois theory plays an indispensable role in algebraic number theory and several other areas of pure
mathematics. It is a subject which (in favourable circumstances) allows one to handle given polynomials
elegantly and with a minimum of algebraic manipulation.
Familiarity with the material concerning field extensions and the polynomial ring K[t] from Part IB
Groups, Rings and Modules is essential, while Part IB Linear Algebra is useful. The most closely related
Part II courses are Representation Theory and Number Fields. The book Galois Theory by I. Stewart
(Chapman and Hall, 1989) gives a very readable introduction to the subject.

Linear Analysis Michaelmas, 24 lectures

Functional Analysis provides the framework, and a great deal of machinery, for much of modern mathe-
matics: not only for pure mathematics (such as harmonic analysis and complex analysis) but also for the
applications of mathematics, such as probability theory, the ordinary and partial differential equations
met in applied mathematics, and the mathematical formulation of quantum mechanics.
The basic idea of Functional Analysis is to represent functions as points in an infinite-dimensional vector
space. Since the space is infinite-dimensional, algebraic arguments are not enough, and it is necessary
and appropriate to introduce the idea of convergence by a norm, which in turn defines a metric on the
space.
In this course, most attention is paid to two sorts of spaces. The first consists of spaces of continuous
functions: here the appropriate convergence is uniform convergence. The second is Hilbert space (partic-
ularly important in Quantum Mechanics) which provides an infinite-dimensional analogue of Euclidean
space, and in which geometrical ideas and intuitions are used.

Riemann Surfaces Lent, 16 lectures

A Riemann surface is the most general abstract surface on which one can define the notion of an analytic
function, and hence study complex analysis. Roughly speaking, a surface is made into a Riemann surface
when the change from one local coordinate system to another system is given by an analytic function.
Not every Riemann surface has a global coordinate system; this accounts for both the interesting and
the difficult parts of the theory.
The course begins with a study of the Riemann sphere (which is just the complex plane with infinity
attached) and of elliptic functions (that is to say, doubly periodic analytic functions) which are the ana-
lytic functions defined on a torus. Abstract Riemann surfaces and holomorphic maps are then introduced
and some of the results already studied in earlier courses on complex analysis are extended to this more
general context.
Another view on Riemann surfaces comes from Riemann’s original idea that the so-called ‘multivalued
functions’ are just considered on a wrong domain: the natural domain is a surface covering the complex
plane several (possibly infinitely many) times. This surface is called the Riemann surface of an analytic
function and is obtained by the process of analytic continuation, extending the function (while keeping
it analytic) in a maximal way from a domain in C.
The last part of the course shows that most Riemann surfaces carry their own intrinsic non-Euclidean
geometry; thus complex analysis is much more closely connected to non-Euclidean geometry than to
Euclidean geometry (despite the fact that it is first studied in the Euclidean plane).
Prerequisite for this course is IB Complex Analysis (some knowledge of IB Analysis and Topology will
also be useful, especially for elliptic functions). Related Part II courses include those on Algebraic
Topology, Algebraic Geometry and Differential Geometry. As a preliminary reading, consider the early
parts of G.A. Jones and D. Singerman, Complex functions CUP, 1987, and of A.F. Beardon, A primer
on Riemann surfaces CUP, 1984.

7
Algebraic Geometry Lent, 24 lectures
Algebraic geometry is a branch of mathematics which, as the name suggests, combines techniques of
abstract algebra, especially commutative algebra, with the language and the problematics of geometry.
It occupies a central place in modern mathematics and has multiple conceptual connections with such
diverse fields as complex analysis, topology and number theory. Initially a study of polynomial equations
in many variables, the subject of algebraic geometry starts where equation solving leaves off, and it
becomes at least as important to understand the totality of solutions of a system of equations, as to
find some solution; this does lead into some of the deepest waters in the whole of mathematics, both
conceptually and in terms of technique.
This course is an introduction to the basic ideas of algebraic geometry (affine and projective spaces,
varieties), followed by a more detailed study of algebraic curves. We will develop the basic tools for
understanding the properties of algebraic curves, and apply these at the end of the course to the beautiful
theory of elliptic curves, which among other things played an essential part in the proof of Fermat’s Last
Theorem! You will find it highly advantageous to have attended the Part IB course Groups, Rings and
Modules. Part II courses with which this course is related include Galois Theory, Differential Geometry,
and Algebraic Topology. Students wishing to do some preliminary reading could browse the books of
Reid or Kirwan noted in the schedules.

Differential Geometry Lent, 24 lectures

A manifold is a space that looks locally like euclidean space. The surface of a sphere or the surface of
a torus are natural examples, but manifolds often arise indirectly, for example, as the space of solutions
of some set of conditions, the parameter space of a family of mathematical objects, configuration spaces
in mechanical systems and so on. Manifolds provide the appropriate arena on which one can explore
interactions between various branches of Mathematics and Theoretical Physics.
Manifolds often come endowed with geometric structures, for example, with a way of measuring the length
of a curve (Riemannian metrics) as in the case of a surface in 3-space. One can then define geodesics
and curvature and study how these objects influence one another and interact with the topology of the
manifold. A key illustration of this interplay, and a central result in this course, is the Gauss-Bonnet
theorem, which shows that the average curvature determines the topological type of the surface.
Rather than worrying about how to define abstract manifolds (which you will see in Part III), we will
study manifolds as objects already embedded in euclidean space. This will allow us to have a very short
working definition of manifold and get fairly quickly into examples and the basic notions in Differential
Topology (such as regular values, degree and transversality), giving a measure of how much one space
folds onto another. Once we have set up the framework we will study the (Riemannian) geometry of
curves and surfaces in euclidean space and we will prove at the end of the course that curvature can
detect knottedness.
IB Geometry provides useful examples and an introduction to some of the ideas that we will develop and
IB Analysis and Topology will be very useful when we set up the manifold framework in the first few
lectures.

Analysis of Functions Lent, 24 lectures

The analysis of functions has its roots in the rigorous study of the equations of mathematical physics,
and is now a key part of modern mathematics. This course builds on the Part II courses Linear Analysis
and Probability & Measure, which are pre-requisites, applying the theory of integration and the tools
of functional analysis to explore such topics as Lebesgue and Sobolev spaces, the Fourier transform and
the generalised derivative.
These topics are important and interesting in themselves, but the emphasis is on their use in other areas
of mathematics (for instance in the representation of functions and in partial differential equations),
rather than their maximal generalisation. You can get an idea of the flavour of the course by browsing
Analysis by Lieb & Loss (Springer).

8
Logic and Set Theory Lent, 24 lectures
The aim of this course is twofold: to provide you with an understanding of the logical underpinnings
of the pure mathematics you have studied in the last two years, and to investigate to what extent, if
any, the ‘universe of sets’ can be considered as a structure in its own right. As such, it has few formal
prerequisites: some familiarity with naive set theory, as provided by the IA Numbers and Sets course,
is helpful, but no previous knowledge of logic is assumed. On the other hand, the course has links to
almost all of pure mathematics, and examples will be drawn from a wide range of subjects to illustrate
the basic ideas.
The course falls into three main parts. One part develops the notions of validity and provability in formal
logic, culminating in the Completeness Theorem, which asserts that these two notions coincide. Another
part is concerned with ordinals and cardinals: these are notions that generalise the ideas of size and
counting to the infinite. The final part is an introduction to formal set theory, where one makes precise
the idea of a ‘universe of sets’, and studies its structure.
The book ‘Notes on Logic and Set Theory’ by P.T. Johnstone (C.U.P., 1987) covers most of the material
of the course, and is suitable for preliminary reading.

Number Fields Lent, 16 lectures

Number theory studies properties of the integers and the rationals. The questions that arise are usually
very simple to state but their solutions are often very deep and involve techniques from many branches
of mathematics. It is also a subject where numerical experiments have proved useful as a guide to the
sort of result one might seek to establish. Diophantine equations constitute a central theme; they are
basically polynomial in form and lead to the study of integer or rational points on algebraic varieties.
Two particularly famous problems here, Fermat’s Last Theorem and the Catalan Conjecture, have only
recently proved amenable to solution.
The course provides an introduction to algebraic number theory – it arose historically from investigations
of reciprocity laws and attempts to solve the Fermat problem and it now forms one of the nicest and
most fundamental topics in mathematics. Knowledge of the course on Groups, Rings and Modules is
desirable.
The book Problems in algebraic number theory by J. Esmonde and R. Murty contains most of the material
covered in the course. For a historical introduction to the subject, see also Chapter 1 of D. Cox’s Primes
of the form x2 + ny 2 .

Probability and Measure Michaelmas, 24 lectures

Measure theory is basic to some diverse branches of mathematics, from probability to partial differential
equations. This course combines a systematic introduction to measure theory with an account of some of
the main ideas in probability. You will be familiar with the Riemann integral from Parts IA and IB and
have done some elementary probability in IA. The expectation operator of probability behaves somewhat
like the integral, and in this course we see that they are both examples of some more general integral.
These general integrals and the measures which underlie them have advantages over the Riemann integral,
even for functions defined on the reals. In Part IA the definition and properties of expectation were only
partially explored and here we do it more fully.
If you like to see how a substantial and coherent mathematical theory is put together, you will enjoy the
measure theory part of this course, and this will be essential to any further work you do in analysis. It
also underpins the probability which provides motivation and application throughout the course. The
course ends with the Strong Law of Large Numbers and Central Limit Theorem, both of which are of
real practical importance, being the mathematical basis for the whole of statistics.
A good book to read for the early part of the course is Probability with Martingales, by D. Williams
(CUP, 1991).

9
Principles of Statistics Michaelmas, 24 lectures
In Part IB Statistics, an introduction to the main statistical problems and inference techniques was given,
including parameter estimation, hypothesis testing and the construction of confidence sets in a variety
of examples, including specific families of distributions and models. The Principles of Statistics course
aims to give a unified perspective on these problems, and develops the main mathematical theory that
underpins these basic principles of statistical inference.
The first pillar will be the inferential paradigm surrounding the likelihood function of the observations,
and the associated maximum likelihood estimator, providing a conceptually unified and in most situations
also practical solution to the problem of statistical inference. The distribution of this estimator will be
shown to have a universal limiting normal distribution, permitting the use of the estimator for statistical
inference. A generalisation of the Gauss–Markov theorem from the linear model can be proved for this
estimator, establishing that it is in a certain sense the best among all estimators. Related to the likelihood
principle, but in other respects fundamentally different, is the Bayesian approach to statistical inference.
This approach, likewise, will be developed for general families of parametric statistical models.
The study of the notion of optimality of certain statistical procedures from a general perspective is known
as statistical decision theory, of which the main ideas will be presented in the course. The course will also
develop the main ideas of some related classical fields in statistics that are crucial in applications, such
as inference methods for multivariate data, nonparametric techniques, and resampling (Monte Carlo)
procedures.

Requirements: Probability IA and Statistics IB are essential.

Stochastic Financial Models Michaelmas, 24 lectures

This is concerned with the pricing of financial assets under uncertainty. It builds towards a presentation
of the celebrated Black-Scholes formula for the price of an option on stocks. The holder of a call
option on a stock has the right to purchase one unit of that stock at a specified ‘strike’ price within a
designated time period. The holder hopes that within the period the stock price will go above the strike
price whereupon the option may be exercised with the stock being bought at the strike price and sold
immediately at the higher current price to yield a profit. What is the fair price to charge for such an
option? In seeking an answer to this question, the course introduces some important ideas of probability
theory including martingales and Brownian motion. Deciding when the holder should exercise the option
leads to the techniques of dynamic programming and optimal stopping which are applicable throughout
applied probability and statistics.
The main prerequisite for this material is Part IA Probability – if you liked that course then you should
enjoy this one. Probability and Measure is recommended as a companion course, but it is not strictly
necessary. No previous knowledge of economics or finance is necessary. It complements, but does not
rely on, Markov Chains. To get a better idea of the sort of problems the course is seeking to tackle it is
worth browsing in the book Option, Futures and Other Derivative Securities by J. Hull (Prentice-Hall,
2nd Ed. 1993).

Applied Probability Lent, 24 lectures

This course provides an introduction to some of the probabilistic models used to study phenomena
as diverse as queueing, insurance ruin, and epidemics. The emphasis is on both the mathematical
development of the models, and their application to practical problems. For example, the queueing
models studied will be used to address issues that arise in the design and analysis of telecommunication
networks.
The material is likely to appeal to those who enjoyed Part IA Probability; Markov Chains is useful, but
the style of the course, involving a mix of theory and applications, will more closely resemble the earlier
course. Probability and Measure is a loosely related Part II course.

10
Mathematics of Machine Learning Lent, 16 lectures
Suppose we are given a collection of emails labelled as either being spam or legitimate (known as ham).
With this data, we aim to build an algorithm to annotate new unlabelled emails as spam or ham. This
is known as a classification problem and is a central problem studied in Machine Learning, a field whose
broader remit, like Statistics, is about how to learn from data. The focus of this course is the classification
problem, other instances of which include predicting whether or not patients have a particular disease
based on their medical history, or predicting whether a user will click an internet advertisement or not
(a problem of great interest to search engines).
The first part of the course will deal with the theory of empirical risk minimization (ERM), a simple
but powerful general strategy for building classification algorithms. We will work with a mathematical
framework where each data instance (i.e. each email and label pair in the case of our first example)
is viewed as a realisation of random elements, all sharing the same distribution. Studying ERM in
this context will require us to develop important probabilistic tools, so-called concentration inequalities.
Part IA Probability is a necessary prerequisite and in practice it is also advisable to have attended
Part IB Statistics. The results we will derive were essentially state-of-the-art until the late 1990’s and
early 2000’s; despite being suboptimal in some cases, studying them is a useful starting point for more
sophisticated results.
The second part of the course studies computational aspects of ERM, building on parts of the IB
Optimisation course (relevant material will be reviewed in the course). In particular, we will introduce
stochastic gradient descent, which is now the most popular optimisation strategy for machine learning
problems and currently the centre of a great deal of research. The last part of the course will illustrate
some of the concepts and ideas developed in the earlier parts by studying random forests, boosting and
neural networks, perhaps the three most successful methods in machine learning.

Asymptotic Methods Michaelmas 16 lectures

There are many instances, arising not only in mathematical physics, but also in analysis and number
theory, where one needs an approximation to a function for which no usable convergent series expan-
sion is available. Typically, the function is given as an integral or else as the solution of a differential
equation. It turns out that excellent approximations can be obtained using certain series, called asymp-
totic expansions, which are normally non-convergent. Such an expansion might describe, for example,
the behaviour of an integral depending on a parameter, as the parameter becomes large; alternatively,
it might describe the behaviour of a solution of an ordinary differential equation, as the independent
variable becomes large.
A certain amount of familiarity with the basics of complex-variable theory is essential, either through
Part IB Complex Methods or Part IB Complex Analysis. This would be reinforced by the Part II Further
Complex Methods course, which is desirable but not essential. An introduction to the course material is
given in A. Erdelyi Asymptotic Expansions (Dover 1956).

Numerical Analysis Michaelmas, 24 lectures

Many mathematical problems, e.g. differential equations, can be solved generally only by computation,
using discretisation methods. In other cases, e.g. large systems of linear equations, calculation of the
exact solution is impractical and, again, we need to resort to numerical methods. Numerical analysis
concerns itself with the design, implementation and mathematical understanding of computational algo-
rithms. The course will address iterative techniques for linear equations, the calculation of eigenvalues
and eigenvectors, and the solution of partial differential equations by finite differences (following the
treatment of ordinary differential equations in Part IB). The last section of the course deals with Fourier
expansions and their generalisations.
Part IB Numerical Analysis is an obvious prerequisite but Part IB Analysis courses, Complex Methods
and Linear Algebra are also highly relevant. Mathematical ability is sufficient for understanding the
course, while computational experience provides only a useful advantage.

11
Dynamical Systems Michaelmas, 24 lectures
Contrary to the impression that you may have gained, most differential equations can not be solved
explicitly. In many cases, however, a lot can still be said about the solutions. For example, for some
systems of differential equations one can show that every solution converges to an equilibrium, while for
others one can prove that there is a subset of solutions which are equivalent to infinite sequences of coin
tosses (‘chaos’). In this course, we study differential equations which can be written in the form ẋ = v(x)
with x in some (mainly two or three-dimensional) ‘state’ space. We take the “dynamical systems”
viewpoint, concentrating on features which are invariant under coordinate change and time rescaling.
We will find that two-dimensionality imposes severe restrictions though many interesting ‘bifurcations’
are possible: ways that the behaviour of a system ẋ = vµ (x) can change as external parameters µ are
varied. We shall also study nonlinear maps, which can be thought of either as difference equations or
as a way of investigating the stability of periodic solutions of differential equations. We conclude with a
discussion of chaotic behaviour in maps and differential equations, including a treatment of the famous
logistic map. The treatment is ‘applied’ in flavour, with the emphasis on describing phenomena, though
key theorems will be proved when needed.
If you browse P Glendinning Stability, instability and chaos, CUP, 1994 you will be well prepared.
The material contained in this course is relevant to any subject involving a modern treatment of differ-
ential equations. This includes most areas of Theoretical Physics, but usually not at the undergraduate
level.

Electrodynamics Michaelmas, 16 lectures

Electrodynamics is the most successful field theory in theoretical physics and it has provided a model
for all later developments. This course develops from Part IB Electromagnetism. It starts by developing
electromagnetism as a classical relativistic field theory, showing how the relativistic form of Maxwell’s
equations can be derived from a variational principle, and presenting the covariant treatment of the
energy and momentum carried by the electromagnetic field. Such a treatment is essential for later use
in quantum field theory. The remainder of the course shows how Maxwell’s equations describe realistic
phenomena. In particular, the production of electromagnetic waves by accelerated charges is discussed,
and mechanisms for scattering radiation are briefly introduced. The course ends with the treatment of
electromagnetism in continuous media, which allows one to understand some of the tremendous diversity
in the electric and magnetic properties of real materials. The propagation of electromagnetic waves in
materials is a particular focus.
The main prerequisite is familiarity with the basic ideas (especially those involving Maxwell’s equations
and their relativistic formulation) of Part IB Electromagnetism and knowledge of special relativity. As
often for a theoretical physics course, the Feynman Lectures provide good introductory reading.

Principles of Quantum Mechanics Michaelmas, 24 lectures

This course develops the principles and ideas of quantum mechanics in a way which emphasizes the
essential mathematical structure, while also laying the foundations for a proper understanding of atomic
and sub-atomic phenomena. In contrast to the introductory treatment given in Part IB, which is based
entirely on wavefunctions and the Schrödinger equation, observables are presented as linear operators
acting on vector spaces of states. This new approach has practical as well as aesthetic advantages,
leading to elegant and concise algebraic solutions of problems such as the harmonic oscillator and the
quantum theory of angular momentum. Some of the other key aspects of quantum behaviour that are
treated include: intrinsic spin, multi-particle systems, symmetries, and their implications. Perturbation
theory techniques, which are indispensable for realistic applications, are also discussed. The course ends
by examining in more detail the inherently probabilistic nature of quantum mechanics, as illustrated by
Bell’s inequality and related ideas.

12
Applications of Quantum Mechanics Lent, 24 lectures
This course develops the ideas and methods introduced in Part IB Quantum Mechanics and Part II
Principles of Quantum Mechanics and uses them to explain how we probe and understand the structure
of atoms and solids. The various material objects that surround us in the everyday world exist as
vast collections of particles (electrons and nuclei) making up atoms, molecules and various crystalline
substances. Quantum mechanics is essential for an understanding of how this happens.
An important tool for probing the structure of matter (finding out where the particles are, how the
electric charge is distributed) is the scattering of a beam of particles of appropriate energy on targets of
interest. The course develops the theory of scattering in a form applicable to both atomic and crystalline
targets.
There are two particularly important aspects of crystalline materials: the elastic vibrations of the atoms
in the crystal matrix and the dynamics of electrons moving through the crystal. In quantum theory the
elastic vibrations are understood as particle-like excitations known as phonons. In travelling through
a crystal both phonons and electrons exhibit a band structure in their permitted energies. The role of
phonons and electrons in condensed matter physics and the significance of this energy band structure is
explained by means of simple but physically significant quantum mechanical models. Energy bands are
used to understand the properties of semiconductors and some simple devices such as the pn junction
are explained.
Some idea of the material of the course can be gained by consulting a book such as Principles of the
Theory of Solids by J. M. Ziman, (CUP, 1972).

Statistical Physics Lent, 24 lectures

Statistical mechanics is the art of turning the microscopic laws of physics into a description of Nature
on a macroscopic scale. This requires the development of tools to understand the properties of systems
which contain a very large number of particles.
The course starts by defining new concepts such as as entropy, temperature and heat, all viewed from
a microscopic perspective. These ideas are then used to understand a wide array of phenomena, from
gases, to the vibrations of solids (phonons), to the behaviour of light (blackbody radiation). In each case,
the laws governing these phenomena are derived from first principles. The course goes on to describe
more subtle quantum effects that occur at low temperature such as Bose-Einstein condensation and the
formation of Fermi surfaces. The course ends with a discussion of thermodynamics and phase transitions,
in which the properties of a system change discontinuously.
Ideas from Part IB Quantum mechanics are essential and Part II Principles of Quantum Mechan-
ics is highly desirable. Lecture notes from David Tong’s course are available to download at https:
//www.damtp.cam.ac.uk/user/tong/statphys.html, while a shorter set of notes representing almost
exactly the material presented in the classes is available at https://www.damtp.cam.ac.uk/user/
us248/Lectures/lectures.html.

General Relativity Lent, 24 lectures

General Relativity is a relativistic theory of gravitation which supersedes the Newtonian Theory. This
course shows how the theory can be built up on the foundations of Part IA Special Relativity. The
necessary ideas from differential geometry will be taught ab initio, relying on the methods courses in
Parts IA and IB. As an extended example, the course includes a careful treatment of the Schwarzschild
spacetime and its interpretation as a black hole.
An elegant informal treatment of much of the material is contained in chapters 1,2,7 and 8 of W. Rindler
Essential Relativity (Springer, 1977). A slightly more formal introduction is chapters 5,6,8,9,10,14-16 of
R. d’Inverno Introducing Einstein’s Relativity (Oxford, 1992).

13
Integrable Systems Lent, 16 lectures
A soliton was first observed in 1834 by a British experimentalist, J. Scott Russell. It was mathematically
discovered in 1965 by Kruksal and Zabusky, who also introduced this name in order to emphasise the
analogy with particles (‘soli’ for solitary and ‘ton’ for particles).
Solitons appear in a large number of physical circumstances, including fluid mechanics, nonlinear optics,
plasma physics, elasticity, quantum field theory, relativity, biological models and nonlinear networks.
This is a consequence of the fact that a soliton is the realisation of a certain physical coherence which
is natural, at least asymptotically, to a variety of nonlinear phenomena. The mathematical equations
modelling such phenomena are called integrable. There exist many types of integrable equations including
ODEs, PDEs, singular integrodifferential equations, difference equations and cellular automata.
The mathematical structure of integrable equations is incredibly rich. Indeed soliton theory impacts on
many areas of mathematics including analysis, algebraic geometry, differential geometry, group theory
and topology. However, it must be emphasised that the basic concepts of the integrable theory can be
introduced with only minimal mathematical tools. This course will give an introduction to soliton theory
with emphasis on the occurrence of solitons in nonlinear dispersive PDEs.

Fluid Dynamics Michaelmas, 24 lectures

How does a hummingbird hover? How does a bumblebee fly? How, for that matter, does a Boeing 747
defy the pull of gravity? Does the bath-tub vortex really rotate anti-clockwise – or is it clockwise – in
the Northern Hemisphere? How can a flow which exhibits an infinite sequence of eddies in a confined
space satisfy a ‘minimum dissipation’ theorem? How can a flow that is strictly reversible have irreversible
consequences?
Such questions lie within the domain of Fluid Dynamics, a subject that contains the seeds of chaos (and
indeed provides the main stimulus for much of the current intense interest in chaos). The course will
address the above questions, among others, in a progression from phenomena on very small scales (‘low
Reynolds number problems’) to phenomena on very large scales (‘large Reynolds number problems’).
The course thus encompasses, at one extreme, flows that arise at the biological level (e.g. the swimming
of microscopic organisms) and, at the other, flows on the scale of the Earth’s atmosphere and oceans,
or even larger. And, in between of course, it encompasses the bath-tub! Mathematical techniques,
further to those developed in IB, will be used to determine solutions to the nonlinear, time-dependent
Navier–Stokes equations.
Part II Asymptotic Methods covers some material which would be useful for this course. The course
has natural links with Part II Waves and less obvious links with Dynamical Systems. A number of the
Computational Projects are directly relevant. Introductory reading: Elementary Fluid Dynamics by
D.J. Acheson, chapters 1-4.

Laboratory Demonstrations in Fluid Dynamics Non-examinable

A series of laboratory demonstrations and experiments is used to expose you to material covered by
the Part IB and Part II Fluid Dynamics lecture courses. The emphasis is on understanding the physics
behind the mathematics, along with the limitations of the simple analytical models. Attending this
course will help you develop the physical insight necessary to derive and evaluate mathematical models,
and to determine whether their predictions are reasonable. Specific topics covered include potential flow,
surface waves, Reynolds experiment, Stokes flow, Kelvin’s circulation theorem, spin-up, boundary layers
and bubbles. Student participation is encouraged but not required.
There are no prerequisites for this course and the course is not examined.
Further details will be announced during the Part II Fluid Dynamics lectures.

14
Waves Lent, 24 lectures
Waves occur in almost all physical systems including continuum mechanics, electromagnetic theory and
quantum mechanics. In this course examples will be drawn from fluid and solid mechanics, although
much of the theory has application in other contexts. In the first part of the course sound waves in a gas
are studied (after which you will understand why you can hear the lecturer). Small amplitude acoustic
waves are described by the wave equation (see IB Methods); however at larger amplitudes nonlinear
effects must be included. The change in the governing equations caused by nonlinearity leads to the
formation of shocks, i.e. sonic booms. Applications of the underlying theory to both traffic flow and
blood flow are mentioned.
Linear elastic waves, e.g. seismic waves, split into two types: the faster-travelling compressional waves
(cf. sound waves) and the slower-travelling shear waves. The surface waves that cause most destruction
in an earthquake are also studied.
Not all linear waves have a wavespeed that is independent of wavelength. In such systems it is important
to distinguish the speed of wavecrests from the speed at which energy propagates; indeed, the wavecrests
and energy can propagate in opposite directions. As a consequence, (a) if you throw a stone into a
pond to generate a circular wave packet, you will see that the wavecrests propagate outward through the
wave packet and disappear, and (b) atmospheric waves generated near ground level can appear to the
eye as if they are propagating down from the heavens! Finally, the ray tracing equations are derived.
These are used to describe, inter alia, why you can go surfing (i.e. why waves tend to approach a beach
◦
perpendicularly), why the wave pattern behind a ship (or a duck) subtends a half-angle of 19 21 , and
why sound can travel long distances at night.
The mathematical techniques assumed are those covered in the IA and IB Methods courses. While the
course is otherwise self-contained, there is a small amount of complementary material in Asymptotic
Methods, Electrodynamics, and Fluid Dynamics.
A good book to look at is Wind waves: their generation and propagation on the ocean surface by B.
Kinsman (Prentice-Hall).

Computational Projects (CATAM)

This course is similar in nature to the Part IB course. There are a variety of projects to choose from;
some are closely related to Part II courses and others are not. As in Part IB, the projects are listed
in a CATAM manual which is available on the Faculty website at https://www.maths.cam.ac.uk/
undergrad/catam/II.
The course is examined by means of work handed in close to the beginning of the Easter Term.

5 Lecture Timetable: Clashes

Each term, there are only 8 slots for lecture courses1 This year there are a total of 37 courses to be fitted
into a total of 16 slots, so there will be 11 slots with double clashes and 5 slots with triple clashes.
The general policy is to avoid triple clashing any C-courses, and to avoid clashing any applied courses,
any pure courses and any applicable courses. Further, by using examination statistics to gauge the
popularity of combinations of courses, the aim is to minimise the effect of the clashes (for example, by
not clashing a very popular pure course with a very popular applied course). However, there are also
significant constraints associated with the fact that the most popular courses in Parts II and III will fit
only in the larger lecture theatres (MR2, MR3 and MR9). In addition there are constraints concerning
lecturer availability. At the end of the day, it is unfortunately inevitable that some students would have
wanted to attend courses that clash.
A provisional lecture timetable should be available early in September, or thereabouts, from which you
will be able to see the clashes.

1 Or 9 if 16-lecture courses were distributed asynchronously.

15