Computer Intensive Methods - Lent Term 2002

This course will run in weeks 11-15, and will involve approximately 15 contact hours. There are two weekly lectures, Thursday 2-3pm and Friday 1-2pm in B9, and one weekly lab session, Tuesday 2-3pm in the John Nelder lab.

The course begins on Tuesday the 15th of January 2002 at 2pm in the John Nelder lab.

The notes for the course are here.

S-PLUS users:
Corresponding Splus functions and datasets are also available. The functions can be read into Splus using the `source' command. The datasets were created using the `data.dump' command and should be restored using the `data.restore' command. For further details, see here.

R users:
The functions and datasets are also available as a package for R, see here.


Assessment: 30% coursework and 70% examination.

Some other useful things

Markov chain Monte Carlo techniques seem to be everywhere in statistics these days. For some kind of idea of the scope of this fast moving area, see the MCMC preprint server.

A superb software package for Gibbs sampling called BUGS, and an accompanying program for convergence diagnosis and output analysis called BOA (previously known as CODA).

An interesting site about random number generators, tests for random numbers and further links can be found here.
For non-uniform random variate generation see also Luc Devroye's homepage here.

Coursework

Coursework sheet 1
Solutions to WS1 Q2-Q5
Solutions to WS1 Q6-Q8
Coursework sheet 2
Solutions to WS2 Q1-Q5
Coursework sheet 3
Solutions to WS3 Q4-Q5