Lubetzky, E (Microsoft Research)
Wednesday 26 March 2008, 
Markov-chain Monte Carlo Methods

Rapid mixing of Gibbs sampling on graphs that are sparse on average

The past half-decade has seen an increasing interaction between combinatorialists, probabilists, computer scientists and theoretical physicists concerned broadly with the study of "probability theory on graphs" or "statistical mechanics on graphs".

The programme will build on this cross-fertilisation. It is particularly timely for a number of reasons:
* methods from mathematical physics are beginning to make their mark on previously intractable combinatorial problems;
* increasing computer power, together with the wide availability of symbolic-algebra packages, has brought the possibility of exploration of non-trivial examples; 
* phase transitions are increasingly being investigated on a wide variety of combinatorial structures, including matroids, set partitions and constraint satisfaction problems, as well as graphs.

Read more at: www.newton.ac.uk/programmes/CSM/