Thursday 26 October, 14.00, room S4.36, Strand Building, King’s College London, WC2R 2LS.
Speaker: Nick Whiteley (Bristol)
Title: Log-concavity and importance sampling in high dimensions
Abstract:
Direct application of importance sampling can perform badly in high dimensions, in the sense that for even very simple examples involving densities on R^d, computational effort much be increased exponentially with d in order to prevent the Monte Carlo variance of importance sampling approximations from exploding.
The purpose of this talk is to illustrate that when the basic importance sampling idea is applied in a more sophisticated manner, which involves an importance sampling correction between the laws of certain diffusion processes constructed in terms of the target distribution known as Jarzynski's identity, the exponential growth of computational cost with dimension can be avoided. The assumption of log-concavity of the target distribution furnishes the diffusions in question with some appealing "dimension-free" ergodic properties.
This is joint work with Christophe Andrieu and James Ridgeway.
All welcome, no need to register. If stopped by Security, please give my name.
George Deligiannidis
Lecturer in Statistics
Department of Mathematics
King’s College London
Strand | London | WC2R 2LS
UK
Tel +44 (0)20 7848 2853
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|