2000 SIAM Annual Meeting

Tuesday, July 11

The John von Neumann Lecture:
Geometry of Markov Chains

2:00 PM-3:00 PM
Room: Rio Mar 5
Chair: Gilbert Strang, SIAM President, and Massachusetts Institute of Technology, USA

Markov chain Monte Carlo has revolutionized statistical computing and is widely used in applications in physics, chemistry and biology. Analyzing these algorithms requires good bounds on the eigenvalues of large, sparse, symmetric matrices. A geometric theory has emerged based on the diameter and various discrete curvatures of the underlying graph of the matrix. In recent work, these ideas have been adapted to continuous problems. There, operators have curious spectrum, but the geometric bounds can be pushed through to give useful results.

Persi Diaconis
Departments of Mathematics and Statistics
Stanford University, USA

©2000, Society for Industrial and Applied Mathematics
Designed by Donaghy's Web Consulting
Created 4/12/00; Updated 4/12/00