PIMS-UBC Statistics Constance van Eeden Lecture Series: Jeffrey Rosenthal
- Date: 04/06/2017
- Time: 16:00
University of British Columbia
Adaptive MCMC For Everyone
Markov chain Monte Carlo (MCMC) algorithms, such as the Metropolis Algorithm and the Gibbs Sampler, are an extremely useful and popular method of approximately sampling from complicated probability distributions. Adaptive MCMC attempts to automatically modify the algorithm while it runs, to improve its performance on the fly. However, such adaptation often destroys the ergodicity properties necessary for the algorithm to be valid. In this talk, we first illustrate MCMC algorithms using simple graphical Java applets. We then discuss adaptive MCMC, and present examples and theorems concerning its ergodicity and efficiency. We close with some recent ideas which make adaptive MCMC more widely applicable in broader contexts.
Jeffrey Rosenthal is a professor in the Department of Statistics at the University of Toronto. He received his BSc from the University of Toronto at the age of 20, his PhD in Mathematics from Harvard University at the age of 24, and tenure at the University of Toronto at the age of 29. He has received teaching awards at both Harvard and Toronto, the 2006 CRM-SSC Prize, the 2007 COPSS Presidents' Award, and the 2013 SSC Gold Medal for fundamental contributions to the Markov chain theory and to the convergence analysis of Markov chain Monte Carlo algorithms. Rosenthal's book for the general public, Struck by Lightning: The Curious World of Probabilities, was published in sixteen editions and ten languages, and was a bestseller in Canada, leading to numerous media and public appearances, and to his work exposing the Ontario lottery retailer scandal. He has also dabbled as a computer game programmer, musical performer, and improvisational comedy performer. His web site is www.probability.ca.
Time: 4pm - 5pm, with a pre-talk reception at 3.30 pm
Location: Michael Smith Labs, Room 102