On Friday, October 27 from 4 PM - 6 PM, Professor Jefferey Rosenthal of the Department of Statistical Sciences will be joining the David Sproutt Distinguished Lecture Series at the University of Waterloo and giving attendees a lecture on theoretical results which can help improve the Metropolis algorithm's convergence speed.
These results represent Rosenthal’s main research effort over the years on the topic of Markov chain Monte Carlo (MCMC): special computer algorithms which are designed to converge to complicated high-dimensional target distributions. MCMC plays a significant role in our world, as they are used in finance, medicine, and physics applications. It is important to note that the time required for their convergence is essential for practical use, and through topics including diffusion limits, optimal scaling, optimal proposal shape, tempering, adaptive MCMC, the Containment property, and the notion of adversarial Markov chains, Rosenthal hopes to demonstrate the large role that theoretical mathematical analysis has played in understanding complicated statistical computation algorithms. He also hopes to inspire other researchers to continue and expand these analyses, ultimately leading to new approaches and deeper insights in the years ahead.
A photo taken in the 90s, this photo of Professor Rosenthal giving a lecture illustrates how he has been influencing the world of Statistics for decades. (Photo was provided courtesy of Professor Rosenthal.)
If you are interested in learning more about MCMC and its relationship with theoretical mathematical analysis, please check out this interactive program written by Professor Rosenthal!