Speeding sampling and molecular dynamics simulations using random batches


主讲人:李磊  上海交通大学副教授




内容介绍:In the first part of the talk, we will have a brief introduction to Markov Chain  Monte Carlo methods for sampling from a given distribution, especially those  methods involving Stochastic Differential Equations (SDEs). In the second part,  we focus on speeding sampling from the Gibbs distribution for many body systems.  We will introduce both an MC and an MD algorithm that we propose recently, in  which we use random batch ideas to speed up the computation. In the Random Batch  Monte Carlo method, a singular potential is split into a smooth long range part  and a singular short range part. The smooth part with random batch strategy is  used to generate a proposal sample, and the singular part is used for a  Metropolis rejection. This reduces the computational cost for sampling from O(N)  to O(1) in one iteration. In the random batch Ewald method, we apply the random  batch idea in frequency domain to obtain an efficient molecular dynamics method.