Fall 2021

Quantitative Convergence Analysis of Hypocoercive Sampling Dynamics

Thursday, Sep. 30, 2021 10:20 am11:00 am PDT

Add to Calendar


Jianfeng Lu (Duke University)


Calvin Lab Auditorium and Zoom

In this talk, we will discuss some recent advances on quantitative analysis of convergence of hypocoercive sampling dynamics, including underdamped Langevin dynamics, randomized Hamiltonian Monte Carlo, zigzag process and bouncy particle sampler. The analysis is based on a variational framework for hypocoercivity which combines a Poincare-type inequality in time-augmented state space and an energy estimate. Time permitting, we will also discuss computational complexity analysis of the zigzag sampler as an application of the convergence analysis. Based on joint works with Yu Cao (NYU) and Lihan Wang (CMU).