Unbiased sampling using reversibility checks
Data: Ar, Urt 21 2025
Ordua: 16:00
Lekua: Maryam Mirzkhani Seminar room at BCAM
Hizlariak: Prof. Tony Lelievre
We will present recent results concerning the importance of using a reversibility check in some Markov Chain Monte Carlo algorithms based on a Metropolis Hastings procedure. More precisely, we will discuss two situations: sampling measures supported on submanifolds and Hamiltonian Monte Carlo with non-separable Hamiltonians. In both cases, the numerical procedure requires to solve an implicit problem at some point, which induces some numerical difficulty concerning the actual reversibility of the proposed move. Special care should be taken in the rejection procedure to avoid biases. These reversibility checks can be seen as generalizations of a procedure suggested by Goodman, Holmes-Cerfon and Zappa for Metropolis random walks on submanifolds.
Hizlari baieztatuak:
Tony Lelièvre is a Professor at École des Ponts and the Institut Polytechnique de Paris. He conducts his research at CERMICS, the applied mathematics laboratory of École des Ponts, and is a member of the Inria project-team Matherials.
After earning his PhD in 2004, he spent a year in Montreal as a postdoctoral researcher in 2005 and was an invited professor at Imperial College London in 2020. Tony began his career focusing on computational fluid dynamics, particularly viscoelastic fluids and free-surface flows. His research interests later shifted to numerical methods in molecular dynamics, including free energy calculations, rare event sampling, and coarse-graining techniques. He is in particular collaborating on these subjects with companies such as Sanofi and IFPEN.
He has received several distinctions, including an ERC Consolidator Grant in 2013. To date, he has published two books and authored more than 110 articles.
Related events