Stochastic Numerics and Inverse Problems

Home > What's on > Stochastic Numerics and Inverse Problems

Stochastic Numerics and Inverse Problems

 Jul 07 2021

13:00 - 14:00

Organisers:

    •    Charles-Edouard Bréhier, CNRS & Université Lyon 1

    •    Evelyn Buckwar, Linz University

    •    Erika Hausenblas, Loeben University

    •    Ray Kawai, Tokyo University

    •    Gabriel Lord, Radboud University

    •    Mikhail Tretyakov, Nottingham University

    •    Kostas Zygalakis, University of Edinburgh

This is a One World Seminar
. The seminars occur bi-weekly on a Wednesdays 13.00-14.00 BST
Zoom is the online platform being used to deliver this seminar series.
Register Here
Recordings from this seminar series are available here.

Wednesday July 7
Gabriel Stoltz, Ecole des Ponts and Inria Paris
Computation of transport coefficients in molecular dynamics: methods and numerical analysis

Abstract: In computational statistical physics, transport coefficients relate an external forcing applied to a system to its response in terms of a current. They are also known as sensitivity estimates in the field of uncertainty quantification. I will first recall, in a hopefully pedagogical way, how these quantities are defined for typical ergodic stochastic dynamics such as Langevin dynamics. I will then present numerical methods to estimate them using both linear response theory and Green-Kubo formulas. I will provide elements of numerical analysis, both to estimate the bias arising from finite time integration and the use of finite timesteps, and to quantify the statistical error in terms of variance. I will in particular present a newly developed method based on Girsanov's change-of-measure theory in the linear response regime. 

This talk is based on joint work with David Cohen, Erik Jansson, Mihály Kovács, and Mike Pereira.

Previous Seminars

A full list of past seminars in the series can be found here 

This seminar series is supported as part of the ICMS Online Mathematical Sciences Seminars.