Titre : Error estimates in molecular dynamics
Résumé : Average properties in molecular systems are typically obtained through ergodic averages of discretizations of certain stochastic differential equations. The timestep discretization leads to biases on the invariant measure which is sampled. I will provide an introduction, aimed at an audience familiar with techniques from analysis, on how to quantify these errors — by working with generators of stochastic differential equations, (discrete) semigroups, Fokker-Planck equations, etc.
Depending on the time I will briefly hint at more recent results on error estimates for transport coefficients and/or nonlinear dynamics à la Feynman-Kac.