Sampling from Rough Energy Landscapes

This lecture by the instructor covers the challenges and strategies for sampling from rough energy landscapes, focusing on Monte Carlo integration and tuning algorithms. The content includes motivation, background, managing roughness, numerical experiments, and considerations for optimizing acceptance rates and mixing. Various classical proposals like RWM and MALA are discussed, along with alternative strategies like Preconditioned MALA and Metropolis-Hastings. The lecture also delves into explicit test problems, stability criteria, and the performance of different sampling methods in equilibrium and out of equilibrium scenarios.