De Manon Michel
Apparaît dans la collection : Point configurations: from statistical physics to potential theory / Configurations de points : de la physique statistique à la théorie du potentiel
This one-hour introductory course surveys core ideas in computational statistical mechanics through the lens of stochastic sampling algorithms. Beginning with classical Monte Carlo methods and the foundations of Markov chain Monte Carlo (MCMC), the course introduces global and detailed balances, ergodicity, and practical sampling strategies for equilibrium systems. It then discusses quantitative diagnostics for convergence and mixing, including autocorrelation times and spectral considerations. The course concludes with modern approaches to accelerating sampling by moving beyond reversible dynamics toward non-reversible Markov processes, highlighting how broken detailed balance and irreversible flows can substantially improve convergence efficiency in high-dimensional and metastable systems.
De Kazumasa Takeuchi
De Satya N. Majumdar
