Reproduce “On sampling from a log-concave density using kinetic Langevin diffusions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Arnak S.Dalalyan, Lionel Riou-Durand
The paper analyzes kinetic Langevin Monte Carlo (KLMC), adding noise-driven velocity to position updates, and proposes a Hessian-aware discretization. At equilibrium, position follows the target while velocity is independent Gaussian. For strongly log-concave targets, first-order KLMC needs at least 17√κ fewer iterations than Cheng et al.’s bound when κ ≤ p/(16mε²). This conditional comparison is not runtime evidence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.