Reproduce “Rényi Tracking Bounds for Langevin Dynamics with Moving Targets” 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
Yuchen Xin, Jingxin Zhan, Zhihua Zhang
The paper analyzes Langevin sampling for changing targets. It smooths a nonsmooth potential with a Moreau envelope, decreases the smoothing parameter while using Langevin updates. Under uniform log-Sobolev and quadratic Rényi-stability assumptions, the method reaches order-4/3 Rényi divergence at most ε, with continuation complexity O(d C_LSI² C_Rényi/ε²). This is an update-count bound, not a runtime result, and does not cover nonsmooth targets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.