Reproduce “Improved Langevin Monte Carlo for stochastic optimization via landscape modification” 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
Michael C. H. Choi, Youjia Wang
The paper modifies a non-convex objective above an energy threshold to reduce barriers hindering low-temperature Langevin Monte Carlo sampling. The transformation preserves critical points and the low-energy region while compressing energies above the threshold. Under assumptions, the original Log-Sobolev bound is exponential in βM, whereas the modified bound is polynomial when H(m₂)-c stays positive. This is a theoretical result, not a measured speedup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.