Reproduce “Numerical Integration on Bounded Domains via Learned Reflected 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
Kirill Korolev, Artur Goldman, Timofei Gritsaev, Tigran Ramazyan, Nikita Morozov
Reflected diffusion samplers (RDS) estimate integrals of nonnegative functions over bounded domains by learning a reverse diffusion proposal whose paths reflect at the boundary, then applying importance weights. On the 20-dimensional Genz Corner Peak task, RDS-RP reported mean absolute error 5.00×10⁻¹⁰ ± 1.18×10⁻¹⁰ with 2²⁰ samples, over three runs; this is one benchmark result, not universal superiority.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.