Reproduce “Parallel Tempering for Diffusion-Based Combinatorial Optimization” 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
Arman Mielke, Uwe Bauknecht, Thilo Strauss, Mathias Niepert
PT-Denoise coordinates candidate trajectories in pretrained graph denoisers to improve solutions without retraining. During denoising, it exchanges states between sampling-temperature settings, directing promising states toward more focused sampling while others explore more randomly. On evaluated problems, it improves solution quality at the same denoiser-call budget; in small-graph MIS, it lowers relative gaps, with added computation and results limited to tested settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.