Reproduce “Windowed thinning and query complexity for the bouncy particle and Zigzag samplers” 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
Jianfeng Lu, Yinchen Luo
The authors develop windowed thinning to simulate Bouncy particle sampling (BPS) and Zigzag using gradient-based event-rate bounds refreshed at regular intervals. For smooth, strongly log-concave targets, they prove expected query bounds from an explicit Gaussian cold start: BPS requires O(κ^1/2 d(d log κ + log(1/ε))) full-gradient queries, while Zigzag requires O(κ d^1/4(d log κ + log(1/ε))) full-gradient equivalents. These are theoretical oracle-call guarantees, not runtime comparisons.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.