Reproduce “A Jump-Modified Wasserstein Gradient Flow for Maximum Mean Discrepancy with Energy Kernels” 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
Ansh Desai, Jason Liu, Srujan Roplekar, Matthew Rosenzweig
The authors study distribution matching by combining continuous transport with jumps that move mass toward lower kernel-energy potential values. Jump rates depend on the positive potential drop and a full-support proposal. Under stated assumptions, zero jump dissipation identifies the target, and the continuum flow converges to it in MMD, without a quantitative convergence rate.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.