Reproduce “Relative Entropy Estimation in Function Space: Theory and Applications to Trajectory Inference” 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
Chao Wang, Luca Nepote, Giulio Franzese, Pietro Michiardi
The authors introduce Functional KL (FKL) to compare distributions over trajectories rather than only snapshots, using discrepancies between learned velocity fields under Functional Flow Matching. In a synthetic branching task, TIGON fits snapshot metrics better than MSBM at τ = 0.75, but its reverse path-space KL is higher: 35.505 versus 17.055. Real-data comparisons use inferred reference paths, not known biological truth.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.