Reproduce “Permutation-Invariant Spectral Learning via Dyson Diffusion” 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
Tassilo Schwarz, Cai Dieball, Constantin Kogler, Renaud Lambiotte, Arnaud Doucet, Aljaž Godec, George Deligiannidis
Dyson Diffusion Model (DyDM) learns graph spectra by diffusing ordered adjacency-matrix eigenvalues, whose dynamics include Brownian motion, a restoring force, and pairwise repulsion. Because spectra are invariant to vertex relabeling, the method does not require an equivariant score architecture. On the WL-Bimodal test, DyDM improves spectrum-distance measures over EDP-GNN; this evidence concerns tested spectra, not general graph generation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.