Reproduce “P-Tensors: a General Formalism for Constructing Higher Order Message Passing Networks” 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
Andrew Hands, Tianyi Sun, Risi Kondor
The authors introduce P-tensors, a framework for passing messages between selected graph substructures while respecting atom relabeling. Each tensor keeps atom-indexed information, and linear maps transfer it across shared atoms under specified symmetry conditions. On ZINC-12K molecular regression, first-order P-tensors report 0.071 ± 0.004 MAE versus CIN’s 0.079 ± 0.006. This is a result on one benchmark, not evidence of broader wins or that these maps caused the gain.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.