Idleness Functions for Ollivier-Ricci Curvature on Hypergraphs

What the math gives to ML

The paper gives a strong finite-dimensional structure for Ollivier–Ricci curvature as a function of random-walk idleness: for every adjacent pair, the curvature is concave and has at most three affine regimes. This means a curvature-dependent message-passing coefficient need not be tuned over a dense continuous range; it can be estimated from a few strategically chosen idleness values, with exact interpolation on the guaranteed high-idleness interval. The most practical transfer is a curvature-aware graph or hypergraph message-passing layer that chooses self-loop strength from local transport geometry, while exploiting the exact reduction from linear uniform hypergraphs to the random walk on their 2-section.