AI RESEARCH
A Geometric Perspective on the Difficulties of Learning GNN-based SAT Solvers
arXiv CS.AI
•
ArXi:2508.21513v3 Announce Type: replace-cross Graph Neural Networks (GNNs) have gathered increasing interest as learnable solvers of Boolean Satisfiability Problems (SATs), operating on graph representations of logical formulas. However, their performance degrades sharply on harder and constrained instances, raising questions about architectural limitations. In this paper, we work towards a geometric explanation built upon graph Ricci Curvature (RC