quadrilateral-mesh

This paper presents a reinforcement learning agent trained to generate provably optimal quadrilateral block decompositions for planar domains, demonstrating superior performance over existing tools like Gmsh in terms of usability and optimality.