The numerical solution of partial differential equations (PDEs) is difficult,
having led to a century of research so far. Recently, there have been pushes to
build neural--numerical hybrid solvers, which piggy-ba
“Message Passing Neural PDE Solver” by Brandstetter et al. designs a graph neural network that outperforms both Fourier Neural Operator and classical PDE solvers in generalization capabilities and performance, addressing instability in autoregressive models.