Reproduce “Finite Basis Physics-Informed Neural Networks (FBPINNs): a scalable domain decomposition approach for solving differential equations” 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

Ben Moseley, Andrew Markham, Tarje Nissen-Meyer

The authors introduce FBPINNs, which divide a problem domain into overlapping subdomains, normalize each locally, and combine window-weighted predictions from smaller neural networks. On a high-frequency test, a 30-subdomain FBPINN achieved higher accuracy in fewer training steps than the best tested standard PINN, using multiple orders fewer forward-inference FLOPS. This does not establish lower total training cost or elapsed time.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.