Reproduce “Random Neural Network Expressivity for Non-Linear Partial 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

Muhammed Ali Mehmood, Lukas Gonon

The authors analyze random-feature neural networks for approximating time-dependent nonlinear PDE solutions. The method freezes randomly sampled hidden features and fits output weights. Under dimension-dependent smoothness assumptions, the theorem bounds expected squared mixed Sobolev error by a constant times 1/N. In a shock-profile experiment, error slopes were -0.49 versus -0.50 for the reference, without establishing practical PINN training or universal dimension-free error.

Summary by alphaXiv.

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