Reproduce “Warm-starting PDE solvers with any-dimensional machine learning” 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
Wilson G. Gregory, George A. Kevrekidis, Ben Blum-Smith, Soledad Villar
Nepo transfers geometric filters from a lower-dimensional solver to initialize a model. In the Navier–Stokes test, it reaches 22.8% median relative error on 3D data without 3D fine-tuning, versus 27.2% for a baseline trained on eight 3D trajectories. This comparison uses 12% of estimated FLOPs and 20% of data size; it does not establish exact transfer for arbitrary 3D inputs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.