Reproduce “Asymptotic-Preserving Tensor Neural Networks for 3D High-Field Kinetic Equations with Applications to Uncertainty Quantification” 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

Shi Jin, Tian-ai Zhang

APTNN solves high-field kinetic equations using coordinate-separable neural representations. It combines one-dimensional quadratures to evaluate loss terms and adds a first-moment constraint on the limiting flux. In a 1D ablation, this constraint improves VPFP high-field errors from O(10⁻²) to O(10⁻³). A 2D2V benchmark shows lower per-step cost and memory than full-grid evaluation, not total training time.

Summary by alphaXiv.

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