Reproduce “CI-PINN: Causal Integral Physics-Informed Neural Network for Solving Evolution 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
Xiaodong Feng, Ziyu Sun, Tao Tang, Xiaoliang Wan, Tao Zhou
CI-PINN predicts time-dependent physical fields while enforcing governing equations and initial and boundary conditions. It combines current-time features with learned features from earlier times through a gated history pathway, retaining continuous predictions at arbitrary query times. On the sparse Allen–Cahn benchmark, its mean relative L2 error was 1.35×10−2, versus 6.98×10−1 for standard PINN; this result is specific to that setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.