Reproduce “DynaSolver: A Neural Operator for Autoregressive 4D Flow Prediction with Dynamic Boundaries” 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
Rui Qian, Shibo Dai, Mengjun Lin, Yuk-Kit Hou, Saining Zhang, Bohan Li, Junhao Chen, Hao Zhao and 1 more
DynaSolver predicts fluid flow around articulated robots whose moving boundaries make flow depend on motion history. It conditions on boundary evolution and recent flow, processes surface and volume tokens, and uses replay-augmented parallel training. In closed-loop RoboCFD-4D Base rollouts, its overall relative L2 error is 17.71%, below Transolver’s 19.99%; it does not lead initially, and results are simulated, prescribed-motion only.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.