{"id":"arxiv/2604.08002","url":"https://arxiv.org/abs/2604.08002","tags":["physics.flu-dyn","math.NA"],"score":0,"title":"Invariant Guided PINN for Fluid Flow Computation","domain":"physics.flu-dyn","source":"arxiv","authors":["Zheng Lu","Jiwei Jia","Bora Aniruddha","Xingyu An","Young Ju Lee"],"reviews":0,"subject":"physics.flu-dyn","summary":"Physics-informed neural networks (PINNs) often become difficult to optimize for incompressible flow problems with large spatial domains, multiscale stresses, or long-time invariant dynamics. We propose an invariant-guided PINN (IG-PINN) framework that uses partitioned training as a conservative preconditioning stage rather than as the final piecewise representation. A globally defined architecture is trained successively on spatial subdomains or temporal slabs; selected field traces, structural information, and conservative diagnostics are then transferred to a final global correction,…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-04-09","reactions":{}}