{"id":"arxiv/2601.18575","url":"https://arxiv.org/abs/2601.18575","tags":["math.NA"],"score":0,"title":"Moving sample method for solving time-dependent partial differential equations","domain":"math.NA","source":"arxiv","authors":["Beining Xu","Haijun Yu","Jiayu Zhai","Kejun Tang","Xiaoliang Wan"],"reviews":0,"subject":"math.NA","summary":"Solving time-dependent partial differential equations (PDEs) that exhibit sharp gradients or local singularities is computationally demanding, as traditional physics-informed neural networks (PINNs) often suffer from inefficient point allocation that wastes resources on regions already well-resolved. This paper presents an adaptive sampling framework for PINNs aimed at efficiently solving time-dependent partial differential equations with pronounced local singularities. The method employs a residual-driven strategy, where the spatial-temporal distribution of training points is iteratively…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-01-26","reactions":{}}