{"id":"arxiv/2611.09339","url":"https://arxiv.org/abs/2611.09339","tags":["math.NA","nlin.PS"],"score":0,"title":"A scaled TW-PINN framework for soliton solutions of the Korteweg--de Vries equation","domain":"math.NA","source":"arxiv","authors":["Seungwan Han","Jiaxi Gu","Daniel Olmos-Liceaga","Jae-Hun Jung"],"reviews":0,"subject":"math.NA","summary":"Accurate approximation of soliton solutions of the Korteweg--de Vries (KdV) equation remains challenging for physics-informed neural networks (PINNs) due to localized wave structures, high-order spatial derivatives, nonlinear interactions between solitons, and difficulties in long-time prediction. In this work, we address these challenges using a scaling PINN framework to approximate KdV soliton solutions. The framework transforms the governing equation into a normalized form, approximates its solution using a PINN solver, and reconstructs the solution in the original variables through…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-06","reactions":{}}