{"id":"arxiv/2610.16040","url":"https://arxiv.org/abs/2610.16040","tags":["math.NA","cs.LG"],"score":0,"title":"CLNO: Conservative Local Neural Operator for Compressible Flow Prediction","domain":"math.NA","source":"arxiv","authors":["Buxuan Wang","Rongzu Li","Qingsong Zou"],"reviews":0,"subject":"math.NA","summary":"We propose a conservative local neural operator (CLNO) for accurate and generalizable autoregressive prediction of compressible flows. Building upon the locality and translation invariance of the local neural operator (LNO), CLNO introduces a conservative evolution framework based on local subdomains of fixed physical size, separating local mean-state evolution from fine-scale reconstruction. The flow is encoded into subdomain-integrated conservative states and fine-scale features, while a shared local network predicts interface fluxes for finite-volume updates that enforce discrete…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-12","reactions":{}}