{"id":"arxiv/2609.26843","url":"https://arxiv.org/abs/2609.26843","tags":["math.NA","stat.ML"],"score":0,"title":"Gaussian-process surrogate indicators for residual-based adaptive GMsFEM","domain":"math.NA","source":"arxiv","authors":["Siqing Liu","Eric Chung","Yiran Wang"],"reviews":0,"subject":"math.NA","summary":"Residual-based adaptive GMsFEM for high-contrast elliptic problems repeatedly evaluates local weighted $H^{-1}$ indicators on every coarse neighborhood, making indicator evaluation a recurring cost in repeated-query settings. We introduce a non-intrusive Gaussian-process (GP) surrogate for the indicator scores used in Dörfler marking. The operational predictor is the GP posterior mean, algebraically equivalent to a kernel ridge regression (KRR) estimator under the stated convention; it uses compressed local solution and spectral features without changing the multiscale solve, local spectral…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-22","reactions":{}}