{"id":"arxiv/2611.10071","url":"https://arxiv.org/abs/2611.10071","tags":["math.NA","math.OC"],"score":0,"title":"Operator-Adapted Multiscale Kernels for High-Contrast Elliptic Inverse Problems","domain":"math.NA","source":"arxiv","authors":["Siqing Liu","Eric Chung"],"reviews":0,"subject":"math.NA","summary":"We consider the joint recovery of the state and diffusion coefficient in a high-contrast elliptic equation from sparse, noisy state measurements. We seek the coefficient in a parameterized family defined relative to a prescribed reference coefficient. We construct a generalized multiscale finite element basis from the reference coefficient and use it to define a finite-rank kernel for the state. A weak Galerkin residual enforces the equation without pointwise differentiation across material interfaces. For each trial coefficient, the state is obtained by solving a linear least-squares…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-06","reactions":{}}