20
1

Multivariate change estimation for a stochastic heat equation from local measurements

Abstract

We study a stochastic heat equation with piecewise constant diffusivity θ\theta having a jump at a hypersurface Γ\Gamma that splits the underlying space [0,1]d[0,1]^d, d2,d\geq2, into two disjoint sets ΛΛ+.\Lambda_-\cup\Lambda_+. Based on multiple spatially localized measurement observations on a regular δ\delta-grid of [0,1]d[0,1]^d, we propose a joint M-estimator for the diffusivity values and the set Λ+\Lambda_+ that is inspired by statistical image reconstruction methods. We study convergence of the domain estimator Λ^+\hat{\Lambda}_+ in the vanishing resolution level regime δ0\delta \to 0 and with respect to the expected symmetric difference pseudometric. Our main finding is a characterization of the convergence rate for Λ^+\hat{\Lambda}_+ in terms of the complexity of Γ\Gamma measured by the number of intersecting hypercubes from the regular δ\delta-grid. Implications of our general result are discussed under two specific structural assumptions on Λ+\Lambda_+. For a β\beta-H\"older smooth boundary fragment Γ\Gamma, the set Λ+\Lambda_+ is estimated with rate δβ\delta^\beta. If we assume Λ+\Lambda_+ to be convex, we obtain a δ\delta-rate. While our approach only aims at optimal domain estimation rates, we also demonstrate consistency of our diffusivity estimators.

View on arXiv
Comments on this paper