English

On infinite-dimensional hierarchical probability models in statistical inverse problems

Statistics Theory 2009-07-31 v2 Functional Analysis Statistics Theory

Abstract

In this article, the solution of a statistical inverse problem M=AU+EM = AU+\mathcal{E} by the Bayesian approach is studied where UU is a function on the unit circle T\mathbb{T}, i.e., a periodic signal. The mapping AA is a smoothing linear operator and E\mathcal{E} a Gaussian noise. The connection to the solution of a finite-dimensional computational model Mkn=AkUn+EkM_{kn} = A_k U_n + \mathcal{E}_k is discussed. Furthermore, a novel hierarchical prior model for obtaining edge-preserving conditional mean estimates is introduced. The convergence of the method with respect to finer discretization is studied and the posterior distribution is shown to converge weakly. Finally, theoretical findings are illustrated by a numerical example with simulated data.

Keywords

Cite

@article{arxiv.0907.5322,
  title  = {On infinite-dimensional hierarchical probability models in statistical inverse problems},
  author = {Tapio Helin},
  journal= {arXiv preprint arXiv:0907.5322},
  year   = {2009}
}

Comments

28 pages, 14 figures

R2 v1 2026-06-21T13:30:48.657Z