Inverse Problems and a Lévy Process Solution

Martin B. Hansen, Katja Ickstadt and Robert L. Wolpert

Aalborg Universitet, Duke University and Technische Universität Darmstadt

August 2000

We propose a new method of making inference about an unknown measure F(ds) upon observing some values of the integral g(x)=\int k(x,s) F(ds) of a known kernal k(x,s), using Lévy processes as Bayesian prior distributions for modeling uncertainty about F(ds), using simulation-based MCMC methods. The methods are illustrated with two problems in polymer chemistry.

Keywords: Lévy process, inference, inverse problem, polymer.


A preliminary version of the manuscript is available in postscript and pdf formats.