Linear and non-linear approaches to solve the inverse problem: applications to positron annihilation experiments

Abstract

We present linear and non-linear filters to solve the ill-posed inverse problem and we use them to extract relevant information from positron lifetime and 2D-angular correlation of the annihilation radiation of positrons in solids. A general optimal linear filter is first derived. Then a second linear approach, based on Bayes' theorem, is described. We show that these two linear approaches are indeed equivalent. Two non-linear methods are then discussed. The first is a Bayesian approach which makes use of the maximum entropy principle. The second is an iterative method derived from the general optimal linear filter. Applications of these filtering techniques to positron lifetime decay curves illustrate how lifetimes shorter than the instrumental resolution can be extracted. Finally, we apply the iterative non-linear filter to the problem of the ridge-like Fermi surface on the high temperature superconducting compound YBa 2Cu 3O 7−δ. For the first time a direct measurement of the ridge width through a Brillouin zone is obtained. It is compared with results of band structure calculations.