Generalized Monte Carlo perturbation algorithms for correlated sampling and a second-order Taylor series approach

Abstract

A generalized formulation of two different Monte Carlo perturbation algorithms is presented for the collision and the track-length estimator. They are based on correlated sampling and a second-order Taylor series approach. In a synopsis their mutual advantages are discussed, especially in view of uncertainty behaviour. A promising scheme for their application in eigenvalue (criticality) problems is included.