Monte Carlo algorithms for expectation values of coordinate operators (Q1176463)

This paper deals with two efficient Monte Carlo approaches, quantum Monte Carlo (QMC) and variational Monte Carlo (VMC), to obtain the desired accuracy of quantum mechanical expectation values of coordinate operators, that do not commute with the Hamiltonian. First, a tagging algorithm is developed for efficiently counting the descendants of a walker during the QMC walk. A variant of the usual branching generally used in QMC walks is explored to reduce the statistical error in sampled values. Secondly, VMC approach with QMC side walks is described. It enables the computation of trial, mixed, and pure expectation values in a single calculation. Then, an approach is discussed for reducing statistical error in averages of odd powered coordinate operators. For purposes of evaluating the different approaches, a number of simple moments of charge distribution is computed. The appendix gives a comparison of the variance obtained for the weights when it is integer rounded versus when its full real value is kept.