Weight optimization in multichannel Monte Carlo

The improvement in the accuracy of a Monte Carlo integration that can be obtained by optimization of the a-priori weights of the various channels is discussed, where an effective increase in program speed by almost an order of magnitude is observed.

Abstract

We discuss the improvement in the accuracy of a Monte Carlo integration that\ncan be obtained by optimization of the `a-priori weights' of the various\nchannels. These channels may be either the strata in a stratified-sampling\napproach, or the several `approximate' distributions such as are used in event\ngenerators for particle phenomenology. The optimization algorithm does not\nrequire any initialization, and each Monte Carlo integration point can be used\nin the evaluation of the integral. We describe our experience with this method\nin a realistic problem, where an effective increase in program speed by almost\nan order of magnitude is observed.\n