Exponential convergence for Monte Carlo particle transport

Monte Carlo transport calculations are usually preferred to deterministic transport calculations whenever computing resources permit. Unlike most deterministic methods, the Monte Carlo method is easily applicable in arbitrary geometries and can use the nuclear data without approximation. If complete and exact data are supplied, Monte Carlo calculations are exact as N ..-->.. infinity; N is the number of samples. The problem with most Monte Carlo calculations is that the convergence rate is only N/sup -1/2/. This paper demonstrates a simple Monte Carlo learning technique that has exponential-like convergence.