A method for Monte Carlo importance sampling with parametric dependence is proposed. The method depends on scaling of the particle tracks for a certain number of generated histories, to determine the position of the optimal biasing parameters. Computational results for a model problem show that difficulties pointed out by previous authors as to the occurrence of infinite variances or under-sampling can be easily understood and spotted by the careful user, and eliminated so as to obtain dependable results with moderate numbers of particle histories.
|Original language||English (US)|
|Number of pages||9|
|State||Published - 1981|
ASJC Scopus subject areas