L.I. Dvoyris - Dr. Sc. (Eng.), Professor M.V. Kobzar K.D. Galev

Deficiency of experimental data compels recognition of the problems solved by simulating the distribution of signs. For one-dimensional random variables, this problem is solved quickly thanks to the simple algorithms or ready-made generators. But for spaces of large dimensions and distributions, for which there is no ready-made generators, this task requires the use of special algorithms. The method is based on the fact that some of the values of the random variable modeling the process are of great importance (prob-ability) evaluated for function (parameter) than others. If these «more probable» values will appear in the selection of a random variable often estimated variance function decline. The method is the choice of the distribution, which contributes to the emergence of «more probable» random variable.

 

1. Prigarin S.M. Metody chislennogo modelirovanija sluchajjnykh processov i polejj. Novosibirsk: Izd-vo IVM i MG SO RAN. 2005. 259 s.
2. Ermakov S.M. Metod Monte-Karlo v vychislitelnojj matematike. Sankt-Peterburg. 2009. 192 s.
3. Calvetti D., Somersalo E. Introduction to Bayesian Scientific Computing. Ten Lectures on Subjective Computing. Springer. 2007. 196 c.
4. Bishop C. Pattern Recognition and Machine Learning. Springer. 2006. 728 c.