Accession Number : AD0674240

Title :   APPROXIMATE GENERATION OF RANDOM VARIABLES.

Descriptive Note : Technical rept.,

Corporate Author : TEXAS A AND M UNIV COLLEGE STATION INST OF STATISTICS

Personal Author(s) : Suharto,S.

Report Date : AUG 1968

Pagination or Media Count : 13

Abstract : In many situations when a Monte Carlo procedure is used to solve some statistical problems, the computational work can be very excessive, since the number of samples, N, must usually be very large in order to obtain an adequate degree of accuracy. Let x sub 1, x sub 2,...,x sub n be a random sample drawn from the parental distribution f(x) with cumulative distribution F(x) and let h(x sub i) be a statistic which is a function of the observations x sub i. Then, the Monte Carlo procedure for evaluating P(h = or < H) may be described as follows: (a) generate random uniform variates, u sub i; (b) transform the uniform variate u sub i into x sub i with the help of x sub i = (F superscript -1)(u sub i); (c) evaluate the statistic h(x sub i) and determine the proportion for which h(x sub k) = or < H. The present paper is concerned with part (b) only.

Descriptors :   (*STATISTICAL DISTRIBUTIONS, MATHEMATICAL PREDICTION), (*MONTE CARLO METHOD, EFFICIENCY), RANDOM VARIABLES, APPROXIMATION(MATHEMATICS), DISTRIBUTION FUNCTIONS, NUMERICAL ANALYSIS, SAMPLING

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE