MONTE CARLO DISTRIBUTED COMPUTATION FOR GENERAL DISTRIBUTION FUNCTION TABLE OF PROBABILITY DISTRIBUTION OF STATISTICS
This paper firstly proposes a new method of generating pseudo-random numbers that enormously breaks through the limitation of pseudo-random number period. Distributed computing is very suitable for this method. Then pseudo-random numbers generated by distributed computing are utilized to construct the distribution function table of statistics that can solve the problem of precision controlling and improving. Finally, this method is applied to calculate distribution function table for computing unit root process test statistics in time series. Meanwhile, a new testing method is proposed to compare with famous Dickey-Fuller test, which have been proved the new method to be better.
Monte Carlo Method Distributed Computing Distribution Function Table Dickey-Fuller Test
Hengqing Tong Tianzhen Liu Yang Liu
Department of Mathematics, Wuhan University of Technology Wuhan, Hubei Province 430070, P.R.China
国际会议
杭州
英文
124-127
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）