X~2分布、t分布和F分布的近似计算

当自由度很大时，三大抽样分布的P值无表可查，而且如果编程计算P值时，需要计算一个无穷区间上的广义积分，还会面临数值溢出的问题，解决这些问题需要较高的计算机编程技巧，这些不利于开发实用程序，限制了它们的实际应用。本文利用χ^2分布、一元t分布和一元F分布的一致渐进正态性、标准正态分布的分布函数近似计算公式和软件Mathematica 5．0，通过计算机仿真，指出当自由度较大时，可以用正态分布进行近似计算，避免了数值溢出和无穷区间的积分问题，为使用常用开发语言编程解决实际问题提供了一种可行方案，具有一定的应用价值。

The Approximate Computation of Chi-Square,Student and F Distribution

There are no tables to use for the p-values of Chi-Square, Student, and Fisher Distribution as the freedom is very large. If the P-values are computed, the integration on an infinity interval should be computed and the overflow problem may happen. All of these restrict the applications of the three distributions, as it needs advanced programming skills to solve them. Based on the uniformly asymptotic normality the three distributions, the standard normal distribution approximate computation formula, the software Mathematica 5.0 and simulations, the normal distribution can take place when the freedom is large, which overcomes the shortcomings of traditional methods. The approximate computation methods are proposed and the comparisons between exact and approximate p-values show that the methods are solvable.