模拟伽玛随机变量的一种有效方法

模拟产生伽玛随机变量，在统计理论和应用上都有很重要的意义．现有产生伽玛随机变量的方法，或需要借助比较复杂的优化函数，或依赖于中间参量的取值，方法上都有一定的局限性，这一点尤其体现在形参α∈（0，1）情形中．论文分析了伽玛随机变量形状参数在模拟过程中所起的作用，及“拒绝方法”在模拟过程中的局限性，利用截尾伽玛分布，成功模拟出形参α∈（0，1）的伽玛随机变量．模拟结果完全通过检验，效率较高，效果理想．

An Effective Method for Generating Gamma Random Variables

Simulation methods for generating Gamma random variables play an important role in both statistical theories and application. Methods available presently for generating Gamma random variables are limited in some way for its dependence on optimizing functions or values of parameters （especially for Gamma parameter α∈（0, 1 ） ）. This paper first analyses the effect that a takes simulating Gamma variables with α〈 1 , secondly it and the restriction which is caused by Rejection-Method when combines the Rejection-Method with truncated Gamma distribution, at last, generates ideal Gamma random variable with α 〈 1 successfully. The result shows that this new method is more effective.