Chapter 2 RANDOM NUMBERS USNA（第二章随机数后）.pdf

Chapter 2 RANDOM NUMBERS Simulation using an electronic device requires algorithms that produce streams of numbers that a user cannot distinguish from a similar string of numbers generated randomly. This is normally done using algorithms that generate numbers between 0 and 1, called a random number generator. This chapter will look at some of the properties and applications of such generators. 2.1 Pseudo-Random Numbers To be useful in simulation, a sequence of random numbers , , … must have two important properties: uniformity and independence. That is, each random number is an independent sample drawn from a continuous uniform distribution between 0 and 1 (mean 1/2, standard deviation 1/12). Some consequences of the uniformity and independence properties are the following: 1. If the interval [0, 1] is divided into subintervals of equal length, then the expected number of observed values in each interval is where is the total number of observations. 2. The probability of observing a value in a particular interval is independent of the previous values observed. Simulation using a computer must use some method of generating numbers that appear to be random. These are usually called pseudo-random numbers. “Pseudo” means having a deceptive resemblance to, so the numbers generated must deceive the user to the extent that statistical tests cannot determine that they are not random. In practice, the pseudo is usually dropped, and they are (incorrectly) referred to as random numbers. There are numerous methods that have been devised and used on computers as random number generators. The following properties have been found to be useful for such routines. 1. The routine used to generate the numbers should be fast. A simulation often requires large numbers of random values, and a slow method of generation can slow the execution of the sim