Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site lasspvax.UUCP Path: utzoo!linus!philabs!cmcl2!seismo!harvard!talcott!panda!genrad!decvax!tektronix!uw-beaver!cornell!lasspvax!norman From: norman@lasspvax.UUCP (Norman Ramsey) Newsgroups: net.math Subject: Re: deviates from various distr Message-ID: <433@lasspvax.UUCP> Date: Mon, 5-Aug-85 15:15:53 EDT Article-I.D.: lasspvax.433 Posted: Mon Aug 5 15:15:53 1985 Date-Received: Mon, 12-Aug-85 01:47:29 EDT References: <2238@utcsstat.UUCP> Reply-To: norman@lasspvax.UUCP (Norman Ramsey) Organization: LASSP, Cornell University Lines: 23 Keywords: random numbers In article <2238@utcsstat.UUCP> anthony@utcsstat.UUCP (Anthony Ayiomamitis) writes: > > I am looking for references to papers describing how one may generate >random deviates from various distributions (for example, normal, Poisson) >using a uniform random number generator. As many of you may know, micro- >computers have built-in random number generators. However, these only >produce deviates which are uniform and in the range [0,1]. Thus, one must >use some "trick" to get deviates from other distributions. This is real easy. I assume you have a probability density function on [-infinity, infinity] that you want to duplicate. What you do is integrate this up to yield an integral probability distribution P(x). This is a monotonic function of x and is zero at - infinity and one at infinity. You can think of it as the probability of a random variable's being less than x. Anyway you have this beautiful monotonic thing with a range of [0,1], so you just invert it, and voila! you have your uniform-to-whatever conversion function. It's a theorem. -- Norman Ramsey ARPA: norman@lasspvax -- or -- norman%lasspvax@cu-arpa.cs.cornell.edu UUCP: {ihnp4,allegra,...}!cornell!lasspvax!norman BITNET: (in desperation only) ZSYJARTJ at CORNELLA Never eat anything with a shelf life of more than ten years