Article View: sci.math.stat
Article #124Re: random number generator
From: Don Martin
Date: Mon, 03 Aug 1992 17:21
Date: Mon, 03 Aug 1992 17:21
17 lines
1349 bytes
1349 bytes
There are a number of known problems with linear congruential generaters, mostly due to the hyperplane problem. However, these generaters are still usefull for many projects where speed is important. I sometimes compare runs using a good slow generator with runs using an LCG. If the LCG is giving the same results, then I cut back on the slower generater. My usual practice is to use a shuffle generator based on several LCG's. See Press et a. Numerical Recipes for an example. This is still pretty fast. I have not heard of any falures of shuffle generaters yet. Comments would be welcome. Choose a prime for the shuffle table length and try to have most of the factors relatively prime in the periods. Don Martin Biostatistics, University of Washington
Technical Details
Message-ID:
<1992Aug3.172141.6416@u.washington.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!usc!sol.ctr.columbia.edu!destroyer!ubc-cs!uw-beaver!news.u.washington.edu!martinReferences:
<1992Jul29.092934.23335@cl.cam.ac.uk> <1992Jul29.090353.237@nutmeg.hnrc.tufts.edu> <1992Jul30.020515.16652@u.washington.edu>Show Full Headers
Complete Article Headers
Newsgroups: sci.math.stat Path: arc1-ca.pugleaf.net!sparky!uunet!usc!sol.ctr.columbia.edu!destroyer!ubc-cs!uw-beaver!news.u.washington.edu!martin From: martin@biostat.washington.edu (Don Martin) Subject: Re: random number generator Message-ID: <1992Aug3.172141.6416@u.washington.edu> Sender: news@u.washington.edu (USENET News System) Organization: Biostatistics, University of Washington, Seattle References: <1992Jul29.092934.23335@cl.cam.ac.uk> <1992Jul29.090353.237@nutmeg.hnrc.tufts.edu> <1992Jul30.020515.16652@u.washington.edu> Date: Mon, 3 Aug 1992 17:21:41 GMT Lines: 18