From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: ** X-Spam-Status: No, score=2.1 required=5.0 tests=BAYES_20,INVALID_MSGID, REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4 X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 103376,c76113b004e50a06 X-Google-Attributes: gid103376,public From: xanthian@well.com (Kent Paul Dolan) Subject: Re: Gnat Chat, Random Numbers in GNAT Date: 2000/02/07 Message-ID: #1/1 X-Deja-AN: 582491672 References: <389CBCBF.1DE4F0F@earthlink.net> <389D1B45.973F3A30@earthlink.net> X-Complaints-To: news@wenet.net X-Trace: news.wenet.net 949892132 208.178.101.2 (Sun, 06 Feb 2000 18:55:32 PST) Organization: Birthright Party "The birthright of humankind is the stars!" Reply-To: xanthian@well.com (Kent Paul Dolan) NNTP-Posting-Date: Sun, 06 Feb 2000 18:55:32 PST Newsgroups: comp.lang.ada Date: 2000-02-07T00:00:00+00:00 List-Id: Robert Iredell Eachus wrote: Nothing quite so confuses the mind as two authors with the same email box. >If Kent has a large processor farm, That a bit overstates my glorious CPU cycle surplus. We subcontract the manufacture of our InterJets, which are pretty mean FreeBSD x86 computers deliberately disguised as internet appliances, so I suppose if the need arose and I were very persuasive we could probably hook up the hundred or so usually cluttering up an annex to the meeting room as ready to ship stock, but what I have at present is more an engineering lab with two or three (or occasionally seven or more) leased computers on/at every desk, plus more in freestanding racks, most of their processing cycles of necessity going to the generation of heat and little else for lack of a typist per keyboard matchup, or of a 24x7 operations model, or of enough data to keep them busy. Besides the split between "productivity software" (hack, spit) and machines on which real work can be done, most of the rest are wired up in dedicated networks for testing updated build images, since a large part of what we do has to do with making really icky things work well (to the point of "transparently") across a network. > he may want to splurge and buy one. Again, for the kind of experimentation I want to do (and will soon be doing), a genuinely random seed is not particularly interesting, a long period and nice distribution statistics are what are important, so that a single run will provide enough appropriately independent random numbers to keep the testing valid. And again, thanks for help from the Eachus family, if that isn't already assuming too much from a shared email box and last name. [For the morbidly curious, my current single test run of The Blind Travelling Salesman Problem, a Java applet from Scott Robert Ladd's pages at www.coyotegulch.com, has 500 cities, a population size of 40 chromosomes, the most I could stand to watch trudge along, the last time I benchmarked it had been running at 300MHz or so for 11.5 days and 4.28 million generations, was about 2/3rds of its way toward a "'final' solution" in terms of excess distance removed, had dropped by a factor of about 6,000 in rate of progress since the run began, continues to "ignore" (fail to find) changes that would be huge wins instead finding changes that are part in 10,000 improvements, and looks, optimistically, to have about another 50 days to run to achieve at least a non-self-intersecting approximate solution. My reasons for wanting to replace this code with code of my own should be achingly obvious, when you consider that TSPs, though not, to my knowledge, the more difficult supremely masochistic BTSPs, are routinely solved in the chip fabrication industry with node counts in the hundreds of millions by other techniques, if I understand correctly.] ===== random archival quality quote ===== Walk to an unused portion of diskspace and create. And if you don't like where you are, uncreate. -- Robert Kelly -- Kent Paul Dolan.