Author: Heiner Marxen
Date: 14:07:21 05/22/01
Go up one level in this thread
On May 22, 2001 at 09:43:49, Robert Hyatt wrote: >On May 22, 2001 at 08:03:51, Pham Minh Tri wrote: > > >>3) Which set of random numbers is "bad" for chess? How to generate a "good" set >>of random numbers? Is it necessary to filter (prune) some "bad" numbers? > > >Not really. You can try to optimize the hamming distance between each pair, >but that is computationally expensive at setup time and I'm not sure it is worth >the trouble. Lots of duplicates will wreck things of course. I would definitely prefer to use a fixed set of numbers for all runs. Otherwise one cannot debug that part of the program, since it is not deterministic. I have used a little helper program, which spits out the filtered numbers I want to use, in a format that can be used as an C-array initializer. I once ran that helper program and saved the result as a C source, which now is part of the program source. No startup overhead, just initialized data. In general, it is quite an interesting method to _compute_ some sources. Some portability issues can be solved quite nicely that way. Heiner
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.