Author: Uri Blass
Date: 12:08:16 04/24/03
Go up one level in this thread
On April 24, 2003 at 14:32:35, Tim Foden wrote: >On April 24, 2003 at 14:17:06, Uri Blass wrote: > >>On April 24, 2003 at 13:58:10, Will Singleton wrote: >> >>>On April 24, 2003 at 13:37:52, Uri Blass wrote: >>> >>>>On April 24, 2003 at 13:33:16, Will Singleton wrote: >>>> >>>>>Hi, >>>>> >>>>>I'm looking for a program that does perft without underpromotions. Thanks. >>>>> >>>>>Will >>>> >>>>modifying movei to do perft without under promotions is easy only changing one >>>>line in the source code. >>>> >>>>Why do you need it? >>>> >>>>Uri >>> >>>I don't do underpromotions. I'd appreciate if you could send me the modified >>>movei, thanks very much! >>> >>>Will >> >>I will send you it in the next few hours. >> >>I am now busy in writing my evaluation in a more convenient way to change and >>testing for bugs by asking the program to compare new evaluation with old >>evaluation and when they do not agree I know that I have bug in one of my >>evaluation because they are supposed to be the same. > >One of the great ways of catching bugs in the evaluation is to make a random >position, and "mirror" it. I.E. flip the board vertically, and change the >colours of all the pieces to the opposite colours, and change the side to move. >The two boards should have the same evaluation. If they don't there's usually >something wrong. > >Repeat the test something like 100000 times. > >Cheers, Tim. > >P.S. Don't forget to mirror any castling rights or the e-p square if you >randomly generate them. I know that but I did not develop a program to flip the board and at this time other ways detect problems(comparing the evaluation that is done in 2 different methods). I may consider doing it only after there are no problems in another way. Uri
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.