Computer Chess Club Archives




Subject: Re: testing of evaluation function

Author: Uri Blass

Date: 14:49:04 04/21/03

Go up one level in this thread

On April 21, 2003 at 08:51:29, Steven Chu wrote:

>On April 17, 2003 at 14:55:33, Uri Blass wrote:
>>On April 17, 2003 at 12:44:48, William H Rogers wrote:
>>>I would recommend playing your program against another chess program and analyze
>>>the outputs as to the differences. Play as many complete games as you can and
>>>record the moves. Then and only then can you go about making changes to your
>>>program hopefully to make it stronger. I say complete games simple because you
>>>may find weakness or other things wrong with it in the middle game or in the end
>>>game, ie. can it checkmate or does it recognized checkmate.
>>>Good luck
>>I recommend before games to check if you get the expected output
>>by static evaluation in specific positions.
>This is the sort of thing i want to do but can u tell me how i would go about
>doing it please

If you know the expected output in a specific position(not with search) then you
can ask the program simply to print the specific output and to see if you get
the expected result.

I sometimes even start with printing specific varaibles and I detect bugs when I
see that the varaibles are not what I expect.


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.