Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Does your program resign here? (or better: is it evaluated correctly?)

Author: Lieven Clarisse

Date: 07:19:44 01/11/03

Go up one level in this thread


On January 11, 2003 at 10:12:56, Jeroen Noomen wrote:

>On January 11, 2003 at 08:57:21, Lieven Clarisse wrote:
>
>>the point was: do not resign if your opponent can't make any progress (material
>>or positional) for as far as you can look.
>
>What is 'no progress'? How do you want to define that? In words it sounds easy,
>for a human it sounds easy, for a computer this can't be done!
>
>Simple example: One side is a piece up, but the positions is very much blocked
>and a pawn sac is needed to open it up. Programs will not like that, until the
>50 move rule will appear and they will be 'forced' to find the winning line.
>Choosing your rule the program has to accept a draw!?
>
>You simply CANNOT define 'making no progress'. The first one who is able to do
>this, deserves a Noble prize :-)
>
>Jeroen

Ok: please send it to me:

progress:
1. evaluate (material+positional+tempo; whatever constitutes your evaluation
function) the current position (score A)
2. search as far as you can, and pick your best line
3. evaluate the end of that position (score B)

Progress in score =B-A
No Progress: if abs(B-A)/A<0.05
Progress: if abs(B-A)/A>0.05

If the engine cannot find any progress (seen from the side of his opponent), he
should not resign.







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.