Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Testing evaluation function

Author: Andreas Herrmann

Date: 13:38:57 08/11/03

Go up one level in this thread


On August 10, 2003 at 23:44:57, Rick Bischoff wrote:

>Is this a "good" way to test an evaluation function--
>
>Algorithm A
>1) If Black is on the move, make the null move
>1.b) If depth <= 0 return static evaluation
>2) If white is on the move, eliminate checks and captures from the move list.

What are you doing if there are only capture and check moves for white possible?

>3) Make each move, call algorithm A recursively, unmake move
>4) return best move, best score
>
>Algorithm B
>1) set best = -INF
>2) int temp = A(depth, best, &bestmove)
>3) if (temp > best) { make move (bestmove), make move(NULLMOVE) } else break;
>4) continue forever.
>
>I.e., when you run this with TSCP's evaluation function, you get the following
>pretty position:
>[D]rnbqkbnr/pppppppp/PPP5/3NN3/3PP3/3BB3/5PPP/2RQR1K1 w kq - 0


What does this result tells you?

Andreas





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.