Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Horizon problems. ( Imagine you were back in the 80s )

Author: Georg v. Zimmermann

Date: 06:58:23 07/20/00

Go up one level in this thread



>I would try some form of selective search last few plys. With my very limited
>skill in bug I find that (as human) I extend on series of checks and if queen >is en prise (other pieces are not worth go hunting after, you simply lose too >many  tempi).

Yes I do extend a full ply on checks and another one on singular extensions.
This does of course mean that since there are so many possible drop-checks my
average depth drops even more.

>In zhouse I would add Rooks+Knights to valuable pieces. With null move
>you should catch threats on those pieces and extend accordingly.

Pieces are worth about 2 x P = N = B = R and 2 x N = Q. Most underestimate the
bishop a lot.

I tried Mate-threat detection in combination with Null Move, but for some reason
this didn't produce good results and I didn't even know whether it worked the
way it was supposed to.

/* Null Move
 *
 */
         NullValue =  -search(-beta, -beta+1, depth - (2 * ONE_PLY), ply + 1);

<snip>

/* We get mated if we don't move,
 * so this is a dangerous position, extend it.
 */
	if (NullValue < -MATE ) 		{ isThreat [ply] = 1 ;}
	else	                                { isThreat [ply] = 0 ;}



>I wonder how
>you handle ever-changing list of pieces to drop in bughouse, btw. (you restart >a search when partner gets you a new piece? Big headache...)

I changed that, it just takes too much time. I now just ignore new pieces :).


Thanks for the feedback !

Georg



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.