Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: late move reductions

Author: Tord Romstad

Date: 08:30:07 03/02/06

Go up one level in this thread


On March 02, 2006 at 07:48:30, Uri Blass wrote:

>On March 02, 2006 at 05:41:18, Tord Romstad wrote:
><snipped>
>>  The biggest
>>recent leap in playing strength for my program recently (except for
>>parallel search, of course) came when I finally decided to drop the use
>>of history counters altogether, and use evaluation data and null move
>>threat detection instead.  This also makes sense from an intuitive point
>>of view, IMHO.  When deciding whether a move should be reduced, the
>>impact of the move on the current position should be much more relevant
>>than how the same move has performed in the often totally different
>>positions we have visited earlier in the search.
>
>The problem is that you have not perfect evaluation so evaluation data may be
>wrong so intuitively combination of both is probably best.

That is possible, and something I definitely intend to try sooner or later.
It does not have very high priority at the moment, though.  Right now I am
experimenting with a more advanced null move threat detection scheme:

Assume that move m1 was reduced, and that at the node following the reduction,
the null move was refuted by the move m2.  I then cancel the reduced depth
search and re-search move m1 with full depth if one of the following conditions
is satisfied:

a) FROM(m2) == TO(m1).  The moving piece is the same for both moves.

b) FROM(m1) == TO(m2).  The destination square of threat move is the square
   vacated by the reduced move.

c) PIECE(m2) is a slider, and the ray from FROM(m2) to TO(m2) passes through
   FROM(m1).

d) TO(m2) is attacked by the piece on TO(m1).

Previously I used only condition a) above.

The extra conditions look promising so far.  They help in many test positions
(positional as well as tactical), and the cost seems to be very low.  I'll
try to run a few test matches soon.

Tord



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.