Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question to the engine developers

Author: Rafael Andrist

Date: 03:21:22 05/23/02

Go up one level in this thread


On May 23, 2002 at 01:34:08, Les Fernandez wrote:

>I am interested to know an approximation as to how many ply current engines can
>"most" of the time find a mate sequence (assuming it exists).  ie if we know
>there is a mate in 1 then obviously the engines will find this instantly, with
>or without tablebases.  I suspect this is probably also true for mates in 3 or 4
>plies also.  What I am interested to know are most engines able to find mates in
>6,7 ...... + plies.  What do most of you think is a safe ply number (upper
>limit) for current engines to find the mate without the use of tablebases?

That depends of course heavily on the search. If you use many extensions and no
speculative pruning, you may find mates very early (measured in plys) but the
program won't be very good for normal game play.

With a simple MiniMax/AlphaBeta you will find a mate in n at least in 2n-1 plys
(half moves).
With common extensions such as single move or check extensions, you may find
some mate in n also in n plys.
If you use Nullmove, you may miss some mates and you will need e.g. 2n+1 ply to
find a mate in n (depending on your Nullmove implementation)

I think a good guess (using extensions and Nullmove) may be 1.5n plys for a mate
in n.

regards
Rafael B. Andrist



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.