Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How deep should I search?

Author: Robert Hyatt

Date: 07:10:18 07/12/99

Go up one level in this thread


On July 12, 1999 at 08:47:42, Nobuhiro Yoshimura wrote:

>In program with iterative deeping like:
>   for( iteration=1; i<MAX_ITERATION; i++ )
>        search( alpha, beta);
>
>If we do many search extensions in search(), it will search more deeper than
>"iteration".
>
>Is it a good idea to limit the search depth for each iteration ( for example,
>maximum depth to be 2*iteration) ,  or  not to limit the search depth.
>
>Which idea is true in general?
>   If I limit the search depth,  it will terminate the search-routine soon and
>      1)  prabably it will find a right move in next iteration
>      2)  prabably it will NOT find a right move in next iteration again.


either could be true.  The issue is this.  Without a "limit" you might search
to depth N.  With a limit, you might get to iteration N+1 or N+2.  Which is
better?  Difficult question.  In non-tactical situations, N+1 is always better
than N, and N+2 is always better than N+1.  Because you will see one ply deeper
positionally and have a possible advantage over your opponent.

Best way to answer this is to test it however.  And _not_ on tactical test
suites...



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.