Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: QSearch = Extensions ?

Author: Matthias Gemuh

Date: 15:27:25 04/30/03

Go up one level in this thread


On April 30, 2003 at 17:59:51, Robert Hyatt wrote:

>On April 30, 2003 at 17:50:55, Matthias Gemuh wrote:
>
>>
>>Hi Experts,
>>I have a really basic idea of what QSearch is supposed to be.
>>My QSearch sucks heavily. Can some magical extensions really compensate ?
>>Even at the cost of search depth ?
>>Thanks for the brilliant answers,
>>Matthias.
>
>
>Think of a search as having three components:
>
>1.  A normal search that attempts to following all interesting lines to some
>sort of stable point.
>
>2.  A q-search that is supposed to take the "stable point" and resolve any
>remaining tactical issues dealing with pieces that are hung, overloaded,
>or whatever.
>
>3.  A static evaluation that takes the positions produced by 1 + 2 above,
>and computes a static evaluation.  It depends on the fact that all the tactical
>issues have been resolved by the time 1 and 2 have completed, so that it "knows"
>that the only considerations left are positional ones.
>
>1 and 2 are complementary.  The more you do in 1, the less you have to do in
>2, for example.  But for step 3, you +must+ give it "quiet positions" or it will
>produce bogus evaluations due to overlooked tactical opportunities.



Your answers adresses every important issue.
My middlegame evaluation is best on this planet.
My headaches are therefore 1)extentions and 2)qsearch _independently_.

Tanx a lot,
Matthias.




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.