Author: Robert Hyatt
Date: 05:58:05 10/28/98
Go up one level in this thread
On October 27, 1998 at 01:35:31, James Robertson wrote: >On October 26, 1998 at 17:15:56, John Coffey wrote: > >>On October 26, 1998 at 02:10:37, vincent dichiacchio wrote: >> >>>my question is how programs deal with the possibility of a great non capture >>>move in the middle of a series of captures in a quiescence search. if the >>>quiescence score ignores that non capture, the score is faulty at best. >>>vince >> >> >>How is Q search any different from extending captures out? Could someone >>please define Q search for me? >> >>John Coffey >The q-search is just that; a minimax search that only evaluates capture moves >(maybe also pawn pushes or check evasions). This is a major simplification, as >there are several big differences from the normal minimax search. >I am still trying to think of how to explain them....... um...... er..... can >someone like Bob or Ernst help here? > >James the two main differences between normal search and q-search: 1. in the q-search, we initialize alpha to Evaluate() which gives us the opportunity to either (a) make any move we want to or (b) stop the search at this point and take the static evaluation (this is the "stand pat" option). The point is that you might not have a good capture to make. This lets you say "should I stop the captures here and accept the score/material balance on the board, or should I capture something here?" 2. in the q-search we generally exclude some moves. IE in crafty I exclude all non-captures right up front. Then I exclude captures that seem to lose material based on the SEE procedure. Other than those two things, they are similar. No real need to do repetition checking in the qsearch if you only do captures, for example. Perhaps a few other things like the 50-move rule also don't apply, depending on what you exclude...
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.