Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Quiescent v Normal Node Ratio

Author: Robert Hyatt

Date: 19:53:37 10/06/99

Go up one level in this thread


On October 06, 1999 at 21:42:55, Landon Rabern wrote:

>On October 06, 1999 at 13:53:36, Robert Hyatt wrote:
>
>>You have to be really careful when comparing these two values.  Think about
>>this...  You use Search() so long as depth is > 0.  But as soon as depth ==0
>>you call Quiesce... a node you can not possibly avoid searching (most of us
>>refer to such nodes as 'leaf' nodes.  Below this your searching is optional,
>>but to get here you have no choice at all.  This means that 1:1 is about the
>>best you can do the way you are counting nodes.  Below this point you can do
>>a lot using the so-called "delta pruning" that is being discussed in another
>>thread...  5:1 is high, delta will drop that to well below 2:1
>
>So when you get to depth=0 do you run a q-search always or just when the
>previous move was a capture,check, etc. ?

In Crafty, with depth >= 1, I generate all moves and try them.  For each
move I try, I recurse with depth-1, but since that is zero, I call quiesce
with the position after the move just made.  Here I can try captures or
stand pat.  But I had no choice but to get here, and when I get to quiesce
I try captures if any appear to be reasonable, irregardless of whether the
move at the last full-width ply was a capture, check or normal move.



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.