Author: Tord Romstad
Date: 07:04:08 06/09/04
Go up one level in this thread
Hi Daniel!
On June 09, 2004 at 05:28:28, Daniel Shawul wrote:
>Hi
>
>I read about the botvinnik-markoff extension in ccc archieve
>I have two questions concerning the method
>
>here is how i understood it
>
> do_null
> search(-beta+1,-beta,d-R-1)
> undo_null
> if(score > beta)
> return score
> else
> {
> if(score==-MATESCORE+ply+1)
> matethreat extension
> else if ( score < MARGIN)
> {
> threatmove = bestmove of ply+1
> if(to(threatmove) == to(threatmove of ply-2)
> bm extension
> }
> }
>
> qn1. What is the value of margin. -2,-5 pawns?
> Or should it be something like score < alpha - 2pawns
> or score < material balance - 2pawns
SmarThink doesn't use any margin at all (unless Sergei has added it
recently). Gothmog uses a type of margin which isn't very meaningful
in a non-MTD search, but the closest equivalent in a PVS search would
be alpha - (2 pawns).
> qn2. Should i trigger the extension also when the piece being threatened
> at ply-2 moves and the to square is now threatened?
You should trigger it in one of the two following cases:
1. ThreatMove[Ply] and ThreatMove[Ply-2] both capture the same piece
(which may or may not be on the same square in both cases).
2. ThreatMove[Ply] and ThreatMove[Ply-2] are the same non-capturing
move.
There is lots of scope for experimentation here, of course. It is very
possible that the idea could be improved.
It will be interesting to hear about your results with the extension.
In most engines, it doesn't seem to work very well, but it was a clear
improvement in my engine.
Tord
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.