Author: Bas Hamstra
Date: 08:19:36 02/03/03
Go up one level in this thread
On February 03, 2003 at 06:20:57, Gerd Isenberg wrote: >On February 02, 2003 at 09:06:46, Bas Hamstra wrote: > >>There are some succes stories about replacing nullmove with FH reductions. In >>fact what Ed Schroeder does is the same thing, in my opinion. I am interested in >>other experiences with this technique. So far I find it disappointing. The idea >>is to reduce the search whenever in the FW search the color to move's static >>evaluation (minus a static threat evaluation) is already above beta without >>having moved. Suppose you combine this with nullmove, and you do a nullmove >>first. If the nulmove fails high, you obviously won't profit from FHR. If the >>nullmove fails low, what on earth would you reduce the search for??? It can only >>be wrong... So to really profit from it, you would have to do FHR first. So if >>static-threat > beta, you would reduce the search one ply before doing the >>regular nullmove code. A couple of thoughts: in stead of static eval minus >>static threat, you might as well do a nullmove+qsearch, I would say. However >>neither will be really accurate and you end up using R=4 where you really >>shouldn't. Any thought on this? >> >>Best regards, >>Bas. > >Hi Bas, > >As far as i remember, the original FHR-approach was intruduced by Rainer >Feldmann, one of the authors of Zugzwang (Advances in Computer Chess 96?). It is >based on Nullmove observation, but instead of a possibly risky Nullmove cutoff, >the search depth was simply reduced by one ply. > >What success stories do you mean? There is at least one author that claims to have gained 200 ELO after replacing nullmove with FHR. Further Ed Schroeder uses it in Rebel, with apparent succes. And I suspect some impossibly deep searching programs use it as well. Shredder/Gambit Tiger/Ruffian? Whater Ruffian does, it is tactical as hell. Bas.
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.