Author: Gerd Isenberg
Date: 11:48:51 09/15/04
Go up one level in this thread
On September 15, 2004 at 12:17:14, Robert Hyatt wrote: >On September 15, 2004 at 09:53:53, Stuart Cracraft wrote: > >>Hi, >> >>Anyone know of some code somewhere that implements >>at least part (or all) of the originally described >>singular extension and/or any modifications to it that >>have proven worthwhile (if any)? >> >>I am curious what mediocre (or better) results people >>have gotten with singular extension. Originally Anantharaman >>hypothesized that it wouldn't be good at the slower >>speeds of most programs at the time and would require >>fast speeds to show effect. Has this proven true or >>false in the intervening 15 years? >> >>Is singular extension now generally discredited as a >>non-reproducible singularity in and of itself? >> >>Thanks, >> >>Stuart > > >It worked in Cray Blitz for the last two years it competed. How much it >improves overall chess skill is debatable. Early on the DB guys claimed very >significant improvements. Later more thorough testing showed that there was >definitely an improvement, but not as large as originally thought. > >I have not really spent a lot of time doing this in Crafty although there is a >SE version roaming around that I released, but it didn't do SE like the DB guys >did. Their implementation is not a 2 week deal. Expect to spend a year getting >it working as they did, if not more. > >Some commercial programs use variants. IE Wchess used the PV-singular part of >singular extensions, but not the fail-high part the DB guys defined. There are so many applications of a reduced/modified pre-search to gain more informations for better move sorting, pruning, extensions and reductions. 1. Internal iterative deepening 2. Nullmove oberservation or pre-verification 3. Looking for singularity 4. Looking for multiple beta cuts I wonder whether there is an approach to combine a few of them on cut-nodes. Btw. considering minimal tree node types, pv-, cut- and all nodes (pv most left, cut successor from pv or all, all successor from cut) What information may we gain, for a cut node, where only one successor has to be searched in a minimal tree, if it behaves like an all node? E.g. if static eval <= alpha or even first move doesn't improve alpha? Thanks, Gerd
This page took 0.01 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.