Author: Robert Hyatt
Date: 08:16:31 08/10/98
Go up one level in this thread
On August 10, 1998 at 10:42:28, Mark Young wrote: >On August 10, 1998 at 10:26:14, Robert Hyatt wrote: > >>On August 10, 1998 at 09:03:54, Mark Young wrote: >> >>>On August 10, 1998 at 05:14:36, Robert Hyatt wrote: >>> >>>>On August 10, 1998 at 02:07:54, Mark Young wrote: >>>> >>>>>On August 09, 1998 at 23:09:20, blass uri wrote: >>>>> >>>>>>I know DB and Cray blitz use singular extensions (when one move is clearly >>>>>>better than the others they analyze it more) and I have some questions >>>>>>1) how much time do the programs analyze a move to decide it is singular. >>>>>>Is it a function of the total time you give the program? >>>>>> >>>>>>2)what is exactly a singular move >>>>>>(what is the minimal difference in evaluation between the best move and the >>>>>>second best move to decide that a move is singular). >>>>>> >>>>>>3)In what lines do the programs use singular extensions? >>>>>> >>>>>>4)are there other programs who use singular extensions? >>>>>> >>>>>I had a test position that would test if a program used singular extensions. I >>>>>lost the test position but remember that CM5000 could solve the position in less >>>>>then 1 sec. meaning CM5000 used singular extensions. Other programs I tested >>>>>could not solve the position at all. So they did not use singular extensions. If >>>>>someone has this test position please post it. >>>>>>Uri >>>> >>>> >>>> >>>>Actually a test position can't identify this. It might "suggest" that a program >>>>does or doesn't use SE, but it can't prove it. The singular margin can be >>>>different, on different engines, which will affect what gets extended and what >>>>doesn't... some may only extend when material is being lost or won, while >>>>others may extend on large positional score swings, etc. I don't think *any* >>>>micro can do *full* singular extensions because it is quite expensive, which is >>>>why a couple have tried just PV-singular extensions, or "deferred singular >>>>extensions", to keep the cost contained... >>> >>>Thank you for the info. I also remember a version of the Chessmachine program >>>had singular extensions. I remember an Update disk coming in the mail from >>>inside chess. And the update letter saying this new version uses singular >>>extensions. I did not find the new version stronger. And think the next update >>>did not use it, but Ed should now for sure. >> >> >>I don't know of *anyone* that uses "singular extensions" as explained in the >>paper by Campbell and Hsu. A couple have tried "el cheapo singular extensions" >>as I mentioned. But the term "Singular Extensions" really should be used for >>what Hsu developed as that was the first use for that term, and the algorithm >>was laid out in specific details... >> >>(the only three programs I know of that use "real" singular extensions are >>Cray Blitz, HiTech, and the Deep * programs... It was originally defined by >>Hsu... > >I am not sure if it was true "singular extensions" or not. But the Chessmachine >program claimed that it was true "singular extensions" Because they claimed it >used "sigular Extensions" like Deep Thought. That what the update did over the >other version. Ed just added singular extensions to make the new version. If >they made a bad claim or my memory is not right. I am not sure. But I would bet >money on my memory. There is one sure-fire way to find out. Compare the version "with" them to the version without. If the version with them searches to an average depth that is at least one ply shallower than the one without, then it is possible that the program uses "real singular extensions." If it doesn't lose that ply, then it isn't using them... Note that it may well find solutions at shallower plies, as that is the point. I'm talking about time to do an N ply search in the same position on both programs... Bob
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.