Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Nunn test questions / help please

Author: James T. Walker

Date: 16:18:34 05/04/02

Go up one level in this thread


On May 04, 2002 at 18:24:15, Ferdinand S. Mosca wrote:

>On May 03, 2002 at 17:57:54, James T. Walker wrote:
>
>>On May 03, 2002 at 17:23:34, Ferdinand S. Mosca wrote:
>>
>>>On May 03, 2002 at 15:52:20, James T. Walker wrote:
>>>
>>>>I've been looking at the Nunn test positions and I have some questions.  In the
>>>>Fritz database they are provided as 10 positions.  Why is the first position
>>>>listed with "line" and the rest have either "1-0 or 0-1"?  Are these the
>>>>expected results?
>>>
>>>Hello James,
>>>
>>>I expect it is the expected results.
>>>
>>>>In the Nunn2 test positions they are all listed with "line".
>>>
>>>Yes and according to GM Nunn, those positions are from a practical games where
>>>one side's winning frequency is not greater than 60-40. I still guess that one
>>>side has the upper hand but only minimal, but of course it is interesting, maybe
>>>engines will be able to find a promising plan/novelties through continuous test
>>>and learning, like the following position from a game Chess Tiger 14-Junior 7,
>>>from OP 2002 published by Kurt.
>>>Chess Tiger here played Bc1, which I think is a novelty,
>>>more common here is Bf2 with queenside play to follow.
>>>[D]r1bqnrk1/ppp1n1bp/3p2p1/3Pp3/2P1Pp2/2N1BP2/PP2B1PP/R2QNRK1 w - - 0 12
>>>Is Bc1 reasonable? now look what happened to the game, is that line playable
>>>against a human opponent?
>>>
>>>>If in the first case they are not expected results and they are considered equal
>>>>positions I guess they should all show "line" as a result.  I would like to make
>>>>an opening book for testing using the Nunn positions.  How can I make Chessbase
>>>>programs use each position when I want it to instead of random selection as
>>>>would be the case normally?  Of course the reason for this is to facilitate
>>>>running test using auto232.  Should I make one book out of each position and use
>>>>that book for each test?
>>>
>>>I think so one book for each position.
>>>
>>>Dinan
>>>
>>>Maybe I have answered my own question  for that part!?
>>>> The reason is I don't like "engine/engine" test with pondering off.  Since I
>>>>have 2 computers and auto232 I prefer to test that way.
>>>>Thanks for any help.
>>>>Jim
>>
>>Hello Dinan,
>>Well I tried the one book for each position and can't make it work.  The problem
>>is that Fritz/Hiarcs etc will not use the book.  When starting auto232 I get
>>warning signs that the book is too small and ask if I want to continue.  I click
>>continue and the game starts but the programs will not use the simple book.  I
>>guess the Chessbase GUI has something to do with this nonsense.  I tried
>>increasing the weight of each move from 0 to 100 but it made no difference.  If
>>anyone has any suggestions on how to make the programs use the simple book
>>please let me know.
>>Jim
>
>I have tried this one, but in one computer engine matches.
>1. Make a book named nunn1 intended for position no. 1 from nunn.pgn or
>nunn.cbh.
>2. Import all the 10 games to the book, and modify the book moves by making
>every move on game no. 1, the main move. (right click on the move and choose
>main move say from 1. e4 c5 etc.)
>3. So for opening position 2 make a book named nunn2 etc.
>
>Under book option use optimize and then choose minimum games = 1 for each
>engine.
>
>Regards,
>Dinan

Hello Dinan,
The problem is the Chessbase GUI.  When starting auto232 it resets the book
options for "optimal".  So minimum games goes back to 2 automatically.  I give
up on this project.  Chessbase seems determined to control what you can/can't do
with the programs.  Either that or they simply have no control over the final
product behavior.
Jim



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.