Fizbo 1.5 (29Aug2015) syzygy problems, again!

Discussion of computer chess matches and engine tournaments.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
ymatioun
Posts: 64
Joined: Fri Oct 18, 2013 9:40 pm
Location: New York

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by ymatioun » Tue Sep 08, 2015 12:07 pm

I'll see if i can find a reason for this and fix it. In my testing i have not seen any timeouts. But my testing was at shorter time controls.

Youri.

Frank Quisinsky
Posts: 5165
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Frank Quisinsky » Tue Sep 08, 2015 9:26 pm

Hi Youri,

at the moment after around 310 games I have 11 time loses.
If you like you can look in the games ...

10 games are replayed, not the game vs. Chiron ...

Best
Frank

http://www.amateurschach.de/download/fi ... -games.zip
I like computer chess!

User avatar
Graham Banks
Posts: 34539
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Graham Banks » Wed Sep 09, 2015 1:53 am

Frank Quisinsky wrote:Hi Youri,

at the moment after around 310 games I have 11 time loses.
If you like you can look in the games ...

10 games are replayed, not the game vs. Chiron ...

Best
Frank

http://www.amateurschach.de/download/fi ... -games.zip
I get time losses too, usually on move 40.
gbanksnz at gmail.com

Frank Quisinsky
Posts: 5165
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Frank Quisinsky » Wed Sep 09, 2015 8:15 am

Hi Graham,

I have 16 "time lost games" after 420 = around 3-4%, is to much.

But the interesting style is the same version 1.31 have (last tested version from my site). For the first UCI version really good (stability).

My wishes:

1. Can used 3-4 pieces Syzygy bases.
2. No time loses.
3. Maybe the eval. Often eval is in average around 0.4 points higher after opening book moves compare to other engines. Not easy to find out bad lines ... my work on the opening book which is optimated by games by engines. With ICE I have the same problem, often with Alfil 15.04 too. In case of ICE we know why ... it's a feature ... !!

At the moment 15 of 16 time lost games are replayed. Possible that Rating go over 2800 ELO ... this will be a fight for a League 1 place. Around the same strength Cheng / Shredder have. Really great! Fizbo is on the way ...

Best
Frank
I like computer chess!

Ferdy
Posts: 4296
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Ferdy » Wed Sep 09, 2015 9:17 am

Frank Quisinsky wrote:BTW: Hope a newer Deuterium version will be soon availalbe. Very interested all the time on your development.
Development is active, but with only 6 cores available testing resources, it is really slow. I think I already have around +25 elo improvement compared to last released version. I will try to get around +30 before a new release.
Frank Quisinsky wrote: And ... I think my test-system works fine now. I am very happy with 2050 games test-runs. Only my opening book system is clearly to many work ... need to many time. To complicated what I do in optimizing the opening book again and again. It's works good but it's to many work.

Best
Frank
Regarding your opening book updating, I am not sure how you do it. Perhaps manually. If so and if you like, a tool can be created that will analyze 1 or more positions after the book moves. Then extract these games for manual inspection and verification etc. Something like send the pgn file to the tool and the tool together with strong engines of your choice will analyze the opening positions of every game. Probably do this at night then in the morning you will have a file with games that are not so good in the openings (or it depends on your criteria of what needs to be extracted, like score after opening book goes down reaching below -50 cp after only 4 moves or so, and other conditions based on score and number of moves after the opening book moves). So in your case it is probably convenient since you will directly see the games that needs to be inspected rather than browsing thru the hundred or thousand games looking for bad lines.

Frank Quisinsky
Posts: 5165
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Frank Quisinsky » Wed Sep 09, 2015 9:25 am

Hi Ferdinand,

I checked each game by hand for find out lines I have to set "F" in my book ("F" = deactivated = bad lines for computer chess).

If you have time I need a tool which can do this one ...

1. Sbould be read bigger *.pgn files
2. Find out the first calculated move by engine (eval are in my games included).

Now ...

1. If eval higher as 0.7 for one of the two engines for move 1, 2, 3 after the opening book moves the tool should copy the games in a new *.pgn database.

So I must not checked each of the games (I produced 425 new games daily, you can see how many work I do here) and I can looking in critical games only.

This would help me a lot for my work on my opening book.

But I think such a tool is not possible to create!

Best
Frank
I like computer chess!

Frank Quisinsky
Posts: 5165
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Frank Quisinsky » Wed Sep 09, 2015 9:32 am

Hi Ferdinand,

if you have time for such a tool ...

Optimal is this one ...

1. What I wrote before
2. Tool can find out all games ended with 300 moves (such games I replayed).
3. Tool can find out all games ended with draw undo move 19 (only games draw 20 moves or higher are OK).
4. Tool can find out all time-loses.

This would be perfect.

Best
Frank
I like computer chess!

Ferdy
Posts: 4296
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Ferdy » Wed Sep 09, 2015 9:43 am

Frank Quisinsky wrote:Hi Ferdinand,

I checked each game by hand for find out lines I have to set "F" in my book ("F" = deactivated = bad lines for computer chess).

If you have time I need a tool which can do this one ...

1. Sbould be read bigger *.pgn files
2. Find out the first calculated move by engine (eval are in my games included).

Now ...

1. If eval higher as 0.7 for one of the two engines for move 1, 2, 3 after the opening book moves the tool should copy the games in a new *.pgn database.
This is a lot easier because I only have to see
if score >= 0.7 or score <= -0.7, within 3 moves after book moves, then save the pgn.
I initially thought of analyzing those position with strong engines at longer time, then you will have a reference of whether the engine player has only blundered and the opening is really playable. In this case the move may stay in the book with probably lower priority.
But OK we will do first what you suggest. Then probably later, improve the tool to analyze those opening positions after book moves.

Ferdy
Posts: 4296
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Fizbo 1.5 (29Aug2015) syzygy problems, again!

Post by Ferdy » Wed Sep 09, 2015 9:47 am

Frank Quisinsky wrote:Hi Ferdinand,

if you have time for such a tool ...

Optimal is this one ...

1. What I wrote before
2. Tool can find out all games ended with 300 moves (such games I replayed).
3. Tool can find out all games ended with draw undo move 19 (only games draw 20 moves or higher are OK).
4. Tool can find out all time-loses.

This would be perfect.

Best
Frank
No problem with this I can make one.

Frank Quisinsky
Posts: 5165
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: New tool ...

Post by Frank Quisinsky » Wed Sep 09, 2015 11:26 am

Hi Ferdinand,

OK,

White +0-7
Black -0.4

PS:
That is what I do ...
And looking ... ended with 300 moves, fast draw games undo move 19 (only 20 or higher) and lost on time games. This games are replayed all the time, since I started my work with my rating list.

I collect the 300 move games and collect the fast draw games. This games can be found in my database (at the moment around 185Mb).

But most important ...
Not to look in each of the games.

Example: Tomorrow I drive to Berlin. Can only looking Monday as next in the new games. At this time I have 1.600 new games. To look in all of the 1.600 games Need around 3-4 hours. With such a tool I have all in a very fast time.

For myself ...
You have to know ... with such a tool I will be happy and my girlfriend too.

:-)

Have many thanks Ferdinand ...
Again and again

Best
Frank
I like computer chess!

Post Reply