WOCT2 qualification team EUROPE

[Ciro Vignotto]

Geschrieben von: / Posted by: Ciro Vignotto at 09 January 2004 23:25:37:

1: Deep Sjeng 1.5 2498 33.0 / 34
2: WildCat 2.79 2466 30.5 / 34
3: Gothmog 0.4.3 2367 26.5 / 34
4: Phalanx XXII 2397 26.0 / 34
5: Kaissa2 1.0 2278 24.0 / 34
6: Bestia 0.90 2217 21.0 / 34
7: Asterisk v0.4b 2115 19.5 / 34
8: Cilian 4.13 2152 19.0 / 34
9: Freyr 1.067 2197 18.0 / 34
10: Merlin 1.2 2229 16.5 / 34
11: Booot 2.4 2085 15.0 / 34
12: Crux 05.0j 2050 11.5 / 34
13: Chessterfield CL i5a 2081 11.5 / 34
14: Mustang 2.67 2091 10.5 / 34
15: PolarEngine 1.3 2044 10.5 / 34
16: Replicant 1.2e 2006 9.5 / 34
17: Skaki 1.19 2168 3.5 / 34
18: Sachy 0.2.00 1654 0.0 / 34
Greetings, Ciro



http://vigo.altervista.org

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 January 2004 12:28:46:
Als Antwort auf: / In reply to: WOCT2 qualification team EUROPE geschrieben von: / posted by: Ciro Vignotto at 09 January 2004 23:25:37:

Hi Ciro,
If I understand correctly, this is only the first round of the qualification
for the European team, and a semifinal and a final will be played before
the four participants of the team are chosen.
Although Gothmog 0.4.3 was successful in this tournament, I'm afraid it will
not have much chance to compete with Deep Sjeng, WildCat, Phalanx, Kaissa and
Bestia in the remaining rounds. My latest development version, however,
seems to be about 70 points stronger, despite having a new, horrible bug
which often makes it give away rooks or queens for no compensation at all
in the endgame. How soon will the next round of the qualification be
played? If it starts one of the next few days, I will try to compile
a Windows version of the current development version (hopefully with
the bug fixed, if I am able to find it) and send it to you.
Tord

[Ciro Vignotto]

Geschrieben von: / Posted by: Ciro Vignotto at 10 January 2004 13:14:33:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Tord Romstad at 10 January 2004 12:28:46:

Hi Ciro,
If I understand correctly, this is only the first round of the qualification
for the European team, and a semifinal and a final will be played before
the four participants of the team are chosen.
Although Gothmog 0.4.3 was successful in this tournament
, I'm afraid it will
not have much chance to compete with Deep Sjeng, WildCat, Phalanx, Kaissa and
Bestia in the remaining rounds. My latest development version, however,
seems to be about 70 points stronger, despite having a new, horrible bug
which often makes it give away rooks or queens for no compensation at all
in the endgame. How soon will the next round of the qualification be
played? If it starts one of the next few days, I will try to compile
a Windows version of the current development version (hopefully with
the bug fixed, if I am able to find it) and send it to you.
Tord

correct

Compliments for the improvements of your engine!
I think will be played from 4 or 5 weeks. I hope is enough in order to fix the bug!
Greetings, Ciro

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 January 2004 15:11:04:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Ciro Vignotto at 10 January 2004 13:14:33:

Hi Ciro,
If I understand correctly, this is only the first round of the qualification
for the European team, and a semifinal and a final will be played before
the four participants of the team are chosen.
Although Gothmog 0.4.3 was successful in this tournament
, I'm afraid it will
not have much chance to compete with Deep Sjeng, WildCat, Phalanx, Kaissa and
Bestia in the remaining rounds. My latest development version, however,
seems to be about 70 points stronger, despite having a new, horrible bug
which often makes it give away rooks or queens for no compensation at all
in the endgame. How soon will the next round of the qualification be
played? If it starts one of the next few days, I will try to compile
a Windows version of the current development version (hopefully with
the bug fixed, if I am able to find it) and send it to you.
Tord

correct

Compliments for the improvements of your engine!
I think will be played from 4 or 5 weeks. I hope is enough in order to fix the bug!

Thanks!
I've already found and fixed the bug. It turned out that my evaluation function
sometimes returned a score with a wrong sign in endgames where one player had
only the king left. For instance, the engine would often believe that it was
winning when it had a king against king and queen!
I'm playing a test match with the bugfixed version against Arasan 7.3 right
now, and the results so far are almost too good to be true: 12.5-3.5 for
Gothmog after 16 games (12 wins, 1 draw, 3 losses). I expect that the end
result will be closer, but it still looks very promising.
Unless I find some new problems, I will probably release Gothmog 0.4.5
already next week.
Tord

[Marcus Prewarski]

Geschrieben von: / Posted by: Marcus Prewarski at 10 January 2004 20:20:34:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Tord Romstad at 10 January 2004 15:11:04:

Bestia in the remaining rounds. My latest development version, however,
seems to be about 70 points stronger, despite having a new, horrible bug
which often makes

I've already found and fixed the bug. It turned out that my evaluation function
sometimes returned a score with a wrong sign in endgames where one player had
only the king left. For instance, the engine would often believe that it was
winning when it had a king against king and queen!
I'm playing a test match with the bugfixed version against Arasan 7.3 right
now, and the results so far are almost too good to be true: 12.5-3.5 for
Gothmog after 16 games (12 wins, 1 draw, 3 losses). I expect that the end
result will be closer, but it still looks very promising.
Unless I find some new problems, I will probably release Gothmog 0.4.5
already next week.
Tord

Tord could you slow down with the improvements, you're making the rest of us look bad.
-Marcus

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 January 2004 22:51:07:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Marcus Prewarski at 10 January 2004 20:20:34:

Tord could you slow down with the improvements, you're making the rest of us look bad.

:-)
To be honest, I haven't made many improvements lately. I have experimented with several
new ideas, but most of them didn't work very well. I have made a few minor improvements
in my evaluation function, but the main reason why Gothmog has suddenly made a jump
in playing strength the last few days is that I have discovered a serious and extremely
stupid bug which has been in my engine for several months. Passed pawns and backward
pawns were given a bonus for *not* being able to advance safely. Fixing this bug seems
to have given me at least 50 rating points for free!
I'm running some further tests right now (three different machines are working, running
matches against Arasan, Phalanx, PostModernist, Dragon, Abrok, AnMon, Taktix and
SmarThink). So far it looks very promising. If I don't discover any unexpected problems
when I play through the games tomorrow, I'll release the new version.
Tord

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 10 January 2004 23:15:51:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Tord Romstad at 10 January 2004 22:51:07:

Tord could you slow down with the improvements, you're making the rest of us look bad.

:-)
To be honest, I haven't made many improvements lately. I have experimented with several
new ideas, but most of them didn't work very well.
in my evaluation function, but the main reason why Gothmog has suddenly made a jump
in playing strength the last few days is that I have discovered a serious and extremely
stupid bug which has been in my engine for several months. Passed pawns and backward
pawns were given a bonus for *not* being able to advance safely. Fixing this bug seems
to have given me at least 50 rating points for free!

It seems that you work harder than other programmers.
I believe that most programmers do not experiment with several new ideas in a short time.
How much time do you work on Gothmog in a day?
I am surprised that you have time to work a lot on it when you show interest on other things like teaching Gothmog to play different chess games with different board.
I have made a few minor improvements
I guess that this is the reason for part of the stupid sacrifices of gothmog
because it may get for a piece pawns that are not able to advance and they get a bonus.
If I understand correctly the problem is only of giving bonus to pawns that are not able to advance and pawns that are able to advance get the correct bonus.
Uri

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 11 January 2004 15:44:03:
Als Antwort auf: / In reply to: Re: WOCT2 qualification team EUROPE geschrieben von: / posted by: Uri Blass at 10 January 2004 23:15:51:

Tord could you slow down with the improvements, you're making the rest of us look bad.

:-)
To be honest, I haven't made many improvements lately. I have experimented with several
new ideas, but most of them didn't work very well.
in my evaluation function, but the main reason why Gothmog has suddenly made a jump
in playing strength the last few days is that I have discovered a serious and extremely
stupid bug which has been in my engine for several months. Passed pawns and backward
pawns were given a bonus for *not* being able to advance safely. Fixing this bug seems
to have given me at least 50 rating points for free!

It seems that you work harder than other programmers.
I believe that most programmers do not experiment with several new ideas in a short time.
How much time do you work on Gothmog in a day?
I am surprised that you have time to work a lot on it when you show interest on other things like teaching Gothmog to play different chess games with different board.
I have made a few minor improvements
I guess that this is the reason for part of the stupid sacrifices of gothmog
because it may get for a piece pawns that are not able to advance and they get a bonus.
If I understand correctly the problem is only of giving bonus to pawns that are not able to advance and pawns that are able to advance get the correct bonus.

I probably don't work harder, but I'm less patient and systematic than most other
programmers. When I find some new idea I add code for it in my engine very quickly,
and the new code frequently contains several bugs. If the first results are not promising,
I tend to give up almost immediately. Instead of debugging the new code or trying to
improve the idea, I proceed to implement one of the many other ideas I would like to try
out.
Hard to say, because it is a bit ill-defined. On average, I think I spend something like
2-5 hours per week writing code. But of course, I also spend a lot of time just thinking
and looking for new ideas and possible improvements. I usually do this while sitting on
the train or bus, and sometimes before falling asleep at night. In addition, I almost always
have several test matches between slightly different versions of Gothmog running, where
I experiment with various search and evaluation parameters.
Yes, it is always a problem that I try to do too many things simultaneously (not only
related to chess programming), and that I change my mind about what should have the
highest priority at least a dozen times per day.
But I believe that spending some time working on other chess-like games is actually
very productive for a chess programmer. You face new and unexpected problems, and
are forced to invent some new tricks and techniques in order to solve them. Some of
these can hopefully also be applied to normal chess.
It probably explains some of the stupid sacrifices, but I promise that you will still see
a lot of them even from the new version.
No, it was worse than that. Pawns which could advance safely were not given any bonus at
all.
Tord