WBEC Ridderkerk new results.

[Leo Dijksman]

Geschrieben von:/Posted by: Leo Dijksman at 15 May 2004 17:55:53:

Premier Division:
Round 14:
Aristarch 4.41 ½1½1 3.0/4
Tao 5.6 ½0½0 1.0/4
------------------------------------
Crafty 19.12 ½½11 3.0/4
SmarThink 0.18a-r119 ½½00 1.0/4
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------
Dragon 4.6b 11½ 2.5/3 = 1 game to go!
Quark 2.35 Paderborn 00½ 0.5/3
------------------------------------

2nd Division:
Round 14:
The CrazyBishop 0050 11½½ 3.0/4
Resp 0.19 00½½ 1.0/4
------------------------------------
SlowChess 2.89b 1½½0 2.0/4
Frenzee 146 0½½1 2.0/4
------------------------------------
Round 15:
Amateur 2.80 ½111 3.5/4
The CrazyBishop 0050 ½000 0.5/4
------------------------------------
Scidlet 3.6 - Amy 0.8.7 1-0 = 3 games to go!
------------------------------------

New engine qualify:
Round 4:
Murderhole 1.0.10 - Borer 2.01a 0-1 70
Hokus Pokus 0.43 - Elephant 1.05 0-1 40
Geko 0.4.3 - Excelsior 2.32b 0-1 55
Round 5:
Fruit X0505 - Naum 1.0 1-0 68
AnMon 5.32 - Bruja 1.5 1-0 56
Hermann 1.1.1 - Borer 2.01a 0-1 66
Jonny 2.62 - Spike 0.2b 0-1 35
Alfil 403.1 - GNUChess 4TM 0-1 39
Delphil 0.7g - DanChess 1.04d beta =-= 99
Nesik 0.6.4 - Marvin 1.0.0 1-0 65
Elephant 1.05 - Praetorian 2.1 1-0 30
Needle 0.53.1 - Ajedrez Tactico 2.15 0-1 60
Excelsior 2.32b - Murderhole 1.0.10 0-1 23
Geko 0.4.3 - Hokus Pokus 0.43 0-1 36
Round 6:
Fruit X0505 - AnMon 5.32 =-= 71

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
15/05/2004:
Updated: Frenzee v157
Updated: Bruja v1.6
Updated: Crafty v19.13 (exe by Dann Corbit (egtbs fixed!))
Best wishes,
Leo.


WBEC Ridderkerk homepage.

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 15 May 2004 18:06:31:
Als Antwort auf:/In reply to: WBEC Ridderkerk new results. geschrieben von:/posted by: Leo Dijksman at 15 May 2004 17:55:53:

Premier Division:
Round 14:
Aristarch 4.41 ½1½1 3.0/4
Tao 5.6 ½0½0 1.0/4
------------------------------------
Crafty 19.12 ½½11 3.0/4
SmarThink 0.18a-r119 ½½00 1.0/4
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------
Dragon 4.6b 11½ 2.5/3 = 1 game to go!
Quark 2.35 Paderborn 00½ 0.5/3
------------------------------------

2nd Division:
Round 14:
The CrazyBishop 0050 11½½ 3.0/4
Resp 0.19 00½½ 1.0/4
------------------------------------
SlowChess 2.89b 1½½0 2.0/4
Frenzee 146 0½½1 2.0/4
------------------------------------
Round 15:
Amateur 2.80 ½111 3.5/4
The CrazyBishop 0050 ½000 0.5/4
------------------------------------
Scidlet 3.6 - Amy 0.8.7 1-0 = 3 games to go!
------------------------------------

New engine qualify:
Round 4:
Murderhole 1.0.10 - Borer 2.01a 0-1 70
Hokus Pokus 0.43 - Elephant 1.05 0-1 40
Geko 0.4.3 - Excelsior 2.32b 0-1 55
Round 5:
Fruit X0505 - Naum 1.0 1-0 68
AnMon 5.32 - Bruja 1.5 1-0 56
Hermann 1.1.1 - Borer 2.01a 0-1 66
Jonny 2.62 - Spike 0.2b 0-1 35
Alfil 403.1 - GNUChess 4TM 0-1 39
Delphil 0.7g - DanChess 1.04d beta =-= 99
Nesik 0.6.4 - Marvin 1.0.0 1-0 65
Elephant 1.05 - Praetorian 2.1 1-0 30
Needle 0.53.1 - Ajedrez Tactico 2.15 0-1 60
Excelsior 2.32b - Murderhole 1.0.10 0-1 23
Geko 0.4.3 - Hokus Pokus 0.43 0-1 36
Round 6:
Fruit X0505 - AnMon 5.32 =-= 71

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
15/05/2004:
Updated: Frenzee v157
Updated: Bruja v1.6
Updated: Crafty v19.13 (exe by Dann Corbit (egtbs fixed!))
Best wishes,
Leo.

I see that Fruit X0505 does very well.
Is it better than the public version?
Uri

[Johan Havegheer]

Geschrieben von:/Posted by: Johan Havegheer at 15 May 2004 18:08:12:
Als Antwort auf:/In reply to: WBEC Ridderkerk new results. geschrieben von:/posted by: Leo Dijksman at 15 May 2004 17:55:53:

Premier Division:
Round 14:
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------

Ruffian now at full force
Best
Johan

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 15 May 2004 18:18:12:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Johan Havegheer at 15 May 2004 18:08:12:

Premier Division:
Round 14:
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------

Ruffian now at full force
Best
Johan

No
It seems that it got tired after the first 2 wins(otherwise it could be 4-0).
Uri

[Tom Likens]

Geschrieben von:/Posted by: Tom Likens at 15 May 2004 19:50:31:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Uri Blass at 15 May 2004 18:18:12:

Premier Division:
Round 14:
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------

Ruffian now at full force
Best
Johan

No
It seems that it got tired after the first 2 wins(otherwise it could be 4-0).
Uri

Hey Uri,
Congratulations so far in division 2. I know there is plenty of chess left,
but after all your worry Movei is hanging tough and still in contention for
the top spot.
regards,
--tom

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 16 May 2004 01:16:22:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Tom Likens at 15 May 2004 19:50:31:

Premier Division:
Round 14:
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------

Ruffian now at full force
Best
Johan

No
It seems that it got tired after the first 2 wins(otherwise it could be 4-0).
Uri

Hey Uri,
Congratulations so far in division 2. I know there is plenty of chess left,
but after all your worry Movei is hanging tough and still in contention for
the top spot.
regards,
--tom

Thanks but if you look at the table you can see that movei lost against resp that is in the last place.
It means that there is not a big gap between movei and the weakest program in this division.
I think that Movei was lucky not to lose games on time(after all it has bugs and can lose games on time based on my tests).
I am surprised to see it happens with other programs like Amy when Movei is supposed to have more bugs because I do not use a lot of computer time for testing.
previous versions of movei had also bugs but for some reason Movei was lucky in previous tournaments and did not crush or lose on time inspite of the fact that I know that it could happen.
Uri

[Joachim Rang]

Geschrieben von:/Posted by: Joachim Rang at 16 May 2004 15:54:46:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Uri Blass at 15 May 2004 18:06:31:

Premier Division:
Round 14:
Aristarch 4.41 ½1½1 3.0/4
Tao 5.6 ½0½0 1.0/4
------------------------------------
Crafty 19.12 ½½11 3.0/4
SmarThink 0.18a-r119 ½½00 1.0/4
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------
Dragon 4.6b 11￾ 2.5/3 = 1 game to go!
Quark 2.35 Paderborn 00￾ 0.5/3
------------------------------------

2nd Division:
Round 14:
The CrazyBishop 0050 11½½ 3.0/4
Resp 0.19 00½½ 1.0/4
------------------------------------
SlowChess 2.89b 1½½0 2.0/4
Frenzee 146 0½½1 2.0/4
------------------------------------
Round 15:
Amateur 2.80 ￾ 3.5/4
The CrazyBishop 0050 ￾ 0.5/4
------------------------------------
Scidlet 3.6 - Amy 0.8.7 1-0 = 3 games to go!
------------------------------------

New engine qualify:
Round 4:
Murderhole 1.0.10 - Borer 2.01a 0-1 70
Hokus Pokus 0.43 - Elephant 1.05 0-1 40
Geko 0.4.3 - Excelsior 2.32b 0-1 55
Round 5:
Fruit X0505 - Naum 1.0 1-0 68
AnMon 5.32 - Bruja 1.5 1-0 56
Hermann 1.1.1 - Borer 2.01a 0-1 66
Jonny 2.62 - Spike 0.2b 0-1 35
Alfil 403.1 - GNUChess 4TM 0-1 39
Delphil 0.7g - DanChess 1.04d beta =-= 99
Nesik 0.6.4 - Marvin 1.0.0 1-0 65
Elephant 1.05 - Praetorian 2.1 1-0 30
Needle 0.53.1 - Ajedrez Tactico 2.15 0-1 60
Excelsior 2.32b - Murderhole 1.0.10 0-1 23
Geko 0.4.3 - Hokus Pokus 0.43 0-1 36
Round 6:
Fruit X0505 - AnMon 5.32 =-= 71

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
15/05/2004:
Updated: Frenzee v157
Updated: Bruja v1.6
Updated: Crafty v19.13 (exe by Dann Corbit (egtbs fixed!))
Best wishes,
Leo.

I see that Fruit X0505 does very well.
Is it better than the public version?
Uri

I as a betatester think it is better but not much (20-30 Elopoints). But I have no definite number, since we are still tuning and not yet comparing with the older version.
regards joachim

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 16 May 2004 16:10:34:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Joachim Rang at 16 May 2004 15:54:46:

Premier Division:
Round 14:
Aristarch 4.41 ½1½1 3.0/4
Tao 5.6 ½0½0 1.0/4
------------------------------------
Crafty 19.12 ½½11 3.0/4
SmarThink 0.18a-r119 ½½00 1.0/4
------------------------------------
Ruffian 2.1.0 11½½ 3.0/4
DeepSjeng 1.6 00½½ 1.0/4
------------------------------------
Dragon 4.6b 11￾ 2.5/3 = 1 game to go!
Quark 2.35 Paderborn 00￾ 0.5/3
------------------------------------

2nd Division:
Round 14:
The CrazyBishop 0050 11½½ 3.0/4
Resp 0.19 00½½ 1.0/4
------------------------------------
SlowChess 2.89b 1½½0 2.0/4
Frenzee 146 0½½1 2.0/4
------------------------------------
Round 15:
Amateur 2.80 ￾ 3.5/4
The CrazyBishop 0050 ￾ 0.5/4
------------------------------------
Scidlet 3.6 - Amy 0.8.7 1-0 = 3 games to go!
------------------------------------

New engine qualify:
Round 4:
Murderhole 1.0.10 - Borer 2.01a 0-1 70
Hokus Pokus 0.43 - Elephant 1.05 0-1 40
Geko 0.4.3 - Excelsior 2.32b 0-1 55
Round 5:
Fruit X0505 - Naum 1.0 1-0 68
AnMon 5.32 - Bruja 1.5 1-0 56
Hermann 1.1.1 - Borer 2.01a 0-1 66
Jonny 2.62 - Spike 0.2b 0-1 35
Alfil 403.1 - GNUChess 4TM 0-1 39
Delphil 0.7g - DanChess 1.04d beta =-= 99
Nesik 0.6.4 - Marvin 1.0.0 1-0 65
Elephant 1.05 - Praetorian 2.1 1-0 30
Needle 0.53.1 - Ajedrez Tactico 2.15 0-1 60
Excelsior 2.32b - Murderhole 1.0.10 0-1 23
Geko 0.4.3 - Hokus Pokus 0.43 0-1 36
Round 6:
Fruit X0505 - AnMon 5.32 =-= 71

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
15/05/2004:
Updated: Frenzee v157
Updated: Bruja v1.6
Updated: Crafty v19.13 (exe by Dann Corbit (egtbs fixed!))
Best wishes,
Leo.

I see that Fruit X0505 does very well.
Is it better than the public version?
Uri

I as a betatester think it is better but not much (20-30 Elopoints). But I have no definite number, since we are still tuning and not yet comparing with the older version.
regards joachim

What exactly do you tune?
Do you try to find better personality and if yes what parameters do you change?
Is the difference between original Fruit and FruitX0505 is changing weights or are there more differences?
The main difference in the next movei that I am going to release is possibility to create personality by editing a file with some hundreds of numbers and I may add some boolean features of fruit like open files with default value of 0.
I have not time to test changing the evaluation so I will probably still not evaluate things like open files or knight outposts in the default next version but I plan to give the user the possibility to test by changing the default value of 0(I need to write the code for evaluating it but I think that the hard part is to test if it is better and I prefer to test search changes that are not about changing parameters).
Uri

[Fabien Letouzey]

Geschrieben von:/Posted by: Fabien Letouzey at 17 May 2004 11:39:00:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Uri Blass at 16 May 2004 16:10:34:


Hi Uri,

What exactly do you tune? Do you try to find better personality and
if yes what parameters do you change?
Is the difference between original Fruit and FruitX0505 is changing
weights or are there more differences?
The main difference in the next movei that I am going to release is
possibility to create personality by editing a file with some hundreds
of numbers and I may add some boolean features of fruit like open
files with default value of 0.
I have not time to test changing the evaluation so I will probably
still not evaluate things like open files or knight outposts in the
default next version but I plan to give the user the possibility to
test by changing the default value of 0(I need to write the code for
evaluating it but I think that the hard part is to test if it is
better and I prefer to test search changes that are not about changing
parameters).

I have added UCI options for some eval parameters. Joachim tries to
improve the default values. For now I am mostly interested in
removing features that don't really help, to simplify the code.
The addition of mobility is the main difference. It seems to help
despite a ~20% slow down (straightforward implementation).
I am slowly re-writing the evaluation function, and am providing UCI
options as I go along.
Yes, Beowulf and some other engines do that. With UCI I don't even
need a file, but I must be careful about the number of options.
In the next public version I will try to provide only main options
(e.g. weights for centrality/mobility).
I think adding eval options is very interesting for users as search
options usually don't affect playing style.
I can't say I am delighted by working on the evaluation but Fruit 1.0
knows near to nothing ...
Fabien.

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 17 May 2004 12:24:06:
Als Antwort auf:/In reply to: Re: Fruit X geschrieben von:/posted by: Fabien Letouzey at 17 May 2004 11:39:00:

Hi Uri,

What exactly do you tune? Do you try to find better personality and
if yes what parameters do you change?
Is the difference between original Fruit and FruitX0505 is changing
weights or are there more differences?
The main difference in the next movei that I am going to release is
possibility to create personality by editing a file with some hundreds
of numbers and I may add some boolean features of fruit like open
files with default value of 0.
I have not time to test changing the evaluation so I will probably
still not evaluate things like open files or knight outposts in the
default next version but I plan to give the user the possibility to
test by changing the default value of 0(I need to write the code for
evaluating it but I think that the hard part is to test if it is
better and I prefer to test search changes that are not about changing
parameters).

I have added UCI options for some eval parameters. Joachim tries to
improve the default values. For now I am mostly interested in
removing features that don't really help, to simplify the code.
The addition of mobility is the main difference. It seems to help
despite a ~20% slow down (straightforward implementation).
I am slowly re-writing the evaluation function, and am providing UCI
options as I go along.
Yes, Beowulf and some other engines do that. With UCI I don't even
need a file, but I must be careful about the number of options.
In the next public version I will try to provide only main options
(e.g. weights for centrality/mobility).
I think adding eval options is very interesting for users as search
options usually don't affect playing style.
I can't say I am delighted by working on the evaluation but Fruit 1.0
knows near to nothing ...
Fabien.

Same for movei.
Movei still does not know about backward pawns or knight outposts.
It also does not know to increase the value of pawns in the endgame.
I implemented only yesterday knowledge about open files(similiar to fruit).
I still did not test it in games(I only gave it to analyze positions from one of movei's games to see if it give different scores than previous version for the right direction and it does it)

It's mobility function is relatively primitive and is based only on counting number of legal moves and the stage of the game and if one of the kings are in check it goes 2 plies backward until it finds a king that is not in check.
Note that I plan to give the user not only to change eval numbers but
to change also search numbers and I do not think that my numbers are optimal.
There is no significant speed change from doing it and movei became even faster from some of the cahnges.
Uri