TRUUS (draughts program) in a normal tournament
Joost de Heer
I know this isn't about chess computers, but I thought you might find this
interesting.
Worlds strongest draughts program 'TRUUS' (by Stef Keetman) participated
last week in a 'normal' tournament. Her results were 2 won, 5 draws. She
played against 1 IGM and 2 IM. In the final ranking she was third. I had
to play myself against the compueter, and I don't really like it. It's
just so weird. You can't do the 'usual' tricks : Stare into the other's
eyes, stand behind his back when he's thinking, make the position very
complicated .... If anyone's interested in the games TRUUS played, I am
posting all games in rec.games.abstract.
Joost
--
Think about all the good in your life - It's only temporary
Think about all the positive sides in life - They never last forever
So drink to forget and drown all your sorrow SENTENCED
Bury your dreams and choose Catharsis NEPENTHE
Simon Read
How does this compare with Chinook?
Marcel van Kervinck
Chinook plays checkers, a trivial game when compared to draughts.
Marcel
-- _ _
_| |_|_|
|_ |_ Marcel van Kervinck
|_| bue...@urc.tue.nl
Rein Halbersma
Checker vs draughts, draughts vs chess, chess vs go; can one really say
that one game is trivial when compared to another? IMHO, one cannot!
Afterall it are the players who make the game difficult, not just the
game tree complexity.
To answer Simon Read's question: Chinook is far better in checkers than
Truus is in draughts. Chinook plays (slightly?) stronger than the world
champion whereas Truus is at most a strong IM/weak GM in normal gameplay
and a weak/intermediate GM in blitz/rapid games.
One of the reasons is that Truus isn't using huge parallel machines and
8 piece databases. As far as I know, only all 4 piece databases and some
5 piece databases have been created so far.
--
+---------------------------------------------------+
| Rein S. Halbersma |
| |
| e-mail: R.S.Ha...@cpedu.rug.nl |
| URL: www.cpedu.rug.nl/~N0769940/ |
+---------------------------------------------------+
Jonathan Schaeffer
bue...@asterix.urc.tue.nl (Marcel van Kervinck) writes:
>Simon Read (s.r...@cranfield.ac.uk) wrote:
>> How does this compare with Chinook?
>Chinook plays checkers, a trivial game when compared to draughts.
5 x 10^20 positions is not "trivial". Yes, there may be more
possible positions in 10X10 draughts, but so what? The 8x8 game
is difficult enough.
And when 10X10 checkers really becomes "trivial", then 12x12 checkers
(so-called Canadian checkers) awaits you.
Joost de Heer
>As far as I know, only all 4 piece databases and some
>5 piece databases have been created so far.
Stef Keetman (author of TRUUS) finished the 5-piece database about 2 years
ago. He started on the 6-piece database (I think he started with the '4 to 2'
endgames since they are the most frequent 6-piece positions to appear in
normal play) but stopped working on it when he left Maastricht University.
Marcel van Kervinck
Jonathan Schaeffer (jona...@cs.ualberta.ca) wrote:
> bue...@asterix.urc.tue.nl (Marcel van Kervinck) writes:
> >Chinook plays checkers, a trivial game when compared to draughts.
> 5 x 10^20 positions is not "trivial". Yes, there may be more
> possible positions in 10X10 draughts, but so what? The 8x8 game
> is difficult enough.
No offence intended. Just wanted to point that they are
completely different games. Checkers is 'trivial' in the
sense that the best player is far closer to perfect play
than the best draughts player. At least, a former world
champion (10x10) once told me that. You are the authority on
checkers, I guess.
Rein Halbersma
Joost de Heer wrote:
[snip]
> Stef Keetman (author of TRUUS) finished the 5-piece database about 2 years
> ago. He started on the 6-piece database (I think he started with the '4 to 2'
> endgames since they are the most frequent 6-piece positions to appear in
> normal play) but stopped working on it when he left Maastricht University.
>
> Joost
> --
[snip]
Are there any statistics published on this by Keetman? Like drawing
percentage, longest win etc. ? In 1993, Bert van Oortmarssen published
all the 4 piece endgames in the Dutch draughts magazine 'Hoofdlijn'. It
would be nice if someone could get a hold of Keetman's figures.
Alan Tomalty
Jonathan Schaeffer (jona...@cs.ualberta.ca) writes:
> bue...@asterix.urc.tue.nl (Marcel van Kervinck) writes:
>
>>Simon Read (s.r...@cranfield.ac.uk) wrote:
>>> How does this compare with Chinook?
>
>>Chinook plays checkers, a trivial game when compared to draughts.
>
> 5 x 10^20 positions is not "trivial". Yes, there may be more
> possible positions in 10X10 draughts, but so what? The 8x8 game
> is difficult enough.
>
> And when 10X10 checkers really becomes "trivial", then 12x12 checkers
> (so-called Canadian checkers) awaits you.
I must protest this thread. Checkers and all similar games should only be
discussed
discussed in relation to computer chess. If we make an exception
for checkers on this newsgroup, then other board games will want
in. Komputer Korner
--
Komputer Korner
Jonathan Schaeffer
Rein Halbersma <R.S.Ha...@cpedu.rug.nl> writes:
>Joost de Heer wrote:
>> Stef Keetman (author of TRUUS) finished the 5-piece database about 2 years
>> ago. He started on the 6-piece database (I think he started with the '4 to 2'
>> endgames since they are the most frequent 6-piece positions to appear in
>> normal play) but stopped working on it when he left Maastricht University.
>>
>> Joost
>> --
>Are there any statistics published on this by Keetman? Like drawing
>percentage, longest win etc. ? In 1993, Bert van Oortmarssen published
>all the 4 piece endgames in the Dutch draughts magazine 'Hoofdlijn'. It
>would be nice if someone could get a hold of Keetman's figures.
Correction. I computed the 5-piece 10x10 draughts databases, using Stef
Keetman's move generator. We used the Chinook database code, merely modifying
our data structures to support 10x10 instead of 8x8 boards. I have all the
5-piece databases and a few of the 6-piece. Completing the 6-piece and
doing the 7-piece databases is doable, but I have found no reason to do
so. There is a suggestion that we complete the 10X10 draughts databases
and make them available on a CD, complete with the 7-piece 8X8 checkers
databases and the 8X8 pool checkers databases.
I have all the W/L/D percentages somewhere, but the log files that contain
the longest win sequences are gone as far as I know. On the other hand,
building the 5-piece databaes takes only a month or two so they can be easily
reconstructed.
Chris Whittington
Disagree. For programmers there are enough similarities between
checkers and chess in terms of search etc. to make it worthwhile
studying what happens in the 'other' field.
Chris Whittington
Peter W. Gillgasch
In article <83426443...@cpsoft.demon.co.uk>, Chris Whittington
<chr...@cpsoft.demon.co.uk> wrote:
>dc...@FreeNet.Carleton.CA (Alan Tomalty) wrote:
[snip]
>> I must protest this thread. Checkers and all similar games should only be
>> discussed
>> discussed in relation to computer chess. If we make an exception
>> for checkers on this newsgroup, then other board games will want
>> in. Komputer Korner
>> --
>>
>> Komputer Korner
>
>Disagree. For programmers there are enough similarities between
>checkers and chess in terms of search etc. to make it worthwhile
>studying what happens in the 'other' field.
I second Chris. Keep in mind that the playing strength of the best checkers
programs (and especially Chinook) is much closer to being "perfect" than
the best chess programs. So checkers may provide us with some interesting
insights what needs to be done to make chess programs even better. Good
example is the usage of endgame databases in general and what to do with
"draw" scores in particular. Actually I wouldn't mind if some other guys
that write programs playing games like Othello or Go (he he...) jump into
this group and discuss algorithmic stuff.
IMHO this newsgroup is more relevant to the chess programmer than to the
user of chess programs and while checkers discussions may be crossly off
topic for the "user" it is right on target for the "programmer".
-- Peter
p.s.: Why don't you protest when some mortals post their latest insight
on some human vs. human game without any computer context ? They
have rec.games.chess.analysis, right ?
Joost de Heer
>Correction. I computed the 5-piece 10x10 draughts databases, using Stef
>Keetman's move generator.
Excuses for giving the wrong person the credit for the work. I misinterpreted
his words probably. But I knew he was working with you at that time.
M D. Grimminck
jona...@cs.ualberta.ca (Jonathan Schaeffer) writes:
>>Are there any statistics published on this by Keetman? Like drawing
>>percentage, longest win etc. ? In 1993, Bert van Oortmarssen published
>>all the 4 piece endgames in the Dutch draughts magazine 'Hoofdlijn'. It
>>would be nice if someone could get a hold of Keetman's figures.
There is a set of 4 piece endgames available on the net, which
includes the distance to win. I have computed most of the
5 piece endings with my own program, I will make the data
available when I have time.
>Correction. I computed the 5-piece 10x10 draughts databases, using Stef
>Keetman's move generator. We used the Chinook database code, merely modifying
>our data structures to support 10x10 instead of 8x8 boards. I have all the
>5-piece databases and a few of the 6-piece. Completing the 6-piece and
>doing the 7-piece databases is doable, but I have found no reason to do
>so. There is a suggestion that we complete the 10X10 draughts databases
>and make them available on a CD, complete with the 7-piece 8X8 checkers
>databases and the 8X8 pool checkers databases.
>I have all the W/L/D percentages somewhere, but the log files that contain
>the longest win sequences are gone as far as I know. On the other hand,
>building the 5-piece databaes takes only a month or two so they can be easily
>reconstructed.
--
Michel Grimminck, Computational Physics, University of Amsterdam.
draughts/checker page: http://carol.fwi.uva.nl/~grimmink/draughts.html