Re: [computer-go] Former Deep Blue Research working on Go
terry mcintyre wrote: IIRC, a few Microsoft researchers did some interesting work with SVMs and the prediction of pro-level moves. I've always wondered whether that could be integrated with UCT to narrow the search tree. Hi, This is what I do in Crazy Stone: http://remi.coulom.free.fr/Amsterdam2007/ Mango does something similar, too. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
Andrés Domínguez wrote: 2007/10/10, Don Dailey [EMAIL PROTECTED]: Andrés, You are right about null move of course. The assumption that other moves are = to the value of a pass is much stronger in GO than in Chess, yet ironically it's not as effective in Go. That was what i was trying to say. Pass is one of the worst moves (except final) is good for null-move on Go. Of course you have reduced depth, probably bad with alpha-beta with a bad evaluation function, but looks interesting with UCT reducing the number of simulations and increasing the % value. I don't use UCT, so I haven't tried it. Andrés Hi, UCT does no alpha-beta pruning, so null-move pruning cannot be used. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
Rémi Coulom wrote: Andrés Domínguez wrote: 2007/10/10, Don Dailey [EMAIL PROTECTED]: Andrés, You are right about null move of course. The assumption that other moves are = to the value of a pass is much stronger in GO than in Chess, yet ironically it's not as effective in Go. That was what i was trying to say. Pass is one of the worst moves (except final) is good for null-move on Go. Of course you have reduced depth, probably bad with alpha-beta with a bad evaluation function, but looks interesting with UCT reducing the number of simulations and increasing the % value. I don't use UCT, so I haven't tried it. Andrés Hi, UCT does no alpha-beta pruning, so null-move pruning cannot be used. Rémi Hi again, I did not read your reply carefuly before answering, sorry. I still don't believe your approach could work. You would waste a lot of simulations searching a bad move, and it would be extremely difficult to determine how much the % value should be increased. In alpha-beta tree search, you only need to determine that one move is better than another, regardless of the difference in value. In UCT, it is very important to also determine how much better one move is. I cannot see any reasonable approach to determine how much the null move is worse than the others. Depending on very subtle details of the position, it could be a lot or very little. Regarding the question of null move in Go, I remember that some programmers who tried it in alpha-beta programs did not manage to make it work (Peter MacKenzie comes to mind, maybe others). As Don wrote, the main problem of null move is the depth reduction. It hides long-term threats that the evaluation function might not be able to evaluate. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
As Don wrote, the main problem of null move is the depth reduction. It hides long-term threats that the evaluation function might not be able to evaluate. even with a very good evaluation function, i would think that another problem (this is likely just restating what you and others have already said) is that your opponent can quite readily often crush you if you pass, even if he plays what would otherwise be a fairly substandard move. the sheer advantage of having sente for free can be huge. at the beginning of the game it's an entire handicap stone, and near the endgame it can mean several new ko threats. in the middle game it means winning any reasonable liberty race, turning many reasonable kills into sekis, blocking any ladder, etc. so it wouldn't, generally, ever generate any cutoffs, and yet you'd be checking it with every move for effectively no reason. there is a related concept that go players actually do use, and it has to do with reordering a set of moves that have been played to see if it changes the position. tewari analysis -- this is probably more useful than null-move pruning, as it should be able to make a relatively weak evaluation function act stronger. s. Take the Internet to Go: Yahoo!Go puts the Internet in your pocket: mail, news, photos more. http://mobile.yahoo.com/go?refer=1GNXIC ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
On 10/10/07, Don Dailey [EMAIL PROTECTED] wrote: In GO, threats tend to be very indirect and distant, at least from the point of view of a naive search algorithm and this is a real killer to the idea - my feeling is that null move in GO is not workable. I have the same feeling. Some years ago in Magog I did quite a lot of experiments with tricks like (recursive) null move pruning. Although it provided significant reductions in the search tree it consistently made the program play weaker. The only trick that (almost) seemed to work was Multi-Cut. Erik ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
I get: 500 - Internal Server Error I've wanted something like what you describe for running bots on CGOS and KGS. When I do see the script, I'll see what I can do about hacking in KGS support to it. On 10/10/07, Urban Hafner [EMAIL PROTECTED] wrote: Hej all, for those of you that also have the problem that they have more than one bot they want to run on CGOS, but only have one bot available: I've written a small script that automatically rotates your bots (so that each plays N games). It's available at http://bettong.net/articles/2007/10/10/cgos-rotation-script Cheers, Urban ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] CGOS rotation script
Hej all, for those of you that also have the problem that they have more than one bot they want to run on CGOS, but only have one bot available: I've written a small script that automatically rotates your bots (so that each plays N games). It's available at http://bettong.net/articles/2007/10/10/cgos-rotation-script Cheers, Urban PGP.sig Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
It works fine for me here. I hope you don't mind, Urban, I'll post the
script here so that Jason can see it, since he seems to have problems
accessing your server:
#!/usr/bin/env ruby
#
# Run bots one after the other on CGOS. The BOTS variable is an array
# if directories of the bots. In each of these directory there has to
# be an executable file called 'cgos.sh' that connects the bot to CGOS.
# This executable is also required to set the sentinel file name to the
# one given in the TERM variable below.
# Number of games for each bot
GAMES = 1
# The bots that should play
BOTS = [libEGO-AMAF, libEGO-AMAF-2, libEGO-AMAF-3]
# Sentinel filename to be used by CGOS
TERM = stop.txt
round = 0
if ARGV.length 0 and ARGV.first == stop
File.new(TERM, w)
exit
end
# Clean up (if script wasn't terminated correctly)
(BOTS+[.]).each do |dir|
Dir.chdir do
File.delete(TERM) if File.exists?(TERM)
end
end
until(File.exists?(TERM))
dir = BOTS[round % BOTS.length]
puts [#{Time.now}] Round #{round+1}
puts [#{Time.now}] Running '#{dir}' for #{GAMES} games
last_was_print = false
Dir.chdir(dir) do
games = 0
IO.popen('./cgos.sh 2/dev/null') do |pipe|
until(pipe.eof?)
line = pipe.gets
if line =~ /setup/
games += 1
File.new(TERM, w) if games = (GAMES-1)
end
case line
when /gameover/
line =~ /gameover\s+\d\d\d\d-\d\d-\d\d\s+(\S+)/
print \n if last_was_print
puts [#{Time.now}] #{$1}
last_was_print = false
when /setup/
line =~ /setup\s+.*?(\S+\(.*\)\s+\S+\(.*\))/
print \n if last_was_print
puts [#{Time.now}] #{$1}
last_was_print = false
when /info/
print .
STDOUT.flush
last_was_print = true
end
end
end
File.delete(TERM)
end
round += 1
end
File.delete(TERM)
On 10/10/07, Jason House [EMAIL PROTECTED] wrote:
I get:
500 - Internal Server Error
I've wanted something like what you describe for running bots on CGOS and
KGS. When I do see the script, I'll see what I can do about hacking in KGS
support to it.
On 10/10/07, Urban Hafner [EMAIL PROTECTED] wrote:
Hej all,
for those of you that also have the problem that they have more than one
bot they want to run on CGOS, but only have one bot available: I've
written a small script that automatically rotates your bots (so that
each plays N games). It's available at
http://bettong.net/articles/2007/10/10/cgos-rotation-script
Cheers,
Urban
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
On Oct 10, 2007, at 17:32 , Jason House wrote: I get: 500 - Internal Server Error Sorry about that. It should work now (I hope). I've wanted something like what you describe for running bots on CGOS and KGS. When I do see the script, I'll see what I can do about hacking in KGS support to it. I guess it shouldn't be too hard. Though I don't know if kgsgtp supports the use of a sentinel file to stop after the current page. So you might need to explicitly kill it. Urban PGP.sig Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
Quoting Rémi Coulom [EMAIL PROTECTED]: Regarding the question of null move in Go, I remember that some programmers who tried it in alpha-beta programs did not manage to make it work (Peter MacKenzie comes to mind, maybe others). As Don wrote, the main problem of null move is the depth reduction. It hides long-term threats that the evaluation function might not be able to evaluate. I used null-moves in my old program Viking which used alpha-beta with lazy MC-Evaluation. It worked in the sense that it searched deeper, but I never observed an increase in playing strength. This might of course mean that the implementation was buggy or could be improved somehow. -- Magnus Persson Berlin, Germany ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
On Oct 10, 2007, at 18:01 , Adrian Petrescu wrote: It works fine for me here. I hope you don't mind, Urban, I'll post the script here so that Jason can see it, since he seems to have problems accessing your server: Sure! Thanks Adrian. Urban PGP.sig Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
It looks like the script cycles through directories, kicking off bots via a script file. Making that script file do some other server seems trivial. The trick in my mind is figuring out how to get things to terminate properly. I forsee two issues: 1. Supporting # of games to play before disconnecting This may require parsing data dumps while connected to a server. While the script currently handles CGOS, it becomes tougher to extend. kgsGtp can write a log file, and parsing of that log file is possible (even though in a different format). It may be possible to mix checks that are specific to different servers without conflict as long as the checks match... e.g. /gameover/ for CGOS and /kgs-game_over/ for KGS. It may be easier to delay writing a sentinel file until some set time after kicking off the bot. #2. Causing the bots to disconnect You're right that kgsGtp does not support a sentinel file, but kgs does send kgs-game_over following each game. It's relatively trivial to make the bot check a sentinel file and exit at the end of a game. I also wonder about when the CGOS client checks the sentinel file. This would affect the easy alternative in #1. Checks at the start of a game could make it disconnect just as the first round it'd play in begins. Checks at the end of the game would be more convenient. PS: Is there a bug in the script, it looks like GAMES=1 and GAMES=2 would cause the same effect. It may be correct to swap the following two lines to fix this (or GAMES-1 to GAMES) games += 1 File.new(TERM, w) if games = (GAMES-1) On 10/10/07, Urban Hafner [EMAIL PROTECTED] wrote: On Oct 10, 2007, at 17:32 , Jason House wrote: I get: 500 - Internal Server Error Sorry about that. It should work now (I hope). I've wanted something like what you describe for running bots on CGOS and KGS. When I do see the script, I'll see what I can do about hacking in KGS support to it. I guess it shouldn't be too hard. Though I don't know if kgsgtp supports the use of a sentinel file to stop after the current page. So you might need to explicitly kill it. Urban ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS rotation script
On Oct 10, 2007, at 18:30 , Jason House wrote: It looks like the script cycles through directories, kicking off bots via a script file. Making that script file do some other server seems trivial. The trick in my mind is figuring out how to get things to terminate properly. I forsee two issues: I guess we should take this off-list. I'll gladly help you set it up to work for KGS, too. PS: Is there a bug in the script, it looks like GAMES=1 and GAMES=2 would cause the same effect. It may be correct to swap the following two lines to fix this (or GAMES-1 to GAMES) games += 1 File.new(TERM, w) ifgames = (GAMES-1) You're right. Thanks so much. Urban PGP.sig Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Mertin's p rivate 9×9 19×19 tournament
From: Nick Wedd nick at maproom.co.uk Subject: Re: [computer-go] Mertin's private 9×9 19×19 tournament Date: 2007-10-09 10:28:38 GMT Edward de Grijs grey.matter at hotmail.com writes Hello all, is something known about this tournament yet? In http://www.computer-go.info/events/future.html is stated that this tournament would take place this year in Germany, but so far no news. Does somebody have more information? This is Stefan Mertin. He runs tournaments by getting copies of go-playing programs and playing them against each other. He has told me of his plans to run one this year, but he may have decided that with CGOS working so well, it would be superfluous. He has posted to this list, so let's hope he sees this thread. Nick YES! It´s me - Stefan Mertin in Osnabrueck, Germany I am very happy about the great succes of Don Daileys Computer GO Server although with it my own private tournaments have become by far less important but my home PC is still running (nearly night and day!) playing new games between all the Go programs I have - So I hope to present first results still this year! I started playing private tournaments on my home PC in 2000 with a first 9x9 contest of all the Go-programs I could get for it - Nick Wedd published the results on the Go sites at the BGA. Then from 2002 to 2005 I played a new huge 13x13 tournament with at the end 25 programs and nearly 1000 games played - results and games are still here: http://www.geocities.com/comp_go/13x13.html Winner was Go++ (prototype of what later was available as Go++4.0) by Nick Reiss Second was GoeMate (a last tournament-version of 2001) by Chen Zhixing Third was the newest GnuGo version of that time 3.2 playing on level 15 and fourth was ManyFaces (v.10.2 shortly before v.11.0) by David Fotland together with GoProfessional-III, an earlier commercial version of Go++ here is a complete result-list of all the participants: rank programstatus version year author country --- - -- 01. GO PROF.(GO4++) xbeta72 Aug.2001 Dr. Michael Reiss England 02. GOEMATE x,c:jap Apr.2001 Prof. Chen, Zhixing China 03. GNU-GOfv3.2lev15 2002 FSF(D.Bump +div.) [EMAIL PROTECTED] 04. GO PROFESSIONAL cvIII 1999 Dr. Michael Reiss England 04. MANY FACES OF GO x,c v10.2 2001 David Fotland USA 06. GO-AHEAD xv4.2 2001 Dr. Peter Woitke Germany 07. HANDTALK cv99.04 (98.10E) Prof. Chen, Zhixing China 07. WINHONTE fv0.94 2003 Fredrik A. Dahl Norway 07. WULU c,d v2.38 2001 Lei Xiuyu Wulu-TeamChina 10. GO INTELLECT xv10.0Win 1998 Prof. Ken Chen USA 11. DARIUSH xv2.6.8.2 2002 E.Marchand + F.Boissac France 12. GNU-GOfv3.0 Aug.2001 FSF(D.Bump +div.) [EMAIL PROTECTED] 13. GOLIATH c,d v3.5 1996 Mark Boon Netherlands 14. KATSUNARI d,x v8.0 Aug.2001 Sei, ShinichiJapan 15. STAR OF POLANDc? v6.0 1993 Janusz Kraszek Poland 15. AYA fv4.65 2001 Hiroshi YamashitaJapan 17. EGO c,d samurai1995 Bruce WilcoxHawaii 18. EXPLORER fv3.3(Mac) 1995 Martin Müller Austria-CH-USA 19. SMARTGO x,c v1.0.1 2002 Anders Kierulf USA 20. INDIGOx,fSept.2001 Dr. Bruno Bouzy France 20. GOLOISx,c v1.0 2001 Dr. Tristan CazenaveFrance 22. GO-STAR xv4.0 Aug.2001 Jimmy Lu (Lu Jinqiang) China 22. PROGO ?v1.4 1991 Robert RehmNetherlands 24. TS-GO xv9.3 2001 Ivo Tonkes Netherlands 24. TURBO-GO sv5.0 2000 Arnoud v.d. Loeff Netherlands (f=free; s=shareware; d=demo free; c=commercial; x=tourn./experimental version) In 2007 I now first started playing hundreds of games on the big board 19x19 including Go++5.0 / GoeMate2001 / FunGo2000 / ManyFaces11.0 and GoAhead5.2 but then I changed again to the small board 9x9 and I hope to finish soon a first set of rounds and want to publish the results and all the 9x9 games played so far between these programs: ManyFacesOfGo v.11.0 Go Intellect v.10 WinHonte v.2.01 Go++ v.5.0 GoeMate 2001 GnuGo v3.6 level 15 Aya v5.99 CrazyStone0005 and Suzie v0.26 This will be nearly round robin with 150 games each pairing. Next perhaps SmartGo 2.7 and Dariush 5.2 may be included (then without playing each other program) and of course MoGo - but all the above participating programs are older versions and I think it could be most interesting to include newer versions of most of them to see the progress and I hope very much on all the programmers to
Re: [computer-go] Former Deep Blue Research working on Go
At 02:33 PM 10/7/2007, you wrote: Found this link and thought you all might find it interesting. http://www.spectrum.ieee.org/oct07/5552 thread on slashdot: http://slashdot.org/article.pl?sid=07/10/10/1758244 --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
I'm just now reading the article. Monte Carlo techniques have recently had success in Go played on a restricted 9-by-9 board. My hunch, however, is that they won't play a significant role in creating a machine that can top the best human players in the 19-by-19 game. The author loses credibility with this statement. On 10/10/07, Ray Tayek [EMAIL PROTECTED] wrote: At 02:33 PM 10/7/2007, you wrote: Found this link and thought you all might find it interesting. http://www.spectrum.ieee.org/oct07/5552 thread on slashdot: http://slashdot.org/article.pl?sid=07/10/10/1758244 --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
Of no particular importance I suppose, but did any one else get the impression after looking at the picture (and the way he is holding the stone) that he is not a regular go player? Chris Fant wrote: I'm just now reading the article. Monte Carlo techniques have recently had success in Go played on a restricted 9-by-9 board. My hunch, however, is that they won't play a significant role in creating a machine that can top the best human players in the 19-by-19 game. The author loses credibility with this statement. On 10/10/07, Ray Tayek [EMAIL PROTECTED] wrote: At 02:33 PM 10/7/2007, you wrote: Found this link and thought you all might find it interesting. http://www.spectrum.ieee.org/oct07/5552 thread on slashdot: http://slashdot.org/article.pl?sid=07/10/10/1758244 --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Chris Fant wrote: I'm just now reading the article. Monte Carlo techniques have recently had success in Go played on a restricted 9-by-9 board. My hunch, however, is that they won't play a significant role in creating a machine that can top the best human players in the 19-by-19 game. The author loses credibility with this statement. Monte Carlo is the best thing going right now and the most probable future direction, software or hardware - that's my hunch anyway! - - Don On 10/10/07, Ray Tayek [EMAIL PROTECTED] wrote: At 02:33 PM 10/7/2007, you wrote: Found this link and thought you all might find it interesting. http://www.spectrum.ieee.org/oct07/5552 thread on slashdot: http://slashdot.org/article.pl?sid=07/10/10/1758244 --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFHDZZMDsOllbwnSikRAj1JAJ94Msw1bcN0Iu4gpAR3XuQuCkpkKQCfeuwc T7o/PxRxGxSanLOc7kug3Wg= =6fTh -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Former Deep Blue Research working on Go
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 He is clearly posing for a picture, this is not a spontaneous photograph. Notice the Thinker pose. I'm not a good go player at all, but the board position seems a little unnatural to me. But it could be my lack of experience. Over the last few decades, there have been may movies and television shows where a chess board appears in some scene with perhaps someone player a game. These are almost always WRONG in some obvious way. For instance someone plays a move and announces check. Then the response is a checkmate!Possible, but highly improbably. Very common is the king and queen on the wrong squares or a pawn on the 1st rank or something else really silly. Although a king and queen could move to these squares, it's extremely unlikely, especially near the opening. - - Don Richard J. Lorentz wrote: Of no particular importance I suppose, but did any one else get the impression after looking at the picture (and the way he is holding the stone) that he is not a regular go player? Chris Fant wrote: I'm just now reading the article. Monte Carlo techniques have recently had success in Go played on a restricted 9-by-9 board. My hunch, however, is that they won't play a significant role in creating a machine that can top the best human players in the 19-by-19 game. The author loses credibility with this statement. On 10/10/07, Ray Tayek [EMAIL PROTECTED] wrote: At 02:33 PM 10/7/2007, you wrote: Found this link and thought you all might find it interesting. http://www.spectrum.ieee.org/oct07/5552 thread on slashdot: http://slashdot.org/article.pl?sid=07/10/10/1758244 --- vice-chair http://ocjug.org/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFHDZWKDsOllbwnSikRAjGzAKDKUOHaEPnme19+d/UxJkSsNbJrzwCgiJeH /CvKCzEEo8Ds5e8+ZFA1BbU= =t0zW -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
