Re: [computer-go] Long superko example
Brian, this is interesting. The sequence is as interesting as the superko, really, because it's the sort of strange sequence that would only occur to monte carlo programs. When black plays B9, a human white player would answer H4. This gives the highest win. If he's feeling obliging, he might answer A5. What he certainly would never play is A9. As for black, looks like he should resign instead of B9. I'm very curious what black thought the win rate was before B9 Christian On 04/10/2009 22:21, Brian Sheppard wrote: 1 2 3 4 5 6 7 8 9 A O O O X - X X O - B - O O X X O O - - C O - O X X - O O O D - O X X O O O X O E O O O O X X O X X F X O O X X X X X - G X X O O X O X - X H - X X - O O O X - J X X - O - X X - X X: B9 (hole-of-three) O: A9 (atari) X: B8 (capture) O: A8 (atari) X: A9 (capture) O: A8 (capture) ___ 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] [Fwd: Announcement ICGA Events 2010]
---BeginMessage--- 18th World Computer Chess Championship World Chess Software Championship World Computer Chess Blitz Championship JAIST Computer Olympiad International Conference on Computers and Games 2010 Kanazawa, Japan : September 24th to October 2nd 2010 The ICGA is delighted to announce that our events for 2010 will be held in Kanazawa, Japan, hosted by the Japan Advanced Institute of Science and Technology (JAIST). The contract for hosting this event was signed on October 5th in Kanazawa by ICGA President David Levy and JAIST President Takuya Katayama. The provisional dates are September 24th to October 2nd 2010. 2010 is the 20th anniversary of the founding of JAIST and our events will form an important role in their anniversary celebrations. Kanazawa is a city with a strong cultural identity. One of the most important sights in the city is Kenroku-en Garden, which is one of the three most beautiful gardens in Japan. JAIST have generously made available funding to enable us to provide financial support of 1,000 Euro for 10 of the participating teams in the World Computer Chess Championship, in order to defray their travel and accommodation costs. Full details will be made available in due course. This announcement serves also as a call for papers for the International Conference on Computer Games (2010). Full details of the conference will be available in due course. ---End Message--- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] [Fwd: Announcement ICGA Events 2010]
In message 4ac9e4f4.9000...@univ-lille3.fr, Rémi Coulom remi.cou...@univ-lille3.fr writes 18th World Computer Chess Championship World Chess Software Championship World Computer Chess Blitz Championship JAIST Computer Olympiad International Conference on Computers and Games 2010 Kanazawa, Japan : September 24th to October 2nd 2010 Thank you, Rémi, for reporting this. I wonder if anyone here has a URL for either the Tainan Computer Go Tournament, to take place on October 30th and 31st, or the GPW Cup, November 13th and 14th? I cannot read Chinese or Japanese, and so have failed to find anything by using a search engine. Nick -- Nick Weddn...@maproom.co.uk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] [Fwd: Announcement ICGA Events 2010]
Nick Wedd: vgf6vg99igykf...@maproom.demon.co.uk: In message 4ac9e4f4.9000...@univ-lille3.fr, Rémi Coulom remi.cou...@univ-lille3.fr writes 18th World Computer Chess Championship World Chess Software Championship World Computer Chess Blitz Championship JAIST Computer Olympiad International Conference on Computers and Games 2010 Kanazawa, Japan : September 24th to October 2nd 2010 Thank you, Rémi, for reporting this. Please come to Japan, all! I wonder if anyone here has a URL for either the Tainan Computer Go Tournament, to take place on October 30th and 31st, or the GPW Cup, November 13th and 14th? I cannot read Chinese or Japanese, and so have failed to find anything by using a search engine. http://ai.csie.ndhu.edu.tw:9898/eng/ is the URL. It's not Tainan but Taichu tournament this year. It's not two day but one day tournament at 30th. MoGoTW, Zen and some (six?) domestic programs have registered, I've heared. I'll attend and operate Zen via KGS. GPW Cup will be held in November 14th and 15th at Hakone, Japan as usual. 9x9 only, Chinese rules, komi is 7 points (thank you, Erik). Since I'm the director of the tournament, please freely ask me if you have any question. Hideki -- g...@nue.ci.i.u-tokyo.ac.jp (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] [Fwd: Announcement ICGA Events 2010]
Hi! On Tue, Oct 06, 2009 at 12:08:06AM +0900, Hideki Kato wrote: Nick Wedd: vgf6vg99igykf...@maproom.demon.co.uk: I wonder if anyone here has a URL for either the Tainan Computer Go Tournament, to take place on October 30th and 31st, or the GPW Cup, November 13th and 14th? I cannot read Chinese or Japanese, and so have failed to find anything by using a search engine. http://ai.csie.ndhu.edu.tw:9898/eng/ is the URL. It's not Tainan but Taichu tournament this year. It's not two day but one day tournament at 30th. MoGoTW, Zen and some (six?) domestic programs have registered, I've heared. I'll attend and operate Zen via KGS. GPW Cup will be held in November 14th and 15th at Hakone, Japan as usual. 9x9 only, Chinese rules, komi is 7 points (thank you, Erik). Since I'm the director of the tournament, please freely ask me if you have any question. Is it possible to participate in these tournaments remotely? I think I would like to start participating in these, but I certainly have no budget (or time) to travel outside of Europe. :-( -- Petr Pasky Baudis A lot of people have my books on their bookshelves. That's the problem, they need to read them. -- Don Knuth ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] [Fwd: Announcement ICGA Events 2010]
In message 4aca0bcf.14%hideki_ka...@ybb.ne.jp, Hideki Kato hideki_ka...@ybb.ne.jp writes Nick Wedd: vgf6vg99igykf...@maproom.demon.co.uk: In message 4ac9e4f4.9000...@univ-lille3.fr, Rémi Coulom remi.cou...@univ-lille3.fr writes 18th World Computer Chess Championship World Chess Software Championship World Computer Chess Blitz Championship JAIST Computer Olympiad International Conference on Computers and Games 2010 Kanazawa, Japan : September 24th to October 2nd 2010 Thank you, Rémi, for reporting this. Please come to Japan, all! I wonder if anyone here has a URL for either the Tainan Computer Go Tournament, to take place on October 30th and 31st, or the GPW Cup, November 13th and 14th? I cannot read Chinese or Japanese, and so have failed to find anything by using a search engine. http://ai.csie.ndhu.edu.tw:9898/eng/ is the URL. It's not Tainan but Taichu tournament this year. It's not two day but one day tournament at 30th. MoGoTW, Zen and some (six?) domestic programs have registered, I've heared. I'll attend and operate Zen via KGS. GPW Cup will be held in November 14th and 15th at Hakone, Japan as usual. 9x9 only, Chinese rules, komi is 7 points (thank you, Erik). Since I'm the director of the tournament, please freely ask me if you have any question. Thank you for the updates. Nick -- Nick Weddn...@maproom.co.uk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Long superko example
The sequence is as interesting as the superko, really, because it's the sort of strange sequence that would only occur to monte carlo programs. Yes, that is why I posted it. I'm very curious what black thought the win rate was before B9... I do not know. This position wasn't from a game. It was from a playout. I was searching for instances of hole of three situations, following MoGo's recommendation. I was building a regression suite of test cases. I don't know whether Pebbles could play that sequence. I just noticed the possibility of a repetition while I was validating that the situation was a genuine hole of three. As for black, looks like he should resign instead of B9. You are right that X dies first in alternating play. But the only way to be sure of that is to play B9 and see what happens. So hole of three rule does generate a critical play here. 1 2 3 4 5 6 7 8 9 A O O O X - X X O - B - O O X X O O - - C O - O X X - O O O D - O X X O O O X O E O O O O X X O X X F X O O X X X X X - G X X O O X O X - X H - X X - O O O X - J X X - O - X X - X X: B9 (hole-of-three) O: A9 (atari) X: B8 (capture) O: A8 (atari) X: A9 (capture) O: A8 (capture) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] end game analysis
A quick question: What programs are useful for coaching a player by analysing the moves that have been played in the endgame of some 19x19 game? What one would want to do is to input the position, say 30 moves from the end, and get a ranking of the remaining moves. It would be nice if it would not be too cumbersome to explore optimal follow up moves for any one of the moves, i.e. to select a move and see what the winning statistics for the followup moves is. It also should be possible to add more and more time to the analysis to see how stable it is if more time is available. The program should be able to use large computing resources (e.g. computing nodes with 32 CPU sharing 128GB RAM would be available). Thanks, Thomas ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] [Fwd: Announcement ICGA Events 2010]
Petr Baudis: 20091005153323.go6...@machine.or.cz: Hi! On Tue, Oct 06, 2009 at 12:08:06AM +0900, Hideki Kato wrote: Nick Wedd: vgf6vg99igykf...@maproom.demon.co.uk: I wonder if anyone here has a URL for either the Tainan Computer Go Tournament, to take place on October 30th and 31st, or the GPW Cup, November 13th and 14th? I cannot read Chinese or Japanese, and so have failed to find anything by using a search engine. http://ai.csie.ndhu.edu.tw:9898/eng/ is the URL. It's not Tainan but Taichu tournament this year. It's not two day but one day tournament at 30th. MoGoTW, Zen and some (six?) domestic programs have registered, I've heared. I'll attend and operate Zen via KGS. GPW Cup will be held in November 14th and 15th at Hakone, Japan as usual. 9x9 only, Chinese rules, komi is 7 points (thank you, Erik). Since I'm the director of the tournament, please freely ask me if you have any question. Is it possible to participate in these tournaments remotely? I think I would like to start participating in these, but I certainly have no budget (or time) to travel outside of Europe. :-( The tournament in Taiwan allows playing through KGS but according to the rules http://ai.csie.ndhu.edu.tw:9898/eng/p_7.htm, it's preferred to participate at least one person from each team. If not, the entry fee will be doubled. Please ask the organizer for detail (Click Contact us on http://ai.csie.ndhu.edu.tw:9898/eng/). GPW Cup has _NO_ Internet connection because there is no network in the seminor house where the workshop (GPW stands for Game Programming Workshop) is held. Sorry for inconvinience. In addition to above, UEC Cup will allow remote participants, though the rules are not open yet. The registration will start Oct 9th. See http://jsb.cs.uec.ac.jp/~igo/eng/index.html for detail. Hideki -- g...@nue.ci.i.u-tokyo.ac.jp (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] end game analysis
A comment to my own question: I should have formulated it better, of course all MC programs are useful in some sense. The specifics of the request is that the player is not avalable to play live against a normal program and to learn from interactive play. Also, for the analysis to be more accurate and/or to investigate positions that are earlier in the game the computing times may be too long for interactive sessions. Ideal would be a program submitted in batch-mode which is given an sgf file from a game and the program would analyse all positions starting with the last move going backwards and making comments into a file. I realize that MC programs are stronger in close games, so for each analysis the number of prisoners might be adapted to get the best out of MC so that from the analysis one can see where the player lost one or two points. Thomas On Mon, 5 Oct 2009, Thomas Wolf wrote: A quick question: What programs are useful for coaching a player by analysing the moves that have been played in the endgame of some 19x19 game? What one would want to do is to input the position, say 30 moves from the end, and get a ranking of the remaining moves. It would be nice if it would not be too cumbersome to explore optimal follow up moves for any one of the moves, i.e. to select a move and see what the winning statistics for the followup moves is. It also should be possible to add more and more time to the analysis to see how stable it is if more time is available. The program should be able to use large computing resources (e.g. computing nodes with 32 CPU sharing 128GB RAM would be available). Thanks, Thomas ___ 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] end game analysis
Many Faces can do some of this. You can easily try different moves or sequences and have the computer think at each position, although you have the start the analysis by hand. Win rate and PV are shown during the search so you can start a long search and stop it when it looks stable. After the search it shows the PC so you can see the full sequence it likes. It scales up to 8 core single machine, but not multimode. David -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go- boun...@computer-go.org] On Behalf Of Thomas Wolf Sent: Monday, October 05, 2009 9:32 AM To: computer-go Subject: [computer-go] end game analysis A quick question: What programs are useful for coaching a player by analysing the moves that have been played in the endgame of some 19x19 game? What one would want to do is to input the position, say 30 moves from the end, and get a ranking of the remaining moves. It would be nice if it would not be too cumbersome to explore optimal follow up moves for any one of the moves, i.e. to select a move and see what the winning statistics for the followup moves is. It also should be possible to add more and more time to the analysis to see how stable it is if more time is available. The program should be able to use large computing resources (e.g. computing nodes with 32 CPU sharing 128GB RAM would be available). Thanks, Thomas ___ 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: [SPAM] Re: [SPAM] Re: [computer-go] Progressive widening vs unpruning
I tried this yesterday with K=10 and it seemed to make Many Faces weaker
(84.2% +- 2.3 vs 81.6% +-1.7), not 95% confidence, but likely weaker. This
is 19x19 vs gnugo with Many Faces using 8K playouts per move, 1000 games
without and 2000 games with the change. I have the UCT exploration term, so
perhaps with exploration this idea doesn't work. Or perhaps the K I tried
is too large.
David
From: computer-go-boun...@computer-go.org
[mailto:computer-go-boun...@computer-go.org] On Behalf Of Olivier Teytaud
Sent: Saturday, October 03, 2009 1:28 AM
To: computer-go
Subject: Re: [SPAM] Re: [SPAM] Re: [computer-go] Progressive widening vs
unpruning
4) regularized success rate (nbWins +K ) /(nbSims + 2K)
(the original progressive bias is simpler than that)
I'm not sure what you mean here. Can you explain a bit more?
Sorry for being unclear, I hope I'll do better below.
Instead of just number of wins divided by numer of simulations,
we use nb of wins + K divided by nb of simulations + 2K;
this is similar to the even game heuristic previously cited;
it avoids that we 0% of success rate for a move tested just once.
If you apply UCT with constant zero in front of the sqrt{log(N)/N_i)
term, then such a regularization is necessary for showing consistency of UCT
for two-player games; and even with non-zero exploration terms, I guess
this kind of regularization avoids that the program spends a very long time
without looking at a move just because of a few bad first simulations. This
kind of detail is a bit boring, but I think K0 is much better in many
cases... well, maybe not for other implementations, depending on the other
terms you have - our formula is so long now I'm not able of writing it in
closed form :-)
By the way, K0 is in my humble opinion a very good idea if you want to
check that UCT with positive constant has a good effect in your code - I
feel that UCT is great if K=0, just because of the bad first simulation
effect - with K=0 and without exploration term, just loosing the first few
simulations can lead to the very bad situation in which a move is never
tested anymore.
Best regards,
Olivier
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
