[computer-go] Tool for handshaped MC trees?
Hello everyone, does someone of you know if there exists some (software) tool for generating handshaped Monte Carlo trees, for instance for 9x9 go ? Such a tool might be helpful in designing new MC algorithms, for instance new variants of UCT. Ingo Althofer (using MC + variants in his project on computer-aided game inventing) -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser begin:vcard fn;quoted-printable:Ingo Alth=F6fer n;quoted-printable:Alth=F6fer;Ingo adr;home:;;Schubertstr. 16;Lage;;32791; bday:1000-01-01 00:00:00 version:2.1 rev:20080829T113820Z end:vcard ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Goal-directedness of Monte-Carlo
In the last few weeks I have been investigating
Monte-Carlo (MC) game tree search on a rather abstract
level. Especially I was able to "reproduce" the
following behaviour of MC in a very clear model:
MC is playing most "goal-directed" ("zielgerichtet"
in German) when the position is balanced or when
the side of MC is slightly behind. However, when
MC is clearly ahead or clearly behind it is playing rather
lazy.
Does someone here know, if there are or have been
investigations on this topic in existing papers or
projects (for instance in the context of computer go)?
Thanks in advance for all replies!
Ingo Althofer
--
Pt! Schon das coole Video vom GMX MultiMessenger gesehen?
Der Eine für Alle: http://www.gmx.net/de/go/messenger03
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Goal-directedness of Monte-Carlo
Hello Gian-Carlo, Gian-Carlo Pascutto wrote: > There has been discussion here about dynamic komi to keep > the winning rate close to 50%. As far as I saw there was no > clear conclusion about whether that works. Some people argued > that it should not exist and measuring objective winning rates is > always the right thing. > > However, there is less and less doubt in my mind about that the > effect you describe exists and it hurts playing strength. > > What exactly did you do when you say "reproduce in a clear model"? I have just written a quick report on my experiment. You can read or download it from http://www.althofer.de/mc-laziness.pdf Best regards, Ingo. -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Goal-directedness of Monte-Carlo
Olivier Teytaud wrote: > ... I find the following elements interesting: > 1) MoGo was seemingly weaker in handicap games... Was this (feeled) weakness on both sides (giver and taker) of handicap tables? > 2) MoGo has become stronger (and the difference is huge for long time > settings) with more randomization; ... This strong improvement with > randomization > might be also an improvement in the case of handicap games, but I've not > checked that. Of course, more randomization in sense of "indenpendent runs" and also in sense of "different heuristics" should help. But I do not see, how it might overcome the laziness problem of MC. > 3) By the way, we have positive results (small improvements, but > improvements) by using different Monte-Carlo simulations for different > cores. This is consistent with 2) above - by performing N simulations of > type 1 and M simulations of type 2, you cover more > markov models than with one Markov model - you can cover different > scenarios. It also fits well with the philosophy of "the wisdom of the crowds", where it is stated that an important source of this wisdom is that different people have different knowledge/heuristics. * Here comes a proposal to check for the laziness of Monte Carlo go programs: Take some (a dozen?) typical board positions. For each such position P do the following: For "each" komi value k make 100 runs of your program on (P; k), say with fixed thinking time of 10 seconds. Monte Carlo programs are random, and thus also have random outcomes. Make statistics of the 100 moves proposed by the program, and determine the entropy H(P; k) = minus sum of (h_i * log(h_i) ), where the h_i are the frequencies of the different move proposals. Small entropy would mean that the algorithm is rather sure about the best move. Large entropy means that there are several moves with similar merits (in the opinion of the algorithm). When you find a central value k' = k'(P), such that ... > H(P; k'-2) > H(P; k'-1) > H(P; k') < H(P; k'+1) < H(P; k'+2) < ... this would be an indication that k' is near to the likely value of the position. Ingo. -- Pt! Schon das coole Video vom GMX MultiMessenger gesehen? Der Eine für Alle: http://www.gmx.net/de/go/messenger03 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Bernd Bruegmann picture
In November 2007, there was a "long night of science" in Jena. My group presented games and research on games. One of the visitors was Dr. Bernd Bruegmann (professor for gravitational theory in theoretical physics), who had been the first to apply Monte Carlo search in computer go back in 1993. Here is a photo from the event, showing him. http://www.althofer.de/bremer-walter-bruegmann.jpg Bernd Bruegmann is on the right, just talking with Lars Bremer (with the red card, journalist from c't magazine and computer games afficionado). In the background Rico Walter (Ph.D. student in discrete optimization) is listening. Ingo. >From the very same event is this poster on MC and UCT, designed by Jakob Erdmann: http://www.althofer.de/poster--monte-carlo.jpg -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] sgf format for non-quadratic board sizes ?
Hello, does there exist a generalisation of the sgf-format for rectangular board sizes? The background of my question is that I would like to try how MCTS works on small boards, and there are not so many quadratically ones... (For instance, I am eager to see what happens on 7x5-board.) By the way, what do MCTS programs "say" to Erik van der Werf's optimal solution for Go on 5x5 board? Even for komi=24.5 Black should win. However, (MCTS-) programs may have problems with such a high komi-value. Ingo. http://erikvanderwerf.tengen.nl/5x5/5x5solved.html -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: sgf format for non-quadratic board sizes?
Dear Urban, thx for the quick reply. >> does there exist a generalisation of the sgf-format >> for rectangular board sizes? > > What exactly do you mean by generalization? You can > use SZ[1] to define the size of your rectangular board. > > [1] http://www.red-bean.com/sgf/properties.html#SZ I did not know about the notion SZ[m:n] for rectangular boards. And neither do at least some publicly available Go programs. I tried to enter SZ[3:5] into ManyFaces (version 10) and in Leela, without success. They simply show a 3x3 board... Ingo. -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] MyungWan vs MoGo
Hello, on September 21, there was a new exhibition match between MyungWan (8p) and MoGo (on massive hardware). They played two 19x19 games with 7 stones handicap, first a short warm-up with 15 min per side, and then a long one with 90 min per side. The short one turned into a loose-ladder catastrophy for MoGo, the longer one ran more or less "normal", with the pro winning clearly in the end. sgf of both games may be found at http://www.gokgs.com/gameArchives.jsp?user=mogotitan Some interesting analysis on the second (long) game has been done by Thomas Redecker with the help of the analysis/score- functions of "Many Faces of Go 11" and "SmartGo 2.8". He condensed the result into scoresheets, which you can see at http://www.althofer.de/thomas-redecker-on-myungwan-vs-mogo.gif The diagram in the top is by MFgG, the lower one by SmartGo. The fat red bars have been included by Thomas Redecker. Ingo. PS: Thx to Thomas for allowing me to hang in the gif on my homepage. Originally, it was posted in the (German language) forum of the DGoB, but is a bit difficult to find there. -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Analysis of 6x6 Go
Stefan Reisz is the author of the website http://www.reisz.de/gohome.htm There he claims to have a solution for 6x6-Go with Japanese rules. The outcome of his handmade analysis is that komi=3 would be fair. The analysis may be downloaded from the site, as sgf file. Does someone here know of other (documented) attempts to solve 6x6 Go? Ingo Althofer. PS: After some hours of interactive analysis with Leela my impression is that in case of Chinese rules the value of 6x6-Go might be +2. -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Analysis of 6x6 Go
Hello Robert, thx for the feedback. >> Does someone here know of other (documented) attempts >> to solve 6x6 Go? > > Didn't Erik van der Werf do it under his rules? He did it for 5x5-Go, see at http://erikvanderwerf.tengen.nl/5x5/5x5solved.html Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Analysis of 6x6 Go
Erik van der Werf wrote: > ... > Optimal play on 6x6 under Chinese rules is expected > to give a Black win by 4 points. I want to lay open, why my expectation for 6x6-Go under Chinese rules is +2 for Black. With Leela, I played two games (or game fragments) in analysis mode, starting the machine from fresh in each situation, and stopping analysis after 500,000 nodes. Game 1: Whenever it was Black's turn, komi was set to 3.5 . Whenever it was White's turn, komi was set to 4.5 . The numbers in brackets are the win percentages, as shown by Leela after 500,000 nodes. 1.c3 (39.9) 2.d4 (73.7) 3.c4 (46.0) 4.d3 (77.0) 5.d5 (55.3) 6.e5 (85.6) 7.d2 (42.0) 8.e2 (91.8) 9.e4 (24.5) 10.e3 (96.1) 11.e6(17.6) Resigned on behalf of Black. Game 2: Whenever it was Black's turn, komi was set to 1.5 . Whenever it was White's turn, komi was set to 2.5 . The numbers in brackets are the win percentages, as shown by Leela after 500,000 nodes. 1.c3 (71.5) 2.d4 (55.5) 3.d3 (75.4) 4.c4 (60.2) 5.b3 (64.1) 6.e3 (74.5) 7.e2 (59.2) 8.e4 (68.2) 9.b5 (58.9) 10.b4 (73.3) 11.a4 (53.3) 12.a5 (73.6) 13.a3 (65.2) 14.c5 (72.6) 15.a6 (77.1) 16.f2 (81.6) 17.e1 (81.7) 18.c6 (76.6) 19.a5 (81.4) 20.e5 (94.3) 21.c1 (98.6) So, both sides are optimistic to reach their respecitve goals. Conclusion: From the viewpoint of Leela (at 500,000 nodes), komi=2.0 is in the "habitable zone" for the the starting position of 6x6-Go, whereas komi=4.0 is not. When instead fair komi would be 4, each of the games above should contain errors. Ingo. -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Analysis of 6x6 Go
Don Dailey wrote: > I think a better way to do this is to self-play a few hundred games with > various komi values. Do you mean HUMAN self-play or COMPUTER self-/auto-play? When you mean human self-play, I am not sure that this is a safer way for such small boards. > The correct komi will be clear from those games. > This worked on 7x7 Are such 7x7-games documented somewhere, for instance in the internet? > so I assume it would work on 6x6. Of course this > cannot be considered a "proof." Right. What I did is a computer-aided (incomplete) analysis: running repeated Monte-Carlo searches where the komis for both sides differ by 1. Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Results of recent Computer Go events
Gian-Carlo Pascutto wrote: > Apparently it was ruled a loss for Many Faces of Go. I am appealing it - > there is no reason why the refree has to intervene when the players > agree on the score. The result of the game could be very important for > the tournament result. Hello, I looked into the tournament rules at http://www.icga.org/tournaments/olympiadgo.pdf Paragraphs 3.7.2 and 3.7.3 say that the result should be taken on which the programs agree. Greetings, Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Correct Komi for 6x6 is 2.0
Hello Magnus, interesting and strange. I went through your constructed game (it is repeated here without the in-between text) > bC4 wD3 bC3 wD4 bD5 wE5 bD2 wE2 ... > ... bE3 wE4 bE1 wF3!!!... > ... bC1 wC5 bB5 wD6 bB6 wF1!... > ...bF2 wC6 bB4! > ...wF1 bE6 wF6 bF2 wE3 bD1 wF1 bF5 wF4 > bF2 wPass bF1 B+3.5 Here I do not understand your counting: Black has control over 21 sqaures, White over 15. So, the final score of your game would be B+6. But ... when White instead of passing continues wC2, the game should go on with bB2 wF1 bPass wF2, and now the score is B+2. Or did I miss something? By the way, in my experimental runs with Leela it was really strange to see the evaluation switching between "rather extremes" - on this little board. Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Correct Komi for 6x6 is 2.0
Magnus Persson wrote: > Ingo Althofer wrote: >> But ... when White instead of passing continues wC2, >> the game should go on with >> bB2 wF1 bPass wF2, and now the score is B+2. > > Black ignores w C2 and plays F1. F1 by Black would be a huge blunder, because then White plays B1 and kills the black group in the lower right corner. So, Black has to react on the quasi-Ko-thread c2, giving White time to occupy F1. Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Correct Komi for 6x6 is 2.0
Hello Magnus, there was indeed a notation error in your original posting. There you give 13.C1 (see last line below) and not 13.C2 as in the sgf. Ingo. > I have been trying to see what Valkyria does. But it is a little > unstable when it reads deep at 6x6. It should not be a problem for > Valkyria but I have not had any time to search for the bug. > > Anyway the 2.5 komi black should lose if Don is right. So I have the > following very cute complete game sequence: > > bC4 wD3 bC3 wD4 bD5 wE5 bD2 wE2 ... > > The beginning is very natural, and although I did not check carefully > all what Don reported but I think there is agreement that this is a > strong seqence for both colors. > > ... bE3 wE4 bE1 wF3!!!... > > Normally wF2 is played in the corner. But with wF3 white has the > option to play aggresively with wF1 which usually is a bad idea > because the ko fight risk to much. But on a small board things are not > normal... > > ... bC1 wC5 bB5 wD6 bB6 wF1!... -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Congratulations to David Fotland!
His program Many Faces of Go has become winner in the 9x9-Go competition in the "13th International Computer Games Championship", held in Beijing. Rank 2 for MoGo after tiebreak against Leela. http://www.grappa.univ-lille3.fr/icga/tournament.php?id=180 with table and sgf of many games. Today the 19x19 competition has (or should have) started in Beijing. Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Computer-Go on small boards
Gian-Carlo Pascutto replied: > Don Dailey wrote: >> 4. I believe Leela, at a higher level and with a "correction" book >> would play perfect or very close to perfect on 6x6. This may >> depend on seki issues however, it may not be possible for Leela >> (or other Go programs) to play perfectly without some minor >> adjustments to handle the weird corner cases. > > It probably makes sense to force a bit more exploration for this kind of > analysis. I can make a version like that if you want. Very nice proposal. From my point of view it would be nice to generalize Leela that it also can play on rectangular boards: 4x6, 6x5, 6x7, 5x7, ... The sgf specification for these boards are SZ[4:6], SZ[6:5], ... Ingo. PS: Congratulations for the good performance of Leela in 9x9 at Beijing, and good luck for the 19x19-competition. -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] On ranks 2 and 3 of 9x9 in Beijing
Gian-Carlo Pascutto wrote: > the tiebreak is not yet finished! Place 2 and 3 are still undecided. Hmm. In the tournament rules http://www.grappa.univ-lille3.fr/icga/event_info.php?id=20#Rules it reads > Tie-breaking: > (a) if precisely two participants are tied for a medal place, precisely two > play-off games under standard conditions are to be played. > Should the score be equal, the tie-ranking rule in part (b) decides the > winner; > > (b) whenever two or more teams have an equal number of points, a tie-ranking > order is defined as follows. The dominant ranking is by the sum of the > opponents’ > scores. If there is still a tie, the sum of the respective programs’ > cumulative > scores after each round (i.e., score after round 1 + score after round 2 + …. > + > score after last-round) will be used; Two games have been played between MoGo and Leela, with outcome 1-1. So (b) should decide, and this means Mogo (SOS=196) is ahead of Leela (SOS=164). What is done differently in Beijing? Greetings, Ingo. -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] 7.5-komi for 9x9 in Beijing
Gian-Carlo Pascutto wrote: > I'd have some preference for playing the decisive game > with komi = 6.5, but apparently thats not possible on KGS. But that should not be a problem, as long as the operators do not believe in the final verdict of KGS. > I think with komi = 7.5 white > is scoring very high (too high?) in the top games. I made a count for the 9x9-competition in Beijing (komi=7.5: Looking only at games among the top 5 rankers there are 20 games so far (including two tiebreak-games) with 15 wins for White and 5 Wins for Black. Looking at all games among the top 7 rankers there are 40 games (including two tiebrak-games) with 27 : 13 for White. Taking all 164 games of the tournament (including two tiebrak games) White is ahead by 93:71. It is probably not an accident that the quota decrease with decreasing playing strength: 15/20 > 27/40 > 93/164 or translated to decimals 0.75 > 0.675 > 0.567. However, when B+7 is the correct value on 9x9, komi=6.5 might lead into similar problems. Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Congratulations again to David Fotland !
Hello, Many Faces of Go has won also the 19x19 competition in the "13th International Computer Games Championships", with a 100 % score. The silver medal goes to MoGo (only loss against MFoG), Leela achieves Bronze (only two losses, against MFoG and MoGo). Details, including sgf-files, under http://www.grappa.univ-lille3.fr/icga/tournament.php?id=181 Congratulations again to David! Ingo. -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Tiebreak 9x9 in Beijing ?!
Olivier Teytaud wrote here: > ... and good luck for both MoGo and Leela for > the silver medal in 9x9 :-) I saw on KGS that one tiebreak game between MoGo and Leela was played today: starting time 13:00 GMT, MoGo with Black starting with 4,3 move, and Leela winning. When I understood GCP correctly, another game with colors exchanged has to be played ... and one more Amageddon game in case of a 1-1 score after the two. Does someone here know what happened? Thx in advance, Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Tiebreak 9x9 in Beijing ?!
Hello Gian-Carlo, Gian-Carlo Pascutto wrote: >Hideki Kato wrote: >> I don't know the detail but the cluster (or the connection) had some >> trouble and the play-off will be resumed this morning (at Beijing >> time; +0800). > > Leela has been online and ready the whole night but I still see no sign > of the Mogo team. Since it is now 9:25 Beijing time, way over an hour > since the playoff was scheduled to begin, and I have heard nothing at > all from the Mogo people, I will take Leela offline again and from my > side that is pretty much the end of it. It would be a pity when the event ended in such a situation. Isn't there a referee to decide on the case? Nevertheless, congratulations to the nice performance of Leela in both the 9x9- and the 19x19 championships. Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Tiebreak 9x9 in Beijing complete
Hello all, finally the 4th tiebreak game in 9x9 Go has been played in Beijing: Leela won with Black against MoGo, thus getting the silver medal by a 3:1 victory in the 4-game tiebreak. This last game is interesting because it was a win for Black. However, so far it is not completely clear which game it is: * Is it game 4 from http://www.grappa.univ-lille3.fr/icga/round.php?tournament=180&round=10 or is it (more likely in my eyes) from http://www.gokgs.com/gameArchives.jsp?user=leelabot the game that started at 14:08 GMT (on October 5) ? (LeelaBot = Leela, GCP entered the moves for MoGo) * Congretulations to all medal winners (Many Faces of Go 2x gold, MoGo and Leela each 1xsilver and 1xbronze)! Questions concerning 2009: * Will there be a Computer Olympiad/Int.Comp.GamesChamp. again? * Are there already concrete plans, where and when? * Will there be played 9x9-Go again or for instance 11x11? Thanks to all who made the last week so entertaining! Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] komi for 9x9
Hello all, in other games (on other servers) games start with a komi-bidding procedure. Only when both sides propose the same value for komi, colors are given by chance. In my eyes, for Go it would be useful also to allow integral komi (7.0 for 9x9, for instance). Ingo. -- GMX Kostenlose Spiele: Einfach online spielen und Spaß haben mit Pastry Passion! http://games.entertainment.gmx.net/de/entertainment/games/free/puzzle/6169196 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] komi for 9x9
David Fotland wrote: > Integer komi has a problem for many MCTS implementations, > since a playout only returns win or loss. For instance, also in the UTC-implementation of Cameron Browne's Yavalath, as given under "yavalath" on the site http://www.cameronius.com/ Browne had in mind only larger boards (where draws are very unlikely), but Yavalath can also be played on small boards. > This would require playouts to also return drawn. > My playouts work this way. I know Erik's can return draw. > I don’t know about mogo or leela. Leela in the pre-Beijing version was not prepared. GCP's answer on my question was: You can enter integral komi, but I do not know how Leela will react. I would like to see "all" Go programs to be able to live with possible draws (or even with any score spectrum). Ingo. -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Go with modified scores
Some of you may want to stone me for this heresy, but read carfully before. When you have MCST-/UCT for Go that can work for real-valued scores (or at least a version that can work for three-valued scores: winm, draw, loss), you may look at Go with different scoring systems. Example: A win by 0.5+k points is worth 100+k. A loss by 0.5+k points gives score -100-k. Let's call b=100 the base score. Question: How does the playing style change with b? (Of course, for very large B you should have almost the normal playing style.) Ingo. PS: Such evaluations may also help to reduce the problem of MC's laziness. -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Go with modified scores
Some replies on my original mail indicate that I did not make clear the motivation of my proposal. You have to distinguish several scenarii when maximizing the playing strength/value of your Go program: (a) auto-play (or play between different versions of your prog) (a') play against other computer programs (b) play against humans (c) program as tool for human analysis of Go positions or of whole games (d) acceptance by humans I agree, that for (a) and (a') other scoring rules will likely not give better results. But concerning (b) I am not so sure. Here, data from KGS may shed some light. Did some programmers try to test ohter scoring systems there? Concerning (c), I can tell from my experiences in the computer chess world. In the late 1990's strong Shredder and Junior were dominating the computer chess tournaments. However, as analysis tools they were not the best. Concerning (d), in the German go scene several people who tested Leela, were not at all convinced by the playing style, even not on 9x9. And this was not based on performance, but mainly on the "laziness". My claim was (and still is): Other scoring schemes (inside the programs) may prove beneficial with respect to (b), (c), (d). Ingo. *** Denis Fidaali asked: > how can i do so that my response to this mailing-list will be correctly > indented ? > (for example i would have liked to set this one as a response to my previous > post). Not an answer, but I am suffering the same problem -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Questions of Beijing photos
Hello Hiroshi, thanks for the many nice photos from Beijing. I have a few questions (indeed, I have many, but will ask only a few of them): * On which picture(s) can I find you? * Is Feng Hsiung Hsu (guest of honor, when I understand correctly) on some of your pictures? * Is the programmer of HandTalk on some of the pictures? * What was the level of play on (11x11)-Hex? * Who is sitting left of Johan de Koning on 2444? * Who is the beautiful woman on 2501? * The two guys on 2552, I know only one of them. Who is the other? * What is shown on 2559? * What are the lights on 2587? * Who is the left person on 2639 (an uncle of Johan de Koning?) * Is 2711 from the final game of the 9x9-playoff? Thanks again for the pictures, and thanks in advance for your explanations. Best regards, Ingo -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Go/Games with modified scores
During the last few days I have been meditating a lot about the questiion whether taking into account the margin of win into MCTS (UCT) may help or hurt. I do not have a go program by my own, so for the moment I have to believe what programmers are saying, namely that "MCTS with margin of win as a criterion" gives worse performance in play against other computer programs. So, a negative answer... but there may be other points of view from which this phenomenon is not bad but helpful. To give an example: I am active in the scene of "computer-aided game inventing". When a human designer of board games has invented a new game he typically needs many rounds of human test players who find strengths and weaknesses of the game. As it is really a hard job to find enough test players our idea is to let the early phases of test play be done by computer in autoplay mode. (The list of basic criteria for good games include appropriate game length, balance of chances, low drawing rates.) When you now think of a game where the overall goal is to win and to avoid losing, but where also a margin of win/loss can be counted at the end (Go belongs to this class) you may pit two versions of your MCTS algorithm against each other: * MCTS(WinLoss), playing only for a win, independently of the margin * MCTS (margin), playing for win/loss, but also considering the margin. with the philosophy: The game is "interesting", when MCTS(WinLoss) performs better than MCTS(margin), and "boring" when it is the other way. Ingo. PS: Here is an early report on computer-aided game inventing. http://www.minet.uni-jena.de/preprints/althoefer_03/CAGI.pdf -- GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen! Jetzt dabei sein: http://www.shortview.de/[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: computer-go textbook ?
Claus Reinke wrote: > I'm not currently in teaching, but I'd be interested to > hear about uses of computer-go in this context, ie, > not so much as a specialist course but as a focus for > programming and group projects. Let me tell you the experiences in my group at Jena University. I am in a joint faculty for mathematics and computer science. http://www.minet.uni-jena.de/ My chair is for applied mathematics (discrete optimization), but my research interests are rather broad, ranging from number theory and discrete stochastics over experimental mathematics in general to theoretical computer science, artificial intelligence and bio-informatics. I returned to computer-go only after a several-year-long break. In the meantime we were busy in the context of "automatic and computer-aided inventing of board games". The task of the computer typically was/is to play a newly invented game (or game variant) immediately with reasonable strength. For this purpose, "basic Monte-Carlo" (evaluating direct successors of the root position by random games) is a wonderful tool. And it was/is a very easy entry step for students or groups of students: They only have to program the rules of a game, and the computer starts to play (graphics was never a topic; most programs have output in tet mode). I NEVER asked students to use more sophisticated techniques like UCT, "instance-sharing", or tree growth. Interessentingly, several students themselves went such roads: They were not satisfied with the playing strength of basic MC and started to read about UCT, or even to develop their own algorithms: - Joerg Sameith and Stefan Schwarz in the context of "EinStein wurfelt nicht" (result was the vary nice program "Hanfried") - Joerg Guenther, concerning Antonow's connection game "ConHex" (result was the rather strong program GuentHex 2.0; GuentHex 1 was almost pure basic MC) - Jakob Erdmann (Ph.D. student, with a 3-Hirn-grant) has become sort of an expert on UCT for multi-valued games. He also made a program for my game "Seasons of the Sun". - Lisa Schreiber found nice implementations for Reiner Knizia's "Heckmeck am Bratwurmeck (Zoch-Verlag) and for "Treffer" (by Michail Antonow, unpublished yet). - Frank Schneider (not a student in Jena, but interested in our work from outside; chess programmer) invented his own game to have a workbench for his UCT experiments. Probably, earlier or later also some student(s) will catch fire on computer-go. Other students simply made their Monte-Carlo task, got their certificates and were happy. Here are some links: * Long Night of Science, November 2007 http://www.althofer.de/lange-nacht-jena.html * Workshop "Analysis, Design, and Programming of Games; June 2008" http://www.althofer.de/workshop-2008.html * Game "Seasons of the Sun" http://www.althofer.de/seasons-sample.html Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Monte-Carlo and Japanese rules
Hello all, two questions. (i) Do there exist strong 9x9-go programs on Monte-Carlo base for Japanese rules? (ii) Having available only programs for Chinese rules, but playing in a tournament with Japanese rules, which special tricks and settings should be used to maximise winning chances? (This is meant especially in the light of MC's tendency to win games by 0.5 points according to the rules implemented.) Ingo. -- "Feel free" - 10 GB Mailbox, 100 FreeSMS/Monat ... Jetzt GMX TopMail testen: http://www.gmx.net/de/go/topmail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Monte-Carlo and Japanese rules
Hi Remi,
thx for your hints.
I am playing 9x9 interactively, with taking advice from
Leela and ManyFaces. The basic procedure is to run
both in a giving position - and making the final choice
amongst their move proposals with my human brain.
So I have some better potential for corrections then in
full-automatic mode.
I also let run the engines for different values of komi
("komi fan") to get an impression how narrow the position
is.
Ingo.
Remi Coulom wrote:
> The standard trick is to pass as soon as your opponent passes.
> For this to work, you need to take a one-point security margin
> with the komi. You should also score seki the Japanese way.
>
> That is how Crazy Stone played the UEC Cup and the matches
> at FIT2008. In fact, for the UEC cup, Crazy Stone did not understand
> seki, so I took a bigger security margin with the komi. But a bigger
> security margin is not good for 9x9.
--
GMX Download-Spiele: Preizsturz! Alle Puzzle-Spiele Deluxe über 60% billiger.
http://games.entertainment.gmx.net/de/entertainment/games/download/puzzle/index.html
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Opportunity to promote ...
Dear Bob Hearn, it is not what you have been looking for, but nevertheless I want to ask you if the title of your talk "Games Computers Can't Play" is still up-to-date. I would accept something like "Games Computers Could not play well before 2003", but Monte Carlo has changed our world. Ingo Althofer. -- Sensationsangebot nur bis 30.11: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Opportunity to promote ...
Dear Bob, thanks for your explanations. Now I see clearer. > First, the title is deliberately provocative. Accepted. > Also, though, the talk is not just about go: some of it is about > formally undecidable games, that computers provably can't play > well (and of course, that humans can't either!). I see. How much sense does it make to think about Monte-Carlo approaches to certain games that are undecidable? (Probably it does make a lot of sense in certain cases.) > Surprisingly, some of these games can be played with > finite physical resources (unlike, say, an infinite Turing-machine > tape). One real game that is a good candidate for being undecidable > (though it hasn't been proven) is Rengo Kriegspiel: team blindfold go. Here, indeed Monte-Carlo might "work" - for instance against opponents with limited capabilities. * When I read your original mail the following three main steps of game tree search came to my mind: (i) Zermelo's basic tree approach (with minimaxing) in 1911 or 1912. (ii) Shannon's paper from 1950 with the proposal to prune the (large) tree and to evaluate the artificial leaves by heuristic values. (iii) The Monte Carlo approach: getting an evaluation for a position not by applying a heuristic but by playing many "random" games from this position till the very end. These three mile-stones in mind, there are not many games left, where computers are really weak. Ingo. PS: When you meet Elwyn Berlekamp in the symposium you may tell him that looking at the game Clobber in the 2002-Dagstuhl workshop (where he was one of the organizers and I one of the participants) was my starting point for becoming a "game designer with computer help" and thus indrectly the starting point for me to investigate Monte Carlo, first in games with random elements. -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] One-sided 2-inch Rules
Hello, one of the basic problems of go newbies is their tendency to place the next stone near to the latest stone of the opponent. Sometimes this is called the "2-inch heuristic of beginners". What do you think about a formalized variant of Go with one-sided distance-k rule? > Let k be some natural number. > The normal rules of Go hold, except for one thing: > When possible, White has to place his next stone > within distance k (in city-block metric) of the latest > stone of Black. If there is no feasible move of this type > the stone has to be placed within the smallest > possible city-block distance of the latest stone of > Black. White may pass at any time. Example: > On 19x19 board k=36 would mean no restriction at all.) * What should be fair values of komi(k) or fair numbers of handicap stones? * Main question: How strong would MCTS-based programs be in this variant(s)? * Would computers be stronger than humans for certain values of k? Ingo. -- Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten Browser-Versionen downloaden: http://www.gmx.net/de/go/browser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] On Don Dailey's first chess program
Dear Don, sorry to step in here, but I can't believe what you write. So I would like to know some facts. > My first chess program only sold a few copies in Europe. What was the name of your program? In which year was it published? For what platform had it been? > But I came to find out that thousands of people had a copy of it. To which countries of Europe were your contacts? (Europe is NOT one big uniform block...) > I met many people in Europe who said they had a copy and > many of their friends did. I was a very active chess software use for many years. I never had heard that you had written some commercial program. (I only know about Star Socrates - where you were involved - and which was runner-up in the 1995 World Championships.) > Someone pointed me a site where you could download it for free. Which was that site? Ingo, European without pirated software. -- Sensationsangebot nur bis 30.11: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] On Don Dailey's first chess program
Hello Don, thx for all your answers. I think, I found a website where old programs (from the 19_80s and early 90's) are listed: http://www.septober.de/chess/index.htm# There are also screenshots of RexChess http://www.septober.de/chess/pics/9102.gif and "Colossus X" (by Martin Bryant) http://www.septober.de/chess/pics/9001.gif I think, in those days Martin's programs were the best-sold chess software in Europe. I heard, he even bought a Ferrari from the money he earned with Colossus. ** Again my wish: When you speak or write about software piracy in Europe (which really exists): Please, do not throw all regions or countries or people in one big pot. I have many friends in Europe who are honestly paying for their "games software". And some of them (including me) are sensible when someone writes (openly or between the lines) that software piracy is a standard in Europe. Best regards, Ingo -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] 1+23-h mode
Hello all, during my latest reading in the list I found, amongst others, the following mails: Don Dailey wrote: > To summarize, I have found over the years that > just plain CPU/MEMORY performance is the > primary barrier not just to program strength, > but development time. > > You must measure development time 2 ways. One > is how many man hours are spent, and the other > how much "calendar" time is spent... > But you also would have to be willing to wait much > longer between tests and be willing to work > piece-meal on a project while mostly waiting for > tests... David Fotland wrote: > ... I tried more than 400 variations on > the basic UCT algorithm or playout strategy > during 6 months of intensive development... > I couldn't have done this many experiments > (often several per day), without a very fast engine, > because I used 1000 game contests to see if there > was improvement, to get statistically significant > results... Mark Boon wrote: > ... I'm going to run a test overnight, > but so far it looks good. It should have > collected a few thousand samples by tomorrow. This remembered me of my activity some years ago in computer-aided correspondence chess. There I proposed to do "work" in a (1+23)-hours mode: Sitting in front of the computer for about 1 hour per day - and letting the machine investigate interesting lines and (sub)positions in the next 23 hours. Large parts of the "1h" blocks consist in looking at the results of the previous 23 computer hours, interpreting them, and in planning the next 23h block. For development of (game playing) software, especially on MC base, also such 1+23 rhythms seem to be convenient... By the way, Chrilly Donninger (the leading head in the Hydra project in computer chess) once told me that his start to investigate computer poker came as follows: he was doing test runs with Hydra - and needed something constructive for the phases when waiting for new Hydra test results. Ingo. -- Sensationsangebot nur bis 30.11: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] WMSG - Scoring
Hello, in "Deutsche Go-Zeitung", Issue 5/2008, I found a report by Robert Jasiek, on the World Mind Sports Olympiad, which took place in Beijing early in October (directly after the Computer Olympiad). Most interesting I found a paragraph describing the Go rules used in that event: area scoring, super-ko, komi 6.5 and one special rule: When White is the first player to pass than komi is changed from 6.5 to 7.5 . What would such a rule mean for computer-go, especially for programs on MC base? Have a nice weekend, Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: WMSG - Scoring
Michael Williams wrote: > Seems like MC and MCTS programs would cope just fine > after that one line of code is added. Ok, that is a technical answer. But ... ... what does the rule change mean for strengths of programs - especially in play against (strong) humans? Would this rule help the computers or the humans? Ingo. >> ... WMSG-Scoring: >> area scoring, super-ko, komi 6.5 and one special rule: >> When White is the first player to pass than komi is changed >> from 6.5 to 7.5 . -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: WMSG - Scoring
Michael Goetze wrote:
> I doubt that this rule has a significant effect on playing strength,
> either of computers or humans. After all, the average effect is about
> half a point per game, which you probably won't notice below the
> level of amateur 6d or 7d.
You are right, and I did not state clearly what I had in mind:
play on small boards (like 9x9 or even 7x7). Especially,
there Seki situations are (much) more frequent than on
larger boards.
Concerning the next Computer Olympiad and having in mind
the discussion on the last one ("how fair is 7.5 komi for
9x9 computer games?") the WMSG scoring should be worth
to be discussed for 9x9.
Ingo Althofer.
--
Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen:
http://www.gmx.net/de/go/multimessenger
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Rumors on next Computer-Olympiad
Hello to all programmers, in computer chess circles there is an unconfirmed message that the next World Computer Chess Championships may take place May 11-18, 2009, somewhere in Europe. In another message by ICGA president David Levy it is written "alongside with the championship the Computer Olympiad will take place". So, gentlemen, start to prepare you engines! Ingo. PS: Are there any gentlewomen programmers around in computer go ? -- Sensationsangebot verlängert: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K1308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Nullmoves in MCTS and UCT?
Question: Have nullmove-concepts been tried or analysed in MCTS or UCT-settings? Background of the question: Using alpha-beta tree search, the (asymptotic) percentage of nullmove cutoffs may help as an indicator for the "naturality" or "interestingness" of a (newly invented) game. Unfortunately, it is nontrivial to design "traditional" evaluation functions for newly invented games... Ingo. -- Sensationsangebot verlängert: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K1308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: 3-4-5 rule
Hello Don, > The reference bot of course does not build a tree, what I'm actually > looking for is a way to produce a medium strength but really simple bot > that does not build a tree and just has a lot of playout magic. You should have stressed very clearly much earlier in the thread that no tree is built. By the way, there is some in-between parameter between distance 2 and distance 3: * all distance 2-(Manhattan)-paths are allowed: nn, ee, ss, ww, ne, en, es, se, ... * among the distance 3-paths only those are allowed which do not have all three steps in the same direction: this rule would forbid only the four paths , , sss, www would be forbidden. (Here n, e, s, w indicate the four basic directions "north", ..., "west".) Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: 3-4-5 rule
I am sorry for my typos (with and ). Of course I meant: * among the distance 3-paths only those are allowed which do not have all three steps in the same direction: this rule would forbid only the four paths nnn, eee, sss, www. Ingo -- Sensationsangebot verlängert: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K1308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Happy new year ...
Many lifes, always enough memory space and lots of good Monte Carlo ideas ... that are some of my compgo-related wishes to all of you. Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Black/White winning rates with random playout?
ibd asked: > what's an usual winning rate for black/white from an empty > 9x9 board, black playing first, 7.5 komi? I play 50k games > when starting my program, and I usually get around 60% > winning rate for white. This seems rather high to me, and I > suspect a bug somewhere. Do you have any data I could > compare with? I use random playouts, only moves that fill > eyes are pruned. 60% looks rather normal to me. You may repeat the 50k runs with other values of komi to see, if you get an expected slope. Feel free to report the results here, when you do it. Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: GCP on ICGA Events 2009 in Pamplona
Hello Gian-Carlo, Gian-Carlo Pascutto wrote: > The computer chess forums are ablaze with protests, > because ... > [But] the decision seems to have been cast in stone, > with no amount of protest still being able to reverse it. I think, this is indeed the case. But at least YOU would have reason not only to complain all time. You are one of the few strong chess programmers who are in the comfortable situation to have another game at stake. What prevents you from freezing in your chess activities for the next few months and hobbying full (free) time on computer go. There are enough goals left: at least, in Beijing there were still 1.5 teams ahead of Leela ;-) (MFoG and MoGo on 19x19, MFoG on 9x9). And for computer go, there are no hardware limitations given for Pamplona. Best regards, Ingo PS: According to the leading German chess magazine (SchachMagazin 64, Issue January 2009) Jaap van den Herik claimed in a talk during the (traditional) chess olympiad in Dresden (in November 2008) that chess will likely be solved around 2035... -- Sensationsangebot verlängert: GMX FreeDSL - Telefonanschluss + DSL für nur 16,37 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K1308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Is computer Havannah welcome here?
Hello to all, in 1979, Christian Freeling (NL) published "Havannah", a very nice abstract board game for two players. It is a connection game with some territory components. Havannah had (commercially) good years: it was in the shortlist for the (German) "Game of the Year" in 1981 and 1982 and was produced by Ravensburger in several editions. Then it became a bit silent around Havannah. In 2002 Christian proposed a prize: for 10 years (so, until 2012) no computer program would win a single game (on board with side length 10) against him out of ten tries. If a program would achieve such a win, the programmer would get 1,000 Euro. Recently, Havannah has become available for online play on the server www.littlegolem.net . The game is well accepted there, with lots of tournaments, starting almost daily. In the fora of LittleGolem also discussion on computer Havannah has started, and there is now already one program (by Johannes Waldmann; based on UCT) which plays the game online on small boards. There would be even more discussion on computer Havannah in the LG fora but ... the master of LittleGolem is in big fear of spambots. So, you can write into his fora only, when you have played so and so MANY games before on LG. This excludes several competent persons/programmers from participation in the discussions. On the other hand, the computer Havannah scence is still too small to fill with life an own forum. Now, question: Would the computer-go mailinglist accept or welcome when the computer Havannah people uses this mailing list for the next few months? Some links: * The rules of Havannah http://www.littlegolem.net/jsp/blog/blog.jsp?blogid=7 * The fora of Little Golem (for those who want to read what exists) http://www.littlegolem.net/jsp/forum/forum2.jsp?forum=1 http://www.littlegolem.net/jsp/forum/forum2.jsp?forum=50 Registration on Little Golem is for free and simple. The fora can be read without being registered. * The website with the program of Johannes Waldmann http://dfa.imn.htwk-leipzig.de/havannah/ Ingo. -- Jetzt 1 Monat kostenlos! GMX FreeDSL - Telefonanschluss + DSL für nur 17,95 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Taiwan Open 2009
Eric Dunham asked for information on that event. The website is http://go.nutn.edu.tw/2009/English/news_eng.php You should click on the key words in the left coloum. By the way: Good luck for the MoGo team! Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] 19x19: MoGo (handicap 7) vs Jun-Xun Zhou (9p)
Hello, today MoGo (with 7 handicap stones) played two games against the top Taiwan pro Jun-Xun Zhou (9p). MoGo won the first game, Jun-Xun Zhou the second one. Time must have been with about 40 seconds per move in the average. sgf are under http://www.gokgs.com/gameArchives.jsp?user=mogo It are the two "handicap 7" games from February 10, 2009. Cheers, Ingo. -- Jetzt 1 Monat kostenlos! GMX FreeDSL - Telefonanschluss + DSL für nur 17,95 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Human Learning against MoGo
Hello, During the last week (February 10 - 13, 2009) there were several exhibition games between program MoGo and some professional go players from Taiwan (Jun-Xun Zhou 9p; Li-Chen Chien (12 years old) 1p; Shih Chin 2p). First of all congratulations tothe MoGo team for winning one game at handicap 7 (against the 9p) and one more game at handicap 6 (against the 1p)! Each day, two pro games on 19x19 board were played. The results for MoGo were (given here exactly in the order in which the games were played). Feb 10: Win Loss (against 9p; handicap 7) Feb 11: Win Loss (against 1p; handicap 6) Feb 12: Loss Loss (against 2p; handicap 7) Feb 13: Loss Loss (against 9p; handicap 7) >From these data I claim evidence that humans are learning quickly against MoGo: The human with 4 games (Mr. 9p) lost only his very first game. The 1p boy lost only his first game. And the human who had to play on day three (so, just knowing what had happened to the two colleagues on the first two days) won "all" of his games. For those who like evidence by probabilities: * Given a 3:1 score, a random order (which should be implied, if no learning at all were there) of the results means prob 1/4 that the first game was the only loss. * Given a 1:1 score, a random order ... means prob 1/2 that the first game ... * Given three different players who play against MoGo one after the other, "no learning" would mean that the only 100%-scorer would sit in position 3 with prob 1/3. Multiplication of the three terms gives 1/4 * 1/2 * 1/3 = 1/24 < 5 %. When you follow this line of thought, the results of Tainan show that the computer go community will also now (and likely in future, too) have to fight with the problem/phaenomenon of quick human learning (as has been the case already for several decades). Ingo. PS: sgf of the Tainan games can be found in the KGS archive for player/account "mogo". http://www.gokgs.com/gameArchives.jsp?user=mogo -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] April 2009 KGS tournament
So far, only Yamato-san has registered for Nick Wedd's next KGS tournament (to be held on April 5). It would be a shame when Zen19 would remain the only (strong) participant. Come on, gentlemen: register your engines! Other question: Might it be possible to find a volunteer for operating Zen19 in Pamplona? Ingo Althofer. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dsl.gmx.de/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: My ongoing CGoTournament - actual results
Stefan Mertin wrote: > More than ten thousand games 19x19 Go are played > in my computer Go Tournament - time to again publish > some results! > > Please have a look at my newly created site for results > and more informations: www.igosoft.com Hello Stefan, nice to see you posting here, and thanks for all the test-work you are doing. > ... > But first I finally want to let play the 2008 Computer Go Champion, > MANYFACES v12 by David Fotland. .. > and the even much stronger MC-TS engine, level "2 Kyu".. My experience: MC-based programs are playing Go on a much higher level now (since October 2008) than "traditional" programs. So I propose that you directly start "ManyFaces (MC)" versus the top contenders in your list, delaying all other tests. And when - what I expect to happen - ManyFaces-MC bombs the traditional programs below the surface, you should think about your overall strategy. Perhaps it even makes sense to stop the matches between the dinosaurs. However, likely the dominance of MC-based programs will be more clear when running (both sides) on a faster machine, for instance a computer with 2 or better with 4 cores, and when thinking times are not too small. And from my private wishlist: I would like to see (massive) data from MC programs for play on 13x13 boards. > Recently I added a newer version of Peter Woitke´s program > SUZIE for 19x19. This engine works with pure alpha-beta > tree-search and a position evaluation function > like most of the classical chess engines do and I think it > is most remarkable how much SUZIE > improved in last years without use of MonteCarlo TreeSearch > techniques. SUZIE 0.40 finished third and seems to be nearly > as strong as the older version of Go++ 5.0. As far as I know, Chrilly Donninger has been involved in the development of SUZIE. How is SUZIE 0.40 dated? Do you have match for SUZIE on 9x9 board? (The development of SUZIE had been started on 9x9.) Best regards, Ingo -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: April KGS bot tournament: results
Hello Nick, many thanks for your great engagement in these tournaments. Nick Wedd wrote: > I am thinking that the next KGS bot tournament > will be blitz - fast time limits, and probably > small boards. Does anyone have any views? The next Computer Olympiad is in Pamplona, May 11-18. When you want to run the next tounament early in May (so before Pamplona) it might be an idea to play on 13x13 boards. On that size, programmers might have less fear to lose "secrets" so short before the Olympiad - and you might get more participants. Just my 2 Cents, Ingo. -- Pt! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] (no subject)
Hello, during the last weekend I have tried (for the first time) to use commercial go programs to analyse some games played between human players (on KGS). The idea was to use the bots for blunderchecks. So, I was looking for positions where the evaluation (in win percentages) jumped or dropped between consecutive moves. *** Concrete Example *** Amongst others I looked at a game between bwilcox[3d] and harleqin [2d], played on April 20, 2009, starting time 5:31 [CEST]. You can download the sgf for instance from the KGS archive at http://www.gokgs.com/gameArchives.jsp?user=Harleqin (By the way, bwilcox might be Bruce Wilcox, one of the veterans in computer go.) Both programs in my analysis "claimed" that 77.f13 was a big error: * In Leela the score dropped from 50.5 to 41.4 . * In ManyFaces the score dropped from 50.0 to 45.8 . Instead of 77.f13, both programs proposed 77.d14, with the possible continuation 78.e14 79.d13 . I discussed these "findings" with Harleqin, and he agreed that 77.f13 was a move that brought him into problems. *** General Wish *** Unfortunately current, (commercial) go programs (including Leela and ManyFaces) do not have a user interface that allows for COMFORTABLE analysis of games. The user has to click lots of buttons when jumping back and forth in some sgf. My wish is to have something similar to that, what has become standard in computer chess programs in the mid 1990's: * on the left half of the screen the diagram of the board * on the right half a list (not only sgf, but also with move numbers included) * the program is computing in analysis (infinite) mode * when the user clicks (by mouse) on any move in the move list then the program jumps immediately to this position and starts analysing (of course this position is shown in the diagram) * of course the screen should all the time show (in fat font) what the value of komi in the game under investigation is. * Another big wish: the programs should allow for some k-best mode where not only the best but the k best moves are computed (k being some integer specified by the user). Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Analysis mode for human use
First of all sorry for forgetting the title in the previous posting. As David Fotland pointed out to me, sgf is of course not suited for humans to read. See the following example, for the first ten moves of a game. copied from sgf B[pp]; W[qd]; B[cp]; W[fq]; B[dq]; W[jp]; B[mq]; W[kq]; B[hp]; W[hq]; ** translated to Ishi-go B 1 Q4 W 2 R16 B 3 C4 W 4 F3 B 5 D3 W 6 K4 B 7 N3 W 8 L3 B 9 H4 W 10 H3 *** modified Ishi-go 1. q4 2. r16 3. c4 4. f3 5. d3 6. k4 7. n3 8. l3 9. h4 10. h3 * Ishi-go is much better to read, but - at least for people from the west - my "modified Ishi-go-format" should be even better. (The repetitive B W are a bit annoying in Ishi-go, and small letters are better to read than capital ones.) To give you an impression about the size of the market in Germany. Let's assume that in principle a go program with nice analysis mode (and current strength) might be interesting for human players in the range 2k - 3d. The German rating list shows the following ranks: Rank-No rating 80 2301 143 2200 216 2100 342 2000 479 1900 611 1800 1800-1900 is about 2k 1900-2000 1k 2000-2100 1d 2100-2200 2d 2200-2300 3d These German ratings do not fit exactly with KGS ratings, but there is rather strong correlation. So, there should be a market of several hundred potential customers, alone in Germany, for a program with nice analysis mode. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Reply to Lukasz and Don + Roadmap 2020
Topic is the question (how) to use (current) go programs for evaluating human go games. Lukasz wrote: > I like the idea very much. > But the coding effort is mostly in the GUI so > it depends whether gogui's (or other GUIS's) > author will like the idea. > > It has great commercial/popularity potential. > But it is not so important for research. Accepting PROGRESS as the meta goal, we should not artificially distinguish between science/research and commerce/popularitiy. Best progress is achieved by a "free combination" of all forces. At least, this turned out to be true in computer chess. Don wrote: > Yes, this is a powerful feature that all chess interfaces have. > > There is one issue with GTP that will have to be kludged > around - there is no way to stop an engine from thinking > that is provided naturally by gtp... > > I believe the protocol should have a special mode for > this, the ability to accept commands even while busy - > the ability to communicate asynchronously which is > required to build the more sophisticated interfaces. > ... > But it's a necessary step that will have to be taken > sooner or later and I would personally prefer it's hea > > My email got cut off near the end.My final thought > was that it would be preferable to stick with GTP, just > a revised asynchronous version. Seems very reasonable to me. *** I think the computer go community has good chances to achieve the following three goals before 2020. (A) Programs with stable amateur 6-dan level (that would mean ratings like 2,550 on EGF scale). (B) Programs are accepted as analysis tools by professional players. (C) Human playing style (also on pro level, also for games between humans) changes due to experiences made with programs. Explanations: * In chess, (B) with commercial programs happened on Chess World Championship level for the first time in 1990, in the match between Kasparov and Karpov. In those days the best commercial programs had ratings around 2350 (Mephisto Lyon on 68040 Motorola processor). * In correspondence chess, (B) happened on top level already around 1988 (for instance by a member of the East Germany Olympic team, which won the Gold medal in that competition). This help was revealed only much later. * It is likely that (B) will happen for some time only in secret mode before pro players "confess" from whom they get help. * (A) may happen before or after (B). * In chess, computers have shown that defense has in general much more resources than humans believed. * (C) does not necessarily mean that at the same time computers will be stronger than humans. Example: During the Chess World Championship in year 2000, when Kasparov lost to Kramnik, Kramnik used an opening system called "Berlin wall". By intense computer help he had convinced himself that this very passive opening is sufficient to keep a draw balance. Commercial chess programs turned out to be better than the best humans only around 2006. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Reply to Lukasz and Don + Roadmap 2020
Terry Mcintyre wrote: > Is it reasonable to expect pro players to use 6-dan > programs as a tool for analysis? The pro players are > markedly better - at a rough guess, a pro player > could give a 6 dan amateur human or program a 3 stone > handicap. Yes, I believe so. Some tests indicate that 2-kyu programs (Leela or ManyFaces on moderte hardware) seem to be useful to detect "certain" blunders in the games of 2-dans. 2k and 2d are also three stones apart. The other "evidence" is that in chess a rating difference of 300 or 400 points was no problem for computers to be good helpers in analysis of grandmasters. Maybe, when Zen should become commercially available within a few months and with a strength of 1-dan (or even 2-dan), it might be tested to find blunders in games of 4-dan or 5-dan amateur players. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Analysis mode for human use
Don Dailey wrote: > But odd move numbers always mean black to move. That > becomes second nature very quickly and I personally > prefer the less verbose syntax. Darren Cook wrote: > I find the B/W very useful: when playing out a long list of moves it is > very easy to lose track where I am. Most moves are equally likely for > both sides. You may have a look at the notation on the gameserver http://www.littlegolem.net . There no B and W are shown, but the moves themselves are in black and white, on a grey background. Example http://www.littlegolem.net/jsp/game/game.jsp?gid=933716 Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Fuego technical report
Hi Martin, thanks for the information and the report. In the abstract you write "...Fuego includes a Go engine with a playing strength that is competitive with the top programs in 9x9 Go, ..." I want to support this claim. Over the weekend I had the fun to watch some free 9x9 games of Fuego on KGS, and I was really impressed by its playing strength. Good luck for Pamplona! Ingo . -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Playing schedule in Pamplona?
Hello, can someone from the guys in Pamplona please let us know on which days and at which hours the games of the Olympiad are played? Which of those games can be followed on KGS? Thanks in advance, Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Playing schedule in Pamplona?
> See > http://www.grappa.univ-lille3.fr/icga/event_info.php?id=35 > > Hideki Dear Hideki, thank you for the good information. Good luck for you and your bot! Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Cross-Question on Pamplona
Hello, this is not Go, but I feel that some people here should know the answer: What are the results of the Connect6 competition in the Computer Olympiad. Thx in advance, Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Cross-Question on Pamplona
Dear Remi, >> What are the results of the Connect6 competition >> in the Computer Olympiad... > > I forwarded the question to organizers. I'll update the web > site as soon as I have the results. Speed chess is also missing. Thanks for the results, they are online now. > Also: congratulations to Yamato for the impressive victory > of Zen, with the slowest hardware of the whole tournament. Congratulations to Zen author also from me! A question (for the general audience, not especially to Remi): As far as I know, Yamato is only a nickname and not the true name of Zen's author. If this is the case, when will the identity of Zen's author be revealed? Did (or do) the organizers of the Pamplona Olympiad know his or her true identity? Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-surfflat/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Reflections on a disaster
David Fotland wrote: > The last moves in the PV are usually quite weak. > They don’t get a lot of playouts. In principle I like long PVs, therefore (and of course because of its playing strength) Many Faces is my favorite Go program. Several of you may laugh at me/it, but with some training a human can get a lot of information from long PVs. Several years ago I was very active in computer-assisted chess where I took the candidate moves from two different chess programs, and it was up to me (the human) to select amongst them. (I called that system 3-Hirn: 2 computer brains + 1 human mind. "Hirn" is German and can mean both brain and mind. When the computers were at Elo around 2500, me with my Elo 1900 made the 3-Hirn as strong as about 2700.) In 3-Hirn, I did not only look at the candidate moves themselves but also at the corresponding PVs (and at the evaluations). By time I learned to recognize "ill PVs" and such ill PVs were often a useful indicator for me not to select this move. (Of course, the more early the illness occurs in the PV the more serious would the reservations against that move be.) Having in mind the 3-Hirn approach also for Go (it will come: when in a few years bots will have 3-4 dan strength, a good human of low dan strength will be able to bring a 3-Hirn to 5-6 dan), I am pretty sure that PV watching will also be an important tool there. On the other side: Too long PVs are not really helpful. So for me, ManyFaces with maximum PV lengths of 10, 12 or 14 would be more comfortable than the very long lines it shows in its current version. This is a wish only, of course. Keep up your very good work, David! Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-aktionspreis/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Cross-Question on Pamplona
Martin Mueller wrote: > In my view, Zen and CrazyStone are clearly the strongest > 19x19 programs on equal PC-type hardware. This is what we > saw on CGOS a few months ago. I also expected MoGo to still > be a few hundred Elo ahead of Fuego on 19x19, but this is > not how the two games in Pamplona turned out. I don't know > how other top programs such as Many Faces and Leela would > compare on 19x19 on equal hardware. Any suggestions? Before Pamplona, you let Fuego9 run on KGS for several test sessions on 9x9 board. I was the opponent for five games on May 5 (account "goingo"). The very first game of the series I played by myself (it is fantastic that MCTS programs allow even a go-idiot to reach a seemingly good score), and operated ManyFaces/Leela (on a standard Dualcore PC) in the other four games manually. In all the games Fuego did not have any problems to win. > Maybe we can get all these programs running on CGOS at the > same time. That would be interesting to get a snapshot of > the current situation. I would also like to see such an event (and would be willing to write a report on it for "Deutsche Go Zeitung"). > Then there was the rumor that Michael Reiss also had a > strong new program. I would like to see that one, too. And I thought, he had become an old man with large beard and white hair ;-) Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://dslspecial.gmx.de/freedsl-aktionspreis/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Match: MoGo vs Taranu on 9x9
Today an exhibition match of four games was played between program MoGo and European Champion Catalin Taranu (5p). On 9x9 board, with 30 min for each player. Games (with tons of comments from spectators) can be found in KGS archive, under http://www.gokgs.com/gameArchives.jsp?user=mogoRennes Taranu won the first three games and lost the final one. So, the score was 3-1 for him. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Match: MoGo vs Taranu on 9x9
Olivier Teytaud wrote: > >> http://www.gokgs.com/gameArchives.jsp?user=mogoRennes > > ... in games 2 and 3 mogo lost quite late with > some stupid very fast moves - this suggests that perhaps > we should save up time in the beginning. Well, it's a > conclusion based on a sample of 2 games :-) I think your sample is already larger than 2 games: look at MoGo's very last 19x19 game in Pamplona (the loss against Zen). I think, in that game too quick play by MoGo in the end was responsible for the loss. http://www.gokgs.com/gameArchives.jsp?user=mogo Mogo[-] Zen19CO [-] 19×19 May 16, 2009, 18:24 (CEST) B+resignation My general impression (also based on experiences from chess): Distributing time rather balanced over the moves is a stable strategy. Of course you will have cases, where 80 seconds instead of 70 seconds make a big difference. But typically this happens much less frequent than differences at for instance "12 sec vs 2 sec". MCTS programs should be more sensible to corruption by small times than (iterative deepening) alpha-beta tree search. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Time weighting in computer go
* To clarify things: with "rather balanced" I did not mean that the search process for each move should get exactly the same number of milli-seconds. A setting where thinking times in early phases of a game are 5 times that for the final moves fits into this "rather balanced". But - there must be something wrong with a program when "80 seconds for an early move" and "3 seconds for a late move" gives clearly better results than "75 seconds for an early move" and "8 seconds for a late move". * When you want to make timing in computer go clever you should make it "very clever", especially in games against human opposition. Otherwise humans will find the mechanism(s) behind the timing of your bot and will start complicated fights just in those parts of a game, where your bot uses only little time. Ingo. -- Neu: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate + Telefonanschluss für nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Extended AMAF
Hello, maybe this is old stuff for the insiders. In my lecture today a clever student (Hagen Riedel) brought up the following idea to extend AMAF. He counts in four ways moves on a specific square: (1) How often did player A make this move in random games which he won? (2) How often did player A make this move in random games which he lost? (3) How often did opponent B make this move in random games which A won? (4) How often did opponent B make this move in random games which A lost. Let n(1), ... n(4) be these numbers, and q(1), ..., q(4) the respective quota. Would it make sense to combine them by a linear combination, like lamda(1)*q(1) + lamda(2)*q(2), where the lamda's are appropriate positve or negative weights. Any comments, experiences, ...? Ingo. -- Nur bis 31.05.: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Who is stv?
A program named "stv" has won the 9x9-KGS bot tournament today by a laaarge margin, ahead of Fuego, Zen9, Aya, Valkyria and Rango. More details under http://www.gokgs.com/tournEntrants.jsp?sort=s&id=463 Does someone here know who stv's programmer is? Ingo. -- Nur bis 31.05.: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: stv is Steenvreter
Nick Wedd explained: > stv is Steenvreter. Its creator is indeed Erik van der Werf, > whose KGS account is evdw. Its name is Dutch for "stone eater"... Congratulations to Erik van der Warf for the Win! By the way, "Steenvreter" is such a nice name. You should call your baby by full name on KGS. Ingo. PS: "van der Warf" is also nice, but I understand when you want to keep that short. -- Nur bis 31.05.: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: stv is Steenvreter
Hi Erik, >> By the way, "Steenvreter" is such a nice name. You should >> call your baby by full name on KGS. > > When I registered the kgs account for Steenvreter the name was too > long, so I had to shorten it :-( > I've update stv's profile to show Steenvreter under 'Real Name' Thanks. 10 letters seem to be the maximum length for KGS names. Might "Stenvreter" be acceptable for you, or "SteenEater", or ...? > BTW My last name is "Werf" (not "Warf", and even if you wanted to > translate that it would become "Wharf"). Sorry for the typo. I mixed your name up with the german "Warft", which are artificial "high places" on the little Hallig islands in Germany. Ingo. -- Nur bis 31.05.: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: The Shodan Go Bet
Darren Cook wrote: > Back in 1997 I made a $1000 bet with John Tromp that he wouldn't > be beaten my a computer before 2011. I've made a page to publicize > the bet: http://dcook.org/gobet/ Interesting. Can you add John Trump's current/recent view on the bet to your site? Ingo. -- Nur bis 31.05.: GMX FreeDSL Komplettanschluss mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Passive Zen is 2-dan now
Zen19 has not been active on KGS since May 9. But it's rating is soaring. See at http://www.gokgs.com/graphPage.jsp?user=zen19 Just with the start of June 2009 Zen19 has crossed the barrier to 2-dan. Congratulations! Ingo. PS: How can a rating of an inactive player change? It changes because its former opponents may play and get changes in their ratings. These loop back. -- GMX FreeDSL mit DSL 6.000 Flatrate und Telefonanschluss nur 17,95 Euro/mtl.! http://dslspecial.gmx.de/freedsl-aktionspreis/?ac=OM.AD.PD003K11308T4569a ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: US Go Congress
Terry McIntyre wrote: > I can volunteer to organize the computer versus pro demonstration. > Myung-Wan Kim lives in Los Angeles and I can approach him; he > played against Mogo last year. > > Will Mogo be available? > > Mogo team, do you feel that Mogo has improved since the previous > demonstration? > > Which is the strongest computer go program nowadays? I would like to see bot vs pro on 9x9 board. What about Fuego (the 9x9-winner of Pamplona) on strong Microsoft hardware?! Just my 2 cents. Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Changing the wall temporarilly?
Olivier Teytaud wrote: > ... I hope we'll find a similar solution for Go, > but for the moment in spite of many trials it does > not work and I'm not far of being tired of trying > plenty of different implementations of these ideas :-) ). Olivier, you are lucky. You missed the decades-long period of really slow progress in computer go - in the times prior to MCTS. 18 months of running against a wall is not much! At least for the moment, there is the chance to change the target: 9x9 is still far from solved. And, maybe also computer Go research for middle-sized boards with 13x13 squares may help to get new ideas, new courage, new impetus. Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: US Go Congress
Another 2 cents from me: what about inviting good old Bruce Wilcox for a show event against computer(s)? With him you would get all in one: * strong amateur * author of (old) go program * author of one of the best go books ever Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Havannah - Go - LittleGolem
Hello, some time ago I had asked if discussions on "computer Havannah" were welcome here in the list. The reactions were positive, but (by different reasons) actors preferred not to use the opportunity. In the meantime a "computer Havannah" scene has developed on the game server http://www.littlegolem.net There it is possible to play Havannah on many different board sizes (side lengths 4, 5, 6, 7, 8, 9, 10). Active programmers are Thomas Reinhardt (D, with TOBRT), Richard Lorentz (US, with Wanderer), Richard Pijl (NL, with gambler), ab (=anonymous, D, with havai). On small boards (4 and 5) the computers are doing really well in the meantime, but from size 6 on their games look strange. It is also possible to run *** GO bots *** there, for instance Gnugo is doing so. Some more information on LittleGolem: * Registration is required, but it is without cost, easy, and without complications. * Thinking time per move is 36 hours in the average (with a buffer of 240 hours; and 20 days of vacation per year). * It is good style to choose names of the type "xxx_c" for computer accounts. If you do not do this, you should write at least in the profile that a computer is playing. * Some participants on LittleGolem are not playing, but only participating in the interesting fora (one general forum; one special forum for each game). For new players it reqires to have some games completed, otherwise you can not write, but only read. (This is a measure against spam bots.) * At the moment there is, for each player, only one go rating (showing performance on 9, 13, 19) and only one Havannah performance, mixed over all board sizes. But I think there is some hope that this general rating will be split in size-dependent ratings again. Feel free to join LittleGolem. Ingo. PS: My account on Little Golem is "Ingo Althofer". http://www.littlegolem.net/jsp/info/player.jsp?plid=11860 -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Havannah - Go - LittleGolem
Hello Michael, Lukasz, Michael Williams wrote: > Are the games archived? Does the public have access to those archives? Yes, they are. Everybody (with internet access) can see and replay the games. For instance, the games of Lukasz Lew can be found at http://www.littlegolem.net/jsp/info/player_game_list.jsp?gtid=havannah&plid=16210 On that page you have to click a number in the leftmost coloum to get to the notation of the game from that row. To find all games of a certain player, you simply have to enter his name in http://www.littlegolem.net/jsp/info/player_list.jsp Lukasz Lew wrote: > ... > Personally I'm really annoyed by many of these bots, In Havannah, there are not many bots. And, in the meantime the programmers have marked their profiles accordingly. > because they do not resign and play all possible forcing > moves That has happened in some games by TOBRT. You should discuss your complaint in the LG (Hex & Havannah) forum - for instance by including the message that in computer go it is "normal" that MCTS-bots resign against humans when their "evaluation" is below something like 25 percent. You may also include a comment that on large boards it may take a bot very long to "understand" that their position may be lost - and that in such situations the programmer is encouraged to resign on behalf of the bot. > so the games can be really long which is an issue > on turn based servers. It is, but on Little Golem this "delay problem" comes in the majority of cases from undisciplined human players. > Moreover they enter tournaments, so if you want to play in > them you have to play with these bots. Again: make a proposal on this in the LG forum. Ingo. PS: Perhaps your negative view is a consequence of the doubtful honour that you were the first really strong player who lost to a bot on Havannah board of size 8. I think the "funny" final position of that game will become part of the history of Havannah http://www.littlegolem.net/jsp/game/game.jsp?gid=1056604 -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Havannah - Go - LittleGolem
Hello Lukasz and Urban Łukasz Lew wrote: >> In Havannah, there are not many bots. And, in the meantime >> the programmers have marked their profiles accordingly. >> > > Profiles don't help. On LG you just click "register" and you are paired. I see your point. When you definitely want to avoid being paired against a bot, you should not be the first to register for a given board size. Instead, you migth send invitation to "appropriate" other players. Urban Hafner wrote: >> Can you post a link to computer Havannah forum/thread? I was looking >> for it on LG few weeks ago and couldn't find it. > > I'd be interested in Links to the Forum posts, too. Especially to > example code for having a bot play on LG. I guess I will be able to > figure that out on my own, but example code/best practices would be great. The thread http://www.littlegolem.net/jsp/forum/topic2.jsp?forum=50&topic=390 starts with something different, but turns to bot questions soon. You find information on ab's bot at his profile http://www.littlegolem.net/jsp/info/player.jsp?plid=6603 @Lukasz: > I'm not strong at all :) You are almost as strong as Christian Freeling, I believe. > And in this game I confused the rules of > havannah. (it is not interesting) I agree that you must have had confusion with the rule that allows "filled cycles". Otherwise, you were clearly superior in that game. > I lost a much more interesting game here: > http://www.littlegolem.net/jsp/game/game.jsp?gid=1044847 > Have fun with analyzing it :) Nice - and hot - game. It seems that your decisive mistakes were 6.b4 (better g1 ?!) and 12.e5 (better a3 ?!). Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Dynamic komi in commercial programs
One of the difficult questions is if (or better how)
dynamic komi can be used to improve the strength of
MCTS go programs in handicap games (both cases being
"interesting": computer on strong side - and -
computer on weak side).
Especially, there are several normal go players
("non-programmers") who are interested in this topic
and who feel that the current non-activity on this
question is unsatisfactorily.
Stefan Kaitschick (German 5-dan amateur, EGF-rating
2439) is in some sense the most prominent amongst them.
I think that it would be a real chance for the programmers
to use the interest and creativity of such people. This might
happen for instance in the following way: include a
"simple komi-modification button"
in your program (or a pair of buttons, one for +, one for -)
which can be used by simple mouse clicks. Of course, all the
time the true value of komi should also be shown. You programmers
would probably be surprised to see what creative users would find
out when playing around with such a feature.
Ingo.
--
Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate
für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Dynamic komi in commercial programs
Don Dailey wrote: > I think we should open up to other ideas, not > just dynamic komi modification. In fact that > has not proved to be a very fruitful technique > and I don't understand the fascination with it. I was not clear enough in the original posting. My main point is the following: Currently the programmers are developing program, and customers are playing with them. But "we" should try to bring creative customers inside the process of development. Dynamic komi might be one good early try for this approach. You write "that [DyKo] has not proved to be a very fruitful technique" but you are right only concerning the rather narrow community of programmers. (Some) creative customers have lots of time to test strange things and settings and are willing to do so, provided the programs are sufficiently "test-friendly". Concerning dynamic komi I would bet 1:1 that their findings might lead to a breakthrough. Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Dynamic komi in commercial programs
Darren Cook wrote: > Ingo's suggestion (of two buttons to increment/decrement komi by one > point) was to make it easy for strong humans to test out the idea for us. Don Dailey wrote: > There is no question that if you provide a button to push, all kinds > of positions will appear where this idea works. Providing a button is > not nearly the same as providing an actual working algorithm that you > can prove is superior. Right. Especially it may turn out that dynamic komi works well as a tool in computer-aided analysis. To give an example, the very new version 12.013 of "Many Faces of Go" has a feature: the program does not only compute its best move but also shows percentages for alternative "popular moves" (name coined by David Fotland). Here is a screenshot http://www.althofer.de/image-fotland.jpg or in context http://www.althofer.de/k-best-visualisations.html Of course, this feature not at all improves the playing strength of autonomous Many Faces. But it is a very hepful tool for computer- aided analysis of positions. The experience from almost two decades of chess programs on the pc: when you give users such tools to "play" with you will earlier or later get very helpful feedback. Ingo. -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Dynamic komi in commercial programs
thaoeuns at gmail.com wrote: > So changing the komi doesn't actually improve your confidence > interval. If (as Darren said) the win percentage is a crude > estimate of the final score, then changing komi would do nothing > to change the results one got (and at extremes biases it badly). > Moving the ratios closer to 50/50 (by whatever means) at the high > end changes the variance of the data, and in a world where there's > a 1:1 correspondence between score and win ratio does nothing to > change one's confidence that the highest ranked node should be. You are wrong. Look at my experiments with "trivial" double step races http://www.althofer.de/mc-laziness.pdf When a Monte-Carlo player thinks to be behind (for instance in a race 6 vs 3) he plays the weaker move more often than in a 6 vs 6 situation. > Of note there is that the goal, of any method chosen, is to make the > ranking of individual moves as accurate as possible... > > Of course what would be most preferable would be lots of data. > Arguing with guesses instead of data is silly. Very good point. Ingo. -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Mirror Go against Zen
On KGS, some humans players (yakuman, finnes) now have started to play mirror go against Zen19 (Humans as Black; komi=+0.5). So far Zen19 seems to react helpless. In contrast, commercial versions of ManyFaces and Leela seem to have (almost) no problems with mirror go. Ingo. -- Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 - sicherer, schneller und einfacher! http://portal.gmx.net/de/go/chbrowser ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Mirror Go against Zen
Don Dailey wrote: > I thought you played mirror go as white? Or with Black, starting in center. It is possible when komi is only 0.5 and chinese scoring. > I'm not a go player, but it seems like it would be hard to win if > you had the white pieces with 0.5 komi and black mirrored everything > you did.You > essentially start from a losing position anyway, right? Right. But Many Faces and Leela manage to win with White in this situation (in several board sizes). > Does the human play to the center on the first move and thereafter > mirror everything? Right. Very boring. Ingo -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Many Faces Win against Mirror on 9x9
Included is the sgf of a win of Many Faces, when playing on 9x9 board with White, at komi 0.5, with opponent making mirror go all the time. Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 manyfaces-mirror-9x9.sgf Description: application/go-sgf ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Really great cinema by Many Faces
David Fotland wrote in a related thread: > Many Faces does not test for mirror go. But it is able to produce really great cinema against mirror go. Today I ran a test game with MF 12.013. Board size 13 Chinese Rules, but with komi=-45.5 (in words "minus fourtyfive-dot-five"). Many Faces as White (with 20 minuntes for all on my medium PC). Me as Black: First stone on the center (g7), and then only mirror go. It took Many Faces 170 moves, but then I had lost by 3.5 points (including komi). It is unbelievable how the bot managed to turn the table. sgf of the game is attached. Ingo. PS: As a pre-test I had played an analogous game, but with komi only at minus 30.5. Also that game was won by Many Faces against "my" mirror play. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 mf-mirror-13x13-komi-minus-45.5.sgf Description: application/go-sgf ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Mirror Go against Zen
Don wrote: > It could be a matter of style as you say, not a matter of strength. Right. > My main questions is whether it's been established as true that Zen > really plays poorly and Many Faces is brilliant against mirror go. > Or does it just seem that way based on casual observation? Right. Especially, by reasons of time I played only very few 19x19 mirror games. Most of my test games (more than 20 against MF, some against Leela) were on 9x9, 11x11, and 13x13. Perhaps (based on casual observation, again) mirror strategy is more difficult for the bots on larger boards. And for Zen I only know the 2 games from KGS. > ... I think I see a trend or a pattern based on a few games > but it turns out to be just my imagination or low sample size. > Humans have wonderful pattern recognition, but it's well know[n] > that we easily find patterns where they don't exist too. Agreed. Therefore I really like the test series by Yamato (with 30 out of 100 wins for Zen against mirror strategy) > Can we assume that both programs are approximately equal or is MFGO > clearly stronger (or visa versa?) In "normal" go (on KGS) Zen seems to be ahead of all other programs on moderate hardware. Mirror go seems to be a world of its own... Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Mirror Go against Zen
Alain Baeckeroot wrote: > gnugo --mirror will try to play mirror go :) How does it do this? Interesting might be a setting like the following: When gnu-mi (short for "gnugo --mirror") has to make a move in a position, the following procedure is run: (a) Is the position a mirror position and is there a legal mirror move in that position? If the answer is "no", Gnugo makes a normal search process and gives the best move it finds. If the answer is "yes", Gnugo makes two searches: one for the best move (and its score s_opt) and another one for the score s_mirror of the mirror move. Given two parameters THRES and DELTA, the two scores are compared with the following consequence: if "s_mirror > THRES" and "s_opt - s_mirror < DETA", then the mirror move is played. else the optimal move is played. Typical values might be THRES = 48 %, DELTA = 3 %. So, the bot deviates from mirror play when it thinks to be behind or when the best move seems to be clearly better than the mirror move. > maybe fun/interesting to add one on cgos ? And what about playing on KGS, but only with Black in games with komi=0.5 ?! Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: demo with pro at Go Congress in Fairfax
Terry wrote: > I'm trying to set up a demo game between Myungwan Kim and Mogo > at the Go Congress in Fairfax, VA Or, what do you think about the following idea: Make it a "pair-go event" with 2 strong humans versus 2 strong go programs (+ handicap). For the computer team, for instance the two best commercial programs MFoG and Leela (on standard hardware) might be suitable candidates. At least such an event would be a "first". Just my 1.5 cent. Ingo -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: O Meien 9p vs Zen
Yamato San wrote: > The match of O Meien 9p vs Zen will be held on next Monday via KGS. > > Schedule (JST) > Date: August 10 > Time: 14:10 > > Handicap > 9x9: Zen is black, with 0.5 or 3.5 komi (?) > 19x19: Zen is black, with 7 handicap stones. Thanks for the information. * How many games will be played, two on 9x9 and one on 19x19? * What will be the thinking times? Other question: Is it true what I read in KGS chat that Zen wil be commercially availabe in Japan on September 28, for about 9,500 Yen? Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Myungwan Kim 8P versus Many Faces of Go
Terry McI announced: > On Friday, August 8th, at 3 PM Eastern Daylight Time, > Myungwan Kim 8P will play against Many Faces of Go. One question, do you mean Friday, August 7th, 2009 or Saturday, August 8th, 2009 or another year ;-) Thanks for informing us. Cheers, Ingo. -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] O-Meien vs Zen
This "morning" the exhibition games between O Meien (9p) and Zen were played on KGS. On 19x19, at handicap 7, O Meien won by resignation. On 9x9, 3 games were played. In all of them Zen had Black. Two of the games were at komi 3.5 - and won by O Meien. One was at komi 2.5 - and won by Zen. sgf may be downloaded from http://www.gokgs.com/gameArchives.jsp?user=oumeien One question to Yamato San: Why did Zen play so quickly in the 19x19 game? Only very seldomly it used more than 2 seconds to reply. Ingo. -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Dynamic komi at high handicaps
In the last few weeks I have experimented a lot with dynamic komi in games with high handicap. Especially, I used the really nice commercial program Many Faces of Go (version 12.013) with its Monte Carlo level (about 2 kyu on 19x19 board) and its traditional 18-kyu level as the opponent. At handicap 21 I played (manulally!) 8 games with these opponents: 4 games with static komi (0.5) - here MFoG (2-kyu) won 1 of the 4 games. 4 games with dynamic komi - here MFoG (2-kyu) won 3 of the 4 games. I used "dynamic komi" in the following "Rule 42" way. Starting point for this internal artificial komi was a very high value (to compensate for the handicap stones), typically 300.5 or 320.5 . Then, always when the evaluation had climbed up to 42 % or higher, dynamic komi was reduced by 50 or 30 or 20 (or 10 near the end), until finally the true value of 0.5 was reached. After this little sample I also tried a few games with dynamic komi at handicap 25. After some unsuccessful games (the Monte Carlo side died of starvation at komi=40.5 or 30.5) today one win came out: In best Monte Carlo fashion, the MC-level won by half a point. I have included sgf of this game. I am aware that small samples are not enough to prove something. Therefore, I hope that programmers may realize automatic versions of something like "Rule 42" to find out how their programs behave with dynamic komi. Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 handicap25-dynamicKomi.sgf Description: application/go-sgf ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Human vs Computer in IEEE conference
Forthcoming human-vs-computer games in go: http://www.althofer.de/ieee-go-0.jpg http://www.althofer.de/ieee-go-1.jpg http://www.althofer.de/ieee-go-2.jpg http://www.althofer.de/ieee-go-3.jpg http://oase.nutn.edu.tw/FUZZ_IEEE_2009/result.htm Ingo. -- GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT! Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] (no subject)
Jeff Nowakowski wrote: >On Wed, Aug 19, 2009 at 07:27:00AM -0700, terry mcintyre wrote: >>Consider the game when computer is black, with 7 stones against a very >>strong human opponent. >> ... > > Didn't this game actually happen? Didn't MoGo *beat* a pro > with 7 stones? It was long ago: in February 2009, and it was only the first game in a series of 6 games. All other five games in that event were won by the humans. Later, only one more bot-win against a "low pro" at h7. http://www.computer-go.info/h-c/index.html Without special "techniques" the h7 wall will stand for a long time. Ingo. PS: Once again I would like to mention my report on "Laziness of Monte Carlo", at http://www.althofer.de/mc-laziness.pdf In the meantime, a student has found the same phenomenon in UCT search (instead of basic MC). Also in discrete online optimization (so outside of combinatorial games) it has been observed by another Ph.D. student of mine: porcedures on Monte Carlo basis are stronger when they have the impression that the situation is "tense". -- Neu: GMX Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://portal.gmx.net/de/go/dsl02 ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
