Re: [computer-go] Monte-Carlo and 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.) The difference I've noticed most in analyzing strong 9x9 games (in Japanese rules using a bot playing in Chinese rules) is a half-point ko at the end of the game. In Chinese rules, once all the dame have been filled in, if one side has no ko threats he has to pass, and the opponent fills the ko and gets an extra point. In Japanese rules passing does not cost anything. So the score ends up one point different. In these games what tends to happen is the Chinese-rule-monte-carlo bot, playing for its 0.5pt win, ends up choosing the wrong move 20 moves from the end of the game. If that is as clear as mud let me know and I'll try to hunt up an example game. Darren -- Darren Cook, Software Researcher/Developer http://dcook.org/mlsn/ (English-Japanese-German-Chinese-Arabic open source dictionary/semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
I'm sure he meant, "Does your program play optimally in trivial situations where different rules would result in a different winner?" I'm not sure if your last answer also applies to that, more specific question. David Fotland wrote: I score with area, and adjust for Japanese rules. It doesn't play optimally under and rule set :) David -Original Message- From: [EMAIL PROTECTED] [mailto:computer-go- [EMAIL PROTECTED] On Behalf Of Jeff Nowakowski Sent: Thursday, November 06, 2008 11:57 AM To: computer-go Subject: RE: [computer-go] Monte-Carlo and Japanese rules On Thu, 2008-11-06 at 09:43 -0800, David Fotland wrote: Many Faces of Go's Monte Carlo engine plays strongly using Japanese rules. So what do you do in the playouts? Do you score with area or territory? Does your program play optimally where different rules would result in different winner? -Jeff ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
I tried the Tesuji Software CGOS Client. There does seem to be issues on my setup. I'm using IBM jdk 1.6.0 and linux. It connects to the program and to CGOS. It sends the boardsize command with no apparent glitches. Then it sends the clear_board command, which is received by the ending and responded to. However at this point the java client fails to see that the engine responded. It could be a line buffering issue, a unix vs windows line ending issue or something else altogether. The same bot works with the cgos client, so I am at a loss. Has anyone else tried this with Linux? On Thu, 2008-11-06 at 16:38 -0200, Mark Boon wrote: > As part of the open-source project I posted about yesterday, I have > also made a Java client to connect a GTP engine to CGOS. This is > nothing specific to the project, it should work with any GTP > compliant program just as Don's original client does. It does have a > small advantage, and that is it shows a simple go-board with the game > that is being played. So there's no need to start the cgosview > program separately, although that will obviously still have its uses. > And you don't have to open a new window for each game. I figured that > might be of some interest to those who frequently have a bot playing > on CGOS and who like to keep an eye on what it's doing. > > I would welcome it if others used it as well, so they can tell me > what they think. http://plug-and-go.dev.java.net > > Mark > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ signature.asc Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Monte-Carlo and Japanese rules
I score with area, and adjust for Japanese rules. It doesn't play optimally under and rule set :) David > -Original Message- > From: [EMAIL PROTECTED] [mailto:computer-go- > [EMAIL PROTECTED] On Behalf Of Jeff Nowakowski > Sent: Thursday, November 06, 2008 11:57 AM > To: computer-go > Subject: RE: [computer-go] Monte-Carlo and Japanese rules > > On Thu, 2008-11-06 at 09:43 -0800, David Fotland wrote: > > Many Faces of Go's Monte Carlo engine plays strongly using Japanese rules. > > So what do you do in the playouts? Do you score with area or territory? > Does your program play optimally where different rules would result in > different winner? > > -Jeff > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
Those kind of errors usually mean an incompatibility with your Java version and the one I used. Most likely because your default Java installation is 1.4 instead of 1.5 or higher. Go to the menu 'Project -> Properties' and choose 'Java build path'. In the Libraries tab check if there are any errors. And check if it has at least JVM 1.5. If there are errors or you have any doubt, remove the JRE library and click 'Add Library' and choose a JRE installed on your computer that is at least 1.5 Hope that helps. Mark On 6-nov-08, at 17:55, Michael Williams wrote: I get these arrors in Eclipse. I have never used Eclipse before. Sorry, I know this is a bit off topic. The project was not built since its build path is incomplete. Cannot find the class file for java.lang.Object. Fix the build path then try building this project GoEngineTemplate Unknown Java Problem The project was not built since its build path is incomplete. Cannot find the class file for java.lang.Object. Fix the build path then try building this project GoPluginFramework Unknown Java Problem The type java.lang.Object cannot be resolved. It is indirectly referenced from required .class files Point.java GoPluginFramework/ source/tesuji/games/util line 1 Java Problem The type java.lang.Object cannot be resolved. It is indirectly referenced from required .class files RandomGoEngine.java GoEngineTemplate/source line 1 Java Problem Mark Boon wrote: As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Monte-Carlo and Japanese rules
On Thu, 2008-11-06 at 09:43 -0800, David Fotland wrote: > Many Faces of Go's Monte Carlo engine plays strongly using Japanese rules. So what do you do in the playouts? Do you score with area or territory? Does your program play optimally where different rules would result in different winner? -Jeff ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
I get these arrors in Eclipse. I have never used Eclipse before. Sorry, I know this is a bit off topic. The project was not built since its build path is incomplete. Cannot find the class file for java.lang.Object. Fix the build path then try building this project GoEngineTemplate Unknown Java Problem The project was not built since its build path is incomplete. Cannot find the class file for java.lang.Object. Fix the build path then try building this project GoPluginFramework Unknown Java Problem The type java.lang.Object cannot be resolved. It is indirectly referenced from required .class files Point.java GoPluginFramework/source/tesuji/games/util line 1 Java Problem The type java.lang.Object cannot be resolved. It is indirectly referenced from required .class filesRandomGoEngine.java GoEngineTemplate/source line 1 Java Problem Mark Boon wrote: As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
Hi Michael, Sure, should be pretty easy. What we have to think about is what would be the best way to organize some kind of 'engine repository' of engines based on the framework, without things becoming a jungle. If you have any ideas I'd like to hear them. And, if you want developer access to the project, just apply. Mark On Thu, Nov 6, 2008 at 5:15 PM, Michael Williams <[EMAIL PROTECTED]> wrote: > I can take a crack at integrating Don's existing Java reference bot into > your framework, if you want. > > > Mark Boon wrote: >> >>Hi Michael, >> >> I was in fact thinking about that. I just didn't decide yet what would >> be the best way. Most likely I will add at least the reference-bot >> implementation and a straightforward UCT search. And possibly more, I >> have a lot of stuff lying around. But I didn't get around to it yet. >> >>Mark >> >> >> On Thu, Nov 6, 2008 at 4:48 PM, Michael Williams >> <[EMAIL PROTECTED]> wrote: >>> >>> I'm downloading Eclipse now to check it out. Will you be adding a >>> reference >>> implementation (Don's AMAF spec) and/or a basic UCT implementation? >>> >>> >>> Mark Boon wrote: As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ >>> ___ >>> computer-go mailing list >>> computer-go@computer-go.org >>> http://www.computer-go.org/mailman/listinfo/computer-go/ >>> >> ___ >> computer-go mailing list >> computer-go@computer-go.org >> http://www.computer-go.org/mailman/listinfo/computer-go/ >> > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
I can take a crack at integrating Don's existing Java reference bot into your framework, if you want. Mark Boon wrote: Hi Michael, I was in fact thinking about that. I just didn't decide yet what would be the best way. Most likely I will add at least the reference-bot implementation and a straightforward UCT search. And possibly more, I have a lot of stuff lying around. But I didn't get around to it yet. Mark On Thu, Nov 6, 2008 at 4:48 PM, Michael Williams <[EMAIL PROTECTED]> wrote: I'm downloading Eclipse now to check it out. Will you be adding a reference implementation (Don's AMAF spec) and/or a basic UCT implementation? Mark Boon wrote: As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
Hi Michael, I was in fact thinking about that. I just didn't decide yet what would be the best way. Most likely I will add at least the reference-bot implementation and a straightforward UCT search. And possibly more, I have a lot of stuff lying around. But I didn't get around to it yet. Mark On Thu, Nov 6, 2008 at 4:48 PM, Michael Williams <[EMAIL PROTECTED]> wrote: > I'm downloading Eclipse now to check it out. Will you be adding a reference > implementation (Don's AMAF spec) and/or a basic UCT implementation? > > > Mark Boon wrote: >> >> As part of the open-source project I posted about yesterday, I have also >> made a Java client to connect a GTP engine to CGOS. This is nothing specific >> to the project, it should work with any GTP compliant program just as Don's >> original client does. It does have a small advantage, and that is it shows a >> simple go-board with the game that is being played. So there's no need to >> start the cgosview program separately, although that will obviously still >> have its uses. And you don't have to open a new window for each game. I >> figured that might be of some interest to those who frequently have a bot >> playing on CGOS and who like to keep an eye on what it's doing. >> >> I would welcome it if others used it as well, so they can tell me what >> they think. http://plug-and-go.dev.java.net >> >>Mark >> >> ___ >> computer-go mailing list >> computer-go@computer-go.org >> http://www.computer-go.org/mailman/listinfo/computer-go/ >> > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Monte-Carlo and Japanese rules
And of course black should pay 1 point for each extra handicap stone. http://www.britgo.org/rules/compare.html#coun Dave Van: [EMAIL PROTECTED] namens [EMAIL PROTECTED] Verzonden: do 6-11-2008 19:28 Aan: computer-go Onderwerp: RE: [computer-go] Monte-Carlo and Japanese rules What if the playout uses the AGA rule of paying 1 point for a pass and requiring white to pass last (so the game does not end by two passes if black plays the second pass). Wouldn't the score then be equivalent to the japanese score? Dave Van: [EMAIL PROTECTED] namens Mark Boon Verzonden: do 6-11-2008 17:11 Aan: computer-go Onderwerp: Re: [computer-go] Monte-Carlo and Japanese rules Although what Don writes is all correct, I understood the question to be rather different. It's not a matter of being able to determine the right score at the end or the right way to play, it's a matter of determining the right score after each playout. For performance reasons MC programs will cut corners there which could be taken advantage of when playing by Japanese rules because the after the playout it is prone to getting the wrong score in certain situations. Mark On 6-nov-08, at 12:12, Don Dailey wrote: > On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: >> 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.) > > I've thought about those questions myself from time to time. Let me > think out loud concerning this. I am by know means an expert in > Japanese scoring or even GO in general, so I'm just giving some > thoughts > here and a plan for building a Japanese "simple" bot that you can be > free to criticize: > > It seems to me the primary difference between the two is knowing > when to > stop playing and of course scoring dead groups. The Chinese style > bots > do not technically need to know about scoring. > > You can look at the combined statistics at the end of the games for a > given point to get a sense of whether that point is still in play or > whether it's a forgone conclusion. You can do the same to determine > dead groups. I don't know how well that works in all cases, but I > have > used it and it works pretty well. > > But we also want to recognize dame, and not play to dame points early > in the game even if it doesn't affect the final Chinese outcome. So > here is my idea: > > 1. If ALL the stones of a particular group belong to the opponent > with > high certainty, they are dead. > > 2. If there are open spaces that belong to you or the opponent with > high certainty don't move to them. > > 3. If an uncertain point is touching stones of both colors and both > colors have high certainty for the color they belong to, it is > probably > dame and you shouldn't move to them. > > example: White has a stone on d4 that is clearly alive. >Black has a stone on f4 that is clearly alive. >An empty point on e4 is highly uncertain. >Do not play to e4 - it is probably dame. > >question: Is that a reasonably good rule or does it need some > work? > > > 4. If you have no moves other than these cases, you should pass. > > You can test this idea by playing a bot on KGS under Japanese rules. > You may have to tweak what you consider your uncertainty margin. > Also, > I'm not considering seki here but we would want to find a way to cope > with that. > > - Don > > > >> Ingo. > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] CGOS client
I'm downloading Eclipse now to check it out. Will you be adding a reference implementation (Don's AMAF spec) and/or a basic UCT implementation? Mark Boon wrote: As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] CGOS client
As part of the open-source project I posted about yesterday, I have also made a Java client to connect a GTP engine to CGOS. This is nothing specific to the project, it should work with any GTP compliant program just as Don's original client does. It does have a small advantage, and that is it shows a simple go-board with the game that is being played. So there's no need to start the cgosview program separately, although that will obviously still have its uses. And you don't have to open a new window for each game. I figured that might be of some interest to those who frequently have a bot playing on CGOS and who like to keep an eye on what it's doing. I would welcome it if others used it as well, so they can tell me what they think. http://plug-and-go.dev.java.net Mark ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Monte-Carlo and Japanese rules
What if the playout uses the AGA rule of paying 1 point for a pass and requiring white to pass last (so the game does not end by two passes if black plays the second pass). Wouldn't the score then be equivalent to the japanese score? Dave Van: [EMAIL PROTECTED] namens Mark Boon Verzonden: do 6-11-2008 17:11 Aan: computer-go Onderwerp: Re: [computer-go] Monte-Carlo and Japanese rules Although what Don writes is all correct, I understood the question to be rather different. It's not a matter of being able to determine the right score at the end or the right way to play, it's a matter of determining the right score after each playout. For performance reasons MC programs will cut corners there which could be taken advantage of when playing by Japanese rules because the after the playout it is prone to getting the wrong score in certain situations. Mark On 6-nov-08, at 12:12, Don Dailey wrote: > On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: >> 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.) > > I've thought about those questions myself from time to time. Let me > think out loud concerning this. I am by know means an expert in > Japanese scoring or even GO in general, so I'm just giving some > thoughts > here and a plan for building a Japanese "simple" bot that you can be > free to criticize: > > It seems to me the primary difference between the two is knowing > when to > stop playing and of course scoring dead groups. The Chinese style > bots > do not technically need to know about scoring. > > You can look at the combined statistics at the end of the games for a > given point to get a sense of whether that point is still in play or > whether it's a forgone conclusion. You can do the same to determine > dead groups. I don't know how well that works in all cases, but I > have > used it and it works pretty well. > > But we also want to recognize dame, and not play to dame points early > in the game even if it doesn't affect the final Chinese outcome. So > here is my idea: > > 1. If ALL the stones of a particular group belong to the opponent > with > high certainty, they are dead. > > 2. If there are open spaces that belong to you or the opponent with > high certainty don't move to them. > > 3. If an uncertain point is touching stones of both colors and both > colors have high certainty for the color they belong to, it is > probably > dame and you shouldn't move to them. > > example: White has a stone on d4 that is clearly alive. >Black has a stone on f4 that is clearly alive. >An empty point on e4 is highly uncertain. >Do not play to e4 - it is probably dame. > >question: Is that a reasonably good rule or does it need some > work? > > > 4. If you have no moves other than these cases, you should pass. > > You can test this idea by playing a bot on KGS under Japanese rules. > You may have to tweak what you consider your uncertainty margin. > Also, > I'm not considering seki here but we would want to find a way to cope > with that. > > - Don > > > >> Ingo. > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Monte-Carlo and Japanese rules
Many Faces of Go's Monte Carlo engine plays strongly using Japanese rules. It's required for sales in American and japan (as AI Igo). I don't use Remi's trick, since there are sometimes points remaining when your opponent passes when playing against weaker players. David > -Original Message- > From: [EMAIL PROTECTED] [mailto:computer-go- > [EMAIL PROTECTED] On Behalf Of Rémi Coulom > Sent: Thursday, November 06, 2008 1:02 AM > To: computer-go > Subject: Re: [computer-go] Monte-Carlo and Japanese rules > > Ingo Althöfer wrote: > > 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. > > > Hi Ingo, > > 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. > > Rémi > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
On Nov 6, 2008, at 11:09 AM, Don Dailey <[EMAIL PROTECTED]> wrote: On Thu, 2008-11-06 at 10:44 -0500, Jason House wrote: I think simplistic handling of Japanese rules should play dame points that connect chains. This avoids some problems that can arise where ownership probability drops after the opponent plays the dame, and a point of territory must get filled. I'm trying to think of a counter example where my rule will not work. Of course this has to be understood within the context of the playing strength of the bot itself, so we don't expect extremely sophisticated handling of special cases. Consider a miai to connect a chain with two liberties. It's easy to get the ownership correct in all heavy playouts. If one of the two miai points is dame, your rule would say not to play it. If the opponent plays there, the chain ends up in atari... In my example, both black and white have clearly live stones and there is an empty point touching both of them. Assuming that the map is correct of course and both groups are alive, then I suppose a connecting move could cause a weaker group to live. So that suggests another rule - in addition to the original condition of at least 1 white and black live group touching, there should be no connections to a group that is not alive. A group that is "alive" passes some thresehold of certainty. Otherwise, it is not-alive but not dead. The threshold might be something like 90 - 95% or something but some experimentation would be required to find a good value for this. - Don Even if not technically required, I can imagine bots acting like beginners and get nervous over imagined vulnerabilites. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
On Thu, 2008-11-06 at 17:10 +0100, Erik van der Werf wrote: > IIRC under official Japanese rules at the end of the game all groups > with liberties shared between opposing colours are by definition in > seki. Therefore eventually (before counting) all dame have to be > filled. > > Further, playing dame points is almost equally bad under Chinese rules > as it is under Japanese rules. So, if you have a strong 'Chinese' > engine no special tricks are needed at least until you reach the > endgame. The only thing that is severely penalized under Japanese > rules is playing needless defensive moves inside your own territory > while the opponent is passing. The dame rule I assumed was needed for an MC bot playing because of the way MC bots play "badly" when winning or losing. The idea was to focus on what fights were left on the board so that you don't lose due to a Japanese technicality. But it's now not clear to me that my rule would fix that. Maybe it should be left out.I'm also not sure how my rule would affect seki positions. - Don signature.asc Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
Although what Don writes is all correct, I understood the question to be rather different. It's not a matter of being able to determine the right score at the end or the right way to play, it's a matter of determining the right score after each playout. For performance reasons MC programs will cut corners there which could be taken advantage of when playing by Japanese rules because the after the playout it is prone to getting the wrong score in certain situations. Mark On 6-nov-08, at 12:12, Don Dailey wrote: On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: 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.) I've thought about those questions myself from time to time. Let me think out loud concerning this. I am by know means an expert in Japanese scoring or even GO in general, so I'm just giving some thoughts here and a plan for building a Japanese "simple" bot that you can be free to criticize: It seems to me the primary difference between the two is knowing when to stop playing and of course scoring dead groups. The Chinese style bots do not technically need to know about scoring. You can look at the combined statistics at the end of the games for a given point to get a sense of whether that point is still in play or whether it's a forgone conclusion. You can do the same to determine dead groups. I don't know how well that works in all cases, but I have used it and it works pretty well. But we also want to recognize dame, and not play to dame points early in the game even if it doesn't affect the final Chinese outcome. So here is my idea: 1. If ALL the stones of a particular group belong to the opponent with high certainty, they are dead. 2. If there are open spaces that belong to you or the opponent with high certainty don't move to them. 3. If an uncertain point is touching stones of both colors and both colors have high certainty for the color they belong to, it is probably dame and you shouldn't move to them. example: White has a stone on d4 that is clearly alive. Black has a stone on f4 that is clearly alive. An empty point on e4 is highly uncertain. Do not play to e4 - it is probably dame. question: Is that a reasonably good rule or does it need some work? 4. If you have no moves other than these cases, you should pass. You can test this idea by playing a bot on KGS under Japanese rules. You may have to tweak what you consider your uncertainty margin. Also, I'm not considering seki here but we would want to find a way to cope with that. - Don Ingo. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
IIRC under official Japanese rules at the end of the game all groups with liberties shared between opposing colours are by definition in seki. Therefore eventually (before counting) all dame have to be filled. Further, playing dame points is almost equally bad under Chinese rules as it is under Japanese rules. So, if you have a strong 'Chinese' engine no special tricks are needed at least until you reach the endgame. The only thing that is severely penalized under Japanese rules is playing needless defensive moves inside your own territory while the opponent is passing. Erik On Thu, Nov 6, 2008 at 4:44 PM, Jason House <[EMAIL PROTECTED]> wrote: > I think simplistic handling of Japanese rules should play dame points that > connect chains. This avoids some problems that can arise where ownership > probability drops after the opponent plays the dame, and a point of > territory must get filled. > > Even if not technically required, I can imagine bots acting like beginners > and get nervous over imagined vulnerabilites. > > Sent from my iPhone > > On Nov 6, 2008, at 9:12 AM, Don Dailey <[EMAIL PROTECTED]> wrote: > >> On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: >>> >>> 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.) >> >> I've thought about those questions myself from time to time. Let me >> think out loud concerning this. I am by know means an expert in >> Japanese scoring or even GO in general, so I'm just giving some thoughts >> here and a plan for building a Japanese "simple" bot that you can be >> free to criticize: >> >> It seems to me the primary difference between the two is knowing when to >> stop playing and of course scoring dead groups. The Chinese style bots >> do not technically need to know about scoring. >> >> You can look at the combined statistics at the end of the games for a >> given point to get a sense of whether that point is still in play or >> whether it's a forgone conclusion. You can do the same to determine >> dead groups. I don't know how well that works in all cases, but I have >> used it and it works pretty well. >> >> But we also want to recognize dame, and not play to dame points early >> in the game even if it doesn't affect the final Chinese outcome. So >> here is my idea: >> >> 1. If ALL the stones of a particular group belong to the opponent with >> high certainty, they are dead. >> >> 2. If there are open spaces that belong to you or the opponent with >> high certainty don't move to them. >> >> 3. If an uncertain point is touching stones of both colors and both >> colors have high certainty for the color they belong to, it is probably >> dame and you shouldn't move to them. >> >> example: White has a stone on d4 that is clearly alive. >> Black has a stone on f4 that is clearly alive. >> An empty point on e4 is highly uncertain. >> Do not play to e4 - it is probably dame. >> >> question: Is that a reasonably good rule or does it need some work? >> >> >> 4. If you have no moves other than these cases, you should pass. >> >> You can test this idea by playing a bot on KGS under Japanese rules. >> You may have to tweak what you consider your uncertainty margin. Also, >> I'm not considering seki here but we would want to find a way to cope >> with that. >> >> - Don >> >> >> >>> Ingo. >> >> ___ >> computer-go mailing list >> computer-go@computer-go.org >> http://www.computer-go.org/mailman/listinfo/computer-go/ > > ___ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ > ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
On Thu, 2008-11-06 at 10:44 -0500, Jason House wrote: > I think simplistic handling of Japanese rules should play dame > points > that connect chains. This avoids some problems that can arise where > ownership probability drops after the opponent plays the dame, and a > point of territory must get filled. I'm trying to think of a counter example where my rule will not work. Of course this has to be understood within the context of the playing strength of the bot itself, so we don't expect extremely sophisticated handling of special cases. In my example, both black and white have clearly live stones and there is an empty point touching both of them. Assuming that the map is correct of course and both groups are alive, then I suppose a connecting move could cause a weaker group to live. So that suggests another rule - in addition to the original condition of at least 1 white and black live group touching, there should be no connections to a group that is not alive. A group that is "alive" passes some thresehold of certainty. Otherwise, it is not-alive but not dead. The threshold might be something like 90 - 95% or something but some experimentation would be required to find a good value for this. - Don > > Even if not technically required, I can imagine bots acting like > beginners and get nervous over imagined vulnerabilites. signature.asc Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
I think simplistic handling of Japanese rules should play dame points that connect chains. This avoids some problems that can arise where ownership probability drops after the opponent plays the dame, and a point of territory must get filled. Even if not technically required, I can imagine bots acting like beginners and get nervous over imagined vulnerabilites. Sent from my iPhone On Nov 6, 2008, at 9:12 AM, Don Dailey <[EMAIL PROTECTED]> wrote: On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: 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.) I've thought about those questions myself from time to time. Let me think out loud concerning this. I am by know means an expert in Japanese scoring or even GO in general, so I'm just giving some thoughts here and a plan for building a Japanese "simple" bot that you can be free to criticize: It seems to me the primary difference between the two is knowing when to stop playing and of course scoring dead groups. The Chinese style bots do not technically need to know about scoring. You can look at the combined statistics at the end of the games for a given point to get a sense of whether that point is still in play or whether it's a forgone conclusion. You can do the same to determine dead groups. I don't know how well that works in all cases, but I have used it and it works pretty well. But we also want to recognize dame, and not play to dame points early in the game even if it doesn't affect the final Chinese outcome. So here is my idea: 1. If ALL the stones of a particular group belong to the opponent with high certainty, they are dead. 2. If there are open spaces that belong to you or the opponent with high certainty don't move to them. 3. If an uncertain point is touching stones of both colors and both colors have high certainty for the color they belong to, it is probably dame and you shouldn't move to them. example: White has a stone on d4 that is clearly alive. Black has a stone on f4 that is clearly alive. An empty point on e4 is highly uncertain. Do not play to e4 - it is probably dame. question: Is that a reasonably good rule or does it need some work? 4. If you have no moves other than these cases, you should pass. You can test this idea by playing a bot on KGS under Japanese rules. You may have to tweak what you consider your uncertainty margin. Also, I'm not considering seki here but we would want to find a way to cope with that. - Don Ingo. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
On Thu, 2008-11-06 at 09:19 +0100, "Ingo Althöfer" wrote: > 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.) I've thought about those questions myself from time to time. Let me think out loud concerning this. I am by know means an expert in Japanese scoring or even GO in general, so I'm just giving some thoughts here and a plan for building a Japanese "simple" bot that you can be free to criticize: It seems to me the primary difference between the two is knowing when to stop playing and of course scoring dead groups. The Chinese style bots do not technically need to know about scoring. You can look at the combined statistics at the end of the games for a given point to get a sense of whether that point is still in play or whether it's a forgone conclusion. You can do the same to determine dead groups. I don't know how well that works in all cases, but I have used it and it works pretty well. But we also want to recognize dame, and not play to dame points early in the game even if it doesn't affect the final Chinese outcome. So here is my idea: 1. If ALL the stones of a particular group belong to the opponent with high certainty, they are dead. 2. If there are open spaces that belong to you or the opponent with high certainty don't move to them. 3. If an uncertain point is touching stones of both colors and both colors have high certainty for the color they belong to, it is probably dame and you shouldn't move to them. example: White has a stone on d4 that is clearly alive. Black has a stone on f4 that is clearly alive. An empty point on e4 is highly uncertain. Do not play to e4 - it is probably dame. question: Is that a reasonably good rule or does it need some work? 4. If you have no moves other than these cases, you should pass. You can test this idea by playing a bot on KGS under Japanese rules. You may have to tweak what you consider your uncertainty margin. Also, I'm not considering seki here but we would want to find a way to cope with that. - Don > Ingo. signature.asc Description: This is a digitally signed message part ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Monte-Carlo and Japanese rules
Hello Ingo, You (we) have to adjust one point if a game ends by black in usual (no seko etc) cases. As Japanese doesn't count dame while Chinese does, if a game ends by black, black gets one more point under Chinese rules than Japanese. Hideki Ingo Althöfer: <[EMAIL PROTECTED]>: >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. -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] November KGS bot tournament: large boards, slow
Registration is now open for the next bot tournament on KGS, which will be on Sunday November 16th. Each division will be a 5-round Swiss, 19x19 boards, 43 minutes each plus very fast Canadian overtime of 25 moves in 20 seconds. They will start at 16:00 UTC (=GMT) and 16:05 respectively, and end seven hours thirty minutes later. Note that the event is not this weekend, but next. I will be away on holiday this weekend, and will not receive emails until I return on the morning of Wednesday 12th. Registration is as described at http://www.weddslist.com/kgs/how/index.html When you register you should tell me the processor power (number of processors, processor speed, and any other significant details) of the platform that it will be running on. This information will appear in my report of the event, making comparisons between programs more meaningful. I have limited understanding of processor power, and am likely to publish it as given, see for example the final section of http://www.weddslist.com/kgs/past/43/index.html The tournaments are on the KGS site at http://www.gokgs.com/tournInfo.jsp?id=429 and http://www.gokgs.com/tournInfo.jsp?id=430 These pages may give the times of the rounds in your local timezone, depending on your browser and its settings. Nick -- Nick Wedd[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] List of go engines
The list is updated : 32 engines (26 with source) and 10 toolkits Thank's for all programmers eric Eric Marchand a écrit : Hi all, Here is a list of 24 free go engines (18 with source): http://ricoh51.free.fr/go/engineeng.htm Please let me know if there are errors or omissions. eric ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Monte-Carlo and Japanese rules
Ingo Althöfer wrote: 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. Hi Ingo, 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. Rémi ___ 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/
