Re: [computer-go] Crazystone patterns
Chris Fant wrote: I was not able to tell from the CrazyStone paper how the patterns are used in the playouts. Can anyone enlighten me? Does it simply select the move with the highest score? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ No. It selects moves according to the Bradley-Terry probability distribution. Deterministic playouts cannot work, unless you have a super good policy. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [Housebot-developers] [computer-go] ReadyFreddy on CGOS
Another alternatives for testing the performance would be
1. replay a game and count the number of right/near guesses. This
could be pro games, games againt fairly good players, or games played
by the bot before. The benefit of this is that you might be able to
estimate the strength on every few moves rather than an entire game.
2. play against earlier versions of the same bot.
3. keep bots yourself.
4. improve benchmark/regression scores instead of playing strength,
http://www.cs.ualberta.ca/~games/go/cgtc for instance.
See twogtp on how to have bots on your own computer play each other.
This might also be of interest:
http://www.andromeda.com/people/ddyer/go/shape-library.html
On 9/20/07, Jason House [EMAIL PROTECTED] wrote:
Christoph Birk wrote:
// Loop to do #1 above
while (p != singletonSimplePass){
if (numMoves keepMax)
moves[numMoves] = p;
workingCopy.play(c,p);
c = c.enemyColor();
p = randomLegalMove(c, workingCopy, twister);
numMoves++;
}
Do you really stop the simulation after a single pass, ie. when
one side has no more move to play but the other does?
I believe that this would end many games before they are (really) over
and that might lead to false results in the simulations.
Christoph
Actually, that's the only difference between housebot-621-amaf and
hb-amaf-alt. alt is playing games all the way to the end like you
suggest. Looking at the win rate against ego110_allfirst, it looks like
it may be doing a bit worse (but more samples are needed). It's
unfortunate that ranks that low vary so much based on which bots are on
CGOS.
In the future, I'll probably offer the option to do either method. My
logic behind stopping at the first pass is that it's highly unlikely to
form life in the void from captured stones. Since capturing the stones
would increase the length of the game and isn't very likely to change
the outcome of the game, I figured it'd be a good compromise.
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
--
Cenny Wenner
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Update of MoGo binary release, and windows version available!
Hi Gilles, yes they are some problems to use MoGo with Drago. The main issue is the initial message written to stderr as guessed by Dave. Actually, Drago handles incorrectly stdout and stderr in the same way but this is easily corrected. Good news! I have uploaded a patch for using MoGo with Drago: www.godrago.net/DragoForMogoPatch.zip. The zip contains an exe file which should replace the one from the current Drago install. Why is it only a patch for MoGo? Is it only a change to handle stderr in a good way? As MoGo does not implement the GTP command final_result, an error message is displayed at the end of the game and the user must count the result. (By the way, is there a problem with 'final_status_list alive' ?) I don't about all those GTP commands. MoGo implements a subset of GTP which is enough for cgos, KGS and gogui. For the scoring, of course cgos is the minimal :), but the scoring works very well on KGS and gogui. I don't know which commands they send, but it seems enough :). Yet, it would be good for MoGo to support Drago as it seems pretty trivial to do so, I will at some point look into it (without giving a date :)). Cheers, Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Update of MoGo binary release, and windows version available!
Thank you Sylvain! It works fine for me. A very good sparring strong and so different in style! You mean that with the .dll in WINDOWS\system32 that goes further? I thought that the .dll in the same directory as the .exe was enough. Copy/pasting from Microsoft's API reference: The function searches for the file in the following sequence: 1. The directory from which the application loaded. 2. The current directory. 3. Windows 95: The Windows system directory. Use the GetSystemDirectory function to get the path of this directory. Windows NT: The 32-bit Windows system directory. Use the GetSystemDirectory function to get the path of this directory. The name of this directory is SYSTEM32. 4. Windows NT: The 16-bit Windows system directory. There is no Win32 function that obtains the path of this directory, but it is searched. The name of this directory is SYSTEM. 5. The Windows directory. Use the GetWindowsDirectory function to get the path of this directory. 6. The directories that are listed in the PATH environment variable. The priority is only important if there is more than one version of the file. Placing the file that works ok in the same path as the application is good practice. This makes the program work fine without modifying the system anyhow. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Most common 3x3 patterns
Cenny Wenner wrote: Care to elaborate on what you mean by scores here and how they are similar to the 9x9 equivalence? I guess Dave is using Bradley Terry scores. This idea was introduced by Rémi Coulom in Computing Elo Ratings of Move Patterns in the Game of Go The paper is available on the web (finding link left to the reader ;-). This idea outperforms previous approaches. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: Update of MoGo binary release, and windows version available!
Hi Sylvain, Thank you for the releases of the Windows version and the Linux version for older processors. The Windows version, however, seems much weaker than MoGo that running on KGS these days on 19x19, even giving much longer time setting such as --time 300 for example. I guess some other settings than time are necessary. So, my question is what setting allows Windows version being the same strength as KGS version? Regards, Hideki Sylvain Gelly: [EMAIL PROTECTED]: Hi all, you can find here: http://www.lri.fr/~gelly/MoGo_Download.htm an update of MoGo's release, especially binary for non pentium4 compatible processors, some other options explained, and maybe more interesting, an option for time management (I stupidly did not think that people would use MoGo with frontend sending time_left). It is basically adding simple interface to what existed before, but I hope it will be useful for you. In the same time of this update, you can finally download a windows version! Unfortunately, it is not multithread, and still 30% slower than the linux version, but at least it works :). Hopefully a Mac version will come. Please feel free to share the link to any player you know who may be interested. Best, Sylvain PS: I had no time to check everything. Hopefully it works, but if you find problems please report (and don't be upset ;-)) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Crazystone patterns
Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? On 9/20/07, Rémi Coulom [EMAIL PROTECTED] wrote: Chris Fant wrote: I was not able to tell from the CrazyStone paper how the patterns are used in the playouts. Can anyone enlighten me? Does it simply select the move with the highest score? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ No. It selects moves according to the Bradley-Terry probability distribution. Deterministic playouts cannot work, unless you have a super good policy. 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] Crazystone patterns
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Chris Fant wrote: Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? You can use this model in many ways. It is not as expensive to calculate the probabibility of every legal move as you might think because no higher math functions are involved if you put it in the right form. - - Don On 9/20/07, Rémi Coulom [EMAIL PROTECTED] wrote: Chris Fant wrote: I was not able to tell from the CrazyStone paper how the patterns are used in the playouts. Can anyone enlighten me? Does it simply select the move with the highest score? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ No. It selects moves according to the Bradley-Terry probability distribution. Deterministic playouts cannot work, unless you have a super good policy. 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/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFG8mgGDsOllbwnSikRAnnIAJwNG0Lsgy8pELg1QqF746T+GNwkrACggYYy fWo3T0ye+BtxPx1EM92RMLM= =0qbZ -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Update of MoGo binary release, and windows version available!
Hi Hideki, The Windows version, however, seems much weaker than MoGo that running on KGS these days on 19x19, even giving much longer time setting such as --time 300 for example. I guess some other settings than time are necessary. Sorry, you are right the 19x19 settings always put the time management on. So either play with --totalTime xx (instead of --time xx) and use some frontend sending the time_left command, or add --timeManagementMode 0, or wait for me to fix ;). The 9x9 is not affected by that. Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Update of MoGo binary release, and windows version available!
Thank you for the releases of the Windows version and the Linux version for older processors. The Windows version, however, seems much weaker than MoGo that running on KGS these days on 19x19, even giving much longer time setting such as --time 300 for example. I guess some other settings than time are necessary. I'd asked Sylvain this, suggesting he add it to the FAQ. His answer was roughly it plays on KGS with core2 duo 2.66Ghz with 30s/move at 9x9 and 20s/move at 19x19. (Because it actually uses time management it is gradually getting less and less as the game moves to middle then endgame.) I've been playing it at 60s/move on 9x9 (on a celeron 2.8Ghz on linux). It is getting just under 200K simulations in middle-game positions and is definitely dan-level. It is fascinating to see it estimating, say, 0.48 chance of win (for itself, mogo is losing) then I play a move and then it is estimating, say, 0.53 (i.e. mogo is now winning) and almost every time it is right - I made a mistake. Darren ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Crazystone patterns
Chris Fant a écrit : Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? I only keep light-weight local features in the random simulations. So, it is not expensive. A table of move urgencies can be updated incrementally after each move: only a few change. In fact, my program became faster when I introduced 3x3 patterns for the random simulations. I can pre-compute a lot of information indexed by pattern shape, that are useful to detect eyes and move legality. From memory, Crazy Stone does about 5k playouts / second from the empty position on 19x19, on one 3 GHz CPU, which is only about 5 times slower than libego, if I remember correctly. Rémi ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Re: [Housebot-developers] ReadyFreddy on CGOS
On Wed, 19 Sep 2007, Jason House wrote: My logic behind stopping at the first pass is that it's highly unlikely to form life in the void from captured stones. Since capturing the stones would increase the length of the game and isn't very likely to change the outcome of the game But how do you score game, if there are still stones to capture? Do you assume everything's alive? Christoph ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Leksand
The 2008 European Go Congress, to be held in Leksand, Sweden from July 28th to August 9th, will definitely include a 19x19 Computer Go event, with some prize money. It may also include a small-board computer event. There is more information at http://egc2008.eu/en/congress/cg.php, and a form that you can use to register. The computer event(s) will be on Wednesday August 6th, so if you are planning to compete in the Computer Olympiad in Beijing as well, you will have time to get from there to Sweden. 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] MoGo, and computer Go events
On 9/20/07, Nick Wedd [EMAIL PROTECTED] wrote: events as a benchmark. Its source code has been released, and there is It has? Where? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Update of MoGo binary release, and windows version available!
Hello Giles, I downloaded the latest version of Drago (which includes the LibKombilo.dll file that someone mentioned), copied in your patch, and played Mogo for a few moves. It seems to work OK. YAY! I'll mention this here, since others might encounter it. There is a small race condition problem in the latest (patched and unpatched) version of Drago that wasn't there in earlier versions. If you set Mogo to play white, and then quickly make your first move as black, Drago will hang until you select Cancel from the menu. It does the same thing with my engine. Drago blocks the player from trying to move while the engine is thinking about?its own?move. But apparently?he is -not- blocked from moving when Drago should be waiting for the engine to respond to a command like name. So if the player moves while the engine is still busy reading its initialization files, Drago gets stuck. Gnugo initializes so quickly that the player probably couldn't click a button fast enough to win this ?race. - Dave Hillis -Original Message- From: Gilles Arcas [EMAIL PROTECTED] To: computer-go@computer-go.org Sent: Wed, 19 Sep 2007 7:15 pm Subject: [computer-go] Update of MoGo binary release, and windows version available! Hello Dave, bonjour Sylvain,? ? yes they are some problems to use MoGo with Drago. The main issue is the initial message written to stderr as guessed by Dave. Actually, Drago handles incorrectly stdout and stderr in the same way but this is easily corrected.? ? I have uploaded a patch for using MoGo with Drago: www.godrago.net/DragoForMogoPatch.zip. The zip contains an exe file which should replace the one from the current Drago install.? ? As MoGo does not implement the GTP command final_result, an error message is displayed at the end of the game and the user must count the result. (By the way, is there a problem with 'final_status_list alive' ?)? ? I could let the dll and the data files in MoGo folder without any problem.? ? Amicalement? Gilles? ? ___? computer-go mailing list? [EMAIL PROTECTED] http://www.computer-go.org/mailman/listinfo/computer-go/? Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] MoGo, and computer Go events
In message [EMAIL PROTECTED], Chris Fant [EMAIL PROTECTED] writes On 9/20/07, Nick Wedd [EMAIL PROTECTED] wrote: events as a benchmark. Its source code has been released, and there is It has? Where? My mistake. Not the source, only the binary, has been released. Nick -- Nick Wedd[EMAIL PROTECTED] ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] time_left command on CGOS
Hello to all. I have put my program on cgos, but i never receive the time_left command. Is it normal, or am i doing something wrong ? _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Update of MoGo binary release, and windows version available!
Dear Hideki, Dear all, I addressed some of the issues some of you mentioned. Thank you very much for the reports. Now the collision between time management and --time xx in 19x19 is no more here (while you assuming there is no time_left sent). There is also now a --13 option. The --dontDisplay 1 removes all the displays (rather than most of). For the issues reported by Gilles for Drago concerning the final_status and final_status_list alive gtp commands, I read the GTP doc, and I understand now :). MoGo implements a subset of GTP which is sufficient for CGOS, KGS and gogui, so the final_status_list dead works well, and should be enough to compute the score. The others are simply unfortunately not supported. I am very sorry for the inconvenience. Best, Sylvain 2007/9/20, Hideki Kato [EMAIL PROTECTED]: Hi Sylvain, Thank you for the releases of the Windows version and the Linux version for older processors. The Windows version, however, seems much weaker than MoGo that running on KGS these days on 19x19, even giving much longer time setting such as --time 300 for example. I guess some other settings than time are necessary. So, my question is what setting allows Windows version being the same strength as KGS version? Regards, Hideki Sylvain Gelly: [EMAIL PROTECTED]: Hi all, you can find here: http://www.lri.fr/~gelly/MoGo_Download.htm an update of MoGo's release, especially binary for non pentium4 compatible processors, some other options explained, and maybe more interesting, an option for time management (I stupidly did not think that people would use MoGo with frontend sending time_left). It is basically adding simple interface to what existed before, but I hope it will be useful for you. In the same time of this update, you can finally download a windows version! Unfortunately, it is not multithread, and still 30% slower than the linux version, but at least it works :). Hopefully a Mac version will come. Please feel free to share the link to any player you know who may be interested. Best, Sylvain PS: I had no time to check everything. Hopefully it works, but if you find problems please report (and don't be upset ;-)) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ 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] time_left command on CGOS
Thanks, but i dont receive the known_command either :( I added known_command to the list list_commands also, because i thought it could be the problem, but it is still not working. - Message d'origine De : Don Dailey [EMAIL PROTECTED] À : computer-go computer-go@computer-go.org Envoyé le : Jeudi, 20 Septembre 2007, 21h37mn 18s Objet : Re: [computer-go] time_left command on CGOS -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 CGOS does not send that command unless you tell it that it knows the command using the known_commands ivan dubois wrote: Hello to all. I have put my program on cgos, but i never receive the time_left command. Is it normal, or am i doing something wrong ? _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFG8svuDsOllbwnSikRAu0xAKDzYQDU3+8p5LkfgAQgWA4Sh0hw2QCgzED4 IJejdXhKjIa8qZNYBLtVcnQ= =V1ig -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: Re : [computer-go] time_left command on CGOS
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 You are right, It's the list_commands I use. I think GTP shouldn't have both because it's extra work for nothing and has confused me before. Do you implement time_settings? If you don't have that or respond wrongly to it, the server (or perhaps it's the client) will not send the time_left command. - - Don ivan dubois wrote: Thanks, but i dont receive the known_command either :( I added known_command to the list list_commands also, because i thought it could be the problem, but it is still not working. - Message d'origine De : Don Dailey [EMAIL PROTECTED] À : computer-go computer-go@computer-go.org Envoyé le : Jeudi, 20 Septembre 2007, 21h37mn 18s Objet : Re: [computer-go] time_left command on CGOS CGOS does not send that command unless you tell it that it knows the command using the known_commands ivan dubois wrote: Hello to all. I have put my program on cgos, but i never receive the time_left command. Is it normal, or am i doing something wrong ? _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ 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/ _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFG8ts0DsOllbwnSikRAqtQAKCTtqirDxpeTeAK5Dc7r9r1I1OudQCfbqOU cZOu+emQkdF4e4K7AFWggaA= =foHI -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Re: Most common 3x3 patterns
-Original Message- From: Cenny Wenner [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Wed, 19 Sep 2007 11:31 am Subject: Re: [computer-go] Re: Most common 3x3 patterns On 9/19/07, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: I neglected the rather important detail that these patterns are trained on 9x9 games. Training on 19x19 games produces different scores than these. I've tried it both ways (it's much easier to get a large set of 19x19 games for training) and this set is the one I now use for both 9x9 and 19x19. But my program's performance on 19x19 is terrible either way. IIRC, if I train on 19x19 games, but only keep track of patterns with a center within the 5x5 window around the enemy's previous move, then I get scores very similar to those from 9x9 games. Care to elaborate on what you mean by scores here and how they are similar to the 9x9 equivalence? OK. There are many ways to derive a score for a pattern and, naturally, it matters what you're going to use it for. For these patterns, I calculated the score in the obvious way that everyone thinks of first. I took a file with ~22,000 9x9 games from the NNGS. Most of these games were unusable, so I automatically edited most of them out and only actually used those from low KYU or Dan level players where there weren't any obvious problems in the game record, like illegal moves or absurd handicaps. I added a couple hundred pro games to the set. For every move, where the player didn't pass, in every game, I trained the patterns. I (my program of course) looked at every legal move, checked the 3x3 pattern around it and incremented the counter hits for that pattern. If the same pattern showed up 5 times on the board for one turn, I incremented hits 5 times. Then, I looked at the pattern around the move the player actually made, and incremented the counter moves one time for that pattern. (By my convention, it is always white's turn to move, so I mirrored the colors when needed.) After training, the score for each pattern was (100 * moves)/hits. That produces a score between 0 and 100. Because it's difficult to collect a lot of decent 9x9 games by human players, many of the patterns have noisy scores based on very few samples. It's relatively easy to collect 19x19 game records, but their scores, calculated this way, will be fairly different. In some experiments, I used a large set of 19x19 games but only incremented the counters for patterns a small distance away from the opponent's previous move. When I say these scores are similar, I mean that they tended to be within a few points of each other when there was a reasonable sample size for the patterns. For noisy cases, they were at least consistent: the 9x9 games might have a score of 84 based on a sample size (number of hits for the pattern) of only 19 while the 19x19 games gave a (still very high) score of 46 based on 1019 samples. There are much more sophisticated ways to calculate a score. I've used this way for different things. How am I using it now? AntIgo's heavy Monte-Carlo playouts are derived from the description in the first?Mogo paper. If the MC player can't find a move to rescue a threatened group, it looks for any moves with a good pattern close to the foe's previous move. If it finds any, it chooses one at random. My threshold for a good pattern is one with a score = 2, so my score is essentially a binary one. Measured with a micrometer, marked with a piece of chalk, cut with an axe. - Dave Hillis ? Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] 2007 Cotsen Open wants your program to enter
SlugGo entered the first year as a 9 kyu and won 1 game. One other game was clearly won on the board (more than 100 points) but the opponent was clever enough to start playing very complicated moves that I could see were not going to work, but took SlugGo a long time to reply to, and SlugGo lost on time. After that we put in time management code. The second year I made a silly mistake, and the graphical front end (GoBan) was mistakenly set to announce atari. The announcement was something that the multi-processor message passing code did not understand, so SlugGo lost all games by crashing in response to the first atari of the game. I figured this out minutes after the last game. SlugGo had been entered as a 10 kyu the second year. I felt horrible accepting the cash prize for best program, but it did help offset the $1250 cost of a handicap van that I had to rent to transport the cluster to Los Angeles. The Cotsen open is 5 games. I think the bracket ran from 8 to 12 kyu. I did not do anything to narrow my guess of the rating. I will look for the games. Cheers, David On 20, Sep 2007, at 10:19 AM, Mark Schreiber wrote: On Thu Sep 13 16:14:36 PDT 2007, David Doshay wrote: SlugGo won the award for Best Computer Program both of the last 2 years David, thanks for telling us about the Cotsen Open. Where can I find more about SlugGo's games at the Cotsen Open? How many games were played? What were the scores? What were the ratings of the opponents? Did you compute a performance rating for SlugGo? Mark Schreiber ___ 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 : Re : [computer-go] time_left command on CGOS
I found the problem. My response to list_commands was wrong, the commands were separated by spaces, instead of line feed. It is working now, thank you very much for your help :) - Message d'origine De : Don Dailey [EMAIL PROTECTED] À : computer-go computer-go@computer-go.org Envoyé le : Jeudi, 20 Septembre 2007, 22h42mn 29s Objet : Re: Re : [computer-go] time_left command on CGOS -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 You are right, It's the list_commands I use. I think GTP shouldn't have both because it's extra work for nothing and has confused me before. Do you implement time_settings? If you don't have that or respond wrongly to it, the server (or perhaps it's the client) will not send the time_left command. - - Don ivan dubois wrote: Thanks, but i dont receive the known_command either :( I added known_command to the list list_commands also, because i thought it could be the problem, but it is still not working. - Message d'origine De : Don Dailey [EMAIL PROTECTED] À : computer-go computer-go@computer-go.org Envoyé le : Jeudi, 20 Septembre 2007, 21h37mn 18s Objet : Re: [computer-go] time_left command on CGOS CGOS does not send that command unless you tell it that it knows the command using the known_commands ivan dubois wrote: Hello to all. I have put my program on cgos, but i never receive the time_left command. Is it normal, or am i doing something wrong ? _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ 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/ _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFG8ts0DsOllbwnSikRAqtQAKCTtqirDxpeTeAK5Dc7r9r1I1OudQCfbqOU cZOu+emQkdF4e4K7AFWggaA= =foHI -END PGP SIGNATURE- ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ _ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] games per player?
I looked at the cgos game records, and it shows the most recent n games. I was looking for games by Mogo, which did not appear in that list at the time I checked. How do I find games played by a particular program? Thanks! Terry McIntyre [EMAIL PROTECTED] They mean to govern well; but they mean to govern. They promise to be kind masters; but they mean to be masters. -- Daniel Webster Be a better Globetrotter. Get better travel answers from someone who knows. Yahoo! Answers - Check it out. http://answers.yahoo.com/dir/?link=listsid=396545469___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] time_left command on CGOS
ivan dubois wrote: Hello to all. I have put my program on cgos, but i never receive the time_left command. Is it normal, or am i doing something wrong ? When I first put my bot up on cgos, I hit a similar issue, even though I listed time_left in known_commands. They key for me was to add support for a time_settings command. I reported this bug to Don, but I don't know if it was fixed. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Crazystone patterns
-Original Message- From: Rémi Coulom [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Thu, 20 Sep 2007 10:22 am Subject: Re: [computer-go] Crazystone patterns Chris Fant a écrit : Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? I only keep light-weight local features in the random simulations. So, it is not expensive. A table of move urgencies can beupdated incrementally after each move: only a few change. In fact, my program became faster when I introduced 3x3 patterns for the random simulations. I can pre-compute a lot of information indexed by pattern shape, that are useful to detect eyes and move legality. From memory, Crazy Stone does about 5k playouts / second from the empty position on 19x19, on one 3 GHz CPU, which is only about 5 times slower than libego, if I remember correctly. Rémi Chris Fant a écrit : Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? I only keep light-weight local features in the random simulations. So, it is not expensive. A table of move urgencies can beupdated incrementally after each move: only a few change. In fact, my program became faster when I introduced 3x3 patterns for the random simulations. I can pre-compute a lot of information indexed by pattern shape, that are useful to detect eyes and move legality. From memory, Crazy Stone does about 5k playouts / second from the empty position on 19x19, on one 3 GHz CPU, which is only about 5 times slower than libego, if I remember correctly. Rémi Suppose I can generate scores for all of the moves quickly enough. I still face the problem of quickly choosing a move biased by the scores. Tournament selection is fast, but that is a function of relative ranking of the scores, not the values of the scores. Roulette wheel selection gives me an answer, but it is slow slow slow, the way I implement it anyway. Can anybody describe a good way to do this? - Dave Hillis Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Crazystone patterns
Remi keeps a number that is the sum of all the probabilities (I'll call them that, although they are not normalized so they add up to 1) and also one number per row that is the sum of the probabilities of the points in that row. Now you pick from the distribution of rows, and inside the row from the distribution of points, and it can be done fast. On 9/20/07, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: -Original Message- From: Rémi Coulom [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Thu, 20 Sep 2007 10:22 am Subject: Re: [computer-go] Crazystone patterns Chris Fant a écrit : Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? I only keep light-weight local features in the random simulations. So, it is not expensive. A table of move urgencies can beupdated incrementally after each move: only a few change. In fact, my program became faster when I introduced 3x3 patterns for the random simulations. I can pre-compute a lot of information indexed by pattern shape, that are useful to detect eyesand move legality. From memory, Crazy Stone does about 5k playouts / second from the empty position on 19x19, on one 3 GHz CPU, which is only about 5 times slower than libego, if I remember correctly. Rémi Suppose I can generate scores for all of the moves quickly enough. I still face the problem of quickly choosing a move biased by the scores. Tournament selection is fast, but that is a function of relative ranking of the scores, not the values of the scores. Roulette wheel selection gives me an answer, but it is slow slow slow, the way I implement it anyway. Can anybody describe a good way to do this? - Dave Hillis Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ 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] Crazystone patterns
That should do nicely. Thanks! -Original Message- From: Álvaro Begué [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Thu, 20 Sep 2007 10:31 pm Subject: Re: [computer-go] Crazystone patterns Remi keeps a number that is the sum of all the probabilities (I'll all them that, although they are not normalized so they add up to 1) nd also one number per row that is the sum of the probabilities of he points in that row. Now you pick from the distribution of rows, nd inside the row from the distribution of points, and it can be done ast. On 9/20/07, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: -Original Message- From: Rémi Coulom [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Thu, 20 Sep 2007 10:22 am Subject: Re: [computer-go] Crazystone patterns Chris Fant a écrit : Does this mean that you need to calculate the Bradley-Terry probability for every legal move before selecting one on that probability? Isn't that expensive? Have you tried selecting only N legal candidates at random and then selecting one of those based on their Bradley-Terry probability distribution to save time? I only keep light-weight local features in the random simulations. So, it is not expensive. A table of move urgencies can beupdated incrementally after each move: only a few change. In fact, my program became faster when I introduced 3x3 patterns for the random simulations. I can pre-compute a lot of information indexed by pattern shape, that are useful to detect eyesand move legality. From memory, Crazy Stone does about 5k playouts / second from the empty position on 19x19, on one 3 GHz CPU, which is only about 5 times slower than libego, if I remember correctly. Rémi Suppose I can generate scores for all of the moves quickly enough. I still face the problem of quickly choosing a move biased by the scores. Tournament selection is fast, but that is a function of relative ranking of the scores, not the values of the scores. Roulette wheel selection gives me an answer, but it is slow slow slow, the way I implement it anyway. Can anybody describe a good way to do this? - Dave Hillis Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ __ omputer-go mailing list [EMAIL PROTECTED] ttp://www.computer-go.org/mailman/listinfo/computer-go/ Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [Housebot-developers] [computer-go] ReadyFreddy on CGOS
I'm going to echo Cenny's comment. Small samples like this can be very misleading. For this kind of test, I usually give each algorithm 5000 playouts per move and let them play 1000 games on my computer. It takes about a day and a half. - Dave Hillis -Original Message- From: Cenny Wenner [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Tue, 18 Sep 2007 3:33 pm Subject: Re: [Housebot-developers] [computer-go] ReadyFreddy on CGOS By the data in your upper table, the results need to uphold their mean for 40 times as many trials before you even get a significant* difference between #1 and #2. Which are the two methods you used? On 9/18/07, Jason House [EMAIL PROTECTED] wrote: original eye method = 407 ELO alt eye method #1 = 583 ELO alt eye method #2 = 518 ELO While both alternate methods are probably better than the original, I'm not convinced there's a significant difference between the two alternate methods. The cross-tables for both are fairly close and could be luck of the draw (and even which weak bots were on at the time). I put raw numbers below. Since I made one other change when doing the alt eye method, I should rerun the original with that other change as well (how I end random playouts and score them to allow for other eye definitions). While I think the alternate eye definitions helped, I don't think they accounted for more than 100-200 ELO vs ego110_allfirst orig= 33/46 = 71% #1 = 17/20 = 85% #2 = 16/18 = 89% vs gotraxx-1.4.2a orig=N/A #1 = 2/8 = 25% #2 = 3/19 = 16% On 9/17/07, Jason House [EMAIL PROTECTED] wrote: On 9/17/07, Don Dailey [EMAIL PROTECTED] wrote: Another way to test this, to see if this is your problem, is for ME to implement YOUR eye definition and see if/how much it hurts AnchorMan. I'm pretty much swamped with work today - but I may give this a try at some point. I'd be interested in seeing that. It looks like my first hack at an alternate eye implementation bought my AMAF version about 150 ELO (not tested with anything else). Of course, what I did isn't what others are using. I'll do another alteye version either today or tomorrow. It may be possible that some of my 150 was because I changed the lengths of the random playouts. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- Cenny Wenner ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ Check Out the new free AIM(R) Mail -- Unlimited storage and industry-leading spam and email virus protection. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] games per player?
I will get to this next week. Cheers, David On 20, Sep 2007, at 2:50 PM, Don Dailey wrote: I have bzipped them by month and have them on one of Doshay's machines, but they are not visible to the web - don't know how to make them so but I would like to put up some links to them. I will ask David Doshay about this. - - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
