Re: [computer-go] U. of Alberta bots vs. the Poker pros
On 7/26/07, Jeff Nowakowski [EMAIL PROTECTED] wrote: Ah, an opponent model. Where's the poision? http://www.imdb.com/title/tt0093779/quotes#qt0250635 Too much rock, paper, scissors in poker for my tastes. BTW, there's a rather sophisticated Rock Paper Scissors player named Iocane Powder. http://ofb.net/~egnor/iocaine.html - Brian ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] U. of Alberta bots vs. the Poker pros
- Original Message - From: [EMAIL PROTECTED] To: computer-go@computer-go.org Sent: Friday, July 27, 2007 3:26 AM Subject: Re: [computer-go] U. of Alberta bots vs. the Poker pros I don't understand this. For a given hand the odds of winning can be easily calculated for poker and the best play can be formulated accordingly. It's like to program a com[uter to win a coin toss. I would be surprised if any side win big. The only thing a computer can to is to model opponent's behavior, which may deviate from the best play. What did I miss? DL A lot. The chance alone is meaningless. It can be very profitable to play a hand with 20% chance of winning and it can be a desaster. If you have pot odds of 10:1 the 20% are a good deal, with pot-odds 1:1 its a desaster. The direct pot odds are easy. Whats in the pot and whats the money I put it. But the interesting figure are the implied-pot-odds. What money do I have to put in in all betting rounds and what's the pot at the end. This depends of course also on the actions of the opponents. Another point is: The winning-chance depends on the action of the opponents. If one raises and the opponent folds, one wins with every card. Chrilly -- AOL now offers free email to everyone. Find out more about what's free from AOL at AOL.com. -- ___ 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] KGS Tournament Registration
On 7/27/07, Nick Wedd [EMAIL PROTECTED] wrote: I'd like to register HBotSVN for the open tournament. I forget why we ran HB04 in the last tournament as well, but let me know if that's desired for this tournament. It's entirely up to you. I have a slight preference for more entrants rather than fewer, but you should enter the bots that you want to enter. It's nearly zero overhead to also run HB04... both for me to do and on the computer it runs on. all I do is type start_housebot_0.4 and it's up and running! Go ahead and add it to the tournament then. As for the last two events, please send it (with the words KGS Tournament Registration in the title as usual) to me at maproom at gmail dot com (converted to a valid address in the obvious way). Please note the above paragraph, for future events. One day I may run one of these events while away from home, with access to gmail but not to my regular email address. The e-mail title is KGS Tournament Registration. How did gmail handle it? I responded to the past e-mail but then clicked Edit Subject. Should I not do it that way in the future? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Differences..
OK, I see now, with more 1 point eyes for W, W will play into B's 2 areas reducing them to one eye each, and when B can make the capturing moves W can play into its own 1 point eyes, but black can't play into either its own or W's. So, I agree this rule set has very different endgame considerations, and the intuitive Proof on the web page is flawed. Cheers, David On 27, Jul 2007, at 2:30 AM, Nick Wedd wrote: In message [EMAIL PROTECTED], Joshua Shriver [EMAIL PROTECTED] writes What is the difference in Go and Mathematical Go? http://brooklyngoclub.org/jc/rulesgo.html Is Mathamatical Go a subset of Go as the rules look the same to me as regular go. The Mathematical Rules of Go are, like the Chinese rules, Japanese rules, NZ rules, Tromp-Taylor rules, Ing rules, etc., a set of rules by which a game can be played. It is often asserted that the games defined by all these sets of rules are rather similar, at least when played skilfully. This assertion is false. The game defined by the Mathematical Rules of Go is significantly different from the (rather similar) games defined by the other rule sets. You will realise this if you consider the position below: # # # O . Oentire 6x6 board . . # O O .# to play . . # O . O # # # O O . . . # # O O . . . # O . Using the Mathematical Rules of Go, O can win with correct play. Under any other rule set, # has already won, or can win with correct play. This effect of the value of one-point eyes is much more significant than that of the two-stone group tax mentioned by David. 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 mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] KGS Tournament Registration
Dear Jason, In message [EMAIL PROTECTED], Jason House [EMAIL PROTECTED] writes I'd like to register HBotSVN for the open tournament. I forget why we ran HB04 in the last tournament as well, but let me know if that's desired for this tournament. It's entirely up to you. I have a slight preference for more entrants rather than fewer, but you should enter the bots that you want to enter. Unfortunately, I don't anticipate significantly enhanced performance for this next tournament. *Name on KGS: HBotSVN *Name of bot: HouseBot 0.6.2 *Authors: The HouseBot development team *Division: Open *My name: Jason House Thank you. HBotSVN is now registered for the Open division. As for the last two events, please send it (with the words KGS Tournament Registration in the title as usual) to me at maproom at gmail dot com (converted to a valid address in the obvious way). Please note the above paragraph, for future events. One day I may run one of these events while away from home, with access to gmail but not to my regular email address. 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] Differences..
In message [EMAIL PROTECTED], Joshua Shriver [EMAIL PROTECTED] writes What is the difference in Go and Mathematical Go? http://brooklyngoclub.org/jc/rulesgo.html Is Mathamatical Go a subset of Go as the rules look the same to me as regular go. The Mathematical Rules of Go are, like the Chinese rules, Japanese rules, NZ rules, Tromp-Taylor rules, Ing rules, etc., a set of rules by which a game can be played. It is often asserted that the games defined by all these sets of rules are rather similar, at least when played skilfully. This assertion is false. The game defined by the Mathematical Rules of Go is significantly different from the (rather similar) games defined by the other rule sets. You will realise this if you consider the position below: # # # O . Oentire 6x6 board . . # O O .# to play . . # O . O # # # O O . . . # # O O . . . # O . Using the Mathematical Rules of Go, O can win with correct play. Under any other rule set, # has already won, or can win with correct play. This effect of the value of one-point eyes is much more significant than that of the two-stone group tax mentioned by David. 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] Why Poker-GMs don't win at poker.
Poker can be analyzed well by (even naif) Monte Carlo methods. How? Simulation! Whatever evaluation you need and don't know how to compute because it is too complex for easy formulas can be simulated. This applies to the probability of winning, but also on the betting decisions (call, raise, etc.) and the amount in the pots your own and your opponent's. Play it out randomly 10 times and see how many times you win/loose. Whatever estimator you get will be better than what a human can do just guessing. I call it naif because its the first idea, like basic MC. Then, knowing the problem better you can evolve to something like UCT that favors more promising lines, etc. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Go in hardware
Hi Chrilly: You have mentioned go in hardware twice recently and I have, knowing that you have experience in hardware development, some questions: 1. What should be implemented? In your Hydra cluster I have read you implemented mobility, and somewhere you proposed something like influence. Can you explain it? Here are questions for anyone how knows or has ideas: 2. My board system. I have Bradley Terry scores of patterns built of the neighbors of all empty cells that translate to the legal moves sorted by score. I can update the 40 neighbor masks without any conditional jumps in about 3 asm instructions per neighbor, say 2 ns at 3.4GHz. In hardware you could do in parallel what I do sequentially (the 40 neighbors). But few stones have 40 empty neighbors because they may be out of the board (I once explained how I do that) or not empty, so a guesstimate of the gain for doing that in hardware is x20 (i.e. x40). If your hardware technology runs at 3.4 GHz, that's great! But if it runs at 200 MHz it is even. Then, my software is multicore. I test it with 2 cores, but expect to run in on 256 cores machines in the future. Can the hardware support this? 3. My second problem. translating patterns to scores (database search). I call it a database, but it is a set of sorted 32 bit masks where I can find really fast without any conditional jumps using cmovc instructions. So it is very similar to the previous case. Hardware is better only if clock speed is high enough and it supports multiple cores. 4. My bottleneck: Sorting. Its not really a bottleneck because sorting few values is fast, but it is the slowest part of my m-search. (Sorting is not required for MC/UCT methods.) Is there hardware for sorting? In short: a) How fast can we expect hardware to be? b) Can you repeat the hardware kernel x times so that each thread own an independent kernel? c) Can you sort (in flash time) 19^2 integers? d) How many patterns could the hardware store? e) How much would it cost? Of course, implementing in software has a wider market and you should only implement something that proved to be valuable. I am not claiming that my board is, but it is an interesting subject. Jacques. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Differences..
I have pondered about this before however that page's proposal furthermore changes the value of captures. If black captures x stones, he may play at these x spots up to x times (depending on other and size of eyes), avaraging one per capture, at the very most. In both [territory + captures] (japanese) versus [territory + live stones] (chinese) rules the value per capture is 2. In order to make the mathematical go more similar one should force the other player to play an additional [difference in captures + signed komi] stones before being declared winners. This would be equivalent to just playing til neither player has legal moves and then count the live stones and komi. The remaining eyes is what chinese also covers. This would also handle Nick Wedd's excellent problem. On 7/27/07, David Doshay [EMAIL PROTECTED] wrote: OK, I see now, with more 1 point eyes for W, W will play into B's 2 areas reducing them to one eye each, and when B can make the capturing moves W can play into its own 1 point eyes, but black can't play into either its own or W's. So, I agree this rule set has very different endgame considerations, and the intuitive Proof on the web page is flawed. Cheers, David On 27, Jul 2007, at 2:30 AM, Nick Wedd wrote: In message [EMAIL PROTECTED], Joshua Shriver [EMAIL PROTECTED] writes What is the difference in Go and Mathematical Go? http://brooklyngoclub.org/jc/rulesgo.html Is Mathamatical Go a subset of Go as the rules look the same to me as regular go. The Mathematical Rules of Go are, like the Chinese rules, Japanese rules, NZ rules, Tromp-Taylor rules, Ing rules, etc., a set of rules by which a game can be played. It is often asserted that the games defined by all these sets of rules are rather similar, at least when played skilfully. This assertion is false. The game defined by the Mathematical Rules of Go is significantly different from the (rather similar) games defined by the other rule sets. You will realise this if you consider the position below: # # # O . Oentire 6x6 board . . # O O .# to play . . # O . O # # # O O . . . # # O O . . . # O . Using the Mathematical Rules of Go, O can win with correct play. Under any other rule set, # has already won, or can win with correct play. This effect of the value of one-point eyes is much more significant than that of the two-stone group tax mentioned by David. 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 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] Differences..
http://brooklyngoclub.org/jc/rulesgo.html Is Mathamatical Go a subset of Go as the rules look the same to me as regular go. The rules described here are not mathematical go, but no-pass go. In mathematical go a move consists of a board play or by handing the opponent back a captured stone. This is equivalent to stone scoring. -- Barry Phease mailto:[EMAIL PROTECTED] http://barry.phease.org.nz ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] patterns again
Back to bitmaps with most of the bits at zero... Let's take an extreme case, a bitmap with one out of 32 bits set. We're assuming this map is a member of a class with a significant property which we want to recognize. If it were a perfectly random bitmap, the probability of it turning out this way would be quite low. We can't expect that the actual probability that a bit will be 1, in a bitmap with this property, is 1/32. But from what we know so far, that is our best guess. If that's true, is it likely that this particular bit being on is purely by chance. Not likely; someone with more facility with statistics can probably tell us exactly how likely, and that may be useful. Meanwhile, in deciding which bits to mask out for a map intended to test for this particular property, I am guessing that I'd want to make the probability of this bit being masked is (at most) 1/32, while the ideal probability that any particular other bit in the test map should be masked might be 31/32. ? A good conjecture? Or just plausible? Anyone here have ideas on this? Forrest Curo - This email was sent using AIS WebMail. http://www.americanis.net/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Request for comments - Documentation of HouseBot design
Over the last 4 months or so, I've been building up the documentation at http://housebot.sourceforge.net/index.php/Agile_Development I've tried to keep it at a fairly high quality level by polling for feedback from friends and family and hunting down tools for graphical documentation. I think they're at a good enough of a level to solicit feedback from a larger audience. I've been told by another computer go developer that http://housebot.sourceforge.net/index.php/Agile_Development/Block_2 was of particular interest. Just to seed feedback: * Can the information for each block be organized in a better way? What should I do differently in block 3? * Which diagrams are most effective? Least effective? * What additional information should be included? * Is the design sound? Any fatal flaws? Over-engineered? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Engine development for beginners
Are there any really simple engines out there that know just enough to play a legal game of Go? Preferably C, Perl or Java? Some of the open source engines I've looked at are rather complex and not to friendly to a beginner. Kinda looking for the tscp of chess for go :) -Josh ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Engine development for beginners
Since my rewrite, I don't consider my bot (HouseBot) to be too far along... It barely knows how to do more than play a legal game of go (it does 1-ply monte carlo) The class goban tracks the board state, checks for legality, etc... It can be found here: http://housebot.svn.sourceforge.net/viewvc/housebot/trunk/housebot/goban.d?view=markup The relevant code spans lines 711-1227. 500 lines may sound like a lot, but it doesn't really do a heck of a lot. About 40 lines are comments, 100 lines of unit tests and in contracts. The play function, the heart of the class, is ~150 lines, but has 3 helper functions embedded inside for both clarity and profiling. If you poke around, looking at other code in the file, there are a few things that will make it look more complex. I tried to add a generalized code flavor to stuff allowing for different position and board classes. The goban class was written quickly and doesn't use that extra fluff. I plan to refactor this file over the coming week(s). It's written in D, which looks a lot like C++/Java. Joshua Shriver wrote: Are there any really simple engines out there that know just enough to play a legal game of Go? Preferably C, Perl or Java? Some of the open source engines I've looked at are rather complex and not to friendly to a beginner. Kinda looking for the tscp of chess for go :) -Josh ___ 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] U. of Alberta bots vs. the Poker pros
On 7/26/07, chrilly [EMAIL PROTECTED] wrote: This is a remarkable result. I think poker is more difficult than Go and of course chess. I am as surprised by this statement as everyone else. Of course you have to develop some mixed strategies, try go guess implied pot odds, folding equity etc. but assuming you have access to a large database of high level poker games to analyze, why should it be that hard, esp. in 2-person limit Hold'em? Arend ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/