Re: [computer-go] Small wish for Cgos

version is better, but neither is appropriate - it's not the intended use of version or name. I'm a little skeptical of this feature. It seems like this functionality should be handled by a more general mechanism to configure your bot (above and beyond just a name and password.) For instance

Re: [computer-go] Small wish for Cgos

I would prefer to use an optional command such as "describe" or "info" or something e.g. "cgos-info" My program can be invoked with a command line option that describes it - it's basically a change-log entry compiled into the program.This is a similar idea but it probably wouldn't be a chang

Re: [computer-go] Small wish for Cgos (was: use for Monte Carlo on 19X19?)

It's probably better to use the GTP version command instead of the name command. On Thu, 2007-11-08 at 04:11 +0100, Heikki Levanto wrote: > On Wed, Nov 07, 2007 at 03:26:37PM -0800, Christoph Birk wrote: > > CGOS already uses the 'name' feature. You send it (along with a > > password) at login. >

Re: [computer-go] Small wish for Cgos (was: use for Monte Carlo on 19X19?)

On Wed, Nov 07, 2007 at 03:26:37PM -0800, Christoph Birk wrote: > CGOS already uses the 'name' feature. You send it (along with a > password) at login. No, I am talking of two different things. What I send at login is the name cgos uses for my program. This is what I put on the cgos command line.

Re: [computer-go] Rapid action value estimation

On Wed, 2007-11-07 at 14:34 -0800, Christoph Birk wrote: > What is the difference between 'hb-678-UCTRAVE-10k' and 'hb-675-UCT-10k'. It's probably obvious, but UCTRAVE uses RAVE instead of just (tuned) UCT. ___ computer-go mailing list computer-go@com

Re: [computer-go] use for Monte Carlo on 19X19?

on 07.11.2007 07:35 Gunnar Farnebäck wrote: > Stefan Mertin wrote: > > I am using GnuGo scoring in my tournaments. > > But GnuGo 3.7.10 mostly doesn´t score seki correctly, > > has this been revised for v3.7.11 ...?! > > What scoring mode are you using? > > /Gunnar SORRY - I was completely wrong

Re: [computer-go] Small wish for Cgos (was: use for Monte Carlo on 19X19?)

On Wed, 7 Nov 2007, Heikki Levanto wrote: Actually, gtp already has a command 'name' that returns the program name. It would be helpful if the cgos script would ask the programs name (if it supports it), and pass that to the server. The server could then display it on the cross-table page for tha

Re: [computer-go] Rapid action value estimation

On Wed, 2007-11-07 at 14:34 -0800, Christoph Birk wrote: > On Mon, 5 Nov 2007, Jason House wrote: > > I implemented this yesterday. In doing so, I realized I didn't know the > > proper way to initialize new leaves in the UCT tree. MoGo papers seem to > > talk about a progression from always pick

Re: [computer-go] Rapid action value estimation

On Mon, 5 Nov 2007, Jason House wrote: I implemented this yesterday. In doing so, I realized I didn't know the proper way to initialize new leaves in the UCT tree. MoGo papers seem to talk about a progression from always picking an unexplored leaf (AKA using infinity for the upper confidence bo

Re: [computer-go] Small wish for Cgos (was: use for Monte Carlo on 19X19?)

On Wed, Nov 07, 2007 at 09:03:22AM -0500, Jason House wrote: > I've been thinking about the same feature. I wasn't specifically thinking a > hyperlink, but certainly a string with far more than 18 characters. Another > candidate is to have commands that query the engine and display it as > commen

Re: [computer-go] Solving Go

Figured that would be part of the cleaning phase before the evaluation. It's a lot easier to permute the entire set than it is to generate all legal ones at first (at least with smaller sizes). -Josh On 11/7/07, John Tromp <[EMAIL PROTECTED]> wrote: > > I just ran my perm application for 4x4 and

Re: [computer-go] Solving Go

> I just ran my perm application for 4x4 and it's reporting > 43,046,721 unique board states and took 2m6.980s. Will try for 5 and 6. seems you're computing 3**(n*n) 3**16 = 43046721 3**25 = 847288609443 3**36 = 150094635296999121 don't you want to exclude illegal positions? -john

Re: [computer-go] Solving Go

I just ran my perm application for 4x4 and it's reporting 43,046,721 unique board states and took 2m6.980s. Will try for 5 and 6. -Josh ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/

Re: [computer-go] use for Monte Carlo on 19X19?

i literally mean the algorithm that would have nothing to say about any of those stones. (i.e. wouldn't declare any element of any of them to be in either of the two states that i had described). s. - Original Message From: "[EMAIL PROTECTED]" <[EMAIL PROTECTED]> To: computer-go@compute

Re: [computer-go] Go on board with no edges

any unconditional safety for a stone would likely create massively unfair play. sure, both players could play on these safe spots, but my hunch is that it cripples the game. scatter some stones near the edge, now try to prevent me from connecting to them -- this sounds a lot like a game i played

Re: [computer-go] use for Monte Carlo on 19X19?

steve uurtamo said: > i wonder what is known about the set of unconditionally > dead and unconditionally living groups. there must be > something like a small and extremely fast mechanism for > this. what is everyone using? i mean a mechanism that > is independent of any fancy data structure tha

Re: [computer-go] Go on board with no edges

Usually when people do this they play on a torus, made by connecting opposite sides of the board (out one side continues by entering the other side). Otherwise any connection to the "open edge" is automatically alive, and that seems too easy. Cheers, David On 7, Nov 2007, at 12:14 PM, Do

Re: [computer-go] Solving Go

Álvaro Begué wrote: > > > On 11/7/07, *Don Dailey* <[EMAIL PROTECTED] > > wrote: > > [...] > To go beyond 5x5, say 7x7 would require an endgame table with 3**49 > entries or 239299329230617529590083 entries. This can be reduced by > about 8x if you remo

[computer-go] Go on board with no edges

Is go on a board with no edges interesting? I don't mean an infinite board, I just mean a board where moves to the edges are prohibited. So a stone next to these edges are always safe.You could play this game on a regular board and just pretend there was an additional intersection just outs

Re: [computer-go] Solving Go

On 11/7/07, Don Dailey <[EMAIL PROTECTED]> wrote: > > [...] > To go beyond 5x5, say 7x7 would require an endgame table with 3**49 > entries or 239299329230617529590083 entries. This can be reduced by > about 8x if you remove symmetrically equivalent positions.This is > pretty intractable, won

Re: [computer-go] Solving Go

I've put a lot of thought into this. 5x5 is about the largest feasible board size (currently) for creating an "endgame database" which is a table of all possible positions with the resulting score.I don't know if anyone has done this, but I know that this board size can be solved with brute f

Re: Re[2]: [computer-go] use for Monte Carlo on 19X19?

I've been trying to think about what I really want and how it would be achieved. What I'm interested in is the ability for bots to assign an accurate final score to a game as early as possible. When trying to think how to make this into a competitive challenge, I tried to consider both bots submi

Re: [computer-go] Solving Go

On 11/7/07, Joshua Shriver <[EMAIL PROTECTED]> wrote: > > You could go the route chess does with egtbs. That won't work for go. First of all, chess and checkers are games where the position on the board becomes simpler at the end of the game, so many games will end up converging on positions for

Re: [computer-go] Solving Go

You could go the route chess does with egtbs. First permutate every possible board with both stones, and start removing pieces one at a time. I wrote a permutation tool, but even with 3 and 4 size board the end data file was huge. -Josh On 11/7/07, Ben Lambrechts <[EMAIL PROTECTED]> wrote: > >

Re: [computer-go] Solving Go

This was posted a while ago on here by Terry Here are the links: http://senseis.xmp.net/?7x7BestPlay http://erikvanderwerf.tengen.nl/5x5/5x5solved.html I hope that is a little bit helpful... On 11/7/07, Ben Lambrechts <[EMAIL PROTECTED]> wrote: > > I want to create a perfect player on boar

Re: [computer-go] Solving Go

I think how 5x5 was solved was part of a thesis. I don't have the URL or author handy. 7x7 is semi-solved by a bunch of dans, but no formal proof of its correctness is available. I believe that past work used brute force alpha-beta with transposition tables. 5x5 was made tractable by detecting

Re: [computer-go] Re: use for Monte Carlo on 19X19?

and in typical endgame boards on CGOS? s. - Original Message From: Dave Dyer <[EMAIL PROTECTED]> To: computer-go ; computer-go Sent: Tuesday, November 6, 2007 6:50:50 PM Subject: [computer-go] Re: use for Monte Carlo on 19X19? > >the idea is: identify at least one stone from every u

[computer-go] Solving Go

I want to create a perfect player on board sizes 3x3, 5x5 and maybe 7x7 and beyond. But I have no idea how to start. How do I create the move database and how do I select the perfect move for every position? I know Go is solved on boards 5x5 and smaller, but there is no program that plays by t

Re: [computer-go] [OT] All-integer scalable distribution algorithm.

In message <[EMAIL PROTECTED]>, Mike Hill <[EMAIL PROTECTED]> writes Folks... First, let me say how much pleasure my reading of this list has given me. I love that folks are out there cranking on this problem. Truly, it's one of the great problems. I have a rather strange request. I am a

Re: [computer-go] Small wish for Cgos (was: use for Monte Carlo on 19X19?)

I've been thinking about the same feature. I wasn't specifically thinking a hyperlink, but certainly a string with far more than 18 characters. Another candidate is to have commands that query the engine and display it as comments in the games On Nov 7, 2007 2:18 AM, Heikki Levanto <[EMAIL PROTE

[computer-go] Video of Sylvain Gelly talk at NIPS

Hi All, I have been reading this list for a while, but I haven't really posted yet. However, I found a link to a video of Sylvain's Gelly's Mogo talk at NIPS which I think a lot of you would be interested in watching. There are also some other talks on UCT and related topics that are easy to find