Hi Denis,

I wrote a simple java go program a couple of years ago that is very
basic non-tree monte carlo.   Is yours tree based or simple?

I don't quite understand your eye rule but the standard one that I think
most of us are using is this:

  an eye must ...

  1. be surrounded by own stones on orthogonal sides.

  2. if non-edge: opponent must have NO MORE than 1 stone on diagonal.

  3. if edge: opponent must have NO stones on diagonal.

I think there are many posts going back more than a year that give data
on how one of these programs should behave.  

I can check my numbers, but I use a fractional komi value although I can
turn all of this off I suppose.  


- Don





On Tue, 2008-10-07 at 17:33 +0200, Denis fidaali wrote:
> Hi, i have been searching a bit through the archive, and was not able to get 
> a complete list of the characteristics value for the light simulations.
> 
>  By light simulation i mean : Playing randomly the valid moves (No suicide 
> allowed), but not in the "true" eyes. I have a rule for defining a true eye : 
> all vertical and horizontal point around are the player's color. And 
> depending on the position :
>  - no edge around : NOT ( two diagonal or more are the opponent's color).
>  - one or two edge around : NOT ( one diagonal or more is the opponent's 
> color)
> 
> The engine is written in java, and run on a quad core Q9300 @ 2.50 Ghz.
> The code has been lightly optimized, and use pseudo-liberties to detect 
> captures.
> 
> Here is what i get :
> ---------------------------
> 
> mean score =2.086382709065067    (black win on average by 2.08 points)
> 78454.10819786233 Playout/sec       (Trying to take advantage of the 4 cores)
> Time=12.752397841                        (in seconds)
> Number of playout=1000478.0
> Mean moves per sim 111.05742754963127     (Number of move per simulation, 
> considering pass as a move)
> Mean moves(no pass) per sim 107.37764148736903 (Number of move per 
> simulation, not counting passes).
> 
> Note : as there are no seki possible in the simulation, the mean score for 
> the light simulations maybe bugged for seki, and still get the correct value.
> _________________________________________________________________
> Installez gratuitement les 20 émôticones Windows Live Messenger les plus fous 
> ! Cliquez ici !
> http://www.emoticones-messenger.fr/_______________________________________________
> computer-go mailing list
> computer-go@computer-go.org
> http://www.computer-go.org/mailman/listinfo/computer-go/

Attachment: 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/

Reply via email to