Automated Go Analyis(Monte Carlo Simulators)

[SmoothOper]

I wonder if there are any automated Go analysis simulators out there. I was thinking, it would be interesting to take a board position, run the simulations and see what the alternate moves such a simulator would suggest, as well as what the winning propensity for the move I chose was.

[hyperpape]

I believe that Crazy Stone may have some such capacities, including the ability to guess which moves may have been particularly bad. I don't know the details, and suspect it might only be for the desktop versions.

[Boidhre]

Some images from Crazy Stone analysing. You can probably do other stuff with it, I haven't used it that much to be honest.

Placing a board position and then letting analyse run for a few minutes (it's open-ended and this will be heavily dependent on your CPU). You get a "best move" tree as you can see: http://i.imgur.com/7z0VjeI.jpg



Here it's given a game record (a partial in this case from a recent game of mine), here you get alternative's to your move, the delta between the win chance of the AI's best move and your best move, this can be useful for finding potential errors though the AI won't explain why a move was bad, just say this move here is much better): http://i.imgur.com/BwYDsWe.jpg

I have to link to the images, the 800 pixels wide limit renders the text unreadable. Sorry.

[RBerenguel]

With some gtp tricking you can get pachi (or fuego, I don't remember which one could, it's one of these) to do something similar. But of course, it's CPU intensive and not that much useful, given these engines level