[computer-go] Details of AnchorMan

[Don Dailey]

GenericMC_300K     1455.0 
GenericMC_200K     1391.6  
GenericMC_10000    1142.9 
GenericMC_100K     1453.9 

In my implementation I do 10,000 play-outs from the root so
they are not perfectly distributed between moves.  

I do check for simple ko in the simulations but when the
simulations are done and it's time to choose a move, I
veto any moves that are position super ko.

So I think your description is correct except possibly the
simple ko situaiton.

- Don



On Sat, 2007-02-03 at 14:22 -0500, Chris Fant wrote:
> Don,
> 
> I put new player (StonedAgain) on cgos and although it didn't
> establish a rating, it won very few games.  So I'm going to try and
> make it do basically the same thing as some other program on cgos and
> see what happens.  I'll try to replicate GenericMC_10000, (which you
> discussed on this list but I no longer see on cgos).  I just want to
> make sure I have the details right.  It does 10k simulations.  Now,
> that is 10k from the root as opposed to 10k from each leaf of a 1-ply
> search, correct?  And it doesn't fill single point eyes defined as "an
> empty point surrounded by friendly stones on the orthogonals and no
> more than 1 diagonal enemy stone - unless the empty point is on the
> edge, in which case it must not have any diagonal enemies".  It
> doesn't check Ko in the random sims.  I should expect a rating of
> about 1143?  Was there anything else?
> _______________________________________________
> computer-go mailing list
> computer-go at computer-go.org
> http://www.computer-go.org/mailman/listinfo/computer-go/