Hi!
On Mon, Jan 18, 2010 at 07:14:59AM +0000, [email protected] wrote:
> Has anyone tried doing pachi/Fuego + GnuGo hybrid slightly in way
> Many FAces is done?
>
> If I understood correctly Manyfaces is mostly a plausible move
> generator. And serac is widened via the RAVE.
>
> So simplest hybrid could rather simple words that often used before
> huge effort taking for ever. When a new node is initialized:
> Contact GnuGo using GTP and load situation,
> send genmove,
> send topmoves and you have a starting point for simulations.
I'm not aware of such a project but it has been on my mind for some
time as well. I think this would be a great thing to try; I also plan
to add simple support for biasing only root node or nodes that already
have large number of visits, exactly to support expensive evaluation
functions - I guess it would be useful for this as well. Asking GnuGo
for evaluation of all moves would be awfully expensive, I think.
I'm not sure if MCTS ManyFaces uses the old evaluation function only
for "top moves", it would be great to share other information like L&D
and semeai critical moves; perhaps GNUGo even provides interface to get
these as well.
> Obviously this is not what GG was designed for and will do loads of
> overhead but could be interesting. I would have tried I but I could
> not get Pachi to compile under Cygwin and currently I do not a linux
> machine. Also Fuego had some problems building under cygwin -and has
> steeper learning curve. Also it would slow under cygwin. At least
> Pachi multicore due to threading.
I'm sorry, I cannot help much with making Pachi work on Cygwin work as
I'm only very little familiar with Windows or Cygwin - but if you get it
to work (please send a patch if any changes are required!), I can help
you out with troubles you would encounter from there on.
Not too long ago, Ben Lambrechts has been able to compile Pachi on
Cygwin, perhaps he could share some tips? But maybe something changed
in Pachi in the meantime that broke the Cygwin build...
Petr "Pasky" Baudis
_______________________________________________
computer-go mailing list
[email protected]
http://www.computer-go.org/mailman/listinfo/computer-go/
