Re: On Game Space Size (Monte Carlo)

[Mousheng Xu <xu@xxxxxxxxxx>]

At 11:14 AM 4/28/99 +-200, Pieter Cuijpers wrote:
>Is there anyone who can verify the above for me ???

Pieter,

	I don't quite know what you do with the map. But this is what I think:

	We are trying to figure out what portion of all board configurations is
illegal so that we can have an idea about how much calculation we can save
for brute-force searches. In a brute-force search tree, there is no
discrimination about what board configuration is more likely to happen in a
real game. So I would think, say, for a 10 x 10 board, we may sample
	1 stone enough times 
	2 stone enough times
	3 stone enough times
	...
	100 stone enoguh times 

	We do want to know the differences among different numbers of stones,
because they mean different depths on the search tree. The earlier an
illegal position is found, the more calculation you can save.
	So if you could abtain a 3-D table (size_of_board, stones_on_the_board,
%_of_illegal_positions), then you would make a significant contribution to
the computational go theory. :)
	After you are done, please post your table to the list.
	Thanks a lot.

-- Mousheng

	

>
>Greetz. Pieter
>
>
>
>