View Single Post
  #5  
Old February 14th 07, 05:48 PM posted to rec.games.chess.computer
Vincent Diepeveen
external usenet poster
 
Posts: 15
Default Generate and store EGTB's as a data tree

hi,

if you do 1806 * 48 * 47 then you will get positions with a pawn on a square
where a king is located already.

In diep's egtb scheme i'm avoiding that.

Vincent

wrote in message
oups.com...
Vincent,

I am not using 1806 * 24 * 23 ... and never said that I did (and by
the way, I think it should be 1806 * 48 * 47 ... , because you already
have mirrored, or else you wouldn't end up with 1806 KK positions).

The whole idea of my scheme is that there are *no* broken positions,
by definition. Of course I do account for the placement of previous
men. If a King is placed on a 'pawn square', then that square is not
available for pawns. Even more so: the sntmK cannot be in check, which
excludes even more squares for stm pawns, depending on the sntmK
position.

Stef

snip
Actually i might store pawns in less than what you describe.

you would use

1806 * 24 * 23 ...

Where i'm using less than that in fact.
I'm 100% avoiding positions with a 2 figures on the same square.
snip





Ads
 

Cheap Magazine - BabbFest - Property in Spain - Mortgage - AdSense Optimization Tutorials