Computer Chess Club Archives


Search

Terms

Messages

Subject: 6-man ending databases with pawn rams

Author: Bruce Moreland

Date: 23:58:40 11/09/98


If you assume a pair of pawns which are arranged to form a ram, a 6-man table
will be smaller than the corresponding 5-man table with one of the pawns
missing, because the two pawns behave like one pawn, and there are 4 fewer
squares to put pawns on (meaning that with one pawn you can take advantage of
a2..d2..d7..a7 -> h2..e2..e7..h7 symmetry, but with two pawns in a ram you don't
have to put any white ones on the 7th, because you can't have a black pawn on
the 8th).  So it ends up being like 5/6 the size of a normal 5-man table that
includes a pawn, and something like 1/64 the size of a 6-man table including a
pawn, and you don't need any 6-man sub-tables to create the thing.

For example it should be very simple to generate databases such as KRP v KBP
with the pawns rammed, on the order of a few days and under a gigabyte plus
sub-table storage, which would just be KR vs KBP, KP vs KBP, KRP vs KB, and KRP
vs KP, which is maybe another three gigabytes in puffy format (although these
could probably use disk compression).

I know that some of these ram cases have been done, but I'm not sure that the
general cases have been done.  I think there might have been something like
white pawn on a6 and white bishop versus black pawn on a7 and rook, although I'm
not sure exactly without looking it up.  I don't think they let the ram float
all over the board, I think they fixed it in one place, which would of course
mean that this table would be like 1/24 the size of a normal 5-man table with a
pawn.

Have all of the theoretically important cases been done?  And are any of them
likely to have an effect in a real game?

bruce



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.