Author: Dieter Buerssner
Date: 12:59:46 09/10/03
Go up one level in this thread
Not sure, whether this was mentioned already. am and bm upcodes take a list of SAN moves, that should be alphabetically sorted. Why the sorting? Now and then, one sees EPD records, which have both, am and bm. It makes it ambigious for testsuites (When is the position solved?). Perhaps a note, to only have one of those 2 upcodes would be in order. The other one could be used in a comment. For example the testsuite author may mean: this am loses soon, and should be avoided. Best seems this bm. The bm should then be in a comment. Or the author really means: bm is best, am is really horrible. So the am should be in a comment, and the position could be considered solved only when the bm is found. About the scores. It might also be interesting, to say a score is exact or an upper or lower bound. Also, it is not totally clear, what the search depth means. When I got a score for the currently best move at depth 10, but I did not search all moves at that depth yet. The ce scores, that show mate are horrible, IMHO. It is not easy for a human, to translate them to mate in. It would already be an advantage, to not base it on the "odd" numbers 2^^15, but on a decimal number like 30000. For a program, it is no problem to translate between thóse systems, but for a human ... About castling and FRC. Reinhard made some good suggestions, but I would think even cleaner would be, to use the column of the rook. So in normal chess, wqWQ would be ahAH. This would easily generalize to FRC positions. About e.p. target. I think everybody would agree, that it is better, when it is only set, when the pawn has a neighbour, than can capture. One should also define, what to do in the case, that the neighbour is pinned to the K, so that it cannot take. Perhaps, allow in this rare case to have ep set or unset. Regards, Dieter
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.