Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / E.P. standard EPD designation should be altered for clarity
- - By user923005 (****) Date 2019-12-27 02:05
Not only should superfluous e.p. fields be eliminated, but also where the e.p. capture is impossible like these:
rn1q2nr/5pB1/p7/PppP1kpP/6b1/1Q1K3N/1PP1Pb2/RN3B1R w - c6

r3kb1r/Nppn2pp/8/8/3Ppq2/8/PPP1Q1PP/R1B1K2R b KQkq d3

r1b1kb1r/ppp2ppp/2n5/1B1q4/3Pp3/2P5/PP2QPPP/RNB1K2R b KQkq d3

r1b1kb1r/ppp3pp/2N2n2/1B6/3Ppq2/8/PPP1Q1PP/R1B1K2R b KQkq d3

r3kb1r/1ppn2pp/8/1N6/3PpP2/8/PPq1Q1PP/R1B1K2R b KQkq d3

r3kb1r/Nppn2pp/8/6q1/4pP2/8/PPPPQ1PP/R1B1K2R b KQkq f3

r1b1kb1r/ppp3pp/2N2n2/1B4q1/4pP2/8/PPPPQ1PP/R1B1K2R b KQkq f3


The reason is that you get redundant records in the database, where there is no difference whatsoever in game play, so the record is effectively duplicated, causing a doubling of effort to analyze those redundant positions.
Statistically, the number of false e.p. positions is close to 100%, due to nonsense like this:
rnbqkbnr/pppppppp/8/8/3P4/8/PPP1PPPP/RNBQKBNR b KQkq d3
Parent - - By Peter Grayson (****) Date 2019-12-27 19:57
Whether FEN or EPD some GUI's already ignore a non existant E.P. position.The Chessbase GUI's dropped it after Fritz 10
After 1.e4

Fritz 8 GUI
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1"

Fritz 9 GUI
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1"

Fritz 10 GUI
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1"

Fritz 11 GUI
"rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"

Fritz 14 GUI
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - 0 1"

Deep Shredder 10.1 GUI
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq -"

HIARCS Chess Explorer
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - 0 1"

Arena 3.5 (FEN)
"rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1"

Peter
Parent - By user923005 (****) Date 2019-12-28 05:33
Yes, there is a mix of programs that ignore the standard and do the right thing and programs that follow  the standard.
Parent - By bob (Gold) Date 2020-01-07 18:15
When I first started on Crafty, I ran into this from the reverse direction.  Crafty has always been able to output FEN/EPD position descriptions.  And when EP was impossible, I produced a "-".  Stephen contacted me one day telling me that my FEN output was wrong, that anytime one side made a double pawn move, the EP target should be set to the first square the pawn passed over.  I obviously asked "why" and he replied "that's the way it has been since this was formalized and it would be bad to introduce a modification that could break older programs.

I chose to not output the target, and to always check to see if EP is actually possible.  I don't check (IIRC) to make sure it is actually legal (IE no pins, etc) but I do check for the enemy pawn on the right square (and if importing PGN it is even easier).  I also chose to ignore any EP target that was bogus (IE no pawn on the 4th rank on either side of the target).  I've run into a few quirks over the years, but this seemed (to me) to simply be a flaw in the specification.  There was talk of him modifying this several years later.  I don't know whether he fixed this prior to his death or not.

It was one of those "legacy issues" that haunts us forever.

EXACTLY like 0-0, O-O, o-o, Kg1, etc.  Also promotions (e8Q, e8q, e8=q, etc).

I found it better to simply handle all of them quietly.
Up Topic The Rybka Lounge / Computer Chess / E.P. standard EPD designation should be altered for clarity

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill