|(1) Posted by Hauke Reddmann [Thursday, Jan 25, 2018 15:00]|
Nalimov and castling/ep
Evidently it's not implemented - Ke1 Ra1, Kd8 Rb2 is the standard example. (See e.g. http://horizonchess.com/FAQ/Winboard/weaktablebase.html)
What is the lightest position where an e.p. move makes a difference? Of course I mean a retrolegal one, otherwise Kh3 Bc1 Pg4 - Kh5 Ph4 Pg6
or suchlike would count.
|(2) Posted by Jan Hein Verduin [Thursday, Jan 25, 2018 19:12]|
Not sure if this is what you mean, but as an opening bid this one seems unbeatable:
(= 2+2 )
white to move
With e.p. this is a draw, but without it 1.e4 wins.
(I don't have access to the Nalimov dbase and only to Shredder, and that one seems to know about e.p.)
|(3) Posted by Geir Sune Tallaksen Østmoe [Thursday, Jan 25, 2018 19:20]|
Nalimov does support ep. Do you have an example where it doesn’t?
|(4) Posted by Hauke Reddmann [Friday, Jan 26, 2018 01:20]|
But in this example, the double step *happens* and of course
the TB then has the e.p. answer memorized. It was more with
the pawn already on e4. (It would be a nightmare to program
retrolegal e.p.'s...But since those appear only at n>>7,
it's hardly relevant. I just wanted to know min(n).)
|(5) Posted by Branislav Djurašević [Friday, Jan 26, 2018 09:26]|
Lomonosov EGTB does support en passant!!
|(6) Posted by Torsten Linß [Friday, Jan 26, 2018 16:29]|
@Hauke:retrograde analysis is irrelevant for table bases. Retracting moves you can and will run into illegal positions even without ep captures. Positions in tables bases have a future only, but no past.
For example in the Lomonossow TB you find the following position which is clearly illegal
b7/1p6/8/8/8/K7/8/1k6 w - -
And the TB tells you that it is a mate in 19.
|(7) Posted by Joost de Heer [Friday, Jan 26, 2018 23:06]|
Positions in tables bases have a future only, but no past.
But if the past shows that the last move must've been a double step, the future must include the ep capture. And in some cases this will influence the result.
|(8) Posted by Hauke Reddmann [Friday, Jan 26, 2018 23:49]|
Yes, that was exactly what I meant (but I don't think there
is a retrolegal, gamechanging ep with onle 7 men).
|(9) Posted by Marjan Kovačević [Saturday, Jan 27, 2018 00:59]|
To check Nalimov ep you can see if it reports #4 in: 3N4/N7/k1B5/PpK5/8/8/8/8
|(10) Posted by Torsten Linß [Saturday, Jan 27, 2018 11:56]|
@Hauke & Jost: No. With say a white pawn on the fourth rank and black to play there are two positions stored in the TB. One is flagged "ep possible" the other one "ep impossible".
|(11) Posted by Joost de Heer [Saturday, Jan 27, 2018 12:05]|
Castling possible/not possible is not the same as castling legal/not legal.
|(12) Posted by Geir Sune Tallaksen Østmoe [Saturday, Jan 27, 2018 14:17]|
I can’t see a 6- or 7- piece position where it makes a difference in a game/study. This is the best I can do:
(= 5+3 )
|(13) Posted by seetharaman kalyan [Saturday, Jan 27, 2018 18:39]|
This is a smart idea !
|(14) Posted by Siegfried Hornecker [Friday, Feb 2, 2018 08:48]; edited by Siegfried Hornecker [18-02-02]|
EDIT: Oops, this is more about en passant. Still leaving this here as original source for Hauke's position.
(= 3+2 )
A. S. Selezniev, Tidskrift för Schack 1921
White to move and win
No more posts
MatPlus.Net Forum Internet and Computing Nalimov and castling/ep