Author: Ricardo Gibert
Date: 12:09:54 11/17/00
Go up one level in this thread
On November 17, 2000 at 14:58:28, Ricardo Gibert wrote: >On November 17, 2000 at 13:33:50, Heiner Marxen wrote: > >>On November 16, 2000 at 19:18:10, Ricardo Gibert wrote: >> >>>On November 16, 2000 at 17:24:37, Lenard Spencer wrote: >>> >>>>On November 16, 2000 at 04:25:53, Ricardo Gibert wrote: >>>> >>>>>On November 16, 2000 at 02:51:47, Tony Werten wrote: >>>>> >>>>>>On November 15, 2000 at 20:40:16, Lenard Spencer wrote: >>>>>> >>>>>>>This question may probably be best answered by the problemists, but if what I'm >>>>>>>thinking is correct, it may be possible to make looking for double checks go a >>>>>>>lot faster than the brute force approach of looking all over the board for more >>>>>>>than one checker. >>>>>> >>>>>>The way I use it: >>>>>>first, can the piece just moved attack the king (lookup table)? If so get the >>>>>>direction in which it needs to travel (same lookup table) and check if there are >>>>>>any other pieces blocking. >>>>>> >>>>>>second, can a rook or bishop attack the king from the fromsquare of the moved >>>>>>piece. If so get the direction, then travel from the king in the direction of >>>>>>the fromsquare until you go off the board (no discoverd check) or bump into a >>>>>>piece (if piece=rook,bishop,queen then it's a discovered check) >>>>>> >>>>>>if ( first and second) then doublecheck:=true; >>>>>> >>>>>>Tony >>>>> >>>>>How about this position: >>>>> >>>>>[D]8/8/7k/6pP/8/4B3/7R/7K w - g6 >>>>> >>>>>The move 1.hxg6 is double check, but it is not clear to me how your algorithm >>>>>catches this. >>>>> >>>> >>>>In this example, the pawn move delivers a double check, but the pawn itself is >>>>not a checking piece. But it does serve to illustrate just how tricky it can >>>>be. >>> >>>Yes, that was the point of my post. It is unusual, because it is a "double >>>discovered check" to put it more precisely. The loss of the Black pawn discovers >>>a check from the Bishop and the capture by the White pawn discovers a check from >>>the Rook. >>> >>>There is a similar "joker" concerning pins too, which a naive algorithm may >>>miss. >> >>Yes and no: an intended ep capture cannot be illegal by simply uncovering a >>check through the captured pawn: vertically the line (column) is not opened >>(the moving pawn closes it again), and diagonally undiscovering a check >>would imply that in the move before the other side could have captured >>the king instead of moving that pawn. >> >>Horizontally we can have a pin through *both* pawns (as well as an >>undiscovered check through both of them), but I cannot see how we could >>have two simultaneous pins for an ep capture. > >It is the horizontal pin that I had in mind. A naive (mistaken) algorithm would >fail to detect the existence of a pin. > >> >>E.p. captures are somewhat tricky, yes. >> >>Heiner Ah ha! Now I just realized what you thought I had intended. You were thinking I was implying that an analogous example to what I gave might fool a naive algorithm, but that is not true and was *not* what I intended at all! I thought your post was a little odd, but now I see *I* was the source of the confusion. Sorry. I just did not anticipate your interpretation which is very natural one. How funny.
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.