Author: Les Fernandez
Date: 10:17:50 01/01/03
I am interested in hearing comments about an approach that I think has the ability of saving time in EGTB generation. I call it a substitution technique where we substitute various pieces (of the opposing side) at the target square. Target square is defined as the final location of the sliding piece. Let me expalin. If the pv for a particular epd position is Re4 we have 2 possible scenarios. Either e4 is blank in which case we can substitute all of the following pieces (P-N-B-R-Q)onto the target square. Although I have included the pawn it may be best to forget about them in this substitution technique since there appears to be to many rules involved as to whether or not the position is reachable or legal. The above technique implies the following. Take for example the following 3 piece EGTB position assume no piece on target square and that the pv=Re4 as the solution: [D]2k5/8/8/8/8/8/4R3/K7 - - w Now using the above technique you would generate 5 new positions for use in the 4 piece EGTB where the only difference is that captures are involved: [D]2k5/8/8/8/4p3/8/4R3/K7 - - w [D]2k5/8/8/8/4n3/8/4R3/K7 - - w [D]2k5/8/8/8/4b3/8/4R3/K7 - - w [D]2k5/8/8/8/4r3/8/4R3/K7 - - w [D]2k5/8/8/8/4q3/8/4R3/K7 - - w Notice that in all 5 cases above the target square is unchanged but only its contents. Had the target square been occupied by any piece then we do the same thing with a minor difference. Say for example that the e4 square is occupied by a black rook, then the substitution we would do is replace that piece by all of the following: blank-p-n-b-q Notice that additon of the blank square which again does not change the results. By applying this substitution method (even if we do not include pawns) we would be indirectly generating 4 new positions for the n+1 EGTB where n=number of pieces in the EGTB. So if we are generating positions for the 3 piece EGTB for every position we generate for this set we are actually generating 4 new positions for the n+1 EGTB. Be interested in hearing all your thoughts about this apporach and perhaps other techniques might be found. Thanks, Les
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.