Author: Bas Hamstra
Date: 06:41:07 01/11/02
Go up one level in this thread
On January 11, 2002 at 06:33:00, Ulrich Tuerke wrote: >On January 11, 2002 at 04:59:14, Tony Werten wrote: > >>On January 11, 2002 at 04:43:58, Ulrich Tuerke wrote: >> >>>Hi Bas, >>> >>>>... >>>>- only recaps to same square >>>>- the cap-recap sequence must have same material gain >>> >>>Why this ? >>>I'd thought, that the idea is to follow forced lines deeper (e.g. capture and >>>subsequent re-capture of a pawn). Otherwise the program could use these >>>recapture sequences to shift events beyond the horizon. >>> >>>Thus, I would rather think of sequences, which do not disturb the material >>>balance. >> >>That's an addition that limits the extensions even more: Material ballance must >>be higher than alfa-2 pawns and smaller than beta+2 pawns. > >alfa-2 doesn't imply a material gain at all, but could be very well a loss. >So, I guess that Bas does it in another way than you. > >OTOH, your condition seems quite intutive to me, but it was stated: >" the cap-recap sequence must have same material gain " > >> >>> >>>When you restrict these to material wins, it seems to me that you are thinking >>>of avoiding traps ? >> >>He doesn't. He means that the captures must have the same value. ie oppo takes >>knight, I take bishop, or oppo takes rook I take rook. > >But Bas told differently, didn't he ? That's not just a material gain. > >Ciao, >Uli >> No, I meant exactly what Tony describes. Precisely what Crafty does, in fact. But I have experimented with many different versions. You can skip the condition of equality, and solve many combinations faster. But in games I did not like it. I even tried capture extensions (0.25 ply fractional). I remember it scored a lot of draws then, for some reason. Tactically it could not be fooled, but it lacked the depth to accomplish something useful itsself. Even 0.25 for a capture costs enourmously in terms of positional depth. Best regards, Bas.
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.