Author: Antonio Dieguez
Date: 15:59:55 12/29/01
Go up one level in this thread
On December 29, 2001 at 17:13:34, Severi Salminen wrote: >>After 1.e3 e5 2.Nf3 Nc6 3.Bb5 a6 >> >>in the qsearch, the expected material gain for Bxc6 is not at least a pawn? >>and if the d7 pawn doesn't exist, do you prune Nxe5? > >(Position below). SEE sees that material gain for Bxc6 is 0 as expected and it >will _not_ most likely be pruned, so deeper in qsearch the engine will see that >it can capture the pawn at e5. but you know what I mean... it depends on the window of course. > Yes, I prune Nxe5 if there is no pawn at d7 >because my SEE doesn't see pins. It does see batteries and screws (is this the >right word?) but no pins. Luckily pins are not _that_ common and the advantages >of SEE greatly outweight this drawback. yes that could be true. but I heard so much about these drawbacks, while some of them are easily solvable. That two target squares are "horsily connected"..., and one of the pieces is a horsy so you already start getting the identification of two conflictive captures, more deep elaboration is up to you. Amyan for example would not prune Nxe5 in the second case. The example is bad because axB but you know what i mean... let change the example to this one: [D]r1bqkb1r/pppp1ppp/2n2n2/1B2p3/8/4PN2/PPPP1PPP/RNBQK2R w KQkq - 0 4 :) and doesn't think Bxc6 wins nothing because the pawn at e5.(it does if the pawn were on g5 of course) I confess anyway that my implementations tends to be a little optimistic about the captures anyway. I have to review this. >About pins, > And in addition my engine easily >searches over 6 plies so the above situation gets resolved in full-width search >;) i was lazy to give a completely elaborated example :) be well.
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.