Author: Ricardo Gibert
Date: 11:01:29 05/02/02
Go up one level in this thread
On May 02, 2002 at 13:13:45, Robert Hyatt wrote: >On May 02, 2002 at 12:36:13, Peter Fendrich wrote: > >>Please don't try to solve these positions! >>They are only intended to illustrate difficulties with the static exchange >>evaluation (capture on e4). >> >>[D]1k6/1qp5/2b5/8/4P3/3Q4/8/3K4 b - - >>My SEE function,with the help of the XRay function, correctly evaluates that >>Bxe4 earns a pawn. The same function would detect that Bxe4 is no good if white >>had bishop at c2. >> >>There are two different situations that my SEE can't detect because I didn't >>like the extra cost to always try detect these rare cases. >> >>First, my Xray function doesn't care about Xrays through opposite colors. >>[D]1k6/1qp5/2b5/8/4P3/3Q4/2B5/1b1K4 b - - >>In this position my SEE wouldn't detect that Bxe4 earns a pawn. The XRay >>function doesn't care about the bishop at b1. >>Does your SEE functions try to solve that? > >No... It isn't hard to do, but at present when I remove a piece from the >Swap() list, I only look for pieces of the same side behind it, which >breaks this. However, I suspect this is so rare that it doesn't count... >And rather than Bxe4 you can always play BxB first, then after the queen >recaptures, Bxe4 wins the pawn outright... The availability of Bxc2 is incidental e.g. [D]qk6/p1p5/P1b5/8/4P3/3Q4/2B5/1b5K w - - 0 1 Now Bxc2 gets mated, but you are right about such situations being very rare. > >> >>Secondly, my SEE doesn't try to evaluate where to start more than one piece of >>the same type can capture at the same time. i.e. two Pawns, Bishops, Rooks or >>Queens can capture at the same square. >>[D]1k6/1b6/1pq3q1/8/4P3/8/1K2Q3/Q3R3 b - - >>Here my SEE wouldn't know that Qc6xe4 is generally better than Qg6xe4. In the >>first case it opens up for the bishop to recapture the black queen. In the >>second case only the Q at c6 can recapture. >>With the same arguments as above I didn't try to solve the rare case when the >>same piece can capture. >>Does your SEE functions take care of this one? >> >>Peter > >No and I don't think this one is important at all. At least not important >enough to lose sleep over...
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.