Author: Marc Bourzutschky
Date: 19:56:31 01/07/03
Go up one level in this thread
On January 07, 2003 at 22:07:48, GuyHaworth wrote: > >Always good to hear Marc's EGT news, and KRRPKQ is particularly intriguing. > >I recall Marc's 'KBBPKR EGDB' of Dec 14, 2002 [and since the reference number of >the posting is not including _in_ the posting by the system, I can't quote that >and cite the posting more easily]. > >It doesn't include any line showing a 130-move 'record', so maybe I missed >another posting that Marc is referring to. It does show a DTM=150 position >where White only restarts the move-count as it clocks up 107/107 on his move >108. > >I don't know what sort of 'record' this might have been. KRNKNN has the record >maxDTC = maxDTZ depth at 243. Among pawnless endgames KQPKQ and KNNKP have >maxDTC = 114, and I think KQPKQ's maxDTZ is ~70, recalling Ken Thompson's >earliest work. The maxDTZ for KNNKP is not known. > >As RG has just pointed out, a DTM-minimaxing line does not say anything explicit >about the DTZ depth where DTZ == 'Depth to (move-count) Z(eroing) move'. While >the minimaxing lines for different metrics (DTM, DTC, DTZ and DTR) often appear >to be very similar, I don't think one can make any general inferences that they >always are. > >I guess we can be reasonably confident that the 6-man P-endgames are setting >maxDTZ-records, even though we don't know what maxDTZ has been in 3-5-man >P-endgames, and don't know exactly what it is in 6-man P-endgames. Maybe this >is the record Marc is referring to. > >[ Incidentally, it is I think easy enough - for a programmer, not me - to change >EN's code to the DTC instead of the DTM metric, but I don't think anyone has >done it yet. If they have, I'm in the market for KQPKQ DTC. ] > > >What are the '$n' figures after some moves, btw? > >g I did not post the maximal KBBPKR, just a very large one. Using DTM or DTC it is indeed not possible to compute an exact DTZ, but one can set a lower bound by finding a sequence of "only moves" as defined by John Nunn. While I have not done this explicitly, just eyeballing some of the sequences I'm pretty sure there are more than 100 moves. But I think even a long sequence in DTM without pawn moves or captures is an interesting metric to describe the complexity of an ending, even though it is not DTZ (and I did not claim it was). I think a "true" DTZ is of interest mainly in endings that occur reasonably often in practice and are clearly borderline to the 50-move rule. KQPKQ and KQKRP are probably the most relevant examples. The '$n' are comment symbols (!, +-, etc) as defined in the PGN standard. A good PGN reader should be able to translate. -Marc
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.