Subject: Re: when tablebases cause blunders


Post Followup (without quoting) ] [ Post Followup (with quoting) ] [ Computer Chess Resource Center ] [ Message Listing ]

Posted by GuyHaworth (Profile) on May 30, 2002 at 20:13:45:

In Reply to: when tablebases cause blunders posted by Robin Smith on May 29, 2002 at 02:34:34:

I addressed this question of endgame optimisation 'constrained by the 50m rule'
in the ICGA Journal, Vol. 23.1.

One can 'safeguard' use of DTC or DTM EGTs by using a DTZ EGT, ensuring that at
least the first phase is not too long.

However, this - far from being "the only solution" as one contributor put it -
is not always a solution at all.

Cases can be found in KQPKQ where optimising DTZ runs the game into an overlong
2nd phase:  Mednis had one in one of his books [see ICGA_J v24.4 (2001)].

I suspect that KBBKNN harbours examples where safeguarding, but otherwise
optimising on DTC or DTM, does not work either, an overlong 2nd phase being the
result.


My proposal was for EGTs to the so-called 'Depth by The Rule' DTR EGTs, where
'depth' is the minimum k-move rule under which there is an unavoidable win.

Until they arrive - don't hold your breath - certainly the best solution is to
generate DTZ EGTs and use them to safeguard use of DTC and/or DTM EGTs.

G




Post Followup (without quoting) ] [ Post Followup (with quoting) ] [ Computer Chess Resource Center ] [ Message Listing ]