Author: Heiner Marxen
Date: 14:18:20 12/08/01
Go up one level in this thread
On December 08, 2001 at 01:21:25, Mike S. wrote:
>On December 07, 2001 at 23:32:01, Dann Corbit wrote:
>
>>>[D]8/8/8/7p/P5p1/5n1N/1P3k2/7K b - - 0 47
>
>>Taking the knight is a mate in 23 (via tablebase) despite both sides queening
>>(...)
>> 35849404 1:36.4 Mat23 19t 1...gxh3 2.a5 Kg3 3.a6 Ne5 4.a7 Nd3 5.a8=Q Nf2+
>> 6.Kg1 h2+ 7.Kf1 h1=Q+ 8.Qxh1 Nxh1 {EGTB} {340}
>
>Interesting... it may have been unknown so far, that gxh3 would have won (too).
>At least I didn't know it :o).
>
>>The king move (on the other hand) prevents the other side from queening, but
>>takes longer to win.
>>(...)
>>19 2198 14206 84441737 Kf2-f1 Nh3-f4 g3 Nf4-g2 h4 Ng2-e3 Kf1-e2 Kh1-g2 Ke2xe3
>>Kg2-h3 Ke3-f2 Kh3-g4 g2 Kg4-f5 g1=Q Kf5-e6 Qg1-g4 Ke6-f6 Qg4xa4 Kf6-f5 Qa4-c2
>>Kf5-f4 Kf2-g2
Sorry Dann, but that is not correct. According to Chest this is a mate in 13:
PV: Kf1 Nf4 h4 Nd5 g3 Ne3+ Ke2 Kg2 Kxe3 Kh3 Kf4 Kg2 Ne1+ Kf1 Ke3 b3 Kd2 b4 h3
Kg1 Nf3+ Kf1 Ke3 b5 g2#
(K7/600, 350MB hash, 30MB EGTB-cache, 40.2 minutes)
Currently I have only a few 5-piece EGTBs, and all 4-piece tables.
I find this PV somewhat surprising, since there is no promotion in it.
Could someone try to verify this mate in 13, at least partially?
>Look at Black's third move...
>
>(Meanwhile I realise 1...Kf1 is probably one of these "right move, wrong reason"
>moves which programs play seeing only part of the idea. The third move may be
>even more difficult that the first - although many programs actually prefer
>1...Kf1.)
Now, a "shortest mate" is also a good reason :-)
OTOH, I would love to have a program which argues with ideas and plans,
and not search for the smallest solution in terms of depth,
but shortest in terms of the plan/idea/explanation length.
Seems to be out of reach, currently.
>Regards,
>M.Scheidl
Cheers,
Heiner
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.