Author: leonid
Date: 05:14:32 02/13/01
Go up one level in this thread
On February 13, 2001 at 00:16:07, Uri Blass wrote: >On February 12, 2001 at 18:40:38, Heiner Marxen wrote: > >>On February 12, 2001 at 15:47:55, leonid wrote: >> >>>On February 12, 2001 at 13:11:53, Heiner Marxen wrote: >>> >>>>On February 12, 2001 at 07:29:02, leonid wrote: >>>> >>>>>Hi! >>>>> >>>>>If you like to solve a mate, try this position: >>>>> >>>>>[D]rQ1Q1rnk/q4qbp/3nQqbB/q1Q1NnNn/5QQ1/b2Q4/1QB3Q1/1K1R3R w - - >>>>> >>>>>Please, indicate your result. >>>>> >>>>>In dispite that this position is not that deep and difficult, if your program >>>>>don't know how to deel with white crowd it could be in serious trouble. My was >>>>>deadly slow in searching for mate even by selective search. Reason - white have >>>>>142 initial moves and goes beyond 150 during the search for mate. Good challenge >>>>>for every chess program! >>>>> >>>>>Thanks for responding, >>>>>Leonid. >>>> >>>>Hmmm, not deep? Not difficult? "There is no mate in 6" by Chest burns >>>>nearly 18 minutes on a K6-3/400 with 50MB hash. I have started depth=7, >>>>but I expect at least 3 hours to complete it, and may be more. >>>>At least I can confirm the 142 initial legal moves. >>>> >>>>Heiner >>> >>>Thanks, Heiner! You are very close! Mine found mate in 8 by selective. It was >>>long. Took 1 min 17 sec. >>> >>>It is simple since there are not that many moves before the mate and starting >>>move (for 8 moves) is checking one. >>> >>>Leonid. >> >>Phew, that was hard work: after 5.3 hours (K6-3/400, 50MB hash) Chest finds >>a unique key for a mate in 7. Here is a PV: >> >>Ngxf7+ Nxf7 Nxg6+ hxg6 Bxg7+ Nfxg7 Qgxh5+ Nxh5 Qfxf6+ Ngxf6 Qdxf8+ Kh7 Qgxg6# >> >>What about the playing programs? Normally they are much faster to _find_ >>a mate. Did nobody try it, or is this one also hard for them? >> >>Heiner > >My Junior5.9 found mate in 8 in a short time. > >Main line 1.Ngxf7+ Qaxf7 2.Bxg7+ Nfxg7(or 2...Q7xg7 3.Nxg6+ hxg6 4.Qxh5+ gxh5 >5.Rxh5+ Qfh6 6.Rxh6+ Ngxh6 7.Qxf8+ Kh7 8.Qfxg7#) 3.Qfxf6 Rxb8 4.Nfxf7+ Nxf7 >5.Qxg7+ Nxg7 6.Rxh7+ Kxh7 7.Qexg6 Kh8 8.Qh7# > >The part begins with 2...Q7xg7 is the main line of Junior after I give it only >the position after 1.Ngxf7+ Qaxf7 > >I guess that it may be possible to learn from Junior's mistakes in this line >about the rules of the selective search that Junior is using. > >Junior stops to search after finising an iteration when it finds a mate for >itself so I need to give it a position with black to move if I want it to find a >shorter mate. > >Time of Junior on p200MMX 1:40 It look like Junior is very good in mate solving. Its time should be very close to mine. My computer is quicker (400Mhz) but program use no hash. For 8 moves hash is already very efficent. This position also, with its whites moves hitting very often beyond 150 moves limit, say that Junior is very solid. Leonid. >19496 knodes were searched. > >Uri
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.