Author: Uri Blass
Date: 14:40:59 10/18/02
Go up one level in this thread
On October 18, 2002 at 17:36:24, F. Huber wrote: >On October 18, 2002 at 17:09:07, Uri Blass wrote: > >>On October 18, 2002 at 17:00:21, F. Huber wrote: >> >>>On October 18, 2002 at 15:33:11, Uri Blass wrote: >>> >>>>On October 18, 2002 at 14:42:55, Heiner Marxen wrote: >>>> >>>>>On October 17, 2002 at 01:09:45, Jouni Uski wrote: >>>>> >>>>>>And other programs as well. >>>>>> >>>>>>[D]rnbQKbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBqkBNR w - - >>>>>> >>>>>>Alybadix 2002 solves this in unbelievable 21,3 sec (Athlon 1466 Mhz). Look at >>>>>>http://www.saunalahti.fi/~iblom/alybadix/ex2002.htm >>>>>> >>>>>>Jouni >>>>>> >>>>>>PS. Solution of course Nc3! >>>>> >>>>>That is correct. Chest on Athlon 1500+ with 128 MB hash needs 2 hours: >>>>># 5 0.03s [ 3.00] 5kN [ 4.90] 1.36 337- 0 >>>>># 6 0.09s [ 3.00] 19kN [ 3.57] 1.68 1164- 0 >>>>># 7 0.37s [ 4.11] 73kN [ 3.89] 1.93 6250- 0 >>>>># 8 1.92s [ 5.19] 386kN [ 5.29] 2.32 32241- 0 >>>>># 9 17.29s [ 9.01] 3472kN [ 8.98] 2.32 312342- 0 >>>>># 10 110.95s [ 6.42] 22525kN [ 6.49] 2.53 1893435- 14248 >>>>># 11 1229.78s [ 11.08] 229968kN [ 10.21] 3.04 20799132- 17599899 >>>>># 12 7333.75s [ 5.96] 1405765kN [ 6.11] 3.05 117945772- 114746539 >>>>> >>>>>PV: Nc3 Nf6+ Kxf7 Kxf1 Nf3+ Kxg2 Rxd1 Nc6 Rg1+ Kxf2 Nd1+ Kxe2 Nd4+ Nxd4 Qxc7 d6 >>>>>Qc4+ Kf3 Qd3+ Kf4 Qe3+ Kf5 Rg5# >>>>> >>>>>I'm not sure whether the comparison with Alybadix is completely fair. >>>>>On that web page I read (in red) /Q3+ which seems to indicate a search >>>>>restriction. Hence I suspect a shorter mate is not absolutely ruled out >>>>>by Alybadix. >>>> >>>>I understood from the site that it means that white has to play only checks or a >>>>threat of mate in at most 3 moves. >>>> >>>>Uri >>> >>>And that´s exactly the point! With such restrictions you are able to get >>>almost _every_ time (you want) on a special problem - but that doesn´t >>>say _anything_ about the average solving times _without_ giving the program >>>such additional informations. >>>And I really would be interested, _who_ would know exactly _this_ restriction >>>_before_ solving this chess problem? (If knowing the solution before, why >>>using a mate solver at all ;-)) >> >>Knowing is impossible but it is possible to guess. >>I think that having an option to restrict the search is productive. >> >>Of course comparing times of it with times of chest is meaningless. >> >>Uri > >Ok, you´re right with this ´guessing´ and it´s also good for trials, >but if you do so, you _never_ can be sure to get the shortest mate! >And exactly _this_ is what I want, if I´m using a matesolver - otherwise >I could use _every_ good and fast chess engine and would probably also >get a solution (maybe in even faster time). > >Franz. If you do not look for the shortest mate then using a commercial program is not always an advantage. Commercial programs may not care to be fastest in finding mates and prune when they see that they win a lot of material. Chessmaster does not do it but it still cannot find very long mates because it's search is limited. I believe that part of the positions in the site are not going to be solved by the commercial programs. 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.