Author: Uri Blass
Date: 11:27:10 10/09/01
Go up one level in this thread
On October 09, 2001 at 14:21:18, Angrim wrote: >On October 09, 2001 at 13:01:13, Uri Blass wrote: > >>[D]8/1P6/8/5k2/2K4p/7r/1r4p1/6R1 b - - 0 61 >> >>This position is from the game Tao-Deep Junior7 in the last WMCCC >> >>Chest need some hours on p200 to see mate in 10(I used chest by the way that was >>explained by paul some days ago >> >>see http://www.icdchess.com/forums/1/message.shtml?191857 >> >>Solution:61...Rg3 62.Kd4 h3 63.Kd5 h2 64.b8Q Rxb8 65.Re1 Rd3+ 66.Kc4 h1Q 67.Re5+ >>Kxe5 68.Kxd3 Qd1+ 69.Kc3 Rb3+ 70.Kc4 Qc2# >> >>15961 seconds,292,407,791 nodes >> >>In the game Deep Junior7 missed the right move(Rg3) but it did not help Tao that >>resigned after Rxb7. >> >>How much time does your program need to see mate in 10 with the full 5 piece >>tablebases and without tablebases? >> >>Uri > >Well, I don't have tablebases for standard chess, so just results >without tables. This is the sort of mate that pn^2 search has the >most trouble with, since not only are the winning moves not forcing >moves, but there are lots and LOTS of forcing moves that don't win. >It will work out every possible series of pointless checks before it >thinks even 3 ply deep down a non-forcing line like Rg3.. after >about 1/2 hour it is working on lines that have a single non-forcing >move in them, and usually a huge series of forcing moves. > >On the other hand, my minmax searcher(which has piece values + mobility only) >found the right move at ply 2 and never changed its mind. >Guess Junior was too smart for its own good in that position. > >Angrim I suspect that Junior used tablebases and it saw another mate by tablebases. Capturing may be the faster way to get a tablebase mate. 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.