Author: Dann Corbit
Date: 16:56:45 05/19/99
Go up one level in this thread
On May 19, 1999 at 19:15:51, Christoph Fieberg wrote: >I am convinced that with a more extensive use of the available chess engines on >exploring opening moves the whole theory could be rewritten and surprising new >moves could be discovered. > >If you let the computers calculate from the beginning just the 2 best moves in >each position for each side you will arrive at more than 1 billion positions >after the 15th move (2^30 = 1,073,741,824 positions)! The hardest part is finding the two best moves. >That are 100 times more positions than the largest opening books have and yet it >would be a very limited book because some of the most popular lines would not be >covered (e.g. after the 1th move you could arrive - depending on your engine and >time allowed for calculation - at the following four positions: 1.e4 e5; 1.e4 >e6; 1.d4 d5; 1.d4 Sf6 from which the tree would expand. In this case you would >miss the Sicilian completely). > >And if you would take the 3 best moves you would arrive at 205,891 billion >positions after the 15th move!! The hardest part is finding the three best moves. >I think it would be worth to let the computers analyse as much as possible and >probably it would be necessary to organise a search via internet to have the >greatest benefit from their anlysis power and to achieve as much analysis as >possible. Sounds a tiny bit like: ftp://38.168.214.175/pub/Chess%20Analysis%20Project%20FAQ.htm >Let’s say it would be necessary to let the computer calculate at least 1 hour to >find the two best moves than 61,287 years would be necessary to reach 1 billion >positions. Or 61,287 chess enthusiasts would have to let their computers analyse >one full year. A daunting task! For the initial openings, one hour is probably not going to find the best move. Early on, the board is fairly closed. You can analyze some initial positions for 8 hours and only get to 11 plies. If Hsu's boards become a reality, they would speed up the project by several orders of magnitude if everyone bought them. Here are the 1st 400 board positions, analyzed anywhere from 12 minutes to 8+ hours: ftp://38.168.214.175/pub/EPD/1st400.epd We have analyzed about 400K positions in total at 12 minutes or more.
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.