Author: Bob Durrett
Date: 14:04:29 12/11/02
Go up one level in this thread
On December 10, 2002 at 20:59:14, Ricardo Gibert wrote: >On December 09, 2002 at 22:26:11, Bernardo Wesler wrote: > >>Do they consider it as an uthopy? > >By "chess algorithm" I assume you specifically mean one that can solve chess. >Also, perhaps you intend the word "utopian" in place of "as an uthopy." This is >what I will assume. > >>If so, what is the scientific reason? >>If it was possible, how far are they from that point? Did they find any >>evidence? > >There is an algorithm that is proven to guarantee perfect play called minimax. A >trivial, but substantial improvement is called alpha-beta. In either case, the >result is purely of a theoretical nature. The problem is that the algorithm >takes a prohibitive amount of time to execute. For all intents and purposes it >would take forever...with current hardware. Perhaps this "taking forever" problem is sufficient motivation to find an alternative proof that does not "take forever." My gut feel is that a suitable proof must be inductive for the case where 50-move and repetition of position rules [or any similar limitations on the number of moves] are waived. Whenever you have something that can "go to infinity," an inductive proof is indicated. But if the 50-move and repetition of position rules are imposed, my gut feel tells me that there is an upper bound on the size of a solution tree. Maybe in that case an inductive proof is not essential. But if a solution tree is extremely large, it is effectively infinite. So, even in that case, the search for an inductive proof is indicated. I wish I could produce an inductive proof to illustrate my point. Incidentally, giving a position to illustrate my point would also be nice. Maybe next time. Bob D. > >>Anyway, which is the possibility to get a very powerful chess engine or program >>or soft that depends almost nothing upon the hardware? >>Thx. >>Dr Wesler >>blwesler@mail.retina.ar
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.