Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A final stab at big-O

Author: Robert Hyatt

Date: 19:41:41 05/09/01

Go up one level in this thread


On May 09, 2001 at 20:14:41, Ricardo Gibert wrote:

>On May 09, 2001 at 19:36:18, Dann Corbit wrote:
>
>>On May 09, 2001 at 19:34:35, Ricardo Gibert wrote:
>>[snip]
>>>>And it is clear that people who consider intractible problems to be O(1) are
>>>>using a set of definitions that are without value.
>>>
>>>I did not formulate the definition, but I do find it to be of value all the
>>>same.
>>
>>Valuable for WHAT?
>>
>>It certainly can't be used for computation if it delivers up answers like
>>"Chess is O(1)"
>
>
>It delivers the access of chess EGTBs as O(1) and I find that useful. It
>delivers n*n chess as O(exp(n)) and find that useful too. What am I missing?


Since 32 piece egtbs will _never_ be built...  I think it misses a LOT.  IE
I want to know about the complexity of "real" application problems.  Reducing
them all to O(1) is not just useless, worthless, etc...



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.