Author: Omid David Tabibi
Date: 11:24:27 04/12/04
Go up one level in this thread
On April 10, 2004 at 10:29:51, Dan Andersson wrote: > There are reasons for using recursion. One of them is that the generated code >tends to be smaller. And that could trumph any gains made from reducing the >data. > Another one is decreasing complexity. And the ability to use induction to prove >an algorithm. Apart from simple alpha-beta, it is impossible to theoretically prove any algorithm that is of practical use nowadays. For example, can you theoretically prove that null-move search is sane? No, but you can show enough experimental results to establish the point. I know of many researchers who have either deserted the field of computer-chess, or don't want to deal with it, only because it is a field where almost nothing can be theoretically proven. That is the case not only in chess, but in many other games also. My belief is that theory should serve the practice, and not the other way round. If you can't prove something, it doesn't mean that you shouldn't research it. In most cases experimental results are a sufficient alternative for theoretical proofs. > I often use tail recursion as a goto with values. It usually doesn't incur any >overhead. And increases branch prediction for some types of code like parsers, >regexp and interpreters. > >MvH Dan Andersson
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.