home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!munnari.oz.au!goanna!ok
- From: ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe)
- Newsgroups: comp.programming
- Subject: Re: Teaching the basics
- Message-ID: <14195@goanna.cs.rmit.oz.au>
- Date: 26 Aug 92 11:02:17 GMT
- References: <1992Aug19.154830.11787@uwm.edu> <MJN.92Aug23031941@pseudo.uucp> <1992Aug25.174243.10436@uwm.edu>
- Organization: Comp Sci, RMIT, Melbourne, Australia
- Lines: 19
-
- In article <1992Aug25.174243.10436@uwm.edu>, markh@csd4.csd.uwm.edu (Mark) writes:
- > In article <7116@charon.cwi.nl> tromp@cwi.nl (John Tromp) writes:
- > >One of the prime applications of recursion, IMHO,
- > >and where you cannot hope to replace it by anything else,
- > >is in a tree search algorithm like alpha beta or plain old minimax.
-
- > Wrong. In fact, the non-recursive version (tree search) is far smaller and
- > simpler. No stacks either.
-
- The trouble is that (Mark) gives code for a BINARY SEARCH TREE (like
- UNIX "tsearch()), which doesn't even faintly resemble the problems that
- (John Tromp) explicitly named, which are ALPHA-BETA SEARCH and MINIMAX.
- Not the least of the difference is that alpha-beta and minimax are
- algorithms for searching a _game_ tree which is in practice either infinite
- in size or so very much larger than even virtual memory that one just does
- _not_ have (Mark)'s kind of tree stored in memory to walk around.
-
- --
- You can lie with statistics ... but not to a statistician.
-