Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!emory!hubcap!grimlok
From: grimlok@hubcap.clemson.edu (Mike Percy)
Newsgroups: comp.theory
Subject: Re: Help!
Message-ID: <11822@hubcap.clemson.edu>
Date: 25 Nov 90 22:32:00 GMT
References: <11801@hubcap.clemson.edu> <11185@pt.cs.cmu.edu>
Organization: Clemson University, Clemson, SC
Lines: 30


Thanks for the help everyone.  I suppose I should have just kept with
my real problem.  In my simulation, I want to select a random move.
My move generator is a deterministic search for legal moves, but there
is no way of telling how many legal moves there are until we've seen
them all.  I don't want to store them up, nor do I want to run through
the search more than once.  
 
In essence:
 
Start search.
If current move better than previous best, keep this move.
If current move equal in value to the previous best, randomly keep one
of the two moves.
When search is over, I want to have a random move selected from the
best moves.
Simply keeping first, last, or ith is no good as it tends to direct
the search, missing paths that might lead to better solutions.

A couple of people suggested 1/i, which is what I thought of originally,
but dismissed (idiot) as being to easy, and my induction proof kept
going wacky -- must have been at it too long.
 
Thanks to all who offered help.


"I don't know about your brain, but mine is really...bossy."
Mike Percy                    grimlok@hubcap.clemson.edu
ISD, Clemson University       mspercy@clemson.BITNET
(803)656-3780                 mspercy@clemson.clemson.edu