Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!gatech!rutgers!argos.montclair.edu!hubey
From: hubey@pegasus.montclair.edu (H. M. Hubey)
Subject: Re: A question..
Message-ID: <hubey.785391912@pegasus.montclair.edu>
Sender: root@argos.montclair.edu (Operator)
Organization: SCInet @ Montclair State
References: <3abhdq$kht@news.bu.edu> <CzE1y5.680@murdoch.acc.Virginia.EDU>
Date: Mon, 21 Nov 1994 04:25:12 GMT
Lines: 31

mjd4c@uvacs.cs.Virginia.EDU (Michael J. Daniel) writes:

>Give us some examples of really cool problems.

I think I have one :-)..

I have a function R(x1,x2,x3,,,,)=P(x1,x2,,,)/Q)x1,x2,,,,)

I'd like to find the values {x1,x2,x3,,,} for which R(x1,x2,,,)--> oo.

If I try to use L'Hospital's rule I have to try various partial
derivatives from various directions.

I was wondering if something like a "steepest ascent" algorithm
would work to find "a" direction {i.e. say x1 increases, x2 decreases,
etc} so that I could check a good candidate set of {x1,x2,,}
which would show if R(x1,x2,,,}--> oo.

Note that I could define T(x1,x2,,,)=exp(-R(x1,x2,,)) so that
then I'd be looking for values for which T(x1,x2,,}-->0.

Could genetic algorithms work in this case?





--
						-- Mark---
....we must realize that the infinite in the sense of an infinite totality, 
where we still find it used in deductive methods, is an illusion. Hilbert,1925
