constrained minimization -- Minimize/Reduce don't work
- To: mathgroup at smc.vnet.net
- Subject: [mg48576] constrained minimization -- Minimize/Reduce don't work
- From: Sharath <csr at postmark.net>
- Date: Sat, 5 Jun 2004 07:19:36 -0400 (EDT)
- Organization: Washington University in St. Louis
- Sender: owner-wri-mathgroup at wolfram.com
Hi,
Using Mathematica 5.0 I am trying to minimize a non-linear function (for
example),
f(w,u,v,x,y,z) = (w(wx-y)Log(xu)v+3(Log(x)+w(1-z)Log(1/y)))
/(1-y)Log(x)-w^2(x-1)(z-1)Log(1/y)
with some constraints like {w,u} are postive Integers {v,x,y,z} are
Reals with v>=0, x>1, 0<y<1, 0<z<1.
I tried using Minimize[{f, cons}, {x, y,...}] but it gives
"Minimize::mixdom: Exact optimization with mixed real and integer
variables is not yet implemented".
So I removed the restriction of Integers but then it gives me the entire
expression as it is!
I tried using Reduce by using f>0 as the expression. When used with or
without ForAll quantifier for variables either I get
"Reduce::nsmet: This system cannot be solved with the methods available
to Reduce" (even though I tried simplifying the expression and
eliminating some of the varibles some values).
I thought this constrained minimization can be solved using Lagrange
multipliers method (do I need to change < to <=?).
Ideally, I would like x and y expressed in terms of z. Is it possible to
do such a thing -- get symbolic values for x and y to minimize f?
Which function can I use or do I need to put the Lagrange Multiplier
equations myself and try solving.
This is my first use of Mathematica. I went through various examples to
learn the basics.
Any help is appreciated. Thanks.
-Sharath