[prev in list] [next in list] [prev in thread] [next in thread] 

List:       sas-l
Subject:    Constrained Nonlinear Optimization ?
From:       "Tirthankar C.P" <tir () IGIDR ! AC ! IN>
Date:       1999-10-29 19:48:42
[Download RAW message or body]

Dear all,


Is it possible to do constrained nonlinear optimization in SAS ?

What I mean is, can I evaluate functions of the form,


                        min f(x) ,

                s.t   G(x) <= 0

                where x is an n-D vector. And G is a set of constraints,
linear or nonlinear.


        Actually my problem is this. A few days back I'd posted a question
regarding constrained max.likelihood in SAS. I thank Dr.Kattamuri Sarma
for the valuable insights he gave.

But sadly, I wasn't able to do much on it. The fact is, I don't know
much of IML.
But I found that there's a function in it which gives the log-likelihood
of an arma model(SAS/IML software ver6. p-264):


        armalik(lnl,resid,std,x,phi,theta)

        where the input are phi, theta and x.

Theta, and phi are parmeter vectors which have to be prespecified. x is
the the zero-mean time series.

        What I'm hinting at is this, can I put this function in a
nonlinear function with constraints. This would be eqv to doing
constrained ML Estimation. Of course, the elements of phi and theta would
be now parameters which would be changed according to the nonlinear
routine. Am I on the right track ?

This optimizer may be in IML or in some proc( in which case I have to
invoke proc iml inside that proc and is this possible ?

PLEASE help me.


TIA ,

Tirthankar



Tirthankar C.P
Research Student
Indira Gandhi Institute of Development Research    Phone:91-022-840 0919/20/21
Goregaon East                                        Fax:  91-022-8402027
Mumbai 400 065                                     email:-tir@igidr.ac.in
India.

"Reality is merely an illusion, albeit a very persistent one."
                                                              -Albert Einstein
________________________________________________________________________________

[prev in list] [next in list] [prev in thread] [next in thread]