DATA Step, Macro, Functions and more

termination criterion with NLPQN

Reply
Contributor
Posts: 25

termination criterion with NLPQN

[ Edited ]

Hello,

 

I am doing a quasi newton optimization method but I have a problem. The optimization stop before I find the global minimum.

 

I wonder if I could use a termination criterion of 1e-4 in order to stop the optimization after getting an objective function as small as I want and if it is possible, how to do it?

 

Here is my code

proc iml;

         

     use a;

     read all var _NUM_ into data;

     close a;

 

     * Chargement des fonctions ;

     reset storage = &lib..Functions;

     load module=(function1 function2);  

    

     parameters = j(5, 1, .) ;

     parameters[1] = 1.1 ;    

     parameters[2] = 5;       

    parameters[3] = 2;       

     parameters[4] = 1.5;

     parameters[5]= 0.9;      

 

     con = {-10 1E-6 -10 1E-6 1E-6, 10 10 10 10 1}

     opt = {1 4} ;       

    

    

     *Quasi-Newton Method ;

     call NLPQN(rc, parameters, "function1", theta, opt, con);

    

 

quit ;

 Moreover, this is the value of the objective function I get. You can see it is quite neer from each other at 10-4

test.PNG

Thank you for helping

Super User
Posts: 10,610

Re: termination criterion with NLPQN

Post it at IML forum,since it is real IML code. and could try another initial value .

Ask a Question
Discussion stats
  • 1 reply
  • 84 views
  • 0 likes
  • 2 in conversation