Optimizing with Trust-Region Parallel Direct Search (TRPDS)

TRPDS combines the inherent parallelism in PDS with the fast convergence properties of Newton methods. The controlling framework is identical to the standard trust-region algorithm, but the PDS algorithm is used to solve the trust-region subproblem.

In this example, we highlight the steps needed to take advantage of parallel capabilities and to set up TRPDS. Further information and examples for setting up and solving a problem can be found in the Setting up and Solving an Optimization Problem section

First, include the header files and subroutine declarations.

   #ifdef HAVE_CONFIG_H
   #include "OPT++_config.h"

   #include <string>
   #include <iostream>
   #include <fstream>
   #ifdef HAVE_STD
   #include <cstdio>
   #include <stdio.h>

   #ifdef WITH_MPI
   #include "mpi.h"

   #include "NLF.h"
   #include "NLP2.h"
   #include "BoundConstraint.h"
   #include "CompoundConstraint.h"
   #include "OptConstrQNewton.h"
   #include "tstfcn.h"
   #include "ioformat.h"

   using NEWMAT::ColumnVector;
   using std::cerr;
   using namespace OPTPP;

   void SetupTestProblem(string test_id, USERFCN0 *test_problem, 
                         INITFCN *init_problem);
   void update_model(int, int, ColumnVector) {}

After an argument check, initialize MPI. This does not need to be done within an "ifdef", but if you want the option of also building a serial version of your problem, then it should be. (Note: An argument check is used here because this example is set up to work with multiple problems. Such a check is not required by OPT++.)

   int main (int argc, char* argv[])

     if (argc != 3) {
        cout << "Usage: tsttrpds problem_name ndim\n";

     #ifdef WITH_MPI
        int me;

        MPI_Init(&argc, &argv);
        MPI_Comm_rank(MPI_COMM_WORLD, &me);

Define the variables.

     int ndim;
     double time0, opt_time;

     USERFCN0 test_problem;
     INITFCN  init_problem;

     string test_id;

     // Setup the test problem
     // test_problem is a pointer to the function (fcn) to optimize
     // init_problem is a pointer to the function that initializes fcn
     // test_id is a character string identifying the test problem

     test_id = argv[1];
     ndim    = atoi(argv[2]);

     SetupTestProblem(test_id, &test_problem, &init_problem);

Now set up the output file. If you are running in parallel, you may want to designate an output file for each processor. Otherwise, the output from all of the processors will be indiscriminantly intertwined in a single file. If the function evaluation does any file I/O, you should set up a working directory for each processor and then have the each process chdir (or something comparable) into its corresponding directory. Each working directory should have a copy of the input file(s) needed by the function evaluation. If the function evaluation requires file I/O and working directories are not used, the function evaluation will not work properly.

     char status_file[80];
     #ifdef WITH_MPI
        sprintf(status_file,"%s.out.%x", status_file, me);

Set up the problem.

     //  Build a CompoundConstraint object
     ColumnVector bound(ndim );
     bound = -500.0;
     Constraint bc = new BoundConstraint(ndim,bound);

     //  Create an OptppArray of Constraints 
     OptppArray<Constraint> arrayOfConstraints;

     //  Create a compound constraint 
     CompoundConstraint constraints(arrayOfConstraints);  
     //  Create a constrained Nonlinear problem object 
     FDNLF1 nlp(ndim,test_problem, init_problem, &constraints);


Set up a TRPDS algorithm object. TRPDS is a Newton-based method, so it is constructed as a Newton object with a search strategy of TrustPDS. Since it is a blend of two algorithms (Newton and PDS), algorithmic parameters can both be set even though it is all done within the context of the Newton object.

    OptConstrQNewton objfcn(&nlp,update_model);   
    if (!objfcn.setOutputFile(status_file, 0))
      cerr << "main: output file open failed" << endl;
    ostream* optout = objfcn.getOutputFile();
    *optout << "Test problem: " << test_id << endl;
    *optout << "Dimension   : " << ndim    << endl;

Optimize and clean up.

    time0 = get_wall_clock_time();
    opt_time = get_wall_clock_time() - time0;
    *optout << "wall clock time =" << e(opt_time,12,4) << endl;

    objfcn.printStatus("Solution from quasi-newton");

Finally, it is necessary to shut down MPI.

    #ifdef WITH_MPI


Next Section: Generating Set Search Methods | Back to Main Page

Last revised September 14, 2006 .

Bug Reports    OPT++ Developers    Copyright Information    GNU Lesser General Public License
Documentation, generated by , last revised August 30, 2006.