Demonstration of class OptSolver usage:
Solving a one-variable optimization problem

We present a simple program to solve a one-variable optimization problem.

• As any code using OFELI we start by including the main header file of the library. We declare an instance of class OptSolver

  #include "OFELI.h" using namespace OFELI; int main(int argc, char *argv[]) { OptSolver os;

• We next select an optimization method. Here we choose the so-called Truncated Newton method:

  os.setOptMethod(OptSolver::TRUNCATED_NEWTON);

• In order to provide the objective function to minimize and its gradient, we define here the functions by algebraic expressions. For a multivariable optimization problem, the optimization variables are x1, x2, .... Here we are concerned with one variable.

  os.setObjective("x1^2-x1+1"); os.setGradient("2*x1-1");

• We are now ready to run the optimization process. The member function run is used for this. The solution can be retrieved by invoking the member function getSolution

  os.run(); cout << os << endl; cout << "Solution: " << os.getSolution() << endl; return 0; }

   
  Copyright © 1998-2022 Rachid Touzani