Hello Thomas

I am now using your nice Non linear optimization package.

I have some questions. Do not be hard on me, I am not an expert .

For example, read this code:

Code: Select all

   VectorXd x(2);
   
   hybrd_functor functor;
   HybridNonLinearSolver<hybrd_functor> solver(functor);
   solver.parameters.nb_of_subdiagonals = 1;
   solver.parameters.nb_of_superdiagonals = 1;
   solver.diag.setConstant(n, 1.);
   solver.useExternalScaling = true;
   if (solver.solveNumericalDiff(x) != 1)
      Cout() << "No convergence!";

Questions:
- What is the purpose of nb_of_subdiagonals, superdiagonals, setConstant(), useExternalScaling and another parameters?
- Are there other parameters to indicate max. number of iterations and required precision?
- If solveNumericalDiff() returns 1 the solver has converged successfully. Are there other parameters to check the quality of the results?
- You use a "functor" struct which calculates the equations and gives the Jacobian if available. Could have been possible to use a pointer to a function instead?

Best regards
Koldo

I don't know about the HybridNonLinearSolver, you should ask on our ML to reach more people.

Regarding your last question, you can use std::ptr_fun(foo) to build a functor from a function pointer.

ggael wrote:I don't know about the HybridNonLinearSolver, you should ask on our ML to reach more people.

Thank you Gaël

I will move this question to ML.

Best regards
Koldo