Thread Subject: fmincon, interior-point algorithm

Hello!

I have a rather large-scale nonlinear optimization problem with nonlinear equality constraints. I looked trough the documentation of about the optimization toolbox to find an appropriate solver and found that fmincon, using either the active-set or the interior-point methods to be the best options. As the problem is large scale I finally settled for the IP-solver. Now to my question: As I understand it, the "interior-point part" of the IP algorithm is there as a way to handle inequality constraints. So in my case, having only equality constraints, the IP-solver is effective only because of the way the subproblems are solved, while the interior point concept itself is of no importance for me?

Regards
/Arne

You might be better off using an SQP solver, such as SNOPT since you
only have equality constraints.

There seems to be quite a few articles on the web about handling of
inequality/equality constraints for IP methods.

Best wishes, Marcus
http://tomopt.com/
http://tomsym.com/

Thanks for the tip! Actually, I am thinking, by looking at the suggested literature for the IP solver, that in the abscense of inequality constraints, the IP-solver "reduces" to an SQP solver, as apparently the subproblems are solved using an SQP method. But I will, as you suggested, try to look in to the literature a bit more.

Best regards
/Arne