A derivative-free trust-funnel method for equality-constrained nonlinear optimization

In this work, we look into new derivative-free methods to solve equality-constrained optimization problems. Of particular interest, are the trust-region techniques, which have been investigated for the unconstrained and bound-constrained cases. For solving equality-constrained optimization problems, we introduce a derivative-free adaptation of the trust-funnel method combined with a self-correcting geometry scheme and present some encouraging initial numerical results.

Citation

Report NAXYS-08-2014, Namur Center for Complex Systems, University of Namur, Namur, Belgium.

Article

Download

View A derivative-free trust-funnel method for equality-constrained nonlinear optimization