Full text
| |
| Author(s): |
John Lenon Cardoso Gardenghi
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Instituto de Matemática e Estatística (IME/SBI) |
| Defense date: | 2017-08-09 |
| Examining board members: |
Ernesto Julian Goldberg Birgin;
Geovani Nunes Grapiglia;
Elizabeth Wegner Karas;
José Mario Martinez Perez;
Sandra Augusta Santos
|
| Advisor: | Ernesto Julian Goldberg Birgin |
| Abstract | |
In the present work, we have studied and developed algorithms with worst-case evaluation complexity analysis for nonlinear programming problems. For the unconstrained optimization case, we have established two similar algorithms that explore high-order regularization models. We have proposed a computational implementation that preserves the good properties of the evaluation complexity theory, and we made numerical experiments with classical problems from the literature, in order to check the implementation and certify the practical applicability of methods that employ high-order models. For the constrained optimization case, we have established a two phases algorithm that converges to points that meet the unscaled first-order optimality condition for the nonlinear programming problem. (AU) | |
| FAPESP's process: | 13/23494-9 - Large-scale linearly constrained minimization with convergence to second order stationary points |
| Grantee: | John Lenon Cardoso Gardenghi |
| Support Opportunities: | Scholarships in Brazil - Doctorate |