Full text
| |
| Author(s): |
Deise Gonçalves Ferreira
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Matemática, Estatística e Computação Científica |
| Defense date: | 2017-04-19 |
| Examining board members: |
Sandra Augusta Santos;
José Mario Martínez Pérez;
Lucio Tunes dos Santos;
Mael Sachine;
Ernesto Julián Goldberg Birgin
|
| Advisor: | Maria Aparecida Diniz Ehrhardt; Sandra Augusta Santos |
| Abstract | |
In this work we have introduced the algorithm PSIFA - Pattern Search and Implicit Filtering Algorithm - which is a derivative-free algorithm that has been designed for linearly constrained problems with noise in the objective function, combining some elements of the pattern search approach of Lewis and Torczon (2000) with ideas from the method of implicit filtering of Kelley (2011). The global convergence analysis is presented, encompassing the degenerate case, under mild assumptions. Numerical experiments with linearly constrained problems from the literature and also with the feasible set defined by polyhedral 3D-cones with several degrees of degeneration at the solution were performed, including noisy functions that are not covered by the theoretical hypotheses. Furthermore, an instance of a parameter identification problem was considered as an application with inherent noise on the objective function for which linear constraints are present in the model. To put PSIFA in perspective, comparative tests with the Pattern Search and the Implicit Filtering algorithms have been prepared, with encouraging results (AU) | |
| FAPESP's process: | 13/12964-4 - Derivative-free methods for nonlinear programming: constrained and noisy problems |
| Grantee: | Deise Gonçalves Ferreira |
| Support Opportunities: | Scholarships in Brazil - Doctorate |