Newton method with feasible inexact projections for constrained equations and nonsmooth Newton method in Riemannian manifolds

Detalhes bibliográficos
Ano de defesa: 2019
Autor(a) principal: Oliveira, Fabiana Rodrigues de lattes
Orientador(a): Ferreira, Orizon Pereira lattes
Banca de defesa: Prudente, Leandro da Fonseca, Gonçalves, Max Leandro Nobre, Andreani, Roberto, Haeser, Gabriel, Gonçalves, Douglas Soares
Tipo de documento: Tese
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Universidade Federal de Goiás
Programa de Pós-Graduação: Programa de Pós-graduação em Matemática (IME)
Departamento: Instituto de Matemática e Estatística - IME (RG)
País: Brasil
Palavras-chave em Português:
Palavras-chave em Inglês:
Área do conhecimento CNPq:
Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/9507
Resumo: In this thesis, we will study three versions of the Newton method for solving problems in two contexts, namely Euclidean and Riemannian. In the Euclidean context, we will present the Newton method with feasible inexact projections for solving generalized equations subject to a set of constraints. Under local assumptions, the linear or superlinear convergence of a sequence generated by the proposed method is established. Next, a version of the inexact Newton method with feasible inexact projections for solving constrained smooth and nonsmooth equations is presented. Using suitable assumptions, the linear or superlinear convergence of a sequence generated by the method is proved. Furthermore, to illustrate the practical behavior of the proposed method, some numerical experiments are reported. Under another perspective, the last version of the Newton method to be investigated is an extension of the nonsmooth Newton method itself from the Euclidean context to the Riemannian, objecting to find a singularity of a special class of locally Lipschitz continuous vector fields. In particular, this method retrieves the classical nonsmooth Newton method to solve a system of nonsmooth equations. The well-definedness of the sequence generated by the method is ensured and the convergence analysis of the method is made under local and semi-local assumptions.