Orientador: Jose Mario Martinez Perez / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-13T12:07:27Z (GMT). No. of bitstreams: 1 Martinez_AndreLuisMachado_D.pdf: 1457053 bytes, checksum: 6d70bba5dd246c0212765142b6451a5d (MD5) Previous issue date: 2009 / Resumo: Quando resolvemos problemas de programação não linear por meio de algoritmos que utilizam o Lagrangiano Aumentado, um fenômeno chamado voracidade pode ocorrer. Quando isto ocorre o método busca pontos muito infactíveis com valores de função muito pequenos, em geral, nas primeiras iterações, assim o parâmetro de penalidade cresce excessivamente, de tal forma que prejudica o condicionamento do problema. Neste trabalho 'e sugerida uma abordagem de regularização para superar esta dificuldade. Um método de Lagrangiano Aumentado é definido, com a adição de um termo regularizador que inibe a possibilidade do iterando se afastar demasiadamente do ponto de referência. Provamos convergência e apresentamos exemplos numéricos. / Abstract: When one solves Nonlinear Programming problems by means of algorithms that use merit criteria combining the objective function and penalty feasibility terms, a phenomenon called greediness may occur. Unconstrained minimizers attract the iterates at early stages of the calculations and, so, the penalty parameter needs to grow excessively, in such a way that ill conditioning harms the overall convergence. In this work a regularization approach is suggested to overcome this dificulty. An Augmented Lagrangian method is defined with the addition of a regularization term that inhibits the possibility that the iterates go far from a reference point. Convergence proofs and numerical examples are given. / Doutorado / Doutor em Matemática Aplicada

Orientador: Jose Mario Martinez Perez / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-13T22:53:44Z (GMT). No. of bitstreams: 1 Castelani_EmersonVitor_D.pdf: 695936 bytes, checksum: 9434e07a75cde154320a5156daf73684 (MD5) Previous issue date: 2009 / Resumo: Ao resolver problemas de programação não linear usando métodos do tipo Lagrangiano Aumentado, um fenômeno chamado voracidade pode ocorrer. Quando este fenômeno ocorre, o método busca pontos muito infactíveis com valor de função objetivo muito pequeno. Tais fatos ocorrem, em geral, na primeiras iterações e então, o parâmetro de penalidade precisa crescer excessivamente, tornado os subproblemas mal condicionados, prejudicando assim a convergência. Desta forma, o propósito deste trabalho é adicionar restrições de caixas adaptativas (região de confiança) a cada subproblema em cada iteração externa, de modo que, a distância entre dois iterando consecutivos das iterações externas é controlada. O novo método inibe a possibilidade do fenômeno de voracidade. Resultados de convergência, limitação de parâmetro de penalidade e exemplos numéricos são apresentados / Abstract: When we solve nonlinear programming problems by means of algorithms of kind of Augmented Lagrangian, a phenomenon called greediness may occur. Unconstrained minimizers attract the iterates at early stages of the calculations and, so, the penalty parameter needs to grow excessively, in such a way that ill-conditioning harms the overall convergence. In this sense, the proposal of this work is to add an adaptive artificial box constraint (trust-region) to the subproblem at every outer iteration, in such a way that the distance between consecutive outer iterates is controlled. The new method inhibits the possibility of greediness phenomenon. Convergence proofs and numerical examples are given / Doutorado / Otimização / Doutor em Matemática Aplicada

Orientador: Sandra Augusta Santos / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matemática Estatística e Computação Científica / Made available in DSpace on 2018-08-25T00:20:47Z (GMT). No. of bitstreams: 1 Nascimento_IvanXavierMourado_M.pdf: 5587602 bytes, checksum: 769fbf124a59d55361b184a6ec802f66 (MD5) Previous issue date: 2014 / Resumo: Métodos de região de confiança formam uma classe de algoritmos iterativos amplamente utilizada em problemas de otimização não linear irrestrita para os quais as derivadas da função objetivo não estão disponíveis ou são imprecisas. Uma das abordagens clássicas desses métodos envolve a otimização de modelos polinomiais aproximadores para a função objetivo, construídos a cada iteração com base em conjuntos amostrais de pontos. Em um trabalho recente, Scheinberg e Toint [SIAM Journal on Optimization, 20 (6) (2010), pp. 3512-3532 ] mostram que apesar do controle do posicionamento dos pontos amostrais ser essencial para a convergência do método, é possível que tal controle ocorra de modo direto apenas no estágio final do algoritmo. Baseando-se nessas ideias e incorporando-as a um esquema algorítmico teórico, os autores investigam analiticamente uma curiosa propriedade de autocorreção da geometria dos pontos, a qual se evidencia nas iterações de insucesso. A convergência global do novo algoritmo é, então, obtida como uma consequência da geometria autocorretiva. Nesta dissertação estudamos o posicionamento dos pontos em métodos baseados em modelos quadráticos de interpolação e analisamos o desempenho computacional do algoritmo teórico proposto por Scheinberg e Toint, cujos parâmetros são determinados / Abstract: Trust-region methods are a class of iterative algorithms widely applied to nonlinear unconstrained optimization problems for which derivatives of the objective function are unavailable or inaccurate. One of the classical approaches involves the optimization of a polynomial model for the objective function, built at each iteration and based on a sample set. In a recent work, Scheinberg and Toint [SIAM Journal on Optimization, 20 (6) (2010), pp. 3512¿3532 ] proved that, despite being essential for convergence results, the improvement of the geometry (poisedness) of the sample set might occur only in the final stage of the algorithm. Based on these ideas and incorporating them into a theoretical algorithm framework, the authors investigate analytically an interesting self-correcting geometry mechanism of the interpolating set, which becomes evident at unsuccessful iterations. Global convergence for the new algorithm is then proved as a consequence of this self-correcting property. In this work we study the positioning of the sample points within interpolation-based methods that rely on quadratic models and investigate the computational performance of the theoretical algorithm proposed by Scheinberg and Toint, whose parameters are based upon either choices of previous works or numerical experiments / Mestrado / Matematica Aplicada / Mestre em Matemática Aplicada

Cette thèse est située dans le cadre du projet FUI OASIS ayant comme objectif la modélisation d'un processus d'emboutissage optimisé. Le travail consiste essentiellement au développement des algorithmes de contact plus appropriés à ce type de mise en forme. Dans la littérature et pour plusieurs codes de calcul industriels, l'approche NTS (nœud à segment) demeure la plus utilisée pour la résolution d'un problème de contact. Dans certaine configuration, cette méthode présente des insuffisances et un manque de précision. On la remplaçant par l'approche mortier, on arrive à résoudre une gamme assez large de problèmes de contact. La méthode mortier, utilisée au initialement pour un calcul avec décomposition de domaine, a été le centre d'intérêt de plusieurs travaux de recherche pour la modélisation du contact. Dans ce travail, on va regrouper plusieurs méthodes de gestion du contact en les combinant avec l'approche mortier. L'algorithme de résolution, les éléments d'implémentation ainsi quelques exemples de validation présentant une critique des avantages et les limites de chaque techniques sont détaillés dans ce travail afin d'obtenir un support technique pour tous travail ultérieurs avec la méthode mortier. Le principal avantage de la méthode mortier se manifeste dans l'application des conditions de contact sous forme d'intégrale dans l'interface. Bien que cette méthode permette de réduire la différence des contraintes dans l'interface de contact d'un élément à un autre pour obtenir une meilleure continuité de la pression de contact, elle demeure insuffisante dans certaines applications en particulier pour les problèmes en grande déformation. Le lissage des surfaces de contact, qu'on peut appliquer par différentes techniques, présente une solution classique à ce genre de problème en mécanique de contact. L'originalité de ce travail, c'est la combinaison de l'utilisation des courbes B-Spline cubiques pour la description presque exacte de la surface de contact d'un côté avec une formulation avec l'approche mortier pour l'application des conditions de contact d'un autre côté. Cette combinaison forme un duo gagnant permettant de résoudre un problème de contact en grandes déformation. Les termes permettant l'implémentation des différentes techniques de lissage pour la résolution d'un problème de contact sont détaillés. Une attention particulière est accordée au lissage avec les B-Spline Cubiques.Tous les algorithmes détaillés dans ce travail sont implémentés dans un code maison FiEStA. C'est un code de calcul par éléments finis libre en langage C++. Certains développements concernant la loi de comportement hyper-élastique et l'intégralité du module du contact sont développés dans ce travail de thèse. / This thesis is situated in the FUI OASIS project which the objective is the modeling of an optimized stamping process. The work mainly involves the development of the most appropriate contact algorithms such formatting. In the literature and several industrial computing codes, the NTS approach (node to segment) remains the most used for the resolution of a contact problem. In certain configuration, this method has shortcomings and a lack of precision. We replacing it with mortar approach, we manage to solve a broad range of contact problems. The mortar method, used for the initial for calculation using domain decomposition, was the focus of several research projects for the modeling of the contact. In this work, we will consolidate multiple contact formulation methods in combination with mortar approach. The resolution algorithm, the elements of implementation and some examples of validation with a review of the advantages and limitations of each technique are detailed in this work in order to get technical support for subsequent work with the mortar method. The main advantage of the mortar method is in the application of the contact conditions in integral form in the interface. Although this method reduces the difference of the stresses in the contact interface of a component to another to obtain a better continuity of the contact pressure, it is still insufficient in some applications, particularly for large deformation problems. The smoothing of contact surfaces, which can be applied by various techniques, presents a classic solution to this problem in mechanical contact. The originality of this work is the combination of using cubic B-Spline curves for the almost exact description of the contact surface on one side with the use of the mortar approach to the application of the contact conditions on the other hand. This combination forms a winning combination for solving a contact problem in large deformation. The terms allowing the implementation of the different smoothing techniques for solving a problem of contact are detailed. Particular attention is paid to smoothing with Cu bic B-Spline. All algorithms detailed in this work are implemented in a house code 'Fiesta'. This is a free finite elements computer code in C ++. Some developments in the law of hyper-elastic behavior and completeness of the contact module are developed in this thesis.