Submitted by JEFERSON BACK VANDERLINDE null (jefersonbv@yahoo.com.br) on 2017-11-01T16:38:25Z No. of bitstreams: 1 jeferson_tese_final_20171101.pdf: 4860852 bytes, checksum: 2f99c37969be3815f82b1b4455a40230 (MD5) / Approved for entry into archive by LUIZA DE MENEZES ROMANETTO (luizamenezes@reitoria.unesp.br) on 2017-11-13T15:38:34Z (GMT) No. of bitstreams: 1 vanderlinde_jb_dr_ilha.pdf: 4860852 bytes, checksum: 2f99c37969be3815f82b1b4455a40230 (MD5) / Made available in DSpace on 2017-11-13T15:38:34Z (GMT). No. of bitstreams: 1 vanderlinde_jb_dr_ilha.pdf: 4860852 bytes, checksum: 2f99c37969be3815f82b1b4455a40230 (MD5) Previous issue date: 2017-09-06 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Neste trabalho, consideram-se a análise teórica e a implementação computacional dos algoritmos Primal Simplex Canalizado (PSC) e Dual Simplex Canalizado (DSC) especializados. Esses algoritmos foram incorporados em um algoritmo Branch and Bound (B&B) de modo a resolver o problema de Planejamento da Expansão de Sistemas de Transmissão (PEST). Neste caso, o problema PEST foi modelado usando os chamados modelo de Transportes e modelo Linear Disjuntivo (LD), o que produz um problema de Programação Linear Inteiro Misto (PLIM). O algoritmo PSC é utilizado na resolução do problema de Programação Linear (PL) inicial após desconsiderar a restrição de integralidade do problema PLIM original. Juntamente com o algoritmo PSC, foi implementada uma estratégia para reduzir o número de variáveis artificiais adicionadas ao PL, consequentemente reduzindo o número de iterações do algoritmo PSC. O algoritmo DSC é utilizado na reotimização eficiente dos subproblemas gerados pelo algoritmo B&B, através do quadro ótimo do PL inicial, excluindo, assim, a necessidade da resolução completa de cada subproblema e, consequentemente, reduzindo o consumo de processamento e memória. Nesta pesquisa, é apresentada uma nova proposta de otimização, e, consequentemente, a implementação computacional usando a linguagem de programação FORTRAN que opera independentemente de qualquer solver. / In this research, the theoretical analysis and computational implementation of the specialized dual simplex algorithm (DSA) and primal simplex algorithm (PSA) for bounded variables is considered. These algorithms have been incorporated in a Branch and Bound (B&B) algorithm to solve the Transmission Network Expansion Planning (TNEP) problem. In this case, the TNEP problem is modeled using transportation model and linear disjunctive model (DM), which produces a mixed-integer linear programming (MILP) problem. After relaxing the integrality of investment variables of the original MILP problem, the PSA is used to solve the initial linear programming (LP) problem. Also, it has been implemented a strategy in PSA to reduce the number of artificial variables which are added into the LP problem, and consequently reduces the number of iterations of PSA. Through optimal solution of the initial LP, the DSA is used in efficient reoptimization of subproblems, resulting from the B&B algorithm, thus excludes the need for complete resolution of each subproblems, which results reducing the CPU time and memory consumption. This research presents the implementation of the proposed approach using the FORTRAN programming language which operates independently and does not use any commercial solver.

Dans le cadre de cette thèse, nous traitons le problème de flow-shop à deux machines avec temps de transport où l’objectif consiste à minimiser le temps de complétion maximal. Dans un premier temps, nous nous sommes intéressés à la modélisation de ce problème. Nous avons proposé plusieurs programmes linéaires en nombres entiers. En particulier, nous avons introduit une formulation linéaire basée sur une généralisation non triviale du modèle d’affectation pour le cas où les durées des opérations sur une même machine sont identiques. Dans un deuxième temps, nous avons élargi la portée de ces formulations mathématiques pour développer plusieurs bornes inférieures et un algorithme exact basé sur la méthode de coupe et branchement (Branch-and-Cut). En effet, un ensemble d’inégalités valides a été considéré afin d’améliorer la relaxation linéaire de ces programmes et d’accélérer leur convergence. Ces inégalités sont basées sur la proposition de nouvelles règles de dominance et l’identification de sous-instances faciles à résoudre. L’identification de ces sous-instances revient à déterminer les cliques maximales dans un graphe d’intervalles. En plus des inégalités valides, la méthode exacte proposée inclut la considération d’une méthode heuristique et d’une procédure visant à élaguer les nœuds. Enfin, nous avons proposé un algorithme par séparation et évaluation (Branch-and-Bound) pour lequel, nous avons introduit des règles de dominance et une méthode heuristique basée sur la recherche locale. Nos expérimentations montrent l’efficacité de nos approches qui dominent celles de la littérature. Ces expérimentations ont été conduites sur plusieurs classes d’instances qui incluent celles de la littérature, ainsi que des nouvelles classes d’instances où les algorithmes de la littérature se sont montrés peu efficaces. / In this thesis, we study the two-machine flow-shop problem with time delays in order to minimize the makespan. First, we propose a set of Mixed Integer Programming (MIP) formulations for the problem. In particular, we introduce a new compact mathematical formulation for the case where operations are identical per machine. The proposed mathematical formulations are then used to develop lower bounds and a branch-and-cut method. A set of valid inequalities is proposed in order to improve the linear relaxation of the MIPs. These inequalities are based on proposing new dominance rules and computing optimal solutions of polynomial-time-solvable sub-instances. These sub-instances are extracted by computing all maximal cliques on a particular Interval graph. In addition to the valid inequalities, the branch-and-cut method includes the consideration of a heuristic method and a node pruning procedure. Finally, we propose a branch-and-bound method. For which, we introduce a local search-based heuristic and dominance rules. Experiments were conducted on a variety of classes of instances including both literature and new proposed ones. These experiments show the efficiency of our approaches that outperform the leading methods published in the research literature.

Made available in DSpace on 2014-07-29T15:03:39Z (GMT). No. of bitstreams: 1 lilian pureza.pdf: 2774999 bytes, checksum: 2a074d04ee02c7e1c87fdbe8c2c68ef6 (MD5) Previous issue date: 2006-10-20 / The aim of this work is to present a formulation and corresponding computational implementation for sizing optimization of plane frames and cable-stayed columns considering geometric non liner behavior. The structural analysis is based on the finite element method using the updated lagrangian approach for plane frame and cable elements, which are represented by plane truss elements. The non linear system is solved by the Newton-Raphson method coupled to load increment strategies such as the arch length method and the generalized displacement parameter method, which allow the algorithm to transpose any critical point that happen to appear along the equilibrium path. In the optimization process the design variables are the heights of the crosssection of the frame elements, the objective function represents the volume of the structure and the constraints impose limits to displacements and critical load. Lateral constraints impose limits to the design variables. The finite difference method is used in the sensitivity analysis of the displacement and critical load constraints. The optimization process is carried out using three different optimization strategies: the sequential quadratic programming algorithm; the interior points algorithm; and the branch and bound method. Some numerical experiments are carried out so as to test the analysis and the sensitivity strategies. Numerical experiments are presented to show the validity of the implementation presented in this dissertation. / O objetivo deste trabalho é a otimização de dimensões de pórticos planos e de colunas estaiadas planas pela minimização do volume da estrutura, considerando os efeitos da não-linearidade geométrica em seu comportamento. A formulação utiliza, para análise das estruturas, elementos finitos de pórtico e de treliça planos e referencial lagrangeano atualizado. O método de Newton-Raphson foi utilizado como estratégia para solução do sistema de equações não lineares. Foram acopladas estratégias especiais para ultrapassagem de pontos críticos que possam existir ao longo da trajetória de equilíbrio, tais como o comprimento de arco cilíndrico e o controle dos deslocamentos generalizados. Na otimização, as variáveis de projeto são as alturas das seções transversais dos elementos, a função objetivo é o volume do material e as restrições dizem respeito a limitações impostas a deslocamentos e à carga limite, além de limitações impostas aos valores das variáveis. A sensibilidade da função objetivo foi obtida por diferenciação direta e a sensibilidade das restrições pelo método das diferenças finitas. Foram utilizados o algoritmo de programação quadrática seqüencial, PQS, o algoritmo de pontos interiores, PI, e o algoritmo de Branch and Bound, B&B. São apresentados exemplos de validação das estratégias de análise não linear e da análise de sensibilidade, além dos exemplos de validação da formulação empregada para a otimização resolvidos pelos métodos implementados.

No description available.

[pt] Com um número crescente de antenas em sistemas MIMO, o consumo de energia e os custos das interfaces de rádio correspondentes tornam-se relevantes. Nesse contexto, uma abordagem promissora é a utilização de conversores de dados de baixa resolução. Neste estudo, propomos dois novos pré-codificadores ótimos para a sinais de envelope constante e quantização de fase. O primeiro maximiza a distância mínima para o limite de decisão (MMDDT) nos receptores, enquanto o segundo minimiza o erro médio quadrático entre os símbolos dos usuários e o sinal de recepção. O design MMDDT apresetado nesse estudo é uma generalização de designs anteriores que baseiam-se em quantização de 1-bit. Além disso, ao contrário do projeto MMSE anterior que se baseia na resolução de 1-bit, a abordagem proposta emprega quantização de fase uniforme e a etapa de limite no método branch-and-bound é diferente em termos de considerar o relaxamento mais restritivo do problema não convexo, que é então utilizado para um design sub ótimo também. Além disso, três métodos diferentes de detecção suave e um esquema iterativo de detecção e decodificação que permite a utilização de codificação de canal em conjunto com pré-codificação de baixa resolução são propostos. Além de uma abordagem exata para calcular a informação extrínseca, duas aproximações com reduzida complexidade computacional são propostas. Os algoritmos propostos de pré-codificação branch-and-bound são superiores aos métodos existentes em termos de taxa de erro de bit. Resultados numéricos mostram que as abordagens propostas têm complexidade significativamente menor do que a busca exaustiva. Finalmente, os resultados baseados em um código de bloco LDPC indicam que os esquemas de processamento de recepção geram uma taxa de erro de bit menor em comparação com o projeto convencional. / [en] With an increasing number of antennas in multiple-input multiple-output (MIMO) systems, the energy consumption and costs of the corresponding front ends become relevant. In this context, a promising approach is the consideration of low-resolution data converters. In this study two novel optimal precoding branch-and-bound algorithms constrained to constant envelope signals and phase quantization are proposed. The first maximizes the minimum distance to the decision threshold (MMDDT) at the receivers, while the second minimizes the MSE between the users data symbols and the receive signal. This MMDDT design presented in this study is a generalization of prior designs that rely on 1-bit quantization. Moreover, unlike the prior MMSE design that relies on 1-bit resolution, the proposed MMSE approach employs uniform phase quantization and the bounding step in the branch-and-bound method is different in terms of considering the most restrictive relaxation of the nonconvex problem, which is then utilized for a suboptimal design also. Moreover, three different soft detection methods and an iterative detection and decoding scheme that allow the utilization of channel coding in conjunction with low-resolution precoding are proposed. Besides an exact approach for computing the extrinsic information, two approximations with reduced computational complexity are devised. The proposed branch-and-bound precoding algorithms are superior to the existing methods in terms of bit error rate. Numerical results show that the proposed approaches have significantly lower complexity than exhaustive search. Finally, results based on an LDPC block code indicate that the proposed receive processing schemes yield a lower bit-error-rate compared to the conventional design.

De nombreux problèmes en transport et en logistique peuvent être formulés comme des modèles de conception de réseau. Ils requièrent généralement de transporter des produits, des passagers ou encore des données dans un réseau afin de satisfaire une certaine demande tout en minimisant les coûts. Dans ce mémoire, nous nous intéressons au problème de conception de réseau avec coûts fixes et capacités. Ce problème consiste à ouvrir un sous-ensemble des liens dans un réseau afin de satisfaire la demande, tout en respectant les contraintes de capacités sur les liens. L'objectif est de minimiser les coûts fixes associés à l'ouverture des liens et les coûts de transport des produits. Nous présentons une méthode exacte pour résoudre ce problème basée sur des techniques utilisées en programmation linéaire en nombres entiers. Notre méthode est une variante de l'algorithme de branch-and-bound, appelée branch-and-price-and-cut, dans laquelle nous exploitons à la fois la génération de colonnes et de coupes pour la résolution d'instances de grande taille, en particulier, celles ayant un grand nombre de produits. En nous comparant à CPLEX, actuellement l'un des meilleurs logiciels d'optimisation mathématique, notre méthode est compétitive sur les instances de taille moyenne et supérieure sur les instances de grande taille ayant un grand nombre de produits, et ce, même si elle n'utilise qu'un seul type d'inégalités valides. / Many problems in transportation and logistics can be formulated as network design models. They usually require to transport commodities, passengers or data in a network to satisfy a certain demand while minimizing the costs. In this work, we focus on the multicommodity capacited fixed-charge network design problem which consists of opening a subset of the links in the network to satisfy the demand. Each link has a capacity and a fixed cost that is paid if it is opened. The objective is to minimize the fixed costs of the opened links and the transportation costs of the commodities. We present an exact method to solve this problem based on mixed integer programming techniques. Our method is a specialization of the branch-and-bound algorithm, called branch-and-price-and-cut, in which we use column generation and cutting-plane method to solve large-scale instances. We compare our method with CPLEX, currently one of the best solver. Numerical results show that our method is competitive on medium-scale instances and better on large-scale instances.

Le problème de conception de réseaux est un problème qui a été beaucoup étudié dans le domaine de la recherche opérationnelle pour ses caractéristiques, et ses applications dans des nombreux domaines tels que le transport, les communications, et la logistique. Nous nous intéressons en particulier dans ce mémoire à résoudre le problème de conception de réseaux avec coûts fixes et sans capacité, en satisfaisant les demandes de tous les produits tout en minimisant la somme des coûts de transport de ces produits et des coûts fixes de conception du réseau. Ce problème se modélise généralement sous la forme d’un programme linéaire en nombres entiers incluant des variables continues. Pour le résoudre, nous avons appliqué la méthode exacte de Branch-and-Bound basée sur une relaxation linéaire du problème avec un critère d’arrêt, tout en exploitant les méthodes de génération de colonnes et de génération de coupes. Nous avons testé la méthode de Branch-and-Price-and-Cut sur 156 instances divisées en cinq groupes de différentes tailles, et nous l’avons comparée à Cplex, l’un des meilleurs solveurs d’optimisation mathématique, ainsi qu’à la méthode de Branch-and- Cut. Notre méthode est compétitive et plus performante sur les instances de grande taille ayant un grand nombre de produits. / The network design problem has been studied extensively in the field of operational research given its characteristics and applications in many areas such as transportation, communications, and logistics. We are particularly interested in solving the multicommodity uncapacitated fixed-charge network design problem, with the aim of meeting the demands of all the products while minimizing the total cost of transporting commodities and designing the network. This problem is typically modeled as a linear integer program including continuous variables. To solve it, we applied the exact method of Branch-and-bound based on linear relaxation with a stopping criterion, while exploiting the column generation and cutting-plane methods. We tested our Branch-and-Price-and-Cut algorithm on 156 instances divided into five groups of different sizes, and we compared it with Cplex, one of the best mathematical optimization solvers. We compare it also with the Branch-and-Cut method. Numerical results show that our method is competitive and perform better especially on large-scale instances with many commodities.

Abstract: This work is concerned with the development and the implementation of a global optimization method for solving nonlinear nonconvex problems with continuous or mixed integer variables, related to power systems analysis. The proposed method relaxes the problem under study into a linear outer approximation problem by using the concept of special ordered sets. The obtained problem is then successively refined by a branch-and-bound strategy. In this way, the convergence to a global optimum is guaranteed, provided the discrete variables or those appearing nonlinearly in the original problem are bounded. Our method, conceived to solve a specific kind of problem, has been developed in a general framework in such a way that it can be easily extended to solve a large class of problems. We first derive the method theoretically and next present numerical results, fixing some choices inherent to the method to make it as optimal as possible. / Résumé: Ce travail a pour objet la conception et l'implémentation d'une méthode d'optimisation globale pour la résolution de problèmes non linéaires et non convexes, continus ou avec variables mixtes (entières et continues), issus de l'analyse des réseaux électriques. La méthode proposée relâche le problème traité en un problème d'approximation externe linéaire en se basant sur le concept d ensembles spécialement ordonnés. Le problème obtenu est alors successivement raffiné grâce à une stratégie de branch-and-bound. La convergence vers un optimum global est ainsi assurée, pour autant que les variables discrètes ou apparaissant non linéairement dans le problème de départ soient bornées. Notre méthode, mise au point pour résoudre un type de problème bien particulier, a été conçue dans un cadre général permettant une extension aisée à la résolution d'une grande variété de problèmes. Nous développons tout d'abord la méthode théoriquement et présentons ensuite des résultats numériques dont le but est de fixer certains choix inhérents à la méthode afin de la rendre la plus optimale possible.

Top-k and Skyline queries are a long study topic in database and information retrieval communities and they are two popular operations for preference retrieval. Top-k query returns a subset of the most relevant answers instead of all answers. Efficient top-k processing retrieves the k objects that have the highest overall score. In this paper, some algorithms that are used as a technique for efficient top-k processing for different scenarios have been represented. A framework based on existing algorithms with considering based cost optimization that works for these scenarios has been presented. This framework will be used when the user can determine the user ranking function. A real life scenario has been applied on this framework step by step. Skyline query returns a set of points that are not dominated (a record x dominates another record y if x is as good as y in all attributes and strictly better in at least one attribute) by other points in the given datasets. In this paper, some algorithms that are used for evaluating the skyline query have been introduced. One of the problems in the skyline query which is called curse of dimensionality has been presented. A new strategy that based on the skyline existing algorithms, skyline frequency and the binary tree strategy which gives a good solution for this problem has been presented. This new strategy will be used when the user cannot determine the user ranking function. A real life scenario is presented which apply this strategy step by step. Finally, the advantages of the top-k query have been applied on the skyline query in order to have a quickly and efficient retrieving results.

De nombreux problèmes en transport et en logistique peuvent être formulés comme des modèles de conception de réseau. Ils requièrent généralement de transporter des produits, des passagers ou encore des données dans un réseau afin de satisfaire une certaine demande tout en minimisant les coûts. Dans ce mémoire, nous nous intéressons au problème de conception de réseau avec coûts fixes et capacités. Ce problème consiste à ouvrir un sous-ensemble des liens dans un réseau afin de satisfaire la demande, tout en respectant les contraintes de capacités sur les liens. L'objectif est de minimiser les coûts fixes associés à l'ouverture des liens et les coûts de transport des produits. Nous présentons une méthode exacte pour résoudre ce problème basée sur des techniques utilisées en programmation linéaire en nombres entiers. Notre méthode est une variante de l'algorithme de branch-and-bound, appelée branch-and-price-and-cut, dans laquelle nous exploitons à la fois la génération de colonnes et de coupes pour la résolution d'instances de grande taille, en particulier, celles ayant un grand nombre de produits. En nous comparant à CPLEX, actuellement l'un des meilleurs logiciels d'optimisation mathématique, notre méthode est compétitive sur les instances de taille moyenne et supérieure sur les instances de grande taille ayant un grand nombre de produits, et ce, même si elle n'utilise qu'un seul type d'inégalités valides. / Many problems in transportation and logistics can be formulated as network design models. They usually require to transport commodities, passengers or data in a network to satisfy a certain demand while minimizing the costs. In this work, we focus on the multicommodity capacited fixed-charge network design problem which consists of opening a subset of the links in the network to satisfy the demand. Each link has a capacity and a fixed cost that is paid if it is opened. The objective is to minimize the fixed costs of the opened links and the transportation costs of the commodities. We present an exact method to solve this problem based on mixed integer programming techniques. Our method is a specialization of the branch-and-bound algorithm, called branch-and-price-and-cut, in which we use column generation and cutting-plane method to solve large-scale instances. We compare our method with CPLEX, currently one of the best solver. Numerical results show that our method is competitive on medium-scale instances and better on large-scale instances.