Thesis (M. Sc.)--University of Canterbury, 2008. / Typescript (photocopy). "January 2008." Includes bibliographical references (p. [121]-128). Also available via the World Wide Web.

Thesis (M. Sc.)--University of Canterbury, 2009. / Typescript (photocopy). "September 2009." Includes bibliographical references (leaves 89-96). Also available via the World Wide Web.

Thesis (Ph. D.)--University of Canterbury, 2006. / Typescript (photocopy). Includes bibliographical references (p. 246-255). Also available via the World Wide Web.

Les problèmes de tournées de véhicules sont des problèmes d’optimisation combinatoire épineux avec des enjeux économiques et environnementaux importants au sein de la chaîne logistique. Le problème fondamental est de desservir des clients avec un ensemble de véhicules de façon à minimiser la distance totale parcourue. En pratique, il y a une grande variété d’objectifs et de contraintes additionnelles, liées à la législation et à la diversité des domaines d’applications. Ces problèmes de tournées sont très fréquents pour de nombreuses industries et la conception d’approches de résolution génériques est devenue une question de recherche importante.Cette thèse porte sur la conception et le développement d’un nouveau moteur de résolution pour les logiciels de tournées de véhicules proposés par l’entreprise GEOCONCEPT. Le solveur mis au point s’appuie sur la programmation par contraintes (PPC) pour améliorer la flexibilité (prise en compte de contraintes additionnelles), la déclarativité et la maintenance qui sont les limites des solveurs actuels de GEOCONCEPT fondés sur la recherche locale.Dans un premier temps, un modèle de graphe est établi pour la représentation unifiée des données et de nombreuses contraintes métiers. La résolution s’effectue par des approches à base de voisinage large disponibles dans les solveurs de PPC modernes. On peut ainsi traiter des instances de très grandes tailles efficacement tout en conservant une approche déclarative pour exprimer une classe très large de problèmes de tournées de véhicules. Dans un second temps, des modèles PPC s’appuyant sur des représentations redondantes du problème sont proposés afin de renforcer le filtrage. Nous nous intéressons en détails aux mécanismes de filtrage c’est-à-dire aux processus d’élimination des valeurs infaisables ou sous-optimales dans les domaines des variables. Ces algorithmes permettent de simplifier rapidement le problème et de fournir des bornes inférieures afin d’évaluer la qualité des solutions obtenues. Les bornes inférieures sont obtenues en résolvant des relaxations du plus célèbre des problèmes de la Recherche Opérationnelle : le problème du voyageur de commerce (TSP). Ce problème est le cœur de la contrainte globale weightedcircuit permettant de modéliser les problèmes de tournées en PPC. Nous proposons de nouveaux mécanismes de filtrage pour cette contrainte s’appuyant sur trois relaxations du TSP. Ces relaxations sont comparées sur les plans théorique et expérimental. L’originalité de ce travail est de proposer un nouvel algorithme de filtrage permettant de raisonner à la fois sur les successeurs directs d’un client et sur sa position dans la tournée. Ces raisonnements sont particulièrement utiles en présence de contraintes de fenêtres de temps, très communes dans les problèmes industriels.Le nouveau moteur de résolution offre d’excellentes performances sur des problèmes académiques et industriels tout en proposant des bornes inférieures informatives à des problèmes industriels réels. / Vehicle routing problems are very hard combinatorial optimization problems with significant economic and environmental challenges. The fundamental problem is to visit a set of customers with a given fleet of vehicles in order to minimize the total distance travelled. Moreover, these problems arise with a wide variety of objectives and additional constraints, related to the legislation and the diversity of industrial sectors. They are very common for many industries and the design of generic solvers has become an important research issue.This thesis focuses on the design and implementation of a new solver for the vehicle routing services offered by the company GEOCONCEPT. The proposed solver is based on constraint programming (CP) to improve flexibility (ability to take additional constraints into account), declarative modelling and maintenance, which are the limits of current GEOCONCEPT solvers based on local search.Firstly, a graph model is established to provide a common representation of the input-data and the numerous business constraints. The resolution is performed using large neighbourhood search methods available in modern CP solvers. It is thus possible to deal with large instances efficiently with a declarative approach where a broad class of vehicle routing problems can be modelled. Secondly, several CP models based on redundant views of the problem are proposed to strengthen the filtering. We focus on the filtering mechanisms for removing infeasible or suboptimal values in the domains of the variables. These algorithms can quickly simplify the problem and derive lower bounds to assert the quality of the solutions found. The lower bounds are obtained by solving relaxations of the most famous problem in Operations Research: the Traveling Salesman Problem (TSP). This problem is the core of the global constraint WEIGTEHDCIRCUIT for modelling routing problems in CP. We propose new filtering algorithms for this constraint based on three relaxations of the TSP. These relaxations are compared theoretically and experimentally. The originality of this work is to propose a new filtering algorithm for reasoning on the direct successors of a customer as well as his position in the tour. It is particularly useful in the presence of time window constraints, which are very common in industrial problems.The new solver shows excellent performance on academic and industrial problems and can compute informative lower bounds for real-life problems.

Constraint Programming (CP) is a powerful technique for solving a variety of combinatorial problems. Automated modelling using a refinement based approach abstracts over modelling decisions in CP by allowing users to specify their problem in a high level specification language such as ESSENCE. This refinement process produces many models resulting from different choices that can be selected, each with their own strengths. A parameterised specification represents a problem class where the parameters of the class define the instance of the class we wish to solve. Since each model has different performance characteristics the model chosen is crucial to be able to solve the instance effectively. This thesis presents a method to generate instances automatically for the purpose of choosing a subset of the available models that have superior performance across the instance space. The second contribution of this thesis is a framework to automate the testing of a toolchain for automated modelling. This process includes a generator of test cases that covers all aspects of the ESSENCE specification language. This process utilises our first contribution namely instance generation to generate parameterised specifications. This framework can detect errors such as inconsistencies in the model produced during the refinement process. Once we have identified a specification that causes an error, this thesis presents our third contribution; a method for reducing the specification to a much simpler form, which still exhibits a similar error. Additionally this process can generate a set of complementary specifications including specifications that do not cause the error to help pinpoint the root cause.

A incerteza é um elemento fundamental da realidade. Então, torna-se natural a busca por métodos que nos permitam representar o desconhecido em termos matemáticos. Esses problemas originam uma grande classe de programas probabilísticos reconhecidos como modelos de programação estocástica. Eles são mais realísticos que os modelos determinísticos, e tem por objetivo incorporar a incerteza em suas definições. Essa tese aborda os problemas probabilísticos da classe de problemas de multi-estágio com incerteza e com restrições probabilísticas e com restrições probabilísticas conjuntas. Inicialmente, nós propomos um modelo de administração de ativos e passivos multi-estágio estocástico para a indústria de fundos de pensão brasileira. Nosso modelo é formalizado em conformidade com a leis e políticas brasileiras. A seguir, dada a relevância dos dados de entrada para esses modelos de otimização, tornamos nossa atenção às diferentes técnicas de amostragem. Elas compõem o processo de discretização desses modelos estocásticos Nós verificamos como as diferentes metodologias de amostragem impactam a solução final e a alocação do portfólio, destacando boas opções para modelos de administração de ativos e passivos. Finalmente, nós propomos um “framework” para a geração de árvores de cenário e otimização de modelos com incerteza multi-estágio. Baseados na tranformação de Knuth, nós geramos a árvore de cenários considerando a representação filho-esqueda, irmão-direita o que torna a simulação mais eficiente em termos de tempo e de número de cenários. Nós também formalizamos uma reformulação do modelo de administração de ativos e passivos baseada na abordagem extensiva implícita para o modelo de otimização. Essa técnica é projetada pela definição de um processo de filtragem com “bundles”; e codifciada com o auxílio de uma linguagem de modelagem algébrica. A eficiência dessa metodologia é testada em um modelo de administração de ativos e passivos com incerteza com restrições probabilísticas conjuntas. Nosso framework torna possível encontrar a solução ótima para árvores com um número razoável de cenários. / Uncertainty is a key element of reality. Thus, it becomes natural that the search for methods allows us to represent the unknown in mathematical terms. These problems originate a large class of probabilistic programs recognized as stochastic programming models. They are more realistic than deterministic ones, and their aim is to incorporate uncertainty into their definitions. This dissertation approaches the probabilistic problem class of multistage stochastic problems with chance constraints and joint-chance constraints. Initially, we propose a multistage stochastic asset liability management (ALM) model for a Brazilian pension fund industry. Our model is formalized in compliance with the Brazilian laws and policies. Next, given the relevance of the input parameters for these optimization models, we turn our attention to different sampling models, which compose the discretization process of these stochastic models. We check how these different sampling methodologies impact on the final solution and the portfolio allocation, outlining good options for ALM models. Finally, we propose a framework for the scenario-tree generation and optimization of multistage stochastic programming problems. Relying on the Knuth transform, we generate the scenario trees, taking advantage of the left-child, right-sibling representation, which makes the simulation more efficient in terms of time and the number of scenarios. We also formalize an ALM model reformulation based on implicit extensive form for the optimization model. This technique is designed by the definition of a filtration process with bundles, and coded with the support of an algebraic modeling language. The efficiency of this methodology is tested in a multistage stochastic ALM model with joint-chance constraints. Our framework makes it possible to reach the optimal solution for trees with a reasonable number of scenarios.

A incerteza é um elemento fundamental da realidade. Então, torna-se natural a busca por métodos que nos permitam representar o desconhecido em termos matemáticos. Esses problemas originam uma grande classe de programas probabilísticos reconhecidos como modelos de programação estocástica. Eles são mais realísticos que os modelos determinísticos, e tem por objetivo incorporar a incerteza em suas definições. Essa tese aborda os problemas probabilísticos da classe de problemas de multi-estágio com incerteza e com restrições probabilísticas e com restrições probabilísticas conjuntas. Inicialmente, nós propomos um modelo de administração de ativos e passivos multi-estágio estocástico para a indústria de fundos de pensão brasileira. Nosso modelo é formalizado em conformidade com a leis e políticas brasileiras. A seguir, dada a relevância dos dados de entrada para esses modelos de otimização, tornamos nossa atenção às diferentes técnicas de amostragem. Elas compõem o processo de discretização desses modelos estocásticos Nós verificamos como as diferentes metodologias de amostragem impactam a solução final e a alocação do portfólio, destacando boas opções para modelos de administração de ativos e passivos. Finalmente, nós propomos um “framework” para a geração de árvores de cenário e otimização de modelos com incerteza multi-estágio. Baseados na tranformação de Knuth, nós geramos a árvore de cenários considerando a representação filho-esqueda, irmão-direita o que torna a simulação mais eficiente em termos de tempo e de número de cenários. Nós também formalizamos uma reformulação do modelo de administração de ativos e passivos baseada na abordagem extensiva implícita para o modelo de otimização. Essa técnica é projetada pela definição de um processo de filtragem com “bundles”; e codifciada com o auxílio de uma linguagem de modelagem algébrica. A eficiência dessa metodologia é testada em um modelo de administração de ativos e passivos com incerteza com restrições probabilísticas conjuntas. Nosso framework torna possível encontrar a solução ótima para árvores com um número razoável de cenários. / Uncertainty is a key element of reality. Thus, it becomes natural that the search for methods allows us to represent the unknown in mathematical terms. These problems originate a large class of probabilistic programs recognized as stochastic programming models. They are more realistic than deterministic ones, and their aim is to incorporate uncertainty into their definitions. This dissertation approaches the probabilistic problem class of multistage stochastic problems with chance constraints and joint-chance constraints. Initially, we propose a multistage stochastic asset liability management (ALM) model for a Brazilian pension fund industry. Our model is formalized in compliance with the Brazilian laws and policies. Next, given the relevance of the input parameters for these optimization models, we turn our attention to different sampling models, which compose the discretization process of these stochastic models. We check how these different sampling methodologies impact on the final solution and the portfolio allocation, outlining good options for ALM models. Finally, we propose a framework for the scenario-tree generation and optimization of multistage stochastic programming problems. Relying on the Knuth transform, we generate the scenario trees, taking advantage of the left-child, right-sibling representation, which makes the simulation more efficient in terms of time and the number of scenarios. We also formalize an ALM model reformulation based on implicit extensive form for the optimization model. This technique is designed by the definition of a filtration process with bundles, and coded with the support of an algebraic modeling language. The efficiency of this methodology is tested in a multistage stochastic ALM model with joint-chance constraints. Our framework makes it possible to reach the optimal solution for trees with a reasonable number of scenarios.

SPARQL est le langage de requête standard pour les graphes de données du Web Sémantique. L’évaluation de requêtes est étroitement liée aux problèmes d’appariement de graphes. Il a été démontré que l’évaluation est NP-difficile. Les moteurs SPARQLde l’état-de-l’art résolvent les requêtes SPARQL en utilisant des techniques de bases de données traditionnelles. Cette approche est efficace pour les requêtes simples qui fournissent un point de départ précis dans le graphe. Par contre, les requêtes couvrant tout le graphe et impliquant des conditions de filtrage complexes ne passent pas bien à l’échelle. Dans cette thèse, nous proposons de résoudre les requêtes SPARQL en utilisant la Programmation par Contraintes (CP). La CP résout un problème combinatoire enexploitant les contraintes du problème pour élaguer l’arbre de recherche quand elle cherche des solutions. Cette technique s’est montrée efficace pour les problèmes d’appariement de graphes. Nous reformulons la sémantique de SPARQL en termes deproblèmes de satisfaction de contraintes (CSPs). Nous appuyant sur cette sémantique dénotationnelle, nous proposons une sémantique opérationnelle qui peut être utilisée pour résoudre des requêtes SPARQL avec des solveurs CP génériques.Les solveurs CP génériques ne sont cependant pas conçus pour traiter les domaines immenses qui proviennent des base de données du Web Sémantique. Afin de mieux traiter ces masses de données, nous introduisons Castor, un nouveau moteurSPARQL incorporant un solveur CP léger et spécialisé. Nous avons apporté une attention particulière à éviter tant que possible les structures de données et algorithmes dont la complexité temporelle ou spatiale est proportionnelle à la taille de la base dedonnées. Des évaluations expérimentales sur des jeux d’essai connus ont montré la faisabilité et l’efficacité de l’approche. Castor est compétitif avec des moteurs SPARQL de l’état-de-l’art sur des requêtes simples, et les surpasse sur des requête. / SPARQL is the standard query language for graphs of data in the SemanticWeb. Evaluating queries is closely related to graph matching problems, and has been shown to be NP-hard. State-of-the-art SPARQL engines solve queries with traditional relational database technology. Such an approach works well for simple queries that provide a clearly defined starting point in the graph. However, queries encompassing the whole graph and involving complex filtering conditions do not scale well. In this thesis we propose to solve SPARQL queries with Constraint Programming (CP). CP solves a combinatorial problem by exploiting the constraints of the problem to prune the search tree when looking for solutions. Such technique has been shown to work well for graph matching problems. We reformulate the SPARQL semantics by means of constraint satisfaction problems (CSPs). Based on this denotational semantics, we propose an operational semantics that can be used by off-theshelf CP solvers. Off-the-shelf CP solvers are not designed to handle the huge domains that come with SemanticWeb databases though. To handle large databases, we introduce Castor, a new SPARQL engine embedding a specialized lightweight CP solver. Special care has been taken to avoid as much as possible data structures and algorithms whosetime or space complexity are proportional to the database size. Experimental evaluations on well-known benchmarks show the feasibility and efficiency of the approach. Castor is competitive with state-of-the-art SPARQL engines on simple queries, and outperforms them on complex queries where filters can be actively exploited during the search.

Orientadores: Arnaldo Vieira Moura, Cid Carvalho de Souza / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-16T11:05:11Z (GMT). No. of bitstreams: 1 Lopes_TonyMinoruTamura_M.pdf: 1334472 bytes, checksum: c53fc4d807764c31b73609102203a039 (MD5) Previous issue date: 2010 / Resumo: Um conjunto de órgãos distribuidores de derivados de petróleo, incluindo refinarias e terminais, possuem demandas locais e produções de diferentes produtos ao longo de um dado horizonte de tempo. No entanto, pode não haver estoque local de algum produto para satisfazer a demanda correspondente, ou pode não haver espaço nos tanques para estocar uma produção local. Isso leva à necessidade de transporte dos derivados de petróleo entre os órgãos. Dentre os diversos modais, a rede de oleodutos é a melhor opção considerando-se custos e riscos ambientais. Em vista de sua grande complexidade operacional, um uso adequado da rede necessita de um planejamento tático composto mensalmente, e de um agendamento detalhado das operações, cobrindo poucos dias, e que deve ser atualizado diariamente. Tanto o planejamento mensal quanto o agendamento diário devem respeitar um grande conjunto de restrições, envolvendo a capacidade dos tanques, taxas de vazões nos oleodutos, níveis de estoques, dentre outras. Esta dissertação apresenta uma formalização do problema, desenvolvida em dois estágios, representado o planejamento mensal e o agendamento diário. O problema de planejamento recebeu um tratamento inicial heurístico seguido de uma modelagem por fluxo em redes, enquanto o agendamento diário utilizou programação por restrições. Os modelos foram testados sobre dados fornecidos pela companhia brasileira de petróleo Petrobras. Essas instâncias possuem uma das topologias mais complexas quando comparadas a outras redes encontrada na literatura aberta. Os resultados demonstram melhorias significativas sobre a resolução manual desses problemas / Abstract: A set of oil derivative distribution depots, including refineries and terminals, have local demands for and productions of different products in a given time horizon. However, there may be not enough local stock of some product to satisfy the corresponding demand, or there may not be enough tank capacity to stock the local production. This brings the need for transportation of oil derivatives between the depots. Among many transportation modes, the network of pipelines is one of the best options when considerying cost and environment risks. In order to adequately operate the pipeline network, a two phase planning strategy is developed. First, a tactical pumping plan is composed monthly and, secondly, a more detailed operational schedule, spanning a few days, is updated daily. Both the tactical and tghe operational plannings must satisfy a large set of operation constraints, involving many restrictions, such as tanks capacities, pipeline flow rates, and stock levels. This dissertation provides a formalization for the problem along with a decomposition of it in two stages, representing the monthly planning and operational schedule. The tactical stage is solved by applying a heuristic and then with a network flow model, while the operational schedule uses constraing programming. Our model treats the oil pipeline network that is operated by the Brazilian oil company Petrobras. This is one of the most complex and large topologies when compared to other networks treated in the open literature. The model was tested with real-world instances and showed significant improvements over human planning / Mestrado / Ciência da Computação / Mestre em Ciência da Computação

Constraint Programming can be used to solve many problems and this thesis is about getting an overview on the usage of Constraint Programming for Constraint Satisfaction Problems, both interactive and explorative. Many problems can be mathematically modeled as a Constraint Satisfaction Problem but this thesis will focus on the Nurse Rostering Problem since it is a well-studied area. The problem when creating a schedule for nurses is that it can easily be over-constrained and a solution could be hard to find. This thesis will investigate whether if Constraint Programming is a good technique for solving the Nurse Rostering Problem but also if user interaction is considered when solving the problem. The method for this is a literature study where a number of research articles has been reviewed and categorized, and resulted in 27 different kinds of sources that were used. The conclusion is that there exists better ways to solve these problems than the use of pure Constraint Programming. To answer the second part of the thesis, it seems like the solution for the problem is the main focus and therefore is user interaction something that is given less attention.