Due to legal, economic and socio-environmental factors, reverse logistics practices and extended producer responsibility have developed into a necessity in many countries. The end results and expectations may differ, but the motivation remains the same. Two significant components in a reverse logistics system -product recovery options and return channels - are the focus of this thesis. The two main issues examined are allocation of the returned products to recovery options, and selection of the collection methods for product returns. The initial segment of this thesis involves the formulation of a linear programming model to determine the optimal allocation of returned products differing in quality to specific recovery options. This model paves the way for a study on the effects of flexibility on product recovery allocation. A computational example utilising experimental data was presented to demonstrate the viability of the proposed model. The results revealed that in comparison to a fixed match between product qualities and recovery options, the product recovery operation appeared to be more profitable with a flexible allocation. The second segment of this thesis addresses the methods employed for the initial collection of returned products. A mixed integer nonlinear programming model was developed to facilitate the selection of optimal collection methods for these products. This integrated model takes three different initial collection methods into consideration. The model is used to solve an illustrative example optimally. However, as the complexity of the issue renders this process ineffective in the face of larger problems, the Lagrangian relaxation method was proposed to generate feasible solutions within reasonable computational times. This method was put to the test and the results were found to be encouraging.

This thesis is devoted to two $\mathcal{NP}$-complete combinatorial optimization problems arising in computational biology, the well-studied \emph{multiple sequence alignment} problem and the new formulated \emph{interval constraint coloring} problem. It shows that advanced mathematical programming techniques are capable of solving large scale real-world instances from biology to optimality. Furthermore, it reveals alternative methods that provide approximate solutions. In the first part of the thesis, we present a \emph{Lagrangian relaxation} approach for the multiple sequence alignment (MSA) problem. The multiple alignment is one common mathematical abstraction of the comparison of multiple biological sequences, like DNA, RNA, or protein sequences. If the weight of a multiple alignment is measured by the sum of the projected pairwise weights of all pairs of sequences in the alignment, then finding a multiple alignment of maximum weight is $\mathcal{NP}$-complete if the number of sequences is not fixed. The majority of the available tools for aligning multiple sequences implement heuristic algorithms; no current exact method is able to solve moderately large instances or instances involving sequences exhibiting a lower degree of similarity. We present a branch-and-bound (B\&B) algorithm for the MSA problem.\ignore{the multiple sequence alignment problem.} We approximate the optimal integer solution in the nodes of the B\&B tree by a Lagrangian relaxation of an ILP formulation for MSA relative to an exponential large class of inequalities, that ensure that all pairwise alignments can be incorporated to a multiple alignment. By lifting these constraints prior to dualization the Lagrangian subproblem becomes an \emph{extended pairwise alignment} (EPA) problem: Compute the longest path in an acyclic graph, that is penalized a charge for entering ``obstacles''. We describe an efficient algorithm that solves the EPA problem repetitively to determine near-optimal \emph{Lagrangian multipliers} via subgradient optimization. The reformulation of the dualized constraints with respect to additionally introduced variables improves the convergence rate dramatically. We account for the exponential number of dualized constraints by starting with an empty \emph{constraint pool} in the first iteration to which we add cuts in each iteration, that are most violated by the convex combination of a small number of preceding Lagrangian solutions (including the current solution). In this \emph{relax-and-cut} scheme, only inequalities from the constraint pool are dualized. The interval constraint coloring problem appears in the interpretation of experimental data in biochemistry. Monitoring hydrogen-deuterium exchange rates via mass spectroscopy is a method used to obtain information about protein tertiary structure. The output of these experiments provides aggregate data about the exchange rate of residues in overlapping fragments of the protein backbone. These fragments must be re-assembled in order to obtain a global picture of the protein structure. The interval constraint coloring problem is the mathematical abstraction of this re-assembly process. The objective of the interval constraint coloring problem is to assign a color (exchange rate) to a set of integers (protein residues) such that a set of constraints is satisfied. Each constraint is made up of a closed interval (protein fragment) and requirements on the number of elements in the interval that belong to each color class (exchange rates observed in the experiments). We introduce a polyhedral description of the interval constraint coloring problem, which serves as a basis to attack the problem by integer linear programming (ILP) methods and tools, which perform well in practice. Since the goal is to provide biochemists with all possible candidate solutions, we combine related solutions to equivalence classes in an improved ILP formulation in order to reduce the running time of our enumeration algorithm. Moreover, we establish the polynomial-time solvability of the two-color case by the integrality of the linear programming relaxation polytope $\mathcal{P}$, and also present a combinatorial polynomial-time algorithm for this case. We apply this algorithm as a subroutine to approximate solutions to instances with arbitrary but fixed number of colors and achieve an order of magnitude improvement in running time over the (exact) ILP approach. We show that the problem is $\mathcal{NP}$-complete for arbitrary number of colors, and we provide algorithms that, given an instance with $\mathcal{P}\neq\emptyset$, find a coloring that satisfies all the coloring requirements within $\pm 1$ of the prescribed value. In light of our $\mathcal{NP}$-completeness result, this is essentially the best one can hope for. Our approach is based on polyhedral theory and randomized rounding techniques. In practice, data emanating from the experiments are noisy, which normally causes the instance to be infeasible, and, in some cases, even forces $\mathcal{P}$ to be empty. To deal with this problem, the objective of the ILP is to minimize the total sum of absolute deviations from the coloring requirements over all intervals. The combinatorial approach for the two-color case optimizes the same objective function. Furthermore, we use this combinatorial method to compute, in a Lagrangian way, a bound on the minimum total error, which is exploited in a branch-and-bound manner to determine all optimal colorings. Alternatively, we study a variant of the problem in which we want to maximize the number of requirements that are satisfied. We prove that this variant is $\mathcal{APX}$-hard even in the two-color case and thus does not admit a polynomial time approximation scheme (PTAS) unless $\mathcal{P}=\mathcal{NP}$. Therefore, we slightly (by a factor of $(1+\epsilon)$) relax the condition on when a requirement is satisfied and propose a \emph{quasi-polynomial time approximation scheme} (QPTAS) which finds a coloring that ``satisfies'' the requirements of as many intervals as possible.

Dans ce travail nous considérons des problèmes de planification de production sur plusieurs sites avec transport de produits entre ces sites. L’objectif est de synchroniser les deux problèmes (planification et transport) et de construire une solution globale. Le système de production sur chaque site est modélisé comme un problème de Capacitated Lot-Sizing où nous travaillons avec stock et ressources. Le transport de produits entre les sites se ramène à une version simplifiée du Vehicle Routing Problem où le temps est discrétisé. D’abord nous proposons un modèle linéaire en nombres entiers que nous appelons le « Lot-Sizing and Vehicle Routing Problem » (LSVRP). Puis nous présentons deux cas particuliers : le Single-item LSVRP (SLSVRP) et le Single-level LSVRP (1-LSVRP). Les problèmes sont traités ici par six heuristiques que nous avons développé. Quatre de ces méthodes sont des heuristiques qui utilisent la programmation en nombres entiers et prennent en compte la relaxation linéaire de quelques variables du problème. Elles s’appuient sur l’exploration partielle de l’arbre de décision et la fixation de variables. Les deux autres sont spécifiques pour les cas particuliers. La première, qui traite le S-LSVRP, est basée sur la propagation des ordres de production sur chaque site. Puis à chaque itération elle calcule le plan de transport compatible et essaie d’améliorer la solution en modifiant la production sur les sites. L’autre méthode consiste en une relaxation lagrangienne qui travaille sur une modélisation du 1-LSVRP en un problème de flot. Des résultats numériques et des analyses sont présentés pour évaluer l’efficacité de ces heuristiques. / In this work we consider some problems of scheduling both a production distributed on several sites and the transportation of items between those sites. By doing so, the objective is to synchronize the two components and to build a better overall solution. The production system on each site is modeled as a Capacitated Lot-Sizing Problem where stock both on resources and produced items is available. The inter-site items transportation is a simplified version of the Vehicle Routing Problem where time is discretized. We first propose a mixed integer linear programming formulation that we call “The Lot-Sizing and Vehicle Routing Problem” (LSVRP). Then we present two particular cases : The Single-item LSVRP (S-LSVRP) and The Single-level LSVRP (1-LSVRP). All those cases are treated here by the six heuristics we develloped. Four of those methods are MIP based heuristics and take in account the the linear relaxation of some variables of the problem. They rely on partial decision tree exploration along with variable fixing. The other two are specifics for the two particular cases. The one who treats the S-LSVRP is based on production order propagation over the sites. Then, at each iteration, it computes a compatible transportation schedule and it tries to improve the solution by modifying the production on the sites. The other method consists in a lagrangian relaxation that works with an adaptation of the 1-LSVRP into a flow problem. Computational results and analysis are presented to evaluate the efficiency of those heuristics.

Este trabalho aborda o problema de dimensionamento portas lógicas e assinalamento de Vt para otimização de potência, área e temporização em circuitos integrados modernos. O fluxo proposto é aplicado aos conjuntos de circuitos de teste dos Concursos do International Symposium on Physical Design (ISPD) de 2012 e 2013. Este fluxo também é adapatado e avaliado nos estágios pós posicionamento e roteamento global em projetos industriais de circuitos integrados, que utilizam uma ferramenta precisa de análise estática de temporização. As técnicas propostas geram as melhores soluções para todos os circuitos de teste do Concurso do ISPD 2013 (no qual foi a ferramenta vencedora), com em média 8% menos consumo de potência estática quando comparada com os outros concorrentes. Além disso, após algumas modificações nos algoritmos, nós reduzimos o consumo em mais 10% em média a pontência estáticas com relação aos resultados do concurso. O foco deste trabalho é desenvolver e aplicar um algoritmo estado-da-arte de seleção portas lógicas para melhorar ainda mais projetos industriais de alto desempenho já otimizados após as fases de posicionamento e roteamento do fluxo de projeto físico industrial. Vamos apresentar e discutir vários problemas encontrados quando da aplicação de técnicas de otimização global em projetos industriais reais que não são totalmente cobertos em publicações encontradas na literatura. Os métodos propostos geram as melhores soluções para todos os circuitos de referência no Concurso do ISPD 2013, no qual foi a solução vencedora. Considerando a aplicação industrial, as técnicas propostas reduzem a potência estática em até 18,2 %, com redução média de 10,4 %, sem qualquer degradação na qualidade de temporização do circuito. / This work addresses the gate sizing and Vt assignment problem for power, area and timing optimization in modern integrated circuits (IC). The proposed flow is applied to the Benchmark Suites of the International Symposium on Physical Design (ISPD) 2012 and 2013 Contests. It is also adapted and evaluated in the post placement and post global routing stage of an industrial IC design flow using a sign-off static timing analysis engine. The proposed techniques are able to generate the best solutions for all benchmarks in the ISPD 2013 Contest (in which we were the winning team), with on average 8% lower leakage with respect to all other contestants. Also, after some refinements in the algorithms, we reduce leakage by another 10% on average over the contest results. The focus of this work is to develop and apply a state-of-the-art cell selection algorithm to further improve already optimized high-performance industrial designs after the placement and routing stages of the industrial physical design flow. We present the basic concepts involved in the gate sizing problem and how earlier literature addresses it. Several problems found when applying global optimization techniques in real-life industrial designs, which are not fully covered in publications found in literature, are presented and discussed. Considering the industrial application, the proposed techniques reduce leakage power by up to 18.2%, with average reduction of 10.4% without any degradation in timing quality.

Made available in DSpace on 2014-06-11T19:22:18Z (GMT). No. of bitstreams: 0 Previous issue date: 2001-02-17Bitstream added on 2014-06-13T19:07:26Z : No. of bitstreams: 1 fiorotto_dj_me_sjrp.pdf: 485977 bytes, checksum: 8cd2b3ba49a25a9c6a863795f27811c3 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O problema de dimensionamento de lotes é um problema de otimização da produção, em que o objetivo é planejar a quantidade de itens a ser produzida em várias, ou única, máquinas em cada período ao longo do horizonte de tempo, de modo a tender uma demanda e otimizar uma função objetivo. Este trabalho aborda o problema de dimensionamento de lotes em um único estágio em um ambiente com máquinas paralelas distintas. Cada item pode ser produzido em qualquer máquina, acarretando um tempo de preparação que é gasto antes de começar a produção. O objetivo do trabalho consiste em obter limitantes inferiores de boa qualidade para este problema. Para tanto, é desenvolvido um método de solução baseado numa reformulação do problema a e na relaxação lagrangiana de um conjunto de restrições. Alguns resultados computacionais são apresentados algumas propostas futuras para a continuidade do trabalho. / The lot-sizing problem is a production optimization problem, where the objective is to plan the quantity of items to be produced in multiple, or single, machines in each period over a time horizon, in order to satisfy a demand and optimize an objective function. This work addresses the single stage parallel machine lot-sizing problem. Each item can be produced on any machine, and incur a setup time before to start the production. The objective of this work is to lower bounds of good quality for this problem. A solution method is developed based on a reformulation of the problem and the Lagrangian relaxation of a set of constrainsts. Some computational results are presented comparing the proposed method with a method from the literature, and, some future researches are proposed.

Ferramentas de projeto de circuitos integrados (do inglˆes, electronic design automation, ou simplesmente EDA) tˆem um papel fundamental na crescente complexidade dos projetos de circuitos digitais. Elas permitem aos projetistas criar circuitos com um n´umero de componentes ordens de grandezas maior do que seria poss´ıvel se os circuitos fossem projetados `a m˜ao como nos dias iniciais da microeletrˆonica. Neste trabalho, dois importantes problemas em EDA ser˜ao abordados: dimensionamento de portas e posicionamento detalhado dirigido a desempenho. Para dimensionamento de portas, uma nova metodologia de relaxac¸ ˜ao Lagrangiana ´e apresentada baseada em informac¸ ˜ao de temporarizac¸ ˜ao locais e propagac¸ ˜ao de sensitividades. Para posicionamento detalhado dirigido a desempenho, um conjunto de movimentos de c´elulas ´e criado usando uma formac¸ ˜ao ´otima atenta `a forc¸a de alimentac¸ ˜ao para o balanceamento de cargas. Nossos resultados experimentais mostram que tais t´ecnicas s˜ao capazes de melhorar o atual estado-da-arte. / Electronic design automation (EDA) tools play a fundamental role in the increasingly complexity of digital circuit designs. They empower designers to create circuits with several order of magnitude more components than it would be possible by designing circuits by hand as was done in the early days of microelectronics. In this work, two important EDA problems are addressed: gate sizing and timing-driven detailed placement. They are studied and new techniques developed. For gate sizing, a new Lagrangian-relaxation methodology is presented based on local timing information and sensitivity propagation. For timing-driven detailed placement, a set of cell movement methods are created using drive strength-aware optimal formulation to driver/sink load balancing. Our experimental results shows that those techniques are able to improve the current state-of-the-art.

<p>In this thesis a method for solving the Generalized Assignment Problem (GAP) is described. It is based on a reformulation of the original problem into a Set Partitioning Problem (SPP), in which the columns represent partial solutions to the original problem. For solving this problem, column generation, with systematic overgeneration of columns, is used. Conditions that guarantee that an optimal solution to a restricted SPP is optimal also in the original problem are given. In order to satisfy these conditions, not only columns with the most negative Lagrangian reduced costs need to be generated, but also others; this observation leads to the use of overgeneration of columns.</p><p>The Generalized Assignment Problem has shown to be NP-hard and therefore efficient algorithms are needed, especially for large problems. The application of the proposed method decomposes GAP into several knapsack problems via Lagrangian relaxation, and enumerates solutions to each of these problems. The solutions obtained from the knapsack problems form a Set Partitioning Problem, which consists of combining one solution from each knapsack problem to obtain a solution to the original problem. The algorithm has been tested on problems with 10 agents and 60 jobs. This leads to 10 knapsack problems, each with 60 variables.</p>

There is an increasing interest in sustainability and a growing debate about environmental policy measures aiming at the reduction of green house gas emissions across di erent economic sectors worldwide. The transportation sector is one major greenhouse gas emitter which is heavily regulated to reduce its dependance on oil. These regulations along with the growing customer awareness about global warming has led vehicle manufacturers to seek di erent technologies to improve vehicle e ciencies and reduce the green house gases emissions while at the same time meeting customer's expectation of mobility and exibility. Plug-in hybrid electric vehicles (PHEV) is one major promising solution for a smooth transition from oil dependent transportation sector to a clean electric based sector while not compromising the mobility and exibility of the drivers. In the medium term, plug-in hybrid electric vehicles (PHEV) can lead to signi cant reductions in transportation emissions. These vehicles are equipped with a larger battery than regular hybrid electric vehicles which can be recharged from the grid. For short trips, the PHEV can depend solely on the electric engine while for longer journeys the alternative fuel can assist the electric engine to achieve extended ranges. This is bene cial when the use pattern is mixed such that and short long distances needs to be covered. The plug-in hybrid electric vehicles are well-suited for logistics since they can avoid the possible disruption caused by charge depletion in case of all-electric vehicles with tight time schedules. The use of electricity and fuel gives rise to a new variant of the classical vehicle routing with time windows which we call the plug-in hybrid electric vehicle routing problem with time windows (PHEVRPTW). The objective of the PHEVRPTW is to minimize the routing costs of a eet of PHEVs by minimizing the time they run on gasoline while meeting the demand during the available time windows. As a result, the driver of the PHEV has two decisions to make at each node: (1) recharge the vehicle battery to achieve a longer range using electricity, or (2) continue to the next open time window with the option of using the alternative fuel. In this thesis, we present a mathematical formulation for the plug-in hybrid-electric vehicle routing problem with time windows. We solve this problem using a Lagrangian relaxation and we propose a new tabu search algorithm. We also present the rst results for the full adapted Solomon instances.

In this thesis a method for solving the Generalized Assignment Problem (GAP) is described. It is based on a reformulation of the original problem into a Set Partitioning Problem (SPP), in which the columns represent partial solutions to the original problem. For solving this problem, column generation, with systematic overgeneration of columns, is used. Conditions that guarantee that an optimal solution to a restricted SPP is optimal also in the original problem are given. In order to satisfy these conditions, not only columns with the most negative Lagrangian reduced costs need to be generated, but also others; this observation leads to the use of overgeneration of columns. The Generalized Assignment Problem has shown to be NP-hard and therefore efficient algorithms are needed, especially for large problems. The application of the proposed method decomposes GAP into several knapsack problems via Lagrangian relaxation, and enumerates solutions to each of these problems. The solutions obtained from the knapsack problems form a Set Partitioning Problem, which consists of combining one solution from each knapsack problem to obtain a solution to the original problem. The algorithm has been tested on problems with 10 agents and 60 jobs. This leads to 10 knapsack problems, each with 60 variables.

The expansion of supply chains into global networks has drastically increased the distance travelled along shipping lanes in a logistics system. Inherently, the increase in travel distances produces increased carbon emissions from transport vehicles. When increased emissions are combined with a carbon tax or emissions trading system, the result is a supply chain with increased costs attributable to the emission generated on the transportation routes. Most traditional supply chain design models do not take emissions and carbon costs into account. Hence, there is a need to incorporate emission costs into a supply chain optimization model to see how the optimal supply chain configuration may be affected by the additional expenses. This thesis presents a mathematical programming model for the design of green supply chains. The costs of carbon dioxide (CO2) emissions were incorporated in the objective function, along with the fixed and transportation costs that are typically modeled in traditional facility location models. The model also determined the unit flows between the various nodes of the supply chain, with the objective of minimizing the total cost of the system by strategically locating warehouses throughout the network. The literature shows that CO2 emissions produced by a truck are dependent on the weight of the vehicle and can be modeled using a concave function. Hence, the carbon emissions produced along a shipping lane are dependent upon the number of units and the weight of each unit travelling between the two nodes. Due to the concave nature of the emissions, the addition of the emission costs to the problem formulation created a nonlinear mixed integer programming (MIP) model. A solution algorithm was developed to evaluate the new problem formulation. Lagrangian relaxation was used to decompose the problem by echelon and by potential warehouse site, resulting in a problem that required less computational effort to solve and allowed for much larger problems to be evaluated. A method was then suggested to exploit a property of the relaxed formulation and transform the problem into a linear MIP problem. The solution method computed the minimum cost for a complete network that would satisfy all the needs of the customers. A primal heuristic was introduced into the Lagrangian algorithm to generate feasible solutions. The heuristic utilized data from the Lagrangian subproblems to produce good feasible solutions. Due to the many characteristics of the original problem that were carried through to the subproblems, the heuristic produced very good feasible solutions that were typically within 1% of the Lagrangian bound. The proposed algorithm was evaluated through a number of tests. The rigidity of the problem and cost breakdown were varied to assess the performance of the solution method in many situations. The test results indicated that the addition of emission costs to a network can change the optimal configuration of the supply chain. As such, this study concluded that emission costs should be considered when designing supply chains in jurisdictions with carbon costs. Furthermore, the tests revealed that in regions without carbon costs it may be possible to significantly reduce the emissions produced by the supply chain with only a small increase in the cost to operate the system.