This thesis focuses on algorithms solving the on-line Bin-Covering problem, when the items are generated from a known, stationary distribution. We introduce the Prospect Algorithm. The main idea behind the Prospect Algorithm is to use information on the item distribution to estimate how easy it will be to fill a bin with small overfill as a function of the empty space left in it. This estimate is then used to determine where to place the items, so that all active bins either stay easily fillable, or are finished with small overfill. We test the performance of the algorithm by simulation, and discuss how it can be modified to cope with additional constraints and extended to solve the Bin-Packing problem as well. The Prospect Algorithm is then adapted to achieve perfect packing, yielding a new version, the Prospect+ Algorithm, that is a slight but consistent improvement. Next, a Markov Decision Process formulation is used to obtain an optimal Bin-Covering algorithm to compare with the Prospect Algorithm. Even though the optimal algorithm can only be applied to limited (small) cases, it gives useful insights that lead to another modification of the Prospect Algorithm. We also discuss two relaxations of the on-line constraint, and describe how algorithms that are based on solving the Subset-Sum problem are used to tackle these relaxed problems. Finally, several practical issues encountered when using the Prospect Algorithm in the real-world are analyzed, a computationally efficient way of doing the background calculations needed for the Prospect Algorithm is described, and the three versions of the Prospect Algorithm developed in this thesis are compared.

Thesis (PhD (Logistics))--University of Stellenbosch, 2007. / In this dissertation original algorithms are introduced to solve separable resource allocation problems (RAPs) with increasing nonlinear functions in the objective function, and lower and upper bounds on each variable. Algorithms are introduced in three special cases. The first case arises when the objective function of the RAP consists of the sum of convex functions and all the variables for these functions range over the same interval. In the second case RAPs with the sum of convex functions in the objective function are considered, but the variables of these functions can range over different intervals. In the last special case RAPs with an objective function comprising the sum of convex and concave functions are considered. In this case the intervals of the variables can range over different values. In the first case two new algorithms, namely the fraction and the slope algorithm are presented to solve the RAPs adhering to the conditions of the case. Both these algorithms yield far better solution times than the existing branch and bound algorithm. A new heuristic and three new algorithms are presented to solve RAPs falling into the second case. The iso-bound heuristic yields, on average, good solutions relative to the optimal objective function value in faster times than exact algorithms. The three algorithms, namely the iso-bound algorithm, the branch and cut algorithm and the iso-bound branch and cut algorithm also yield considerably beter solution times than the existing branch and bound algorithm. It is shown that, on average, the iso-bound branch and cut algorithm yields the fastest solution times, followed by the iso-bound algorithm and then by die branch and cut algorithm. In the third case the necessary and sufficient conditions for optimality are considered. From this, the conclusion is drawn that search techniques for points complying with the necessary conditions will take too long relative to branch and bound techniques. Thus three new algorithms, namely the KL, SKL and IKL algorithms are introduced to solve RAPs falling into this case. These algorithms are generalisations of the branch and bound, branch and cut, and iso-bound algorithms respectively. The KL algorithm was then used as a benchmark. Only the IKL algorithm yields a considerable improvement on the KL algorithm.

O Problema da Mochila Compartimentada é uma extensão do Problema da Mochila, em que os itens solicitados são divididos em classes, de modo que a mochila deve ser subdividida em compartimentos, os quais têm capacidades limitadas e são carregados com itens da mesma classe. Além disso, a construção de um compartimento tem um custo fixo e ocasiona uma perda no espaço da mochila. O objetivo consiste em maximizar a soma dos valores dos itens, descontado o custo fixo de inclusão de compartimentos. Neste trabalho, são abordados dois métodos de solução. A primeira abordagem é uma heurística, que consiste na combinação de duas heurísticas da literatura. A segunda abordagem é o método Geração de Colunas, que além de fornecer um novo limitante superior para o Problema da Mochila Compartimentada, ao final do método o problema mestre foi resolvido com as variáveis definidas como inteiras, obtendo uma solução factível. Em ambos os métodos, o modelo não-linear é decomposto em dois modelos lineares, no qual, um gera compartimentos e o outro os seleciona. Os resultados obtidos com as duas abordagens foram comparados com um limitante superior e se mostraram bastante satisfatórios / The Compartmentalized Knapsack Problem is an extension of the classical Knapsack Problem, where the ordered items are partitioned into classes, in such way that the knapsack must be divided into compartments, each one having limited capacity. In addition, the building of a compartment has a fixed cost and involves a loss of the overall capacity. The objective is to maximize the sum of the items utility value, minus the fixed costs of the compartments. This dissertation presents two solving methods. The first approach is a heuristic method, which is a combination of two heuristics from the literature. The second approach is a Column Generation method, that apart from it gives a new upper bound to the Compartmentalized Knapsack Problem, in the end of the method the master problem was solved with the variables defined as integer, that supplies a feasible solution. In both methods, the mathematical non linear model is decomposed into two linear models, one generates the compartments, and the other selects them to compose the knapsack. The results obtained with these two approaches were compared with an upper bound and they showed very efficient

Nesse trabalho, estudamos um problema de otimização combinatorial conhecido por Problema da Mochila Compartimentada, que é uma extensão do clássico Problema da Mochila. O problema consiste em determinar as capacidades adequadas de vários compartimentos que podem vir a ser alocados em uma mochila e como esses compartimentos devem ser carregados, respeitando as restrições de capacidades dos compartimentos e da mochila. Busca-se maximizar o valor de utilidade total. O problema é muito pouco estudado na literatura, apesar de surgir naturalmente em aplicações práticas. Nesse estudo, propomos uma modelagem matemática não linear para o problema e verificamos algumas heurísticas para sua resolução. / In this work, we studied a combinatorial optimization problem called the Clustered Knapsack Problem, that is an extension of the standard Knapsack Problem. The problem is to determine the right capacities of several clusters which can be allocated in a knapsack and how these clusters should be placed so as to respect the constraints on the capacities of the clusters and the knapsack. The objective is to maximize a total utility value. The problem has seldom been studied in the literature, even though it appears naturally in practical applications. In this study, we propose a non-linear model for the problem and we insert some heuristics for its resolution.

O Problema da Mochila Compartimentada é uma extensão do Problema da Mochila, em que os itens solicitados são divididos em classes, de modo que a mochila deve ser subdividida em compartimentos, os quais têm capacidades limitadas e são carregados com itens da mesma classe. Além disso, a construção de um compartimento tem um custo fixo e ocasiona uma perda no espaço da mochila. O objetivo consiste em maximizar a soma dos valores dos itens, descontado o custo fixo de inclusão de compartimentos. Neste trabalho, são abordados dois métodos de solução. A primeira abordagem é uma heurística, que consiste na combinação de duas heurísticas da literatura. A segunda abordagem é o método Geração de Colunas, que além de fornecer um novo limitante superior para o Problema da Mochila Compartimentada, ao final do método o problema mestre foi resolvido com as variáveis definidas como inteiras, obtendo uma solução factível. Em ambos os métodos, o modelo não-linear é decomposto em dois modelos lineares, no qual, um gera compartimentos e o outro os seleciona. Os resultados obtidos com as duas abordagens foram comparados com um limitante superior e se mostraram bastante satisfatórios / The Compartmentalized Knapsack Problem is an extension of the classical Knapsack Problem, where the ordered items are partitioned into classes, in such way that the knapsack must be divided into compartments, each one having limited capacity. In addition, the building of a compartment has a fixed cost and involves a loss of the overall capacity. The objective is to maximize the sum of the items utility value, minus the fixed costs of the compartments. This dissertation presents two solving methods. The first approach is a heuristic method, which is a combination of two heuristics from the literature. The second approach is a Column Generation method, that apart from it gives a new upper bound to the Compartmentalized Knapsack Problem, in the end of the method the master problem was solved with the variables defined as integer, that supplies a feasible solution. In both methods, the mathematical non linear model is decomposed into two linear models, one generates the compartments, and the other selects them to compose the knapsack. The results obtained with these two approaches were compared with an upper bound and they showed very efficient

Uma revisão dos algoritmos e conjuntos de instâncias presentes na literatura do Problema da Mochila com Repetições (PMR) é apresentada nessa dissertação de mestrado. Os algoritmos e conjuntos de instâncias usados são brevemente descritos nesse trabalho, afim de que o leitor tenha base para entender as discussões. Algumas propriedades bem conhecidas e específicas do PMR, como a dominância e a periodicidade, são explicadas com detalhes. O PMR é também superficialmente estudado no contexto de problemas de avaliação gerados pela abordagem de geração de colunas aplicada na relaxação contínua do Bin Packing Problem (BPP) e o Cutting Stock Problem (CSP). Múltiplos experimentos computacionais e comparações são realizadas. Para os conjuntos de instâncias artificiais mais recentes da literatura, um simples algoritmo de programação dinâmica, e uma variante do mesmo, parecem superar o desempenho do resto dos algoritmos, incluindo aquele que era estado-da-arte. O modo que relações de dominância é aplicado por esses algoritmos de programação dinâmica têm algumas implicações para as relações de dominância previamente estudadas na literatura. O autor dessa dissertação defende a tese de que a escolha dos conjuntos de instâncias artificiais definiu o que foi considerado o melhor algoritmo nos trabalhos anteriores. O autor dessa dissertação disponibilizou publicamente todos os códigos e conjuntos de instâncias referenciados nesse trabalho. / A review of the algorithms and datasets in the literature of the Unbounded Knapsack Problem (UKP) is presented in this master's thesis. The algorithms and datasets used are brie y described in this work to provide the reader with basis for understanding the discussions. Some well-known UKP-speci c properties, such as dominance and periodicity, are described. The UKP is also super cially studied in the context of pricing problems generated by the column generation approach applied to the continuous relaxation of the Bin Packing Problem (BPP) and Cutting Stock Problem (CSP). Multiple computational experiments and comparisons are performed. For the most recent arti cial datasets in the literature, a simple dynamic programming algorithm, and its variant, seems to outperform the remaining algorithms, including the previous state-of-the-art algorithm. The way dominance is applied by these dynamic programming algorithms has some implications for the dominance relations previously studied in the literature. In this master's thesis we defend that choosing sets of arti cial instances has de ned what was considered the best algorithm in previous works. We made available all codes and datasets referenced in this master's thesis.

One of the most remarkable theories in probability and statistics is the law of large numbers. Law of large numbers describes the behavior of random phenomena when they are reiterated infinitely or in very large trials. Apart from the mathematical exposition of the law of large numbers, its theory and applications have been widely used in gambling houses, financial sectors, and healthcare insurance where uncertainties deteriorate prediction and financial strength. However, the applications of the law of large numbers are not confined to the referred sectors and could be widely applied to industrial organizations and service provider companies in which large number of stochastic phenomena incorporate in their planning. In this thesis, the applications of the law of large numbers are studied in relation to logistics and transportation under conditions of operating in large networks. The results of this study assert that transportation companies can benefit from operating in large networks to increase the filling performance of their vehicles, fleet, etc. Equivalently, according to the law of large numbers the inferior capacity utilization in unit loads, containers, etc. converges to 0 with probability 1 as the size of the network grows. / Uppsatsnivå: D

Dans cette thèse, nous nous intéressons aux problèmes du knapsack multidimensionnel à choix multiple. Ils interviennent essentiellement en télécommunication. Nous proposons de nouvelles méthodes hybrides de résolution exacte et approchée. Dans un premier temps, nous proposons des méthodes heuristiques en se basant sur les techniques de génération de colonnes et d'arrondi. Ensuite, nous abordons une méthode de recherche locale, dite méthode de branchement local, où des contraintes linéaires sont introduites pour intensifier et diversifier la recherche. Cette méthode est ensuite hybridée avec la génération de colonnes et une technique d'arrondi. Concernant la résolution exacte, nous nous basons sur une méthode de "Branch and cut". Nous commençons par proposer de nouvelles contraintes valides pour le problème. Ensuite, nous les associons à des contraintes de couverture locales et globales dans un schéma énumératif. Les approches heuristiques et l'algorithme exact que nous proposons sont comparés à d'autres heuristiques de la littérature et au Solveur de programmes linéaires Cplex . L'ensemble de ces tests numériques ont été menés sur des instances ardues de la littérature ainsi que sur des instances générées aléatoirement de taille modérée.

<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>

This thesis investigates a new approach to lattice basis reduction suggested by M. Seysen. Seysen's algorithm attempts to globally reduce a lattice basis, whereas the Lenstra, Lenstra, Lovasz (LLL) family of reduction algorithms concentrates on local reductions. We show that Seysen's algorithm is well suited for reducing certain classes of lattice bases, and often requires much less time in practice than the LLL algorithm. We also demonstrate how Seysen's algorithm for basis reduction may be applied to subset sum problems. Seysen's technique, used in combination with the LLL algorithm, and other heuristics, enables us to solve a much larger class of subset sum problems than was previously possible.