Orientador: Maria Aparecida Diniz Ehrhardt / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Matematica, Estatistica e Computação Cientifica / Made available in DSpace on 2018-08-10T01:30:21Z (GMT). No. of bitstreams: 1 Lammoglia_Bruna_M.pdf: 679586 bytes, checksum: 221fa89afc7d9f594781baed1dfe6b0e (MD5) Previous issue date: 2007 / Resumo: Neste trabalho o objetivo principal foi a minimização de quadráticas em caixas. Dissertamos sobre os métodos de máxima descida e dos gradientes conjugados, bem como sobre um método mais recente denominado gradiente espectral. O GENCAN, um algoritmo que minimiza funções em caixas, foi estudado em detalhe, particularmente avaliando sua aplicação para quadráticas. O objetivo foi analisar o desempenho do GENCAN, comparado com algoritmos anteriores, como o LANCELOT e o QUACAN. Foram executados experimentos numéricos a fim de avaliar o desempenho das versões de GENCAN sem e com pré-condicionamento. Concluiu-se que pré-condicionar o método dos gradientes conjugados neste caso tornou o GENCAN mais robusto. No entanto, o pré-condicionador usado neste software mostrou-se computacionalmente caro. Em relação à comparação do GENCAN, LANCELOT e QUACÁN, podemos afirmar que o GENCAN. mostrou-se competitivo / Abstract: The focus of this work was the minimization of quadratic functions with box constraints. We were mainly concerned about the steepest descent and conjugated gradient methods, besides a more recent approach called spectral gradient method. The GENCAN, an algorithm that minimizes functions on a box, was studied in details particularly evaluating this algorithm applied to quadratics. The objective was to analyze the efficiency of GENCAN, comparing it to classical algorithms, such as LANCELOT and QUACAN. We executed numerical experiments in order to investigate the efficiency of GENCAN version with and without preconditioning. Evaluating the results we concluded that preconditioning the conjugated gradient method makes the GENCAN work considerably better; despite the fact that the preconditioner used here turned the computational process more expensive. Comparing GENCA'N, LANCELOT, and QUACAN we can state that GENCAN is competitive / Mestrado / Otimização / Mestre em Matemática Aplicada

Orientador: Marcius Fabius Henriques Carvalho / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-17T00:35:22Z (GMT). No. of bitstreams: 1 Pupulin_RicardoPereira_M.pdf: 3494242 bytes, checksum: 2563e0723dce908c7d4957a80a14d641 (MD5) Previous issue date: 2010 / Resumo: Este trabalho tem como objetivo encontrar a melhor programação de colheita para a usina de cana-de-açúcar em estudo. Foram desenvolvidos modelos com base em programação linear que representem o melhor mix varietal do canavial, a melhor programação de colheita com as variedades existentes e o pior cenário, este desenvolvido para quantificar a produção real percebida no período. Estes modelos foram feitos com os rendimentos das variedades divididos em dois grupos que seriam os modelos com base de dados no histórico de colheita e os modelos com base na safra de 08/09 / Abstract: The main purpose of this project is to find the best harvest scheduling program to sugar cane. Models were developed with a linear programming that represent the best variety mix for the reed, the best harvest program with the current sugar cane varieties and the worst scenario, this was done to qualify the real production perceived at the period. These models were done with the varieties income shared in two groups that would be the models based on the harvest historic and the models based on the 08/09 crop / Mestrado / Materiais e Processos de Fabricação / Mestre em Engenharia Mecânica

Orientador: Valéria Abrão de Podestá / Dissertação (mestrado profissional) - Universidade Estadual de Campinas, Instituto de Matemática, Estatística e Computação Científica / Made available in DSpace on 2018-08-16T01:49:31Z (GMT). No. of bitstreams: 1 Trevisan_EbersonPaulo_M.pdf: 1587820 bytes, checksum: 334e76317d2424126dd4bb1fb621c7e4 (MD5) Previous issue date: 2010 / Resumo: Neste trabalho são apresentados alguns assuntos iniciais da teoria de Programação Linear e o método Simplex. Mostramos também como a Programação Linear pode ser utilizada na separação de dois conjuntos de pontos (padrões), através de um modelo linear cuja solução é um hiperplano separador. Finalizamos o trabalho com a apresentação de alguns exemplos de aplicação da Programação Linear na separação de dois conjuntos linearmente separáveis e linearmente inseparáveis / Abstract: In this work we present some introductory issues from Linear Programming theory and the Simplex method. We also show how we can use Linear Programming in two patterns separation by constructing a linear model which solution is a separating hyperplane. Finaly, we also present some examples of Linear Programming application in the linear separability and inseparability of two patterns sets / Mestrado / Programação Linear / Mestre em Matemática

In this work we propose a flexible numerical approach to deal with models of dynamic moral hazard with simultaneous learning about the production function. Because of the complexity of the problem, analytical solutions have so far been limited in scope. The contribution is methodological: through computation, the problem can be studied under few assumptions about functional forms. We depart from a general mechanism, reformulate it as an incentive compatible mechanism, and show how it can be solved by backward induction through a sequence of linear programs. We apply our method to a few cases of interest, and confirm that uncertainty about the production function increases the volatility of the agent\'s utility in order to prevent belief manipulation, as found in the literature. / Neste trabalho, propomos uma estratégia numérica para lidar com modelos de risco moral dinâmico com aprendizado sobre a função de produção. Pela complexidade do problema, soluções analíticas na literatura têm sido limitadas em seu escopo. Nossa contribuição é metodológica: através de métodos computacionais, o problema pode ser estudado sob poucas hipóteses a respeito de formas funcionais. Partindo de um mecanismo geral, reformulamos o problema como um mecanismo compatível em incentivos, e então mostramos como este pode ser resolvido por indução retroativa por meio de uma sequência de programas lineares. Aplicamos o método a alguns casos de interesse, e confirmamos a conclusão da literatura de que a incerteza sobre a função de produção aumenta a volatilidade da utilidade do agente para prevenir manipulação de crenças.

This thesis project will attempt to answer the question if it is possible to gain performance by using SIMD instructions when generating code for scalar computation. The current trend in processor architecture is to equip the processors with multi-way SIMD units to form so-called throughput cores. This project uses the CELL SPE processor for a concrete implementation. To get good code quality the thesis project continues work on the code generator by Mattias Eriksson and Andrzej Bednarski based on integer linear programming. The code generator is extended to handle generation of SIMD code for 32bit operands. The result show for some basic blocks, positive impact in execution time of the generated schedule. However, further work has to be done to get a feasable run time of the code generator.

The school's schedule involves folding lessons within five working days, so that in each class were taught subjects required by the school curriculum in the correct number of hours and for each hour was assigned a teacher and a number of classrooms in which teaching will be realized. However, it is necessary to keep order in the resulting schedule avoid collision or vice versa unnecessary delays in the schedules of students and teachers in order to meet the conditions laid down by legislation and at the same time it was granted as much as possible the requirements of the management, teachers, or students or their parents. In this study it is proposed a mathematical model for the formation of a high school schedule that was created with the help of MPL for Windows and data provided by Gymnázium Chomutov.

The farm record keeping system is frequently used by the farm business manager for historical accounting, mainly to satisfy institutional requirements for information, particularly that of taxation. Little emphasis has been placed on the record system as an aid in forward planning. Farm planning aids frequently ignore the record keeping aspect of farm business management and are generally unrelated to record keeping systems used on the farm. The Canfarm Farm Record System and the various Canfarm Farm Planning Packages are likewise unrelated. This lack of a comprehensive Managerial Information and Decision System for the Canadian farm business manager may, in part, account for the slow rate of adoption of computerized record keeping systems and farm planning aids. The objective of this study was to integrate the Canfarm Farm Record System with a farm planning model, in order to suggest a format for the future development of an extension oriented Managerial Information and Decision System. The specific objectives were: 1) to identify the Canfarm Farm Record System, the statements that are generated from the Record System and the accounting items which make up these statements; 2) to construct, document and validate a linear programming farm model which incorporates the information contained in the Canfarm Farm Record System and generates output in statements that are consistent (identical) with the Record System statements; and, 3) to develop recommendations for standardization of projected financial reports that existing models could adopt, and which could be incorporated into future models. A significant cost of using any system is the investment in time required to become acquainted with that system. Consequently, the fewer the concepts to be learned, the lower would be the cost of learning for both the extension agent and the farm business manager. In addition to reducing the number of new concepts, integrating a record system with the planning function provides the control link through the comparison of planned versus actual values that is not available with fragmented packages. These considerations, in conjunction with management theory and theories of the firm formed the theoretical background to this study. Linear programming was chosen as an appropriate solution method due to its lack of conceptual complexity, the availability of a suitable algorithm and its maximizing capabilities. The Canfarm Farm Record System was described in terms of its accounting concepts. The flow of data was shown, from the farmer's journal entries, through the various detailed reports, to the summary financial statements and to the Balance Sheet. Four reports, the Farm Operating Statement, the Income Statement, the Statement of Assets, Liabilities and Owners Equity (Balance Sheet), plus the Cash Flow Statement, were chosen as the basic reports to be generated by the model. The empirical work was shown to consist of several stages, from the capturing of the historical records, the formulation of the farm plan, the generation of the LP matrix, to the solution of the projected farm plan. A report writer converts the output from the LP solution and produces the four financial statements. The final stage is the control link, in which the projected reports are compared to the actual records as they become available. The model was designed to be applicable to a variety of farm planning situations, in addition to being able to link up with other models. It was therefore subjected to several tests, including a simulation run using case farm data from the Canfarm Farm Record System to test the model structure, integration with two different farm planning models to demonstrate its flexibility and comparison with the same data fed through the Canfarm Farm Record System. Once this stage was reached, other potential applications were outlined and then a recommendation was proposed for the adoption of a standardization base for future models as well as for wider applications of this model. / Land and Food Systems, Faculty of / Graduate

The problems we studied in this thesis are linearly constrained convex programming (LCCP) and nonnegative matrix factorization (NMF). The resolutions of these two problems are all closely related to convex programming with simple convex constraints. The work can mainly be described in the following three parts. Firstly, an interior point algorithm following a parameterized central path for linearly constrained convex programming is proposed. The convergence and polynomial-time complexity are proved under the assumption that the Hessian of the objective function is locally Lipschitz continuous. Also, an initialization strategy is proposed, and some numerical results are provided to show the efficiency of the proposed algorithm. Secondly, the path following algorithm is promoted for general barrier functions. A class of barrier functions is proposed, and their corresponding paths are proved to be continuous and converge to optimal solutions. Applying the path following algorithm to these paths provide more flexibility to interior point methods. With some adjustments, the initialization method is equipped to validate implementation and convergence. Thirdly, we study the convergence of hierarchical alternating least squares algorithm (HALS) and its fast form (Fast HALS) for nonnegative matrix factorization. The coordinate descend idea for these algorithms is restated. With a precise estimation of objective reduction, some limiting properties are illustrated. The accumulation points are proved to be stationary points, and some adjustments are proposed to improve the implementation and efficiency

No description available.

This dissertation develops a mixed integer linear program to establish the upper and lower bounds of the Alphorn of Uncertainty. For a project manager, planning for uncertainty is a staple of their jobs and education. But the uncertainty associated with a catastrophic event presents difficulties not easily controlled with traditional methods of risk management. This dissertation brings and modifies the concept of a project schedule as a bounded “Alphorn of Uncertainty” to the problem of how to reduce the risk of a catastrophic event wreaking havoc on a project and, by extension, the company participating in that project. The dissertation presents new mathematical models underpinning the methods proposed to reduce risk as well as simulations to demonstrate the accuracy of those models. The dissertation further assesses the complexity of the models and thus their practical application. Finally, the dissertation presents strategies to reduce the risk to a project of a catastrophic event using the upper bound of the Alphorn as the measure of risk.