The always best connected paradigm has gain a lot of interest in the research and scientific community. The availability of different wireless technologies and the proliferation of devices supporting multiple connections are open- ing new possibilities for users to share information everywhere and everytime. The multihoming support is being enriched to levels never before established. Indeed, users can configure devices to meet their own requirements, decrease communication costs by choosing links with no expenses associated, or opting for links that provide extended coverage. Such kind of configuration is often limited to static policies that aim the maximization of a single requirement, such as monetary cost or coverage. This approach is not efficient. For instance, if the optimization aims to decrease cost, extend coverage and increase security support simultaneously, static policies do not scale and have narrow support for multihoming goals, namely resilience, ubiquity and load sharing. Multihoming is an important aspect in computer networks. To enable higher levels of availability or optimize recovery processes in the presence of failures are goals that mechanisms improving resilience aim to support. Other goals, in a ubiquity aspect...

Este trabalho apresenta um modelo computacional para otimização energética de sistemas de abastecimento em tempo real. Tal modelo é composto por três módulos principais: (1) um módulo de simulação hidráulica que descreve o comportamento do sistema - EPANET; (2) um módulo de previsão de demandas que realiza a previsão das demandas futuras aplicável à utilização no tempo real (curto prazo), desenvolvido por Odan (2008); e, por fim, (3) um módulo otimizador estruturado em linguagem C++ que implementa a biblioteca de algoritmos genéticos do MIT - Massachusetts Institute of Technology and Matthew Wall, a GAlib, que permite determinar as rotinas operacionais (acionamento de válvulas e bombas) de forma à minimizar o custo de energia elétrica no sistema. O processo de otimização é divido em duas rotinas, nível estratégico e tempo real. Na otimização em nível estratégico, a partir das curvas típicas de demanda para cada nó de demanda do macro-sistema considerado, determina-se o conjunto de controles que minimizam os custos de energia elétrica, respeitando as restrições hidráulicas do sistema. Para cada conjunto de controles otimizados têm-se os níveis que os reservatórios irão atingir ao final de cada hora durante o horizonte de planejamento considerado...

The strut-and-tie models are widely used in certain types of structural elements in reinforced concrete and in regions with complexity of the stress state, called regions D, where the distribution of deformations in the cross section is not linear. This paper introduces a numerical technique to determine the strut-and-tie models using a variant of the classical Evolutionary Structural Optimization, which is called Smooth Evolutionary Structural Optimization. The basic idea of this technique is to identify the numerical flow of stresses generated in the structure, setting out in more technical and rational members of strut-and-tie, and to quantify their value for future structural design. This paper presents an index performance based on the evolutionary topology optimization method for automatically generating optimal strut-and-tie models in reinforced concrete structures with stress constraints. In the proposed approach, the element with the lowest Von Mises stress is calculated for element removal, while a performance index is used to monitor the evolutionary optimization process. Thus, a comparative analysis of the strut-and-tie models for beams is proposed with the presentation of examples from the literature that demonstrates the efficiency of this formulation. © 2013 Elsevier Ltd.

Esta pesquisa teve por finalidade o desenvolvimento de uma metodologia de otimização em duas camadas. A otimização preliminar foi baseada na técnica de planejamento de experimentos junto com a metodologia por superfície de resposta com a finalidade de identificar uma possível região de busca do ponto de operação ótimo, o qual foi obtido através da implementação de métodos híbridos de otimização desenvolvidos mediante associação do modelo determinístico de otimização por programação quadrática sucessiva (SQP) com a técnica dos algoritmos genéticos (GA) no modelo do processo de craqueamento catalítico fluidizado- FCC. Este processo é caracterizado por ser um sistema heterogêneo e não isotérmico, cuja modelagem detalhada engloba as equações de balanço de massa e energia das partículas do catalisador, como também para a fase líquida e gasosa, sendo um dos casos de estudo para a aplicação da metodologia de otimização desenvolvida. Como caso de estudo principal foi considerado o modelo do conversor do processo de FCC desenvolvido por Moro e Odloak (1995). Mediante a metodologia de otimização do processo baseado no uso do modelo determinístico da planta, foram definidas estratégias e políticas operacionais para a operação da unidade de FCC em estudo. Procurou-se alto nível de desempenho e segurança operacional...

Two new powerful mathematical languages, fuzzy set theory and possibility theory, have led to two optimization types that explicitly incorporate data whose values are not real-valued nor probabilistic: 1) flexible optimization and 2) optimization under generalized uncertainty. Our aim is to make clear what these two types are, make distinctions, and show how they can be applied. Flexible optimization arises when it is necessary to relax the meaning of the mathematical relation of belonging to a set (a constraint set in the context of optimization). The mathematical language of relaxed set belonging is fuzzy set theory. Optimization under generalized uncertainty arises when it is necessary to represent parameters of a model whose values are only known partially or incompletely. A natural mathematical language for the representation of partial or incomplete information about the value of a parameter is possibility theory. Flexible optimization, as delineated here, includes much of what has been called fuzzy optimization whereas optimization under generalized uncertainty includes what has been called possibilistic optimization. We explore why flexible optimization and optimization under generalized uncertainty are distinct and important types of optimization problems. Possibility theory in the context of optimization leads to two distinct types of optimization under generalized uncertainty...

The pricing problem in a multi-period setting is a challenging problem and has attracted much attention in recent years. In this thesis, we consider a monopoly and an oligopoly pricing problem. In the latter, several sellers simultaneously seek an optimal pricing policy for their products. The products are assumed to be differentiated and substitutable. Each seller has the option to set prices for her products at each time period, and her goal is to find a pricing policy that will yield the maximum overall profit. Each seller has a fixed initial inventory of each product to be allocated over the entire time horizon and does not have the option to produce additional inventory between periods. There are no holding costs or back-order costs. In addition, the products are perishable and have no salvage costs. This means that at the end of the entire time horizon, any remaining products will be worthless. The demand function each seller faces for each product is uncertain and is affected by both the prices at the current period and past pricing history for her and her competitors. In this thesis, we address both the uncertain and the competitive aspect of the problem. First, we study the uncertain aspect of the problem in a simplified setting...

In this thesis, we investigate theoretical and numerical advantages of a novel representation for Sum of Squares (SOS) decomposition of univariate and multivariate polynomials. This representation formulates a SOS problem by interpolating a polynomial at a finite set of sampling points. As compared to the conventional coefficient method of SOS, the formulation has a low rank property in its constraints. The low rank property is desirable as it improves computation speed for calculations of barrier gradient and Hessian assembling in many semidefinite programming (SDP) solvers. Currently, SDPT3 solver has a function to store low rank constraints to explore its numerical advantages. Some SOS examples are constructed and tested on SDPT3 to a great extent. The experimental results demonstrate that the computation time decreases significantly. Moreover, the solutions of the interpolation method are verified to be numerically more stable and accurate than the solutions yielded from the coefficient method.; by Jia Li Sun.; Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2006.; Includes bibliographical references (leaves 77-79).

In this thesis, we present a modification of an existing methodology to obtain a hierarchy of lower and upper bounds on moments of solutions of linear differential equations. The motivation for change is to obtain tighter bounds by solving smaller semidefinite problems. The modification we propose involves partitioning the domain and normalizing each partition to ensure numerical stability. Using the adjoint operator, linear constraints involving the boundary conditions and moments of the solution are developed for each partition. Semidefinite constraints are imposed on the moments, and an optimization problem is solved to obtain the bounds. We have demonstrated the algorithm by calculating bounds on moments of various one-dimensional case differential equations including the Bessel ODE, and Legendre polynomials. In the two-dimensional case we have demonstrated the algorithm by calculating bounds on various PDEs including the Helmholtz equation, and heat equation. In both cases, the results were encouraging with tighter bounds on moments being obtained by solving smaller problems with domain partitioning.; by Sandeep Sethuraman.; Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2006.; Includes bibliographical references (leaf 95).

We present a further study and analysis of an exponential annealing based algorithm for convex optimization. We begin by developing a general framework for applying exponential annealing to conic optimization. We analyze the hit-and-run random walk from the perspective of convergence and develop (partially) an intuitive picture that views it as the limit of a sequence of finite state Markov chains. We then establish useful results that guide our sampling. Modifications are proposed that seek to raise the computational practicality of exponential annealing for convex optimization. In particular, inspired by interior-point methods, we propose modifying the hit-and-run random walk to bias iterates away from the boundary of the feasible region and show that this approach yields a substantial reduction in computational cost. We perform computational experiments for linear and semidefinite optimization problems. For linear optimization problems, we verify the correlation of phase count with the Renegar condition measure (described in [13]); for semidefinite optimization, we verify the correlation of phase count with a geometry measure (presented in [4]).; by Jeremy Chen.; Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program...

Project scheduling is an important research and application area in engineering management. Recent research in this area addresses resource constraints as well as stochastic durations. This thesis presents a simulation-based optimization model for solving resource-constrained product development project scheduling problems. The model uses design structure matrix (DSM) to represent the information exchange among various tasks of a project. Instead of a simple binary precedence relationship, DSM is able to quantify the extent of interactions as well. In particular, these interactions are characterized by rework probabilities, rework impacts and learning. As a result, modeling based on DSM allows iterations to take place. This stochastic characteristic is not well addressed in earlier literatures of project scheduling problems. Adding resource factors to DSM simulation is a relatively new topic. We not only model the constraints posed by resource requirements, but also explore the effect of allocating different amount of resources on iterations. Genetic algorithm (GA) is chosen to optimize the model over a weighted sum of a set of heuristics. GA is known for its robustness in solving many types of problems. While the normal branch-and-bound method depends on problem specific information to generate tight bounds...

In this thesis, we compare computationally four methods for solving optimization problems under uncertainty: * Robust Optimization (RO) * Adaptive Robust Optimization (ARO) * Data Driven Optimization (DDO) * stochastic Programming (SP) We have implemented several computation experiments to demonstrate the different performance of these methods. We conclude that ARO outperform RO, which has a comparable performance with DDO. SP has a comparable performance with RO when the assumed distribution is the same as the true underlying distribution, but under performs RO when the assumed distribution is different from the true distribution.; by Wang, Yanbo.; Includes bibliographical references (leaf 49).; Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2008.

by Ganesh Gurumurthy.; Thesis (S.M.)--Massachusetts Institute of Technology, Computation for Design and Optimization Program, 2008.; Includes bibliographical references (p. 69-72).

Multiobjective optimization is becoming an increasingly important approach for both the design and operation of water distribution systems (WDSs). Given the multiobjective nature of these problems, multiobjective optimization is expected to provide decision makers with increased insight into the tradeoffs between competing objectives and alternative solutions of WDSs, which might benefit the water industry, society and environment. Due to the advances in computing technology and the development of fast multiobjective sorting algorithms, research activities into the application of multiobjective algorithms to WDS design and operation have increased significantly in the past decade. Minimization of economic cost and maximization of network reliability are the two most commonly considered objectives in WDS optimization. In addition, some environment related issues, such as energy conservation, have been incorporated into the optimization of WDSs. However, the leading environmental concern – Greenhouse gas (GHG) emissions – has not yet been addressed directly in the field of WDS optimization. Consequently, this research incorporates GHG emission minimization as an objective directly into the optimal design of WDSs, together with the economic objective of minimizing cost and the hydraulic reliability objective of maximizing surplus power factor via a multiobjective approach. The major research contributions are presented in six journal publications. These publications describe the motivation and methodology to incorporate GHG emission minimization as an objective of WDS optimization; explore the tradeoffs between the traditional objective of minimizing life cycle cost and the environmental objective of minimizing life cycle GHG emissions; investigate the sensitivity of these tradeoffs to a number of factors...

This is a PhD thesis by publication. It includes five journal papers, three of them already published, and two submitted for publication in a very high quality international journal in the field of Evolutionary Computation (one of which – as an invited paper). Further, the thesis contains also four conference papers presented and published in the top peer reviewed conferences, as well as one peer reviewed chapter book. In this thesis we studied an optimization algorithm called Particle Swarm Optimization (PSO) from theoretical and application point of views. The main focus of the theoretical analysis of the algorithm was towards understanding and addressing its limitations that were related to transformations of the search space, convergence to quality solutions, and stability. Through analysis of the algorithm under transformations of the search space we proposed a modification to the original PSO so that a stable performance was guaranteed when the search space was transformed, i.e., rotated, scaled, and translated. We also studied the ability of the original PSO in locating optimum solutions (local and global optima). Our study showed that this algorithm cannot guarantee to find a local optimum. We introduced a general formulation of topology for the original PSO and identified conditions so that it did not only guarantee local convergence but also transformation invariance. Further...

Structural topology optimization has been extensively studied in aeronautical, civil, and mechanical engineering applications in order to improve performance of systems. This thesis focuses on an optimal design of damping treatment using topology optimization, and the reduction of computational expense of the topology optimization procedure. This thesis presents mainly two works on topology optimization. In the first work, topology optimization is implemented to optimally design damping treatments in unconstrained-layer damping material. Since the damping effect relies on the placement of damping treatment, and the weight of damping material may be an important factor, the placement of damping material is optimally determined using topology optimization with an allowable maximum. Unconstrained-layer plate and shell structures are modeled. The damping layer on the unconstrained-layer structures is considered as the design domain. Using topology optimization, the damping layer is designed numerically, and then experimentally validated by comparing the damping effects. In the numerical example, the topological damping treatment usually provides much higher damping effects compared to other approaches such as strain energy distribution (SED) and an evolutionary structural optimization (ESO). In the second work...

This research utilized topology and size optimization to optimize a lunar rover chassis in order to reduce structural mass while satisfying the required surface and launch vehicle loading criterion. Renewed interest in lunar exploration has provided an opportunity for Canada to participate in a Lunar Prospecting Mission in collaboration with NASA. Queen’s University, in collaboration with Neptec Design Group, has developed methodology to produce the structural design of a next generation lunar rover chassis using systematic design optimization techniques to minimize the structural mass of the chassis. Typical lightweight design can be achieved using lightweight materials, advanced manufacturing processes or systems, and design optimization. Due to the unique requirements for spacecraft, the proposed research is limited to specific materials and processes, therefore weight reduction is achieved exclusively through design optimization. The structural design was completed using a three stage design approach: Conceptual, Preliminary, and Detailed Design Stages. The Conceptual Design Stage developed chassis designs considering component layout and bounding box topology. The generated concepts were evaluated qualitatively to select the best candidates for design optimization. The Preliminary Design Stage utilized Hyperworks© Optistruct commercial software to complete topology optimization to optimize the chassis bounding box topology while considering lunar surface and launch vehicle loading. The topology optimization results were then used to create preliminary optimum designs. In the detailed design stage...

We present and analyze a central cutting surface algorithm for general semi-infinite convex optimization problems, and use it to develop a novel algorithm for distributionally robust optimization problems in which the uncertainty set consists of probability distributions with given bounds on their moments. Moments of arbitrary order, as well as non-polynomial moments can be included in the formulation. We show that this gives rise to a hierarchy of optimization problems with decreasing levels of risk-aversion, with classic robust optimization at one end of the spectrum, and stochastic programming at the other. Although our primary motivation is to solve distributionally robust optimization problems with moment uncertainty, the cutting surface method for general semi-infinite convex programs is also of independent interest. The proposed method is applicable to problems with non-differentiable semi-infinite constraints indexed by an infinite-dimensional index set. Examples comparing the cutting surface algorithm to the central cutting plane algorithm of Kortanek and No demonstrate the potential of our algorithm even in the solution of traditional semi-infinite convex programming problems whose constraints are differentiable and are indexed by an index set of low dimension. After the rate of convergence analysis of the cutting surface algorithm...

Linear optimization is many times algorithmically simpler than non-linear convex optimization. Linear optimization over matroid polytopes, matching polytopes and path polytopes are example of problems for which we have simple and efficient combinatorial algorithms, but whose non-linear convex counterpart is harder and admits significantly less efficient algorithms. This motivates the computational model of convex optimization, including the offline, online and stochastic settings, using a linear optimization oracle. In this computational model we give several new results that improve over the previous state-of-the-art. Our main result is a novel conditional gradient algorithm for smooth and strongly convex optimization over polyhedral sets that performs only a single linear optimization step over the domain on each iteration and enjoys a linear convergence rate. This gives an exponential improvement in convergence rate over previous results. Based on this new conditional gradient algorithm we give the first algorithms for online convex optimization over polyhedral sets that perform only a single linear optimization step over the domain while having optimal regret guarantees, answering an open question of Kalai and Vempala, and Hazan and Kale. Our online algorithms also imply conditional gradient algorithms for non-smooth and stochastic convex optimization with the same convergence rates as projected (sub)gradient methods.

Steepest descent preconditioning is considered for the recently proposed nonlinear generalized minimal residual (N-GMRES) optimization algorithm for unconstrained nonlinear optimization. Two steepest descent preconditioning variants are proposed. The first employs a line search, while the second employs a predefined small step. A simple global convergence proof is provided for the N-GMRES optimization algorithm with the first steepest descent preconditioner (with line search), under mild standard conditions on the objective function and the line search processes. Steepest descent preconditioning for N-GMRES optimization is also motivated by relating it to standard non-preconditioned GMRES for linear systems in the case of a quadratic optimization problem with symmetric positive definite operator. Numerical tests on a variety of model problems show that the N-GMRES optimization algorithm is able to very significantly accelerate convergence of stand-alone steepest descent optimization. Moreover, performance of steepest-descent preconditioned N-GMRES is shown to be competitive with standard nonlinear conjugate gradient and limited-memory Broyden-Fletcher-Goldfarb-Shanno methods for the model problems considered. These results serve to theoretically and numerically establish steepest-descent preconditioned N-GMRES as a general optimization method for unconstrained nonlinear optimization...

Tanner, Herbert G.; A new network topology optimization approach to cellular neural network design, as a method for realizing associative memories using sparser networks is conceptualized. This type of optimization allows recurrent neural networks to be implemented in a spatially distributed fashion, that is, with components of the network residing in different physical locations. This could find application in addressing the problem of dynamic allocation of a team of robots to a collection of spatially distributed tasks which is relevant for large scale environmental monitoring and surveillance. Spatially distributed sensing allows for greater coverage of the environment than a single large vehicle with multiple sensors would permit in many cases. In this work, we try to answer the question of how could the design process be different if the network topology was also part of the design. A sparser cellular neural network topology can be achieved without significantly degrading the performance of the network, by selectively deleting those weights from the optimized network which contribute the least to ability of the network to recall the desired patterns. This approach is particularly useful where neural links incur varying costs, such as implementation of associative memories over wireless sensor networks. The cellular neural networks interconnection topology is diluted...