Um importante problema enfrentado por engenheiros é a busca por soluções ótimas para problemas com um grande número de soluções possíveis. Neste trabalho, estudamos métodos otimização probabilísticos baseados em algoritmos genéticos, propostos inicialmente para o estudo de sistemas biológicos. Propomos algumas alterações do método de otimização por algoritmos genéticos tradicional, utilizando técnicas estatísticas de planejamento experimental, que resultaram em uma melhoria da convergência, percebida, não só na velocidade de convergência, como no número de possibilidades de soluções diferentes analisadas. Como resultado, é proposto um algoritmo que cobre o domínio de atuação dos métodos por algoritmos genéticos e do método por gradientes, permitindo uma melhor sintonização do otimizador com o problema. / A important problem faced by engineers is the search of optimal solutions for problems with a great number of possible solutions. Throughout this work, it is studied stochastic optimizers based on genetic algorithms, applied initially to the study of biological systems. Some alterations on the traditional genetic algorithms based optimizer are proposed through the use of experiment design techniques, which resulted in a improvement of the convergence that can be perceived, not only in the convergence speed, but on the number of solutions analyzed as well. As a consequence, a algorithm is proposed, covering both the traditional genetic algorithms based optimizer and the gradient method domains, allowing a better tuning of the optimizer to the problem.

Oi-Ping Chiu. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 45-46). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Model --- p.4 / Chapter 2.1 --- General Model --- p.4 / Chapter 2.2 --- Constraint --- p.5 / Chapter 3 --- Bayesian Estimation via Gibbs Sampler --- p.7 / Chapter 3.1 --- Conditional Distributions --- p.10 / Chapter 3.2 --- Constraint --- p.15 / Chapter 3.3 --- Bayesian Estimation --- p.16 / Chapter 4 --- Model Comparison using the Bayes Factor --- p.18 / Chapter 5 --- Simulation Study --- p.22 / Chapter 6 --- Real Example --- p.27 / Chapter 6.1 --- Model Selection --- p.29 / Chapter 6.2 --- Bayesian Estimate --- p.30 / Chapter 6.3 --- Sensitivity Analysis --- p.31 / Chapter 7 --- Discussion --- p.32 / Chapter A --- p.34 / Bibliography --- p.45

The purpose of this paper is to compare Analysis of Variance with Contingency Table Analysis when the data being analyzed do not satisfy Analysis of Variance assumptions. The criteria for comparison are the powers of the Standard variance-ratio and the Chi-square test. The test statistic and powers were obtained by Monte Carlo. 1. Calculate test statistic for each of 100 trials, this process was repeated 12 times. Each time different combination of means and variances were used. 2. Powers were obtained for each of 12 combinations of means and variances. Whether Analysis of Variance or Contingency Table Analysis is a better alternative depends on if we are interested in equality of population means or differences of population variances.

No description available.

The current study examined whether elevations on four variables (drive for thinness, depressed mood, maladaptive cognitions, and ineffectiveness) were related to increased risk of developing an eating disorder over a 4-year prospective interval. Subjects (N = 191) were female undergraduates who were assessed with the Anorexia-Bulimia Inventory (ABI), Eating Disorder Inventory (EDI), and a structured clinical interview. Results indicated that individuals with elevated scores on each of the four variables at the initial assessment did not show higher absolute eating disorder incidence rates over the 4-year interval. However, initial scores on the four variables together lll explained approximately 13% of both anorexia and bulimia symptom severity variance at the final assessment. Changes over time in the four variables were more highly related to later symptom severity, explaining 34% of the variance in anorexic severity and 16% in bulimic severity. Thus, the findings suggest that initial scores, and especially changes in scores, on the four variables were related to severity of symptoms 4 years later. However, a large proportion of the variance in eating disorder severity remained unexplained. Examination of the role of each risk variable individually revealed that initial elevations on maladaptive cognitions and drive for thinness subscales were related to higher anorexic symptom severity at the later assessment. Of interest was the absence of significant relationships between early scores on maladaptive cognitions and drive for thinness and subsequent bulimic symptoms, suggesting that anorexia and bulimia may have somewhat different risk pathways. The pretest scores on the depressed mood and ineffectiveness subscales were not significantly correlated with symptom severity at the later assessment, and were not identified by the regression analyses as parsimonious or efficient predictors of eating disorder symptoms. This finding suggests that perhaps early difficulties with depression and low self-esteem are less related to onset of later eating disorders than previously believed. Finally, the overall 4-year incidence rate of .6% found in the current study suggests that as women move through their college years, they are departing the developmental period of high risk for onset of eating disorders, and thus, new cases become increasingly rare.

A consecutive number of studies on the adoption trend of logistics technology since 1988 revealed that logistics organizations are not in the frontier when it comes to adopting new technology and this delayed adoption creates an information gap. In the advent of supply chain management and the strategic position of logistics, the need for accurate and timely information to accompany the logistics executives became more important than ever before. Given the integrative nature of logistics technology, failure to implement the technology successfully could result in writing off major investments in developing and implementing the technology or even in abandoning the strategic initiatives underpinned by these innovations. Consequently, the need to employ effective strategies and models to cope with these uncertainties is rather crucial. This thesis addresses the aspect of uncertainty in implementation success by process and factor research models. Process research approach focuses on the sequence of events in the technology transfer process that occurs over time. It explains the story that explains the degree of association between these sequences and implementation success. Through content analysis, this research gathers, extracts, and categorizes process data of actual stories of logistics technology adoption and implementations in organizations that are published in literature. The extracted event sequences are then analyzed using optimal matching from natural science and grouped using cluster analysis. Four patterns were revealed that organizations follow to transfer logistics technology namely, formal minimalist, mutual adaptation, development concerned, and organizational roles dispenser. Factors that contribute to successful implementation in each pattern were defined as the crucial and necessary events that characterized and differentiated each pattern from others. The factor approach identifies the potential predictors of successful technology implementation and tests empirical association between predictors and outcomes. This research develops a logistics technology success model. In developing the model, various streams of research were investigated including logistics, information systems, and organizational psychology. The model is tested using a questionnaire survey study. The data were collected from Australian companies which have recently adopted and implemented logistics technology. The results of a partial least squares structured equation modeling provide strong support for the model constructs and valuable insights to logistics/supply chain managers. The last study reports a convergent triangulation study using multiple case study of three Australian companies which have implemented logistics technology. A within and a cross case analysis of the three cases provide cross validation for the results of the other two studies. The results provided high predictive validity for the two models. Furthermore, the case study approach was so beneficial in explaining and contextualizing the linkages of the factor-based model and in confirming the importance of the crucial events in the process-based model. The thesis concludes with a research and managerial implications chapter which is devoted for logistics/supply chain managers and researchers.

In a reinforcement learning task an agent must learn a policy for performing actions so as to perform well in a given environment. Policy gradient methods consider a parameterized class of policies, and using a policy from the class, and a trajectory through the environment taken by the agent using this policy, estimate the performance of the policy with respect to the parameters. Policy gradient methods avoid some of the problems of value function methods, such as policy degradation, where inaccuracy in the value function leads to the choice of a poor policy. However, the estimates produced by policy gradient methods can have high variance.¶ In Part I of this thesis we study the estimation variance of policy gradient algorithms, in particular, when augmenting the estimate with a baseline, a common method for reducing estimation variance, and when using actor-critic methods. A baseline adjusts the reward signal supplied by the environment, and can be used to reduce the variance of a policy gradient estimate without adding any bias. We find the baseline that minimizes the variance. We also consider the class of constant baselines, and find the constant baseline that minimizes the variance. We compare this to the common technique of adjusting the rewards by an estimate of the performance measure. Actor-critic methods usually attempt to learn a value function accurate enough to be used in a gradient estimate without adding much bias. In this thesis we propose that in learning the value function we should also consider the variance. We show how considering the variance of the gradient estimate when learning a value function can be beneficial, and we introduce a new optimization criterion for selecting a value function.¶ In Part II of this thesis we consider online versions of policy gradient algorithms, where we update our policy for selecting actions at each step in time, and study the convergence of the these online algorithms. For such online gradient-based algorithms, convergence results aim to show that the gradient of the performance measure approaches zero. Such a result has been shown for an algorithm which is based on observing trajectories between visits to a special state of the environment. However, the algorithm is not suitable in a partially observable setting, where we are unable to access the full state of the environment, and its variance depends on the time between visits to the special state, which may be large even when only few samples are needed to estimate the gradient. To date, convergence results for algorithms that do not rely on a special state are weaker. We show that, for a certain algorithm that does not rely on a special state, the gradient of the performance measure approaches zero. We show that this continues to hold when using certain baseline algorithms suggested by the results of Part I.

"February 1992" Bibliography: leaf 323-344. Develops a general strategy for factorial linear model analysis for experimental and observational studies, an iterative, four-stage, model comparison procedure. The approach is applicable to studies characterized as being structure-balanced, multitiered and based on Tjur structures unless the structure involves variation factors when it must be a regular Tjur structure. It covers a wide range of experiments including multiple-error, change-over, two-phase, superimposed and unbalanced experiments.

<p>There is a big interest for alternative investment strategies than investing in traditional asset classes. Commodities are having a boom dynamic with increasing prices. This thesis is therefore based on applying Modern Portfolio Theory concept to this alternative asset class.</p><p>In this paper we manage to create optimal portfolios of commodities for investors with known and unknown risk preferences. When comparing expected returns to actual returns we found that for the investor with the known risk preference almost replicated the return of the markets. The other investor with unknown risk preference also profited but not as efficient as the market portfolio.</p>

<p>Asian options are an important family of derivative contracts with a wide variety of applications in commodity, currency, energy, interest rate, equity and insurance markets. In this master's thesis, we investigate methods for evaluating the price of the Asian call options with a fixed strike. One of them is the Monte Carlo method. The accurancy of this method can be observed through variance of the price. We will see that the variance with using Monte Carlo method has to be decreased. The Variance Reduction technique is useful for this aim. We will give evidence of the efficiency of one of the Variance Reduction thechniques - Control Variate method - in a mathematical context and a numerical comparison with the ordinary Monte Carlo method.</p>