Esta tese apresenta um desenvolvimento do fator de atrito (f) para escoamentos incompressíveis. O desenvolvimento é baseado no modelo clássico de Colebrook-White e no recente modelo da Entropia Máxima. Este desenvolvimento pode ser considerado como um modelo conceitual, porém não completamente, por causa do relacionamento entre o número de Reynolds (Re) e o parâmetro de entropia (M) determinado através de ajustes numéricos realizados com bons dados experimentais. Quatro algoritmos de cálculo foram criados para simplificar a aplicação do modelo, evidenciando sua eficácia e a eficiência. / This thesis presents a development of friction factor (f) for incompressible pipe flow calculation. The development is based on the classical Colebrook-White model and on the recent maximum entropy model. The development cam be considered as a conceptual one, but not completely, because the relationship that links the Reynolds number (Re) to the entropy parameter (M) was determined by numerical fitting on accurate but experimental data. Four calculation algorithms were produced to simplify the model applications, evidencing efficiency and effectiveness.

No description available.

The focus of this paper is to evaluate if forecast produced by the Central Bank of Sweden (Riksbanken) for the 12 month change in the consumer price index is unbiased? Results shows that for shorter horizons (h < 12) the mean forecast error is unbiased but for longer horizons its negatively biased when inference is done by Maximum entropy bootstrap technique. Can the unbiasedness be improved by strict ap- pliance to econometric methodology? Forecasting with a linear univariate model (seasonal ARIMA) and a multivariate model Vector Error Correction model (VECM) shows that when controlling for the presence of structural breaks VECM outperforms both prediction produced Riksbanken and ARIMA. However Riksbanken had the best precision in their forecast, estimated as MSFE

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

Predicting human behavior from a small amount of training examples is a challenging machine learning problem. In this thesis, we introduce the principle of maximum causal entropy, a general technique for applying information theory to decision-theoretic, game-theoretic, and control settings where relevant information is sequentially revealed over time. This approach guarantees decision-theoretic performance by matching purposeful measures of behavior (Abbeel & Ng, 2004), and/or enforces game-theoretic rationality constraints (Aumann, 1974), while otherwise being as uncertain as possible, which minimizes worst-case predictive log-loss (Gr¨unwald & Dawid, 2003). We derive probabilistic models for decision, control, and multi-player game settings using this approach. We then develop corresponding algorithms for efficient inference that include relaxations of the Bellman equation (Bellman, 1957), and simple learning algorithms based on convex optimization. We apply the models and algorithms to a number of behavior prediction tasks. Specifically, we present empirical evaluations of the approach in the domains of vehicle route preference modeling using over 100,000 miles of collected taxi driving data, pedestrian motion modeling from weeks of indoor movement data, and robust prediction of game play in stochastic multi-player games.

Urban planning has witnessed the profound changes in the methodologies of modelling during the last 50 years. Spatial interaction models have passed from social physics, statistical mechanics to non-spatial and spatial information processing stages of progress that can be designated as paradigm shifts. This thesis traces the Maximum Entropy (MaxEnt) approach in urban planning as pioneered by Wilson (1967,1970) and Spatial Entropy concept by Batty (1974) based on the Information Theory and its developments by Shannon (1948), Jaynes (1957), Kullback (1959) and by Tribus (1962,1969). Information-theoric methods have provided the theoretical foundation for challenging the uncertainty and incomplete information issues concerning the complex urban structure. MaxEnt, as a new logic, gives probabilities maximally noncommittal with regard to missing information. Wilson (1967,1970) has replaced the Newtonian analogy by the entropy concept from statistical mechanics to alleviate the mathematical inconsistency in the gravity model and developed a set of spatial interaction models consistent with the known information. Population density distribution as one of the determinants of the urban structure has been regarded as an exemplar to show the paradigm changes from the analysis of density gradients to the probabilistic description of density distributions by information-theoric methods. Spatial Entropy concept has introduced the spatial dimension to the Information Theory. Thesis applies Spatial Entropy measures to Ankara 1970 and 1990 census data by 34 zones and also obtains Kullback&rsquo / s Information Gain measures for population changes during the two decades. Empirical findings for Spatial Entropy measures show that overall Ankara-1970 and 1990 density distributions are &lsquo / &rsquo / Uneven&rsquo / &rsquo / and the uniform distribution hypothesis is not confirmed. These measures also indicate a tendency towards &ldquo / More Uniformity&rdquo / for density distributions in comparison to 1970. Information Gain measure for population changes also deviates from zero and direct proportionality hypothesis between posterior 1990 and prior 1970 population distributions by zones is not confirmed. Current research is focused on information processing with more engagement in the urban spatial structure and human behavior. This thesis aims to participate with these efforts and concludes that Information Theory has the potential to generate new profound changes in urban planning and modelling processes.

Thesis (Ph.D.)--University of Melbourne, Dept. of Computer Science and Software Engineering, Faculty of Engineering, 2007. / Typescript. Includes bibliographical references (leaves 171-179).

SENA, W. R. de. Teste de validade de métodos de maximização de entropia para construção de modelos com correlação par-a-par. 2017. 60 f. Dissertação (Mestrado em Física) – Centro de Ciências, Universidade Federal do Ceará, Fortaleza, 2017. / Submitted by Giordana Silva (giordana.nascimento@gmail.com) on 2017-04-18T17:57:20Z No. of bitstreams: 1 2017_dis_wrdsena.pdf: 3129941 bytes, checksum: 7bea0c98264c7b69414f8873e73b1291 (MD5) / Approved for entry into archive by Giordana Silva (giordana.nascimento@gmail.com) on 2017-04-18T17:57:42Z (GMT) No. of bitstreams: 1 2017_dis_wrdsena.pdf: 3129941 bytes, checksum: 7bea0c98264c7b69414f8873e73b1291 (MD5) / Made available in DSpace on 2017-04-18T17:57:42Z (GMT). No. of bitstreams: 1 2017_dis_wrdsena.pdf: 3129941 bytes, checksum: 7bea0c98264c7b69414f8873e73b1291 (MD5) Previous issue date: 2017 / In the 21st century humanity has produced more new data (information) than in all its history. Understanding the nature of the various systems that generate this abundance of data has became the great challenge of this century. One way to formally analyze these large databases is to use the information theory developed by Claude Shannon. This theory allows us, using the principle of maximum entropy, to find the distributions of probabilities that best describes the collective behavior of these systems. In this dissertation we discuss the possibility of using Ising models to describe observation of real systems. Due to its limitations, employing the Ising model implies that the elements that constitute the real system can only be in two states, for example active or inactive. In addition, the Ising model counts only interactions between pairs of elements and disregards the possibility of interactions between larger groups of elements. As we will discuss, even with these limitations such a model can well describe results observed in some natural systems, such as networks of neurons. Specifically, we discuss results from earlier work that show that using only the activity averages of each neuron and the correlation between them, using Shannon’s theory, we observe that the states visited by the network follow the Ising distribution. In order to test the applicability of this method in several systems we generate synthetic data, obtained from Ising model in three systems: ferromagnetic, antiferromagnetic and spin glass. We call the system that generate the synthetic data as underlying system. We use methods of maximization of entropy to try to construct model systems that can reproduce the mean and correlations observed in the synthetic data. We thus verify in which situations our methods can actually generate a model system that reproduces the underlying system that generated the data. These results may establish a limit of applicability for the technique discussed. / No século XXI a humanidade produziu mais novos dados (informações) do que em toda sua história. Entender a natureza dos diversos sistemas que geram essa abundância de dados se tornou um dos grandes desafios desse século. Uma forma de analisar formalmente esses grandes bancos de dados é empregando a teoria da informação desenvolvida por Claude Shannon. Essa teoria permite, usando o princípio da máxima entropia, encontrar as distribuições de probabilidades que melhor descrevem os comportamentos coletivos desses sistemas. Nessa dissertação, discutimos a possibilidade de usar modelos tipo Ising para descrever observações de sistemas reais. Devido a suas limitações, empregar o modelo de Ising implica em supor que os elementos que constituem o sistema real só podem estar em dois estados, por exemplo ativo ou inativo. Além disso, o modelo de Ising da conta apenas de interações entre pares de elementos e desconsidera a possibilidade de interações entre grupos maiores de elementos. Como discutiremos, mesmo com essas limitações tal modelo pode descrever bem resultados observados em alguns sistemas naturais, como por exemplo redes de neurônios. Especificamente, discutiremos resultados de trabalhos anteriores que mostram que usando apenas as médias de atividade de cada neurônio e a correlação entre os mesmo, usando a teoria de Shannon, observa-se que os estados visitados pela rede seguem à distribuição de Ising. Para testar a aplicabilidade desse método em diversos sistemas geramos dados sintéticos, obtidos de modelos tipo Ising em três situações: ferromagnético, anti-ferro e vidro de spins (spin glass). Nós chamamos o sistema que gera os dados sintéticos de sistema subjacente. Usamos métodos de maximização de entropia para tentar construir sistemas modelos que consigam reproduzir as média e correlações observadas nos dados sintéticos. Dessa forma, verificamos em que situações nossos métodos conseguem de fato gerar um sistema modelo que reproduza o sistema subjacente que gerou os dados. Esses resultados podem estabelecer um limite de aplicabilidade para a técnica discutida.

Cette thèse propose une nouvelle méthode pour l'estimation non-paramétrique de densité à partir de données censurées par des régions de formes quelconques, éléments de partitions du domaine paramétrique. Ce travail a été motivé par le besoin d'estimer la distribution des paramètres d'un modèle biophysique de décompression afin d'être capable de prédire un risque d'accident. Dans ce contexte, les observations (grades de plongées) correspondent au comptage quantifié du nombre de bulles circulant dans le sang pour un ensemble de plongeurs ayant exploré différents profils de plongées (profondeur, durée), le modèle biophysique permettant de prédire le volume de gaz dégagé pour un profil de plongée donné et un plongeur de paramètres biophysiques connus. Dans un premier temps, nous mettons en évidence les limitations de l'estimation classique de densité au sens du maximum de vraisemblance non-paramétrique. Nous proposons plusieurs méthodes permettant de calculer cet estimateur et montrons qu'il présente plusieurs anomalies : en particulier, il concentre la masse de probabilité dans quelques régions seulement, ce qui le rend inadapté à la description d'une population naturelle. Nous proposons ensuite une nouvelle approche reposant à la fois sur le principe du maximum d'entropie, afin d'assurer une régularité convenable de la solution, et mettant en jeu le critère du maximum de vraisemblance, ce qui garantit une forte attache aux données. Il s'agit de rechercher la loi d'entropie maximale dont l'écart maximal aux observations (fréquences de grades observées) est fixé de façon à maximiser la vraisemblance des données. / This thesis proposes a new method for nonparametric density estimation from censored data, where the censing regions can have arbitrary shape and are elements of partitions of the parametric domain. This study has been motivated by the need for estimating the distribution of the parameters of a biophysical model of decompression, in order to be able to predict the risk of decompression sickness. In this context, the observations correspond to quantified counts of bubbles circulating in the blood of a set of divers having explored a variety of diving profiles (depth, duration); the biophysical model predicts of the gaz volume produced along a given diving profile for a diver with known biophysical parameters. In a first step, we point out the limitations of the classical nonparametric maximum-likelihood estimator. We propose several methods for its calculation and show that it suffers from several problems: in particular, it concentrates the probability mass in a few regions only, which makes it inappropriate to the description of a natural population. We then propose a new approach relying both on the maximum-entropy principle, in order to ensure a convenient regularity of the solution, and resorting to the maximum-likelihood criterion, to guarantee a good fit to the data. It consists in searching for the probability law with maximum entropy whose maximum deviation from empirical averages is set by maximizing the data likelihood. Several examples illustrate the superiority of our solution compared to the classic nonparametric maximum-likelihood estimator, in particular concerning generalisation performance.

Neste trabalho, fazemos um estudo das correntes e das medidas transversas invariantes por holonomia, e mostraremos o resultado de D. Sullivan [23] sobre a correspondência biunívoca entre estes dois objetos. Em particular mostraremos um resultado conhecido de J. Plante [17] sobre a existência de medidas transversas invariantes sob a hipótese de crescimento sub-exponencial. Apresentamos também, o resultado devido a Ruelle-Sullivan [19] de que a medida de máxima entropia de um difeomorfismo topologicamente mixing pode-se expressar como o produto de duas medidas transversas invariantes para as folheações estáveis e instáveis. Por último, mostramos que os difeomorfismos de Anosov topologicamente mixing, que preservam a orientação das folhas estáveis e folhas instáveis induzem elementos da cohomologia de DeRham / In this work, we make a study of currents and holonomy invariant transverse measure, and we will show the result of D. Sullivan [23] about the biunivocal correspondence between these two objects. In particular we show a known result of J. Plante [17] about the existence of invariant transverse measures under the hypothesis of sub-exponential growth. Also we will present, the result due to Ruelle-Sullivan [19] that the maximum entropy measure of a diffeomorphism topologically mixing can be expressed as the product of two invariant transverse measures for stable and unstable foliations. Finally, we show that the Anosov diffeomorphisms topologically mixing, which preserve the orientation of the leaves stable and unstable induce elements DeRham cohomology