Les modèles linéaires latents sont des modèles statistique puissants pour extraire la structure latente utile à partir de données non structurées par ailleurs. Ces modèles sont utiles dans de nombreuses applications telles que le traitement automatique du langage naturel et la vision artificielle. Pourtant, l'estimation et l'inférence sont souvent impossibles en temps polynomial pour de nombreux modèles linéaires latents et on doit utiliser des méthodes approximatives pour lesquelles il est difficile de récupérer les paramètres. Plusieurs approches, introduites récemment, utilisent la méthode des moments. Elles permettent de retrouver les paramètres dans le cadre idéalisé d'un échantillon de données infini tiré selon certains modèles, mais ils viennent souvent avec des garanties théoriques dans les cas où ce n'est pas exactement satisfait. Dans cette thèse, nous nous concentrons sur les méthodes d'estimation fondées sur l'appariement de moment pour différents modèles linéaires latents. L'utilisation d'un lien étroit avec l'analyse en composantes indépendantes, qui est un outil bien étudié par la communauté du traitement du signal, nous présentons plusieurs modèles semiparamétriques pour la modélisation thématique et dans un contexte multi-vues. Nous présentons des méthodes à base de moment ainsi que des algorithmes pour l'estimation dans ces modèles, et nous prouvons pour ces méthodes des résultats de complexité améliorée par rapport aux méthodes existantes. Nous donnons également des garanties d'identifiabilité, contrairement à d'autres modèles actuels. C'est une propriété importante pour assurer leur interprétabilité. / Latent linear models are powerful probabilistic tools for extracting useful latent structure from otherwise unstructured data and have proved useful in numerous applications such as natural language processing and computer vision. However, the estimation and inference are often intractable for many latent linear models and one has to make use of approximate methods often with no recovery guarantees. An alternative approach, which has been popular lately, are methods based on the method of moments. These methods often have guarantees of exact recovery in the idealized setting of an infinite data sample and well specified models, but they also often come with theoretical guarantees in cases where this is not exactly satisfied. In this thesis, we focus on moment matchingbased estimation methods for different latent linear models. Using a close connection with independent component analysis, which is a well studied tool from the signal processing literature, we introduce several semiparametric models in the topic modeling context and for multi-view models and develop moment matching-based methods for the estimation in these models. These methods come with improved sample complexity results compared to the previously proposed methods. The models are supplemented with the identifiability guarantees, which is a necessary property to ensure their interpretability. This is opposed to some other widely used models, which are unidentifiable.

The Burr families (Type III and Type XII) of distributions are traditionally used in the context of statistical modeling and for simulating non-normal distributions with moment-based parameters (e.g., Skew and Kurtosis). In educational and psychological studies, the Burr families of distributions can be used to simulate extremely asymmetrical and heavy-tailed non-normal distributions. Conventional moment-based estimators (i.e., the mean, variance, skew, and kurtosis) are traditionally used to characterize the distribution of a random variable or in the context of fitting data. However, conventional moment-based estimators can (a) be substantially biased, (b) have high variance, or (c) be influenced by outliers. In view of these concerns, a characterization of the Burr Type III and Type XII distributions through the method of L-moments is introduced. Specifically, systems of equations are derived for determining the shape parameters associated with user specified L-moment ratios (e.g., L-Skew and L-Kurtosis). A procedure is also developed for the purpose of generating non-normal Burr Type III and Type XII distributions with arbitrary L-correlation matrices. Numerical examples are provided to demonstrate that L-moment based Burr distributions are superior to their conventional moment based counterparts in the context of estimation, distribution fitting, and robustness to outliers. Monte Carlo simulation results are provided to demonstrate that L-moment-based estimators are nearly unbiased, have relatively small variance, and are robust in the presence of outliers for any sample size. Simulation results are also provided to show that the methodology used for generating correlated non-normal Burr Type III and Type XII distributions is valid and efficient. Specifically, Monte Carlo simulation results are provided to show that the empirical values of L-correlations among simulated Burr Type III (and Type XII) distributions are in close agreement with the specified L-correlation matrices.

[pt] Esta dissertação apresenta um novo tipo de laje mista com uso de perfil de chapa dobrada incorporada. O perfil substitui as barras de aço, a fôrma e as escoras. O sistema estrutural construído consistiu em uma laje mista com perfil C enrijecido preenchido com concreto estrutural. Foram realizados dez testes experimentais em que foram analisados os momentos positivos, os momentos negativos e a influência de séptos. Através da análise teórico-experimental, foi possível perceber a grande influência dos séptos na resistência das lajes mistas testadas. / [en] This work presents a new type of composite deck-slab using incorporate formed steel profiles. The steel profile replaces the reinforcement bars, the form and charry system. The structural system is built by one composite deck-slab with C stiffness steel profile, filed with reinforced concrete. It was performed ten tested and it was analyzed the positive moments, the negative moments and the influence of the diaphragm. Analizing the results it was possible to realize the great influence of the diaphragm over the resistence of composite deck slab tested.

Two different time-resolved spectroscopic techniques are discussed theoretically and demonstrated experimentally in dilute gases. The first technique involves extending the advantages of Stark-effect based time-resolved spectroscopy to non-polar molecules. This involves the development of a stable, TEM₀₀ mode, cw, CO₂ laser capable of switching rapidly and controllably between two frequencies. Design problems and output characteristics are discussed. The frequency switchable laser is applied to the CO₂ 10.6 μm P(16) coincidence with the non-polar molecule SF₆. The population relaxation time, T₁, is measured using two-pulse delayed nutation. The decay of induced dipoles is studied using the phenomenon of photon echoes. It is found that the echoes decay in a manner characteristic of dephasing dominated by velocity-changing collisions. A fit of the data to a model for such decays gives values of γ(ab) ≡ 1/T₂ (the non-velocity-changing contribution to the dipole decay rate), Γ(VC) (the total probability of a velocity-changing collision per unit time), and Δu which is related to the mean velocity change of SF₆ upon a velocity changing collision. A comparison with the published results of the similar Stark experiments on C¹³ H₃F are made. The second technique involves the development of an alternative to the pulsed excitation typically used in time-resolved T₁ studies. This involves inversion of a portion of the velocity distribution by adiabatic rapid passage (ARP) techniques. The center of this portion is then probed in the manner of previous delayed nutation experiments. The system preparation is shown theoretically to be different and simpler than the pulse case. In addition, ARP preparation gives a larger signal than two-pulse delayed nutation experiments. ARP experiments on N¹⁴H₃ and N¹⁵H₃ are described and compared to two-pulse delayed notation experiments. The single exponential decay best fits to the data from the two methods are found to be in agreement. We would expect the N¹⁵H₃ results to be very similar to the N¹⁴H₃ results, though reduced rotational resonance effects in its upper state should give it an overall slower decay. It is indeed found that the decay appears to be a simple exponential as did the N¹⁴H₃ data over the time range studied. The pressure dependent single exponential decay rate for N¹⁵H₃ is however roughly 45% larger than the rate for N¹⁴H₃ in the pressure range from 0.5 to 9 mTorr.

This thesis consists of three parts. The first part is an introduction and referencessome recent work on 2D electromagnetic scattering problems at high frequencies. It alsopresents the basic integral equation types for impenetrable objects. A brief discussionof the standard elements of the method of moments is followed by summaries of thepapers.Paper I presents an accurate implementation of the method of moments for a perfectlyconducting cylinder. A scaling for the rapid variation of the solution improves accuracy.At high frequencies, the method of moments leads to a large dense system of equations.Sparsity in this system is obtained by modifying the integration path in the integralequation. The modified path reduces the accuracy in the deep shadow.In paper II, a hybrid method is used to handle the standing waves that are prominentin the shadow for the TE case. The shadow region is treated separately, in a hybridscheme based on a priori knowledge about the solution. An accurate method to combinesolutions in this hybrid scheme is presented.

A dissertation submitted to the faculty of Engineering, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of science in Engineering Johannesburg 1991 / This dissertation discusses the interaction between shipboard antennas and their environment. The emphasis is on the Use of the Method of Moments to calculate the currents on the structure of the ship. These currents are induced by the antennas mounted on the structure of the ship. The parameters (such as grid spacing and wire radius) to use in creating a wire grid model of the ship is investigated and recommended values given. A sample ship is analyzed and the results obtained compared with measurements done in an unechoic chamber.

Le processus de sélection de portefeuille peut être effectué en deux étapes: la première consiste à évaluer les actifs financiers et la deuxième à déterminer la combinaison d’actifs qui permettrai d‘allouer de façon optimale la richesse. La combinaison des actifs financiers retenue à la fin de ce processus se doit de répondre simultanément et de façon optimale aux différents objectifs de l‘investisseur. Le problème de sélection de portefeuille peut être considéré comme un processus de décision multicritère. Dans cette thèse, plusieurs critères ont été analysés et on a tenté de répondre à la question de combien et où investir ? On a proposé une nouvelle approche multicritère basée sur la méthode d‘enveloppement des données (DEA) et l‘approche de l‘efficacité croisée pour sélectionner un portefeuille d’actifs financier. La première méthodologie proposée consiste à incorporer la méthode d’efficacité croisée dans un espace Moyenne-Variance-Skewness-Kurtosis (MVSK). Le deuxième modèle combine la mesure de l'efficacité croisée par l'enveloppement de données vue comme solution d'un jeu avec la composante risque pour choisir un portefeuille. Finalement, on a proposé d’incorporer la mesure de l'efficacité croisée par l'enveloppement de données vue comme solution d'un jeu dans un modèle d’arbitrage entre profitabilité et efficacité afin de sélectionner un portefeuille. Globalement, ces méthodes ont permis la discrimination entre les actifs financiers et de leur donner un classement unique dans un premier temps, ensuite de sélectionner un portefeuille en prenant en considération les préférences du décideur. / The process of portfolio selection could be divided into two stages: the first one is the evaluation of financial assets and the second is to choose the best ones to construct portfolio. It can be considered as Multi-Criteria-Decision-Making (MCDM) process. It consists in selecting a combination of financial assets that can best meet the investors’ objective. In this dissertation, different criteria are analyzed and the question of where and how much money to allocate to each of the financial asset is processed. We propose a new multi-criteria analysis approach to portfolio selection based on Data Envelopment Analysis (DEA) cross-efficiency model. The first methodology consists in nesting the DEA cross-efficiency model into the Mean-Variance-Skewness-Kurtosis (MVSK) space. The second model combines the DEA game cross-efficiency approach with risk component to select portfolio. Finally, we propose a model incorporating the DEA game cross-efficiency into Profitability-Efficiency. Overall, these methodologies provide more discrimination for financial assets by providing unique ranks in a first step and permit to select portfolio by underlying preferences of the decision-maker in a second step.

"This research concentrates on two major engineering areas associated with biomedical instrumentation that have recently gained significant academic and industrial interest: the gradient coil design for Magnetic Resonance Imaging (MRI) and the high frequency full-wave field simulations with the Method of Moments (MoM). A new computational approach to the design of gradient coils for magnetic resonance imaging is introduced. The theoretical formulation involves a constrained cost function between the desired field in a particular region of interest in space and the current-carrying coil plane. Based on Biot-Savart’s integral equation, an appropriate weight function is introduced in conjunction with linear approximation functions. This permits the transformation of the problem formulation into a linear matrix equation whose solution yields discrete current elements in terms of magnitude and direction within a specified coil plane. These current elements can be synthesized into practical wire configuration by suitably combining the individual wire loops. Numerical predictions and measurements underscore the success of this approach in terms of achieving a highly linear field while maintaining low parasitic fields, low inductance and a sufficient degree of shielding. Experimental results confirm the field predictions of the computational approach. Extending the numerical modeling efforts to dynamic phenomena, a novel MoM formulation permits the computation of electromagnetic fields in conductive surfaces and in three-dimensional biological bodies. The excitation can be provided with current loops, voltage sources, or an incident electromagnetic wave. This method enables us to solve a broad spectrum of problems arising in MRI: full-wave RF coil simulations, eddy currents predictions in the magnet bore, and induced currents in the biological body. Surfaces are represented as triangles and the three-dimensional bodies are subdivided into tetrahedra. This numerical discretization methodology makes the approach very flexible to handle a wide range of practical coil geometries. Specifically, in this thesis the MoM is employed to study the effect of switching gradient coils in the presence of a biological load. "

This papers is divided into numbered sections. The equations are numbered anew in each section, and equation numbers are always enclosed in parentheses. Merely the equation number is given when referring to an equation in the same section as the references; otherwise the section number is prefixed. Thus equation (4) refers to the fourth equation of the same section as the reference, and equation (2.2) refers to the second equation of the second section.

L'évolution de la structure nucléaire loin de la stabilité est un sujet d'actualité de la physique nucléaire expérimentale et théorique.<br> Dans cette étude nous nous sommes intéressés à la structure des noyaux riches en neutrons autour de N=40. L'évolution en énergie de l'orbitale νg9/2 avec le nombre de neutrons est un des points clés pour définir la structure à basse énergie de ces noyaux.<br> Comme outils d'investigation, nous avons utilisé la mesure des moments magnétiques dipolaires et la décroissance β. Pour la mesure du facteur gyromagnétique de l'état 9/2+ du 61mFe nous avons appliqué la méthode TDPAD. Cette méthode (comme la majorité des méthodes de mesure de moments nucléaires statiques) nécessite que le spin du noyau mesuré soit orienté. L'orientation du 61mFe a été obtenue lors de la réaction de fragmentation du 64Ni à 55 MeV/u et par sélection en moment des fragments avec le spectromètre LISE au GANIL. Le dispositif expérimental utilisé nous a permis de garder cette orientation jusqu'au point d'implantation. La valeur mesurée du facteur g a été comparée avec les prédictions des calculs de modèle en couches à grande échelle et Hartree-Fock-Bogoliubov.<br> Pour l'étude de la décroissance β nous avons produit les noyaux riches en neutrons autour de N=40 par la fragmentation du 86Kr à 58 MeV/u. La sélection des produits de réaction a été faite en utilisant pour la première fois le spectromètre LISE2000. Les rayonnements γ ont été détectés avec quatre détecteurs segmentés EXOGAM. Nous avons ainsi déterminé pour la première fois les valeurs de 5 périodes radioactives et amélioré la précision sur 4 autres valeurs. Le peuplement des états excités dans les noyaux fils nous a permis d'identifier de nouvelles transitions et de déterminer pour la première fois la position en énergie du premier état 2+ de 68Fe et de 72Ni. Les résultats expérimentaux ont été comparés avec des prédictions de modèle en couches. D'autres transitions ont été observées pour la première fois dans la décroissance βγ du 60Ti, 70Fe, 71,73Co.