[pt] Com o crescimento das indústrias e a competitividade entre os mercados o transporte de cargas urbano tem-se considerado fundamental para a economia, mas a importância de sua relação direta com a vida das pessoas nas cidades tem ocasionado que exista maior congestionamento nas regiões centrais. É, por isso, que é importante ter o conhecimento sobre o deslocamento das cargas urbanas dentro da Região Metropolitana do Rio de Janeiro – RMRJ. Este trabalho trata de estimar matrizes origem destino (O-D) a partir de contagens de fluxos observadas na rede. Estas contagens foram realizadas para dois tipos de horários considerados importantes por motivo de restrições de circulação, sendo estes o pico da manhã (7:00 às 8:00) e pico da tarde (17:30 às 18:30). Após fazer uma revisão bibliográfica sobre o assunto, o trabalho define entre os métodos conhecidos o que foi considerado mais adequado para a estimativa da matriz OD. Decidiu-se pelo método desenvolvido por Nielsen (1998) denominado como Método de Caminho Único - SPME. O Método foi aplicado na rede viária do plano diretor de transportes urbanos da Região Metropolitana do Rio de Janeiro (PDTU-RMRJ), usando o software TransCAD. / [en] With the growth of industries and competitiveness between markets, the urban freight transport has been considered fundamental for the economy, but the importance of their direct relationship to the lives of people in cities has resulted in more congestion in the central regions. It is therefore important to have knowledge about the movement of urban freight inside the Metropolitan Area of Rio de Janeiro - MARJ. This work try to estimate origin destination matrices (O-D) from flows observed on the network. These counts were done for two types of times considered important for reasons of traffic restrictions, which are the peak of the morning (7:00 - 8:00) and late peak (17:30 - 18:30). After doing a literature review on the issue, the work defines between the known methods the one that was considered most appropriate for the estimation of O-D matrix. Decided for the method developed by Nielsen (1998) termed as Single Path Matrix Estimation - SPME. The method was applied to the road network of the master plan of the urban transport in the Metropolitan Area of Rio de Janeiro (PDTU-RMRJ), using software TransCAD.

The representer theorem from the reproducing kernel Hilbert space theory is the origin of many kernel-based machine learning and signal modelling techniques that are popular today. Most kernel functions used in practical applications behave in a homogeneous manner across the domain of the signal of interest, and they are called stationary kernels. One open problem in the literature is the specification of a non-stationary kernel that is computationally tractable. Some recent works solve large-scale optimization problems to obtain such kernels, and they often suffer from non-identifiability issues in their optimization problem formulation. Many practical problems can benefit from using application-specific prior knowledge on the signal of interest. For example, if one can adequately encode the prior assumption that edge contours are smooth, one does not need to learn a finite-dimensional dictionary from a database of sampled image patches that each contains a circular object in order to up-convert images that contain circular edges. In the first portion of this thesis, we present a novel method for constructing non-stationary kernels that incorporates prior knowledge. A theorem is presented that ensures the result of this construction yields a symmetric and positive-definite kernel function. This construction does not require one to solve any non-identifiable optimization problems. It does require one to manually design some portions of the kernel while deferring the specification of the remaining portions to when an observation of the signal is available. In this sense, the resultant kernel is adaptive to the data observed. We give two examples of this construction technique via the grayscale image up-conversion task where we chose to incorporate the prior assumption that edge contours are smooth. Both examples use a novel local analysis algorithm that summarizes the p-most dominant directions for a given grayscale image patch. The non-stationary properties of these two types of kernels are empirically demonstrated on the Kodak image database that is popular within the image processing research community. Tensors and tensor decomposition methods are gaining popularity in the signal processing and machine learning literature, and most of the recently proposed tensor decomposition methods are based on the tensor power and alternating least-squares algorithms, which were both originally devised over a decade ago. The algebraic approach for the canonical polyadic (CP) symmetric tensor decomposition problem is an exception. This approach exploits the bijective relationship between symmetric tensors and homogeneous polynomials. The solution of a CP symmetric tensor decomposition problem is a set of p rank-one tensors, where p is fixed. In this thesis, we refer to such a set of tensors as a rank-one decomposition with cardinality p. Existing works show that the CP symmetric tensor decomposition problem is non-unique in the general case, so there is no bijective mapping between a rank-one decomposition and a symmetric tensor. However, a proposition in this thesis shows that a particular space of rank-one decompositions, SE, is isomorphic to a space of moment matrices that are called quasi-Hankel matrices in the literature. Optimization over Riemannian manifolds is an area of optimization literature that is also gaining popularity within the signal processing and machine learning community. Under some settings, one can formulate optimization problems over differentiable manifolds where each point is an equivalence class. Such manifolds are called quotient manifolds. This type of formulation can reduce or eliminate some of the sources of non-identifiability issues for certain optimization problems. An example is the learning of a basis for a subspace by formulating the solution space as a type of quotient manifold called the Grassmann manifold, while the conventional formulation is to optimize over a space of full column rank matrices. The second portion of this thesis is about the development of a general-purpose numerical optimization framework over SE. A general-purpose numerical optimizer can solve different approximations or regularized versions of the CP decomposition problem, and they can be applied to tensor-related applications that do not use a tensor decomposition formulation. The proposed optimizer uses many concepts from the Riemannian optimization literature. We present a novel formulation of SE as an embedded differentiable submanifold of the space of real-valued matrices with full column rank, and as a quotient manifold. Riemannian manifold structures and tangent space projectors are derived as well. The CP symmetric tensor decomposition problem is used to empirically demonstrate that the proposed scheme is indeed a numerical optimization framework over SE. Future investigations will concentrate on extending the proposed optimization framework to handle decompositions that correspond to non-symmetric tensors.

[pt] Apresentam-se as formulações, consolidando a nomenclatura e os principais conceitos dos métodos de elementos de contorno: convencional (MCCEC), híbrido de tensões (MHTEC), híbrido de deslocamentos (MHDEC) e híbrido simplificado de tensões (MHSTEC). proposto o método híbrido simplificado de deslocamentos (MHSDEC), em contrapartida ao MHSTEC, baseando-se nas mesmas hipóteses de aproximação de tensões e deslocamentos do MHDEC e supondo que a solução fundamental em termos de tensões seja válida no contorno. Como decorrência do MHSTEC e do MHSDEC, é apresentado também o método híbrido de malha reduzida dos elementos de contorno (MHMREC), com aplicação computacionalmente vantajosa a problemas no domínio da freqüência ou envolvendo materiais não-homogêneos. A partir da investigação das equações matriciais desses métodos, são identificadas quatro novas relações matriciais, das quais uma verifica-se como válida para a obtenção dos elementos das matrizes de flexibilidade e de deslocamento que não podem ser determinados por integração ou avaliação direta. Também é proposta a correta consideração, ainda não muito bem explicada na literatura, de que forças de superfície devem ser interpoladas em função de atributos de superfície e não de atributos nodais. São apresentadas aplicações numéricas para problemas de potencial para cada método mencionado, em que é verificada a validade das novas relações matriciais. / [en] A consolidated, unified formulation of the conventional (CCBEM), hybrid stress (HSBEM), hybrid displacement (HDBEM) and simplified hybrid stress (SHSBEM) boundary element methods is presented. As a counterpart of SHSBEM, the simplified hybrid displacement boundary element method (SHDBEM) is proposed on the basis of the same stress and displacement approximation hypotheses of the HDBEM and on the assumption that stress fundamental solutions are also valid on the boundary. A combination of the SHSBEM and the SHDBEM gives rise to a provisorily called mesh-reduced hybrid boundary element method (MRHBEM), which seems computationally advantageous when applied to frequency domain problems or non-homogeneous materials. Four new matrix relations are identified, one of which may be used to obtain the flexibility and displacement matrix coefficients that cannot be determined by integration or direct evaluation. It is also proposed the correct consideration, still not well explained in the technical literature, that traction forces should be interpolated as functions of surface and not of nodal attributes. Numerical examples of potential problems are presented for each method, in which the validity of the new matrix relations is verified.

Cette thèse est consacrée à l'étude asymptotique d'objets liés au mouvement brownien sur le groupe unitaire en grande dimension, ainsi qu'à l'étude, dans le cadre des probabilités libres, des versions non-commutatives de ces objets. Elle se subdivise essentiellement en trois parties.Dans le chapitre 2, nous résolvons le problème initial de cette thèse, à savoir la convergence de la transformation de Hall sur le groupe unitaire vers la transformation de Hall libre, lorsque la dimension tend vers l'infini. Pour résoudre ce problème, nous établissons des théorèmes d'existence de noyaux de transition pour la convolution libre. Enfin, nous utilisons ces résultats pour prouver que, pareillement au mouvement brownien sur le groupe unitaire, le mouvement brownien sur le groupe linéaire converge en distribution non-commutative vers sa version libre. Nous étudions les fluctuations autour de cette convergence dans le chapitre 3. Le chapitre 4 présente un morphisme entre les mesures infiniment divisibles pour la convolution libre additive d'une part et multiplicative de l'autre. Nous montrons que ce morphisme possède une version matricielle qui s'appuie sur un nouveau modèle de matrices aléatoires pour les processus de Lévy libres multiplicatifs. / This thesis focuses on the asymptotic of objects related to the Brownian motion on the unitary group in large dimension, and on the study, in free probability, of the non-commutative versions of those objects. It subdivides into essentially three parts.In Chapter 2, we solve the original problem of this thesis: the convergence of the Hall transform on the unitary group to the free Hall transform, as the dimension tends to infinity. To solve this problem, we establish theorems of existence of transition kernel for the free convolution. Finally, we use these results to prove that, exactly as the Brownian motion on the unitary group, the Brownian motion on the linear group converges in noncommutative distribution to its free version. Then we study the fluctuations around this convergence in Chapter 3. Chapter 4 presents a homomorphism between infinitely divisible measures for the free convolution, in respectively the additive case and the multiplicative case. We show that this homomorphism has a matricialversion which is based on a new model of random matrices for the free multiplicative Lévy processes.

The central theme of complex systems research is to understand the emergent macroscopic properties of a system from the interplay of its microscopic constituents. The emergence of macroscopic properties is often intimately related to the structure of the microscopic interactions. Here, we present an analytical approach for deriving necessary conditions that an interaction network has to obey in order to support a given type of macroscopic behaviour. The approach is based on a graphical notation, which allows rewriting Jacobi’s signature criterion in an interpretable form and which can be applied to many systems of symmetrically coupled units. The derived conditions pertain to structures on all scales, ranging from individual nodes to the interaction network as a whole. For the purpose of illustration, we consider the example of synchronization, specifically the (heterogeneous) Kuramoto model and an adaptive variant. The results complete and extend the previous analysis of Do et al. (2012 Phys. Rev. Lett. 108, 194102).

This paper is devoted to asymptotic estimates for the condition numbers $\kappa(T_n(a))=||T_n(a)|| ||T_n^(-1)(a)||$ of large $n\cross n$ Toeplitz matrices $T_N(a)$ in the case where $\alpha \element L^\infinity$ and $Re \alpha \ge 0$ . We describe several classes of symbols $\alpha$ for which $\kappa(T_n(a))$ increases like $(log n)^\alpha, n^\alpha$ , or even $e^(\alpha n)$ . The consequences of the results for singular values, eigenvalues, and the finite section method are discussed. We also consider Wiener-Hopf integral operators and multidimensional Toeplitz operators.

This thesis describes improvements of the periodic QZ algorithm and several variants of the Schur algorithm for block Toeplitz matrices. Documentation of the available software is included.

Dissertation ueber die effektive Vorkonditionierung linearer Gleichungssysteme resultierend aus der Diskretisierung eines elliptischen Randwertproblems 2. Ordnung mittels der Methode der Finiten Elementen. Als Vorkonditionierer werden multi-level artige Vorkonditionierer (BPX, Multi-grid, Wavelets) benutzt.

Thesis are aimed to the optimization and validation of the QuEChERS method for determination of polycyclic aromatic hydrocarbons in solid matrices of hydrosphere. The QuEChERS method was also used for determination of polycyclic aromatic hydrocarbons in real samples and for comparison of measured results, time and material costs of the method with currently employed methods: (1) accelerated solvent extraction connected with gel permeation chromatography (ASE/GPC), and (2) ultrasonic extraction connected with solid phase extraction (UZ/SPE). According to the validation criteria, the QuEChERS method is suitable for the determination of polycyclic aromatic hydrocarbons in solid matrices of hydrosphere. The QuEChERS technique provides comparable results to ASE/GPC and UZ/SPE. In terms of price and time for sample preparation, the QuEChERS method allows (unlike the methods ASE/GPC and UZ/SPE) fast and inexpensive determination of polycyclic aromatic hydrocarbons in solid samples of hydrosphere.

The link between speech-in-noise recognition and cognition has been researched extensively over the years, and the purpose of this thesis was to add to this field. Data from a sample of 394 participants from the n200 database (Rönnberg et al., 2016) was used to calculate the correlation between their performance on a speech-in-noise test and their score on a test measuring fluid intelligence. The speech-in-noise test consisted of matrix sentences with 4-talker babble as noise and fluid intelligence was represented by the score on a Raven’s Progressive Matrices test. Around half of the participants (n = 199) had documented hearing loss and were hearing aid users, while the rest were participants with normal hearing. The overall correlation between speech-in-noise recognition and fluid intelligence was -.317, which shows that a better (lower) score on the speech-in-noise test is correlated to a better score in the Raven’s test. The same type of correlation was calculated within the two different groups, and the results showed correlation of -.338 for the group without hearing loss and one of -.303 for the group with hearing loss. The results indicate that there is a weak to moderate correlation between speech-in-noise and fluid intelligence, and they support the theory that cognitive processing is related to speech perception in all people, regardless of hearing status.