Cette présentation examinera le degré de certitude qui peut être atteint dans le domaine scientifique. Le paradigme scientifique est composé de deux extrêmes; causalité et déterminisme d'un côté et probabilité et indéterminisme de l'autre. En faisant appel aux notions de Hume de la ressemblance et la contiguïté, on peut rejeter la causalité ou le hasard objectif comme étant sans fondement et non empirique. Le problème de l'induction et le sophisme du parieur proviennent d’une même source cognitif / heuristique. Hume décrit ces tendances mentales dans ses essais « Of Probability » et « Of the Idea of Necessary Connexion ». Une discussion sur la conception de la probabilité de Hume ainsi que d'autres interprétations de probabilité sera nécessaire. Même si la science glorifie et idéalise la causalité, la probabilité peut être comprise comme étant tout aussi cohérente. Une attitude probabiliste, même si elle est également non empirique, pourrait être plus avantageuse que le vieux paradigme de la causalité. / This presentation examines the degree of certainty which can be attained in science. The scientific paradigm is composed of two extremes; causality and determinism on one end and probability and indeterminism on the other. By appealing to Hume’s notions of resemblance and contiguity, we can dismiss any claim of objective causality or chance as being ungrounded for lack of an empirical basis. The problem of induction as well as the gambler’s fallacy stem from the same cognitive/heuristic source. Hume describes these mental tendencies in his essays ‘Of Probability’ and ‘Of the Idea of Necessary Connexion’. This will necessitate a discussion of Hume’s notion of probability, as well as other interpretations of probability. While science has glorified and romanticized causality, probability can be understood as being just as consistent. While a probabilistic stance is as non-empirical as a causal stance, it will be remarked that we may benefit from a paradigmatic switch to probabilism.

Le but de ce mémoire de maîtrise est de décrire les propriétés de la loi double Pareto-lognormale, de montrer comment on peut introduire des variables explicatives dans le modèle et de présenter son large potentiel d'applications dans le domaine de la science actuarielle et de la finance. Tout d'abord, nous donnons la définition de la loi double Pareto-lognormale et présentons certaines de ses propriétés basées sur les travaux de Reed et Jorgensen (2004). Les paramètres peuvent être estimés en utilisant la méthode des moments ou le maximum de vraisemblance. Ensuite, nous ajoutons une variable explicative à notre modèle. La procédure d'estimation des paramètres de ce mo-\\dèle est également discutée. Troisièmement, des applications numériques de notre modèle sont illustrées et quelques tests statistiques utiles sont effectués. / The purpose of this Master's thesis is to describe the double Pareto-lognormal distribution, show how the model can be extended by introducing explanatory variables in the model and present its large potential of applications in actuarial science and finance. First, we give the definition of the double Pareto-lognormal distribution and present some of its properties based on the work of Reed and Jorgensen (2004). The parameters could be estimated by using the method of moments or maximum likelihood. Next, we add an explanatory variable to our model. The procedure of estimation for this model is also discussed. Finally, some numerical applications of our model are illustrated and some useful statistical tests are conducted.

La présente thèse porte sur différentes questions émanant de la géométrie spectrale. Ce domaine des mathématiques fondamentales a pour objet d'établir des liens entre la géométrie et le spectre d'une variété riemannienne. Le spectre d'une variété compacte fermée M munie d'une métrique riemannienne $g$ associée à l'opérateur de Laplace-Beltrami est une suite de nombres non négatifs croissante qui tend vers l’infini. La racine carrée de ces derniers représente une fréquence de vibration de la variété. Cette thèse présente quatre articles touchant divers aspects de la géométrie spectrale. Le premier article, présenté au Chapitre 1 et intitulé « Superlevel sets and nodal extrema of Laplace eigenfunctions », porte sur la géométrie nodale d'opérateurs elliptiques. L’objectif de mes travaux a été de généraliser un résultat de L. Polterovich et de M. Sodin qui établit une borne sur la distribution des extrema nodaux sur une surface riemannienne pour une assez vaste classe de fonctions, incluant, entre autres, les fonctions propres associées à l'opérateur de Laplace-Beltrami. La preuve fournie par ces auteurs n'étant valable que pour les surfaces riemanniennes, je prouve dans ce chapitre une approche indépendante pour les fonctions propres de l’opérateur de Laplace-Beltrami dans le cas des variétés riemanniennes de dimension arbitraire. Les deuxième et troisième articles traitent d'un autre opérateur elliptique, le p-laplacien. Sa particularité réside dans le fait qu'il est non linéaire. Au Chapitre 2, l'article « Principal frequency of the p-laplacian and the inradius of Euclidean domains » se penche sur l'étude de bornes inférieures sur la première valeur propre du problème de Dirichlet du p-laplacien en termes du rayon inscrit d’un domaine euclidien. Plus particulièrement, je prouve que, si p est supérieur à la dimension du domaine, il est possible d'établir une borne inférieure sans aucune hypothèse sur la topologie de ce dernier. L'étude de telles bornes a fait l'objet de nombreux articles par des chercheurs connus, tels que W. K. Haymann, E. Lieb, R. Banuelos et T. Carroll, principalement pour le cas de l'opérateur de Laplace. L'adaptation de ce type de bornes au cas du p-laplacien est abordée dans mon troisième article, « Bounds on the Principal Frequency of the p-Laplacian », présenté au Chapitre 3 de cet ouvrage. Mon quatrième article, « Wolf-Keller theorem for Neumann Eigenvalues », est le fruit d'une collaboration avec Guillaume Roy-Fortin. Le thème central de ce travail gravite autour de l'optimisation de formes dans le contexte du problème aux valeurs limites de Neumann. Le résultat principal de cet article est que les valeurs propres de Neumann ne sont pas toujours maximisées par l'union disjointe de disques arbitraires pour les domaines planaires d'aire fixée. Le tout est présenté au Chapitre 4 de cette thèse. / The main topic of the present thesis is spectral geometry. This area of mathematics is concerned with establishing links between the geometry of a Riemannian manifold and its spectrum. The spectrum of a closed Riemannian manifold M equipped with a Riemannian metric g associated with the Laplace-Beltrami operator is a sequence of non-negative numbers tending to infinity. The square root of any number of this sequence represents a frequency of vibration of the manifold. This thesis consists of four articles all related to various aspects of spectral geometry. The first paper, “Superlevel sets and nodal extrema of Laplace eigenfunction”, is presented in Chapter 1. Nodal geometry of various elliptic operators, such as the Laplace-Beltrami operator, is studied. The goal of this paper is to generalize a result due to L. Polterovich and M. Sodin that gives a bound on the distribution of nodal extrema on a Riemann surface for a large class of functions, including eigenfunctions of the Laplace-Beltrami operator. The proof given by L. Polterovich and M. Sodin is only valid for Riemann surfaces. Therefore, I present a different approach to the problem that works for eigenfunctions of the Laplace-Beltrami operator on Riemannian manifolds of arbitrary dimension. The second and the third papers of this thesis are focused on a different elliptic operator, namely the p-Laplacian. This operator has the particularity of being non-linear. The article “Principal frequency of the p-Laplacian and the inradius of Euclidean domains” is presented in Chapter 2. It discusses lower bounds on the first eigenvalue of the Dirichlet eigenvalue problem for the p-Laplace operator in terms of the inner radius of the domain. In particular, I show that if p is greater than the dimension, then it is possible to prove such lower bound without any hypothesis on the topology of the domain. Such bounds have previously been studied by well-known mathematicians, such as W. K. Haymann, E. Lieb, R. Banuelos, and T. Carroll. Their papers are mostly oriented toward the case of the usual Laplace operator. The generalization of such lower bounds for the p-Laplacian is done in my third paper, “Bounds on the Principal Frequency of the p-Laplacian”. It is presented in Chapter 3. My fourth paper, “Wolf-Keller theorem of Neumann Eigenvalues”, is a joint work with Guillaume Roy-Fortin. This paper is concerned with the shape optimization problem in the case of the Laplace operator with Neumann boundary conditions. The main result of our paper is that eigenvalues of the Neumann boundary problem are not always maximized by disks among planar domains of given area. This joint work is presented in Chapter 4.

Les chocs thermiques provoquent, en général, l’endommagement et la fissuration des matériaux. Ces phénomènes sont observés, par exemple, dans le revêtement de barrière thermique pour les moteurs des turbines, le traitement des surfaces ou la soudure par laser etc. Plusieurs travaux de recherche ont été réalisés au cours des dernières décennies dans l’objectif d’améliorer les performances thermiques et/ou mécaniques des matériaux sous chargement thermique. L’étude des dommages et de la fissuration des matériaux provoqués par les chocs thermiques, tels que le décollement des interfaces et de décohésion de revêtements, a reçu également une attention considérable par les chercheurs. La majorité de ces travaux utilisent les théories classiques, tels que la loi de Fourier de conduction thermique et l'hypothèse de quasi-statique. Malheureusement ces théories ne sont pas adaptées dans le cas de charges extrêmes provoqués par le choc thermique et dans le cas des matériaux micro-fissurés. En conséquence, les théories conventionnelles doivent être enrichies.L'objectif de la thèse est de montrer le rôle crucial des termes non Fourier et les termes inertiels dans le cas de choc thermique sous conditions sévères et dans le cas où les fissures sont petites. Pour cela nous avons mené des études sur deux structures particulières soumises à des chocs thermiques. Chaque structure contient une fissure parallèle au bord libre de la structure située au voisinage de ce dernier. L’influence de la présence de fissure sur la conductivité thermique est prise en compte. Nous avons utilisé la théorie Hyperbolique de transfert de chaleur par conduction pour les champs thermique et mécanique à la place de la théorie traditionnelle classique de Fourier. Pour mener cette étude, nous avons utilisé les Transformées de Laplace et de Fourier aux équations de mouvement et à l’équation de transfert de chaleur. En s’intéressant en particulier aux champs de contrainte au voisinage de la pointe de fissure et aux facteurs d'intensité de contrainte dynamiques. Le problème se ramène à la résolution d’un système d'équations intégrales singulières dans l'espace de Laplace-Fourier. On utilise une méthode d'intégration numérique pour obtenir les différents champs. Nous résolvons ensuite un système d'équations algébriques linéaires. En effectuant des inversions numériques des transformées, nous obtenons les champs de contrainte de température et les facteurs d'intensité de contrainte dynamiques dans le domaine temporel.Les résultats numériques montrent que la conductivité thermique du milieu est affectée par l’ouverture de la fissure ce qui perturberait fortement le champ de température ainsi que l'amplitude des facteurs d'intensité de contrainte dynamiques. Les amplitudes sont supérieures à celles obtenues à partir de la théorie classique de Fourier ainsi que dans le cadre de l'hypothèse quasi-statique. On constate également qu’elles oscillent au cours du temps. La prise en compte simultanément de l’influence de la fissure sur la conductivité thermique, de l'effet non-Fourier ainsi que les effetsIVd'inertie induit un couplage entre les trois phénomènes qui rendrait le problème de choc thermique très complexe. L'effet de barrière thermique induit par la fissure affecte d’une manière significative les champs de température et des contraintes. Les effets d’inertie, et des termes non-Fourier joueraient également un rôle non négligeable lorsque la longueur de la fissure est petite. Comme dans de nombreux problèmes d'ingénierie, l'initiation et la propagation des micro-fissures sont des mécanismes dont il faut tenir compte dans les prévisions de la rupture des structures. Ces effets non conventionnels ne sont plus négligeables et doivent être inclus dans l'analyse de la fracture des structures soumises à des chocs thermiques. / Thermal shock problems occur in many engineering materials and elements, which are used in high temperature applications such as thermal barrier coatings (TBCs), solid propellant of rocket-engine, pulsed-laser processing of materials, and so on. The thermal shock resistance performances and the thermal shock damages of materials, especially the interface debonding and spallation of coatings, have received considerable attention in both analysis and design. Some conventional theories, such as the Fourier’s law of thermal conduction and the quasi-static assumption of the thermoelastic body, may no longer be appropriate because of the extreme loads provoked by the thermal shock. Therefore, these conventional theories need to be enriched or revised.The objective of this thesis is to develop the solutions of the transient temperature field and thermal stresses around a partially insulated crack in a thermoelastic strip under thermal shock loading. The crack lies parallel to the heated traction free surface. The thermal conductivity of the crack gap is taken into account. Hyperbolic heat conduction theory is used in solving the temperature field instead of the traditional Fourier thermal conduction theory. Equations of motion are applied to obtain the stress fields and the dynamic stress intensity factors of the crack. The Laplace and Fourier transforms are applied to solve the thermal-elastic governing equations such that the mixed boundary value problems are reduced to solving a singular integral equations system in Laplace-Fourier space. The numerical integration method is applied to get the temperature field and stress fields, respectively. The problems are then solved numerically by converting the singular integral equations to a linear algebraic equations system. Finally, numerical inversions of the Laplace transform are performed to obtain the temperature field and dynamic stress intensity factors in the time domain.Numerical results show that the thermal conductivity of the crack gap strongly affects the uniformity of the temperature field and consequently, the magnitude of the dynamic stress intensity factors of the crack. The stress intensity factors would have higher amplitude and oscillating feature comparing to those obtained under the conventional Fourier thermal conduction and quasi-static hypotheses. It is also observed that the interactions of the thermal conductivity of the crack gap, the non-Fourier effect and the inertia effects would make the dynamic thermal shock problem more complex. The magnitude of the thermal barrier, non-Fourier and inertia effects is estimated for some practical cases.

A aplicação de uma sequência de pulsos com tempo de repetição muito menor que os tempos de relaxação Tp << T2; T1, faz com que a magnetização atinja um estado estacionário descrito por H.Y. Carr como Estado Estacionário em Precessão Livre, Steady State Free Precession (SSFP). Nessa condição, o sinal é composto pela complexa sobreposição das componentes FID e eco. Sequências tipo SSFP são utilizadas na aquisição rápida de sinais, resultando em uma boa razão sinal ruído (s/r) em curto intervalo de tempo, porém introduzem fortes anomalias de fase e amplitude devido a complexa interação das componentes que formam o estado estacionário. Neste trabalho, desenvolvemos sequências de pulsos tipo SSFP para RMN em alta e baixa resolução com alternância e incremento de fase. Em alta resolução desenvolvemos as sequências SSFPdx e SSFPdxdt com incremento de fase linear e quadrático respectivamente. Os resultados mostram que espectros de núcleos com baixa sensibilidade podem ser obtidos com mesma razão s/r em menor tempo experimental e as sequências desenvolvidas removem as anomalias espectrais. Em baixa resolução, os resultados mostram que a introdução de alternâncias de fase na Continuous Wave Free Precession (CWFP) possibilita a remoção da dependência da sequência com o offset de frequência e com o tempo entre pulsos. Além disso, mostramos que a sequência CP-CWFPx-x com ângulo de refocalização pequeno (5&deg; a 10&deg;) possibilita a estimativa rápida do tempos de relaxação longitudinal. Apresentamos também resultados dos estudos e desenvolvidos no estágio de pesquisa no exterior, onde as sequências de pulsos no estado estacionário &ndash; DECPMG e Split 180&deg; &ndash; foram estudas numericamente e implementadas nos sistemas magnéticos compactos: mini-Halbach e MOUSE-NMR. Por fim, são apresentados resultados com os métodos de processamento de dados Krylov Basis Diagonalization Method (KBDM) e a Transformada Inversa de Laplace aplicados na análise de sinais SSFP. Resultados mostram que KBDM é uma ferramenta útil no processamento de dados em alta e baixa resolução, tanto na obtenção de espectros como na determinação da distribuição dos tempos de relaxação. / The application of a pulse sequence with repetition time much smaller than the relaxation times, Tp << T2; T1, causes the magnetization to reach a steady state, described by H. Y. Carr as a Steady State Free Precession (SSFP). In this condition, the signal is composed of the complex overlapping of the FID and eco components. SSFP type sequences are used in fast acquisition of NMR signals, resulting in a good signal to noise ratio (s/r) in a short time interval, however, they introduce phase and amplitude anomalies due to the complex interaction between the components of the steady state. In this work, we develop SSFP type pulse sequences for NMR in high and low resolution, with alternation and increment of phase. In high resolution, we develop SSFPdx and SSFPdxdt sequences, with linear and quadratic phase increment respectively. Results show that the low sensitivity nuclei spectra can be obtained with the same s/r ratio in smaller experimental time, about an order of magnitude, and the developed sequences can remove the spectral anomalies. In low resolution, the results show that the introduction of a phase alternation in the Continuous Wave Free Precession (CWFP) allows the elimination of the dependence of the sequence with the offset frequency and the time between pulses. Besides, we show that the CP-CWFPx-x sequence with a small refocalization angle (5° to 10°) allows the fast estimative of the longitudinal relaxation time in a single experiment. The results of the studies conducted during an international research internship are also presented. Steady state pulse sequences &ndash; DECPMG and Split 180° &ndash; were studied and implemented in compact magnetic systems: mini-Halbach and MOUSE-NMR. Finally, the results of the application of the Krylov Basis Diagonalization Method (KBDM) and the Inverse Laplace Transform for the analysis of SSFP signals are presented. The results show that KBDM is a useful tool in data processing for low and high resolution, both for obtaining spectra and determining the relaxation times distribution.

Die vorliegende Arbeit beschäftigt sich mit der Höhe (-log Determinante) kompakter nicht-singulärer heisenbergartiger Nilmannigfaltigkeiten. Heisenbergartige Nilmannigfaltigkeiten sind Verallgemeinerungen von Heisenbergmannigfaltigkeiten, d.h., kompakter Quotienten der Heisenberg-Gruppe, ausgestattet mit einer linksinvarianten Metrik. Zunächst werden explizite Formeln für die spektrale Zeta-Funktion und die Höhe bewiesen. Mithilfe dieser Formeln werden im Weiteren mehrere Resultate zur Existenz unterer Schranken/Minima der Höhe auf verschiedenen Moduli bewiesen. Zum Beispiel ist die Höhe stets von unten beschränkt, wenn man nur Metriken vom Heisenberg-Typ und mit Volumen 1 auf einer gegebenen Nilmannigfaltigkeit betrachtet. Im Gegensatz dazu hängt die Existenz unterer Schranken für die Höhe auf dem Modulraum der heisenbergartigen Metriken mit Volumen 1 von der Dimension Modulo 4 der zugrundeliegenden Mannigfaltigkeit ab. Im letzten Abschnitt werden konkrete Minima der Höhe behandelt. Wir zeigen, dass gewisse 3-, 5-, 9- und 25-dimensionale Nilmannigfaltigkeiten vom Heisenberg-Typ lokale Minima sind. Diese stehen in Zusammenhang mit den Minima der Höhe flacher Tori in der jeweiligen Dimension minus 1. Zum Abschluss werden diejenigen linksinvarianten Metriken charakterisiert, an denen die Höhe ein globales Minimum auf einer gegebenen dreidimensionalen Nilmannigfaltigkeit annimmt, indem sie zur Höhe flacher 2-dimensionaler Tori in Bezug gesetzt werden. / This thesis deals with the height (-log determinant) of compact nonsingular Heisenberg-like nilmanifolds. Heisenberg-like nilmanifolds are generalisations of Heisenberg manifolds, i.e., compact quotients of the Heisenberg group endowed with a left invariant metric. First, an explicit formula for the spectral zeta-function and the height is proved. By means of these formulas, several results concerning the existence of lower bounds/minima for the height on different moduli are proved. For example, while the height is always bounded from below when one considers only volume normalised Heisenberg-type metrics on a fixed nilmanifold, the existence of lower bounds for the height on the moduli space of volume normalised Heisenberg-like metrics depends on the dimension modulo 4 of the underlying nilmanifold. In the last part, we consider concrete minima of the height on Heisenberg manifolds. We show that certain 3-, 5-, 9- and 25-dimensional Heisenberg-type nilmanifolds are (local) minima for the height. These nilmanifolds are related to the minima of the height of flat tori in dimensions one less. Finally, the left invariant metrics at which the height attains a global minimum on any three-dimensional nilmanifold are characterised by relating them to the height of flat 2-dimensional tori.

Abstract The present thesis exploits advanced liquid and gas NMR methods for the characterization of various interesting materials. The methods used to study the structural properties of thermally modified wood, ionic liquids, cements, shales, and porous organic cages include MRI, NMR cryoporometry, Laplace NMR, multidimensional Laplace NMR, as well as ¹²⁹Xe and ¹⁹F NMR. The commonality factor in all the studies is the usage of either inherent or introduced liquid or gas molecules to probe the topical materials. The MRI method was utilized to visualize the water absorption phenomena in the thermally modified pine wood. High-resolution images made it possible to observe the spatial distribution of free water and the changes in the rate of absorption of water in wood samples modified at different temperatures. The images also helped to resolve the individual resin channels. T₂ maps enabled us to observe the changes in the relaxation values of free water in thermally modified wood as compared to their unmodified reference wood samples. The multidimensional Laplace NMR methods were exploited to study the structural and dynamical properties of a novel halogen-free, boron-based ionic liquid (hf-BIL). NMR self-diffusion (D) experiments showed the presence of two coexisting dynamic phases in hf-BIL. Multidimensional D − T₂ correlation experiments made it possible to determine the T₂ relaxation times of the slow and fast diffusing phases. T₂ − T₂ relaxation exchange measurements allowed quantifying the exchange rates of anions and cations between the phases. Moreover, the theoretical modeling of the experimental data revealed that the slow diffusing phase was composed of anion-cation aggregates, while the fast diffusing phase was comprised of free anions and cations. ¹²⁹Xe NMR analysis of the xenon adsorbed in the cements and shales helped us to determine their porous structures. The method exploits the high sensitivity of the chemical shift of ¹²⁹Xe to its local environment. The chemical shift value of ¹²⁹Xe enabled us to estimate the size of the mesopores in the cement samples. The exchange spectroscopy (EXSY) measurements were used to determine the exchange rates between the free gas and mesopores of the cement samples. ¹²⁹Xe NMR spectra of the shale samples provided information about pore sizes and paramagnetic compounds. ¹H NMR cryoporometry measurements of the shale samples immersed in acetonitrile made it possible to analyze the pore size distribution ranging from 10 to over 100 nm. Moreover, T₂ − T₂ exchange measurements helped us to quantify the exchange rates of acetonitrile in the shale samples. Xenon and SF₆ were used as internal reporters to gain versatile information on adsorption phenomena in the cage and window cavities of the crystalline porous organic cages. ¹²⁹Xe NMR analysis of the adsorbed xenon helped us to determine the diffusion coefficients and activation energy of diffusion as well as thermodynamic parameters. With the help of T₂ relaxation time values, it was possible to estimate the exchange rates between cage and window cavities. Chemical exchange saturation transfer (CEST) experiments resolved a window cavity site, which arises from crystal defects in porous organic cages. In addition, ¹⁹F NMR analysis made it possible to estimate the relaxation rates and diffusion coefficients of SF₆ gas in porous organic cages. Modelling of the T₁, T₂ and diffusion data confirmed that the cage to window exchange is the completely dominating mechanism for ¹²⁹Xe T₂ relaxation. T₁ relaxation is dominated by diffusion modulated dipole-dipole relaxation (DDinter) and chemical shift anisotropy (CSA) relaxation due to local cavity mobility. Whereas, in case of SF₆ T₂ data, the dominating mechanism is diffusion modulated dipole-dipole relaxation and for T₁ the local tumbling of SF₆ in cage cavity is the key dynamics behind the dipole-dipole and CSA mechanisms. / Original papers The original publications are not included in the electronic version of the dissertation. Javed, M. A., Kekkonen, P. M., Ahola, S., &amp; Telkki, V.-V. (2015). Magnetic resonance imaging study of water absorption in thermally modified pine wood. Holzforschung, 69(7), 899–907. https://doi.org/10.1515/hf-2014-0183 Javed, M. A., Ahola, S., Håkansson, P., Mankinen, O., Aslam, M. K., Filippov, A., … Telkki, V.-V. (2017). Structure and dynamics elucidation of ionic liquids using multidimensional Laplace NMR. Chem. Commun., 53(80), 11056–11059. https://doi.org/10.1039/c7cc05493a http://jultika.oulu.fi/Record/nbnfi-fe2017102750335 Javed, M. A., Komulainen, S., Daigle, H., Zhang, B., Vaara, J., Zhou, B., &amp; Telkki, V.-V. (2019). Determination of pore structures and dynamics of fluids in hydrated cements and natural shales by various ¹H and ¹²⁹Xe NMR methods. Microporous and Mesoporous Materials, 281, 66–74. https://doi.org/10.1016/j.micromeso.2019.02.034 http://jultika.oulu.fi/Record/nbnfi-fe2019041712678 Komulainen, S., Roukala, J., Zhivonitko, V. V., Javed, M. A., Chen, L., Holden, D., … Telkki, V.-V. (2017). Inside information on xenon adsorption in porous organic cages by NMR. Chemical Science, 8(8), 5721–5727. https://doi.org/10.1039/C7SC01990D http://jultika.oulu.fi/Record/nbnfi-fe201709288804 Håkansson, P., Javed, M. A., Komulainen, S., Chen, L., Holden, D., Hasell, T., … Telkki, V.-V. (2019). NMR relaxation and modelling study of the dynamics of SF₆ and Xe in porous organic cages. Manuscript.

Esta tese visa propor extensões dos modelos de regressão quantílica bayesianos, considerando dados de proporção com inflação de zeros, e também dados censurados no zero. Inicialmente, é sugerida uma análise de observações influentes, a partir da representação por mistura localização-escala da distribuição Laplace assimétrica, em que as distribuições a posteriori das variáveis latentes são comparadas com o intuito de identificar possíveis observações aberrantes. Em seguida, é proposto um modelo de duas partes para analisar dados de proporção com inflação de zeros ou uns, estudando os quantis condicionais e a probabilidade da variável resposta ser igual a zero. Além disso, são propostos modelos de regressão quantílica bayesiana para dados contínuos com um componente discreto no zero, em que parte dessas observações é suposta censurada. Esses modelos podem ser considerados mais completos na análise desse tipo de dados, uma vez que a probabilidade de censura é verificada para cada quantil de interesse. E por último, é considerada uma aplicação desses modelos com correlação espacial, para estudar os dados da eleição presidencial no Brasil em 2014. Nesse caso, os modelos de regressão quantílica são capazes de incorporar essa informação espacial a partir do processo Laplace assimétrico. Para todos os modelos propostos foi desenvolvido um pacote do software R, que está exemplificado no apêndice. / This thesis aims to propose extensions of Bayesian quantile regression models, considering proportion data with zero inflation, and also censored data at zero. Initially, it is suggested an analysis of influential observations, based on the location-scale mixture representation of the asymmetric Laplace distribution, where the posterior distribution of the latent variables are compared with the goal of identifying possible outlying observations. Next, a two-part model is proposed to analyze proportion data with zero or one inflation, studying the conditional quantile and the probability of the response variable being equal to zero. Following, Bayesian quantile regression models are proposed for continuous data with a discrete component at zero, where part of these observations are assumed censored. These models may be considered more complete in the analysis of this type of data, as the censoring probability varies with the quantiles of interest. For last, it is considered an application of these models with spacial correlation, in order to study the data about the last presidential election in Brazil in 2014. In this example, the quantile regression models are able to incorporate spatial dependence with the asymmetric Laplace process. For all the proposed models it was developed a R package, which is exemplified in the appendix.

O objetivo deste trabalho é estudar aproximação na esfera por uma soma com pesos de harmônicos esféricos. Apresentamos condições necessárias e suficientes sobre os pesos para garantir a convergência, tanto no caso contínuo quanto no caso Lp. Analisamos a ordem de convergência dos processos aproximatórios usando um módulo de suavidade esférico relacionado à derivada forte de Laplace-Beltrami. Incluímos provas para vários resultados sobre a derivada forte de Laplace-Beltrami, já que não conseguimos encontrá-las na literatura / The subject of this work is to study approximation on the sphere by weighted sums of spherical harmonics. We present necessary and sufficient conditions on the weights for convergence in both, the continuous and the Lp cases. We analyse the convergence rates of the approximation processes using a modulus of smoothness related to the strong Laplace- Beltrami derivative. We include proofs for several results related to such a derivative, since we were unable to find them in the literature

Dans cette thèse, on propose des méthodes analytiques ou numériques pour l'estimation de la VaR ou l'Expected Shortfall des portefeuilles linéaires, quadratiques, lorsque le vecteur des facteurs de risques suit un mélange convexe de distributions elliptiques. Aussi, on introduit pour la prémière fois la notion de "portefeuille quadratique" d'actifs de bases (ie. actions).