Given a hyperbolic surface, the set of all closed geodesics whose length is minimal form a graph on the surface, in fact a so called fat graph, which we call the systolic graph. The central question that we study in this thesis is: which fat graphs are systolic graphs for some surface -we call such graphs admissible. This is motivated in part by the observation that we can naturally decompose the moduli space of hyperbolic surfaces based on the associated systolic graphs. A systolic graph has a metric on it, so that all cycles on the graph that correspond to geodesics are of the same length and all other cycles have length greater than these. This can be formulated as a simple condition in terms of equations and inequations for sums of lengths of edges. We call this combinatorial admissibility. Our first main result is that admissibility is equivalent to combinatorial admissibility. This is proved using properties of negative curvature, specifically that polygonal curves with long enough sides, in terms of a lower bound on the angles, are close to geodesics. Using the above result, it is easy to see that a subgraph of an admissible graph is admissible. Hence it suffices to characterize minimal non-admissible fat graphs. Another major result of this thesis is that there are infinitely many minimal non-admissible fat graphs (in contrast, for instance, to the classical result that there are only two minimal non-planar graphs).

Any Riemannian manifold possesses a minimal solution of the heat equation for the Dirichlet Laplacian, called the heat kernel. During the last decades many authors investigated geometric properties of the manifold such that its heat kernel fulfills a so-called Gaussian upper bound. Especially compact and non-compact manifolds with lower bounded Ricci curvature have been examined and provide such Gaussian estimates. In the compact case it ended even with integral Ricci curvature assumptions. The important techniques to obtain Gaussian bounds are the symmetrization procedure for compact manifolds and relative Faber-Krahn estimates or gradient estimates for the heat equation, where the first two base on isoperimetric properties of certain sets. In this thesis, we generalize the existing results to the following. Locally uniform integral bounds on the negative part of Ricci curvature lead to Gaussian upper bounds for the heat kernel, no matter whether the manifold is compact or not. Therefore, we show local isoperimetric inequalities under this condition and use relative Faber-Krahn estimates to derive explicit Gaussian upper bounds. If the manifold is compact, we can even generalize the integral curvature condition to the case that the negative part of Ricci curvature is in the so-called Kato class. We even obtain uniform Gaussian upper bounds using gradient estimate techniques. Apart from the geometric generalizations for obtaining Gaussian upper bounds we use those estimates to generalize Bochner’s theorem. More precisely, the estimates for the heat kernel obtained above lead to ultracontractive estimates for the heat semigroup and the semigroup generated by the Hodge Laplacian. In turn, we can formulate rigidity results for the triviality of the first cohomology group if the amount of curvature going below a certain positive threshold is small in a suitable sense. If we can only assume such smallness of the negative part of the Ricci curvature, we can bound the Betti number by explicit terms depending on the generalized curvature assumptions in a uniform manner, generalizing certain existing results from the cited literature. / Jede Riemannsche Mannigfaltigkeit besitzt eine minimale Lösung für die Wärmeleitungsgleichung des zur Mannigfaltigkeit gehörigen Dirichlet-Laplaceoperators, den Wärmeleitungskern. Während der letzten Jahrzehnte fanden viele Autoren geometrische Eigenschaften der Mannigfaltigkeiten unter welchen der Wärmeleitungskern eine sogenannte Gaußsche obere Abschätzung besitzt. Insbesondere bestizen sowohl kompakte als auch nichtkompakte Mannigfaltigkeiten mit nach unten beschränkter Ricci-Krümmung solche Gaußschen Abschätzungen. Im kompakten Fall reichten bisher sogar Integralbedingungen an die Ricci-Krümmung aus. Die wichtigen Techniken, um Gaußsche Abschätzungen zu erhalten, sind die Symmetrisierung für kompakte Mannigfaltigkeiten und relative Faber-Krahn- und Gradientenabschätzungen für die Wärmeleitungsgleichung, wobei die ersten beiden auf isoperimetrischen Eigenschaften gewisser Mengen beruhen. In dieser Arbeit verallgemeinern wir die bestehenden Resultate im folgenden Sinne. Lokal gleichmäßig beschränkte Integralschranken an den Negativteil der Ricci-Krümmung ergeben Gaußsche obere Abschätzungen sowohl im kompakten als auch nichtkompakten Fall. Dafür zeigen wir lokale isoperimetrische Ungleichungen unter dieser Voraussetzung und nutzen die relativen Faber-Krahn-Abschätzungen für eine explizite Gaußsche Schranke. Für kompakte Mannigfaltigkeiten können wir sogar die Integralschranken an den Negativteil der Ricci-Krümmung durch die sogenannte Kato-Bedingung ersetzen. In diesem Fall erhalten wir gleichmäßige Gaußsche Abschätzungen mit einer Gradientenabschätzung. Neben den geometrischen Verallgemeinerungen für Gaußsche Schranken nutzen wir unsere Ergebnisse, um Bochners Theorem zu verallgemeinern. Wärmeleitungskernabschätzungen ergeben ultrakontraktive Schranken für die Wärmeleitungshalbgruppe und die Halbgruppe, die durch den Hodge-Operator erzeugt wird. Damit können wir Starrheitseigenschaften für die erste Kohomologiegruppe zeigen, wenn der Teil der Ricci-Krümmung, welcher unter einem positiven Level liegt, in einem bestimmten Sinne klein genug ist. Wenn der Negativteil der Ricci-Krümmung nicht zu groß ist, können wir die erste Betti-Zahl noch immer explizit uniform abschätzen.

Nous étudions expérimentalement deux mémoires quantiques pour la lumière utilisant la transparence électromagnétiquement induite (EIT) dans des nuages froids de césium.Nous expliquons la pertinence des mémoires quantiques pour le développement de réseaux quantiques à longue distance, et décrivons la théorie de l’EIT en soulignant les paramètres essentiels pour l’implémentation de mémoires quantiques.Notre premier cas d’étude est un piège magnéto-optique en espace libre. Notre principal résultat est la démonstration du caractère multimode de ce système pour le stockage quantique de la lumière. Pour cela, nous utilisons des faisceaux de Laguerre-Gauss (LG), porteurs de moment angulaire orbital (OAM). Dans une première étape, nous avons montré que l’état de moment orbital d’impulsions lumineuses en régime de photons uniques est préservé lors du stockage dans la mémoire. Ensuite, nous avons implémenté un bit quantique comme une superposition de modes LG ayant des hélicités opposées. Nous avons développé un système original pour mesurer ces bits quantiques qui nous a permis de caractériser l’action de la mémoire. Nous avons ainsi pu montrer que le stockage quantique de ces bits quantiques.Le second système, également un nuage d’atomes froids, a la particularité que les atomes sont piégés optiquement autour d’un nano-guide d’onde. Ce design innovant permet une plus grande interaction entre lumière et matière, et facilite l’interfaçage des photons dans et hors de la mémoire. Nous décrivons la construction de ce dispositif et les premiers pas vers son utilisation en tant que mémoire quantique. / We present an experimental study of two optical quantum memory systems based on electromagnetically induced transparency (EIT) in cold cesium atoms.We explain the relevance of quantum memories for the development of large-scale quantum networks, we give a comprehensive theory of the EIT phenomenon and underline the role of relevant parameters regarding the implementation of quantum memories.The first system under study is prepared in a free-space magneto-optical trap. The main result of this thesis is the demonstration of the spatial multimode capability of this system at the quantum level. For this, we used Laguerre-Gaussian (LG) light beams, i.e. beams possessing a non-zero value of orbital angular momentum (OAM). In a first step, we showed that the orbital angular momentum of stored light pulses is preserved by the memory, deep in the single photon regime. In a second step, we encoded information in the orbital angular momentum state of a weak light pulse and defined a qubit using two LG beams of opposite helicities. We developed an original setup for the measurement of this OAM qubit and used it to characterize the action of the memory during the storage of such a light pulse. Our results show that the memory performs the quantum storage of such a qubit.The second system under study, also a cloud of cold atoms, has the specificity that the atoms are trapped optically in the vicinity of a nano-waveguide. This innovative design ensures a higher light-matter interaction and facilitates the interfacing of photons into and out of the memory. We describe the building of this setup and the first steps towards quantum memory implementations.

This thesis deals with solutions of nonlinear regression problems using R programming language. The introductory theoretical part is devoted to familiarization with the principles of solving nonlinear regression models and of their applications in the program R. In both, theoretical and practical part, the most famous and used differentiator algorithms are presented, particularly the Gauss-Newton's and of the steepest descent method, for estimating the parameters of nonlinear regression. Further, in the practical part, there are some demo solutions of particular tasks using nonlinear regression methods. Overall, a large number of graphs processed by the author is used in this thesis for better comprehension.

K nahrazení člověka při úloze testování dotykových embedded zařízení je zapotřebí vyvinout komplexní automatizovaný robotický systém. Jedním ze zásadních úkolů je tento systém automaticky zkalibrovat. V této práci jsem se zabýval možnými způsoby automatické kalibrace robotického ramene v prostoru ve vztahu k dotykovému zařízení pomocí jedné či více kamer. Následně jsem představil řešení založené na estimaci polohy jedné kamery pomocí iterativních metod jako např. Gauss-Newton nebo Levenberg-Marquardt. Na konci jsem zhodnotil dosaženou přesnost a navrhnul postup pro její zvýšení.

The K-distribution method is a fast approximative method used for calculating thermal infrared radiative transfer in the atmosphere, as opposed to the traditional Line-by-line method, which is precise, but very time-costly. Here we consider the atmosphere to consist of homogeneous and plane-parallel layers in local thermal equilibrium. This lets us use efficient upwards recursion for calculating the thermal infrared radiative transfer and ultimately the outgoing irradiance at the top of the atmosphere. Our specific implementation of the K-distribution method revolves around changing the integration space from the wavenumber domain to the g domain by employing Gauss-Legendre quadrature in two steps. The method is implemented in MATLAB and is shown to be several thousand times faster than the traditional Line-by-line method, with the relative error being only 3 % for the outgoing irradiance at the top of the atmosphere.

The essence of face recognition within the image is generally computer vision, which provides methods and algorithms for the implementation. Some of them are described just in this work. Whole process is split in to three main phases. These are detection, aligning of detected faces and finally its recognition. Algorithms which are used to applied in given issue and which are still in progress from todays view are mentioned in every phase. Implementation is build up on three main algorithms, AdaBoost to obtain the classifier for detection, method of aligning face by principal features and method of Eigenfaces for recognizing. There are theoretically described except already mentioned algorithms neural networks for detection, ASM - Active Shape Models algorithm for aligning and AAM - Active Appearance Model for recognition. In the end there are tables of data retrieved by implemented system, which evaluated the main implementation.

In dieser Arbeit befassen wir uns mit einer Reihe von entscheidenden Schritten, um die unpolarisierten Helizitäts- und Trasversitäts-Parton-Verteilungsfunktionen der Nukleonen im Rahmen der Gitter-QCD zu bewerten. Diskretisierungsartefakte werden unter Verwendung eines N_f=2+1+1 Eichensembles von Fermionen mit verdrillter Wilson-Masse untersucht, die bei einer Pionenmasse von ungefähr M=37 MeV simuliert werden. Die unpolarisierten und Helizitäts Partonverteilungsfunktionen weisen eine nicht vernachlässigbare Abhängigkeit vom Gitterabstand auf, und die Kontinuumsextrapolation ergibt eine bessere Übereinstimmung mit Phänomenologie. Die direkte Berechnung der Fourier-Transformation mit diskreten Gitterdaten kann Artefakte verursachen. Daher arbeiten wir mit einer neuen datengesteuerten Methode, die auf Gauß-Prozess-Regression basiert, die sogenannte Bayes-Gauß-Fourier-Transformation, um die Einschränkungen der diskreten Fourier-Transformation zu überwinden. Wir sind der Meinung, dass dieser datengesteuerte Ansatz die durch die Diskretisierung der Fourier-Transformation eingeführten Artefakte drastisch reduzieren kann, jedoch ist der endgültige Effekt auf die Lichtkegel-PDFs gering. Darüber hinaus präsentieren wir die Ergebnisse der ersten ab initio Berechnung der individuellen up, down und strange unpolarisierten, Helizitäts- und Transversitäts-Partonverteilungsfunktionen für das Proton. Die Analyse wird an einem durch N_f=2+1+1 verdrillten Kleeblatt-verbesserten Fermionen-Ensemble durchgeführt, das bei einer Pionenmasse von 260 MeV simuliert wird. Wir verwenden den hierarchischen Sondierungsalgorithmus, um die unzusammenhängenden Quarkschleifen auszuwerten. Dadurch erhalten wir Ergebnisse ungleich Null für den unbegundenen isoskalaren Beitrag und die strange Quark-Matrixelemente. / In this work, we address a number of crucial steps in order to evaluate the nucleon unpolarized helicity and trasversity parton distribution functions within the framework of lattice QCD. Discretization artifacts are investigated using an N_f=2+1+1 gauge ensemble of Wilson twisted mass fermions simulated at a pion mass of approximately M=370 MeV. The unpolarized and helicity parton distribution functions show a non-negligible dependence on the lattice spacing, with the continuum extrapolation producing a better agreement with phenomenology. The direct computation of the Fourier transform using discrete lattice data may introduce artifacts and we, therefore, use a new data-driven method based on Gaussian process regression, the so-called Bayes-Gauss Fourier transform to overcome the limitations of the discrete Fourier transform. We find that this data-driven approach can drastically reduce the artifacts introduced by the discretization of the Fourier transform, however, the final effect on the light-cone PDFs is small. Furthermore, we present results of the first ab initio calculation of the individual up, down, and strange unpolarized, helicity, and transversity parton distribution functions for the proton. The analysis is performed on an N_f=2+1+1 twisted mass clover-improved fermion ensemble simulated at a pion mass of 260 MeV. We employ the hierarchical probing algorithm to evaluate the disconnected quark loops, allowing us to obtain non-zero results for the disconnected isoscalar contribution and the strange quark matrix elements.

<p dir="ltr">This thesis presents advances in numerical methods for partial differential equations (PDEs) and optimization problems, with a focus on improving efficiency, stability, and accuracy across various applications. We begin by addressing 3D Poisson-type equations, developing a GPU-accelerated spectral-element method that utilizes the tensor product structure to achieve extremely fast performance. This approach enables solving problems with over one billion degrees of freedom in less than one second on modern GPUs, with applications to Schrödinger and Cahn<i>–</i>Hilliard equations demonstrated. Next, we focus on parabolic PDEs, specifically the Cahn<i>–</i>Hilliard equation with dynamical boundary conditions. We propose an efficient energy-stable numerical scheme using a unified framework to handle both Allen<i>–</i>Cahn and Cahn<i>–</i>Hilliard type boundary conditions. The scheme employs a scalar auxiliary variable (SAV) approach to achieve linear, second-order, and unconditionally energy stable properties. Shifting to a machine learning perspective for PDEs, we introduce an unsupervised learning-based numerical method for solving elliptic PDEs. This approach uses deep neural networks to approximate PDE solutions and employs least-squares functionals as loss functions, with a focus on first-order system least-squares formulations. In the realm of optimization, we present an efficient and robust SAV based algorithm for discrete gradient systems. This method modifies the standard SAV approach and incorporates relaxation and adaptive strategies to achieve fast convergence for minimization problems while maintaining unconditional energy stability. Finally, we address optimization in the context of machine learning by developing a structure-guided Gauss<i>–</i>Newton method for shallow ReLU neural network optimization. This approach exploits both the least-squares and neural network structures to create an efficient iterative solver, demonstrating superior performance on challenging function approximation problems. Throughout the thesis, we provide theoretical analysis, efficient numerical implementations, and extensive computational experiments to validate the proposed methods. </p>

Cette thèse construit un modèle théorique qui renouvelle l’approche traditionnelle de l’équilibre du marché. En introduisant dans le paradigme néo-classique le principe de préférence pour la quantité, il génère de façon optimale des stocks dans un marché concurrentiel. Les résultats sont très importants, car ils expliquent à la fois l’émergence des invendus et l’existence de cycles économiques. En outre, il étudie le comportement optimal du monopole dont la puissance de marché dépend non seulement de la quantité de biens étalés, mais aussi de celle de biens achetés. Contrairement à l’hypothèse traditionnelle selon laquelle le monopoleur choisit le prix ou la quantité qui maximise son profit, il attire, via un indice de Lerner généralisé la demande à la fois par le prix et la quantité de biens exposés. Quelle que soit la structure du marché, le phénomène d’accumulation des stocks de biens apparaît dans l’économie. De plus, il a l’avantage d’expliquer explicitement les achats impulsifs non encore traités par la théorie économique. Pour vérifier la robustesse des résultats du modèle théorique, ils sont testés sur des données américaines. En raison de leur non-linéarité, la méthode de Gauss-Newton est appropriée pour analyser l’impact de la préférence pour la quantité sur la production et l’accumulation de biens, et par conséquent sur les prévisions de PIB. Enfin, cette thèse construit un modèle à générations imbriquées à deux pays qui étend l’équilibre dynamique à un gamma-équilibre dynamique sans friction. Sur la base de la contrainte de détention préalable d’encaisse, il ressort les conditions de sur-accumulation du capital et les conséquences de la mobilité du capital sur le bien-être dans un contexte d’accumulation du stock d’invendus / This thesis constructs a theoretical model that renews the traditional approach of the market equilibrium. By introducing into the neoclassical paradigm the principle of preference for quantity, it optimally generates inventories within a competitive market. The results are very important since they explain both the emergence of unsold goods and the existence of economic cycles. In addition, it studies the optimal behavior of a monopolist whose the market power depends not only on the quantity of displayed goods but also that of goods that the main consumer is willing to buy. Contrary to the traditional assumption that the monopolist chooses price or quantity that maximizes its profit, through a generalized Lerner index (GLI) it attracts customers’ demand by both the price and the quantity of displayed goods. Whatever the market structure, the phenomenon of inventory accumulation appears in the economy. Furthermore, it has the advantage of explicitly explaining impulse purchases untreated by economics. To check the robustness of the results,the theoretical model is fitted to U.S. data. Due to its nonlinearity, the Gauss-Newtonmethod is appropriate to highlight the impact of consumers’ preference for quantity on production and accumulation of goods and consequently GDP forecast. Finally, this thesis builds a two-country overlapping generations (OLG) model which extends the dynamic OLG equilibrium to a frictionless dynamic OLG gamma-equilibrium. Based on the cash-inadvance constraint, it highlights the conditions of over-accumulation of capital and welfare implications of capital mobility in a context of accumulation of stock of unsold goods.