This thesis is devoted to the study of the applications of the error theory using Dirichlet forms. Our work is split into three parts. The first one deals with the models described by stochastic differential equations. After a short technical chapter, an innovative model for order books is proposed. We assume that the bid-ask spread is not an imperfection, but an intrinsic property of exchange markets instead. The uncertainty is carried by the Brownian motion guiding the asset. We find that spread evolutions can be evaluated using closed formulae and we estimate the impact of the underlying uncertainty on the related contingent claims. Afterwards, we deal with the PBS model, a new model to price European options. The seminal idea is to distinguish the market volatility with respect to the parameter used by traders for hedging. We assume the former constant, while the latter volatility being an erroneous subjective estimation of the former. We prove that this model anticipates a bid-ask spread and a smiled implied volatility curve. Major properties of this model are the existence of closed formulae for prices, the impact of the underlying drift and an efficient calibration strategy. The second part deals with the models described by partial differential equations. Linear and non-linear PDEs are examined separately. In the first case, we show some interesting relations between the error and wavelets theories. When non-linear PDEs are concerned, we study the sensitivity of the solution using error theory. Except when exact solution exists, two possible approaches are detailed: first, we analyze the sensitivity obtained by taking "derivatives" of the discrete governing equations. Then, we study the PDEs solved by the sensitivity of the theoretical solutions. In both cases, we show that sharp and bias solve linear PDE depending on the solution of the former PDE itself and we suggest algorithms to evaluate numerically the sensitivities. Finally, the third part is devoted to stochastic partial differential equations. Our analysis is split into two chapters. First, we study the transmission of an uncertainty, present on starting conditions, on the solution of SPDE. Then, we analyze the impact of a perturbation of the functional terms of SPDE and the coefficient of the related Green function. In both cases, we show that the sharp and bias verify linear SPDE depending on the solution of the former SPDE itself

This study seeks to develop a method that generalises the use of Adams-Bashforth to solve or treat partial differential equations with local and non-local differentiation by deriving a two-step Adams-Bashforth numerical scheme in Laplace space. The resulting solution is then transformed back into the real space by using the inverse Laplace transform. This is a powerful numerical algorithm for fractional order derivative. The error analysis for the method is studied and presented. The numerical simulations of the method as applied to the four-dimensional model, Caputo-Lu-Chen model and the wave equation are presented. In the analysis, the bifurcation dynamics are discussed and the periodic doubling processes that eventually caused chaotic behaviour (butterfly attractor) are shown. The related graphical simulations that show the existence of fractal structure that is characterised by chaos and usually called strange attractors are provided. For the Caputo-Lu-Chen model, graphical simulations have been realised in both integer and fractional derivative orders. / Mathematical Sciences / M. Sc. (Applied Mathematics)

This thesis is concerned with the mathematical analysis of electricity and carbon emission markets. We introduce a novel, versatile and tractable stochastic framework for the joint price formation of electricity spot prices and allowance certificates. In the proposed framework electricity and allowance prices are explained as functions of specific fundamental factors, such as the demand for electricity and the prices of the fuels used for its production. As a result, the proposed model very clearly captures the complex dependency of the modelled prices on the aforementioned fundamental factors. The allowance price is obtained as the solution to a coupled forward-backward stochastic differential equation. We provide a rigorous proof of the existence and uniqueness of a solution to this equation and analyse its behaviour using asymptotic techniques. The essence of the model for the electricity price is a carefully chosen and explicitly constructed function representing the supply curve in the electricity market. The model we propose accommodates most regulatory features that are commonly found in implementations of emissions trading systems and we analyse in detail the impact these features have on the prices of allowance certificates. Thereby we reveal a weakness in existing regulatory frameworks, which, in rare cases, can lead to allowance prices that do not conform with the conditions imposed by the regulator. We illustrate the applicability of our model to the pricing of derivative contracts, in particular clean spread options and numerically illustrate its ability to "see" relationships between the fundamental variables and the option contract, which are usually unobserved by other commonly used models in the literature. The results we obtain constitute flexible tools that help to efficiently evaluate the financial impact current or future implementations of emissions trading systems have on participants in these markets.

Wind turbine noise is a re-emerging issue in the wind industry. As the competition for sites with good wind potential on land is rising, offshore projects in coastal areas seem as a reasonable alternative to onshore. In this context offshore sound propagation is gaining more importance considering that sound will travel over longer distances on water, especially with regard to lower frequencies. Moreover different meteorological conditions that occur on sea may attenuate or enhance sound propagation on water. The prediction tools commonly used by developers are only partially taking these parameters into account. This will be investigated in this thesis.   Hence, different methods for predicting offshore wind turbine noise are going to be assessed. These methods can be divided in two approaches namely algebraic and Partial Differential Equation (PDE) based. The methods evaluated are the ISO 9613-2 standard for outdoor noise prediction, the Danish method and the Swedish method for wind turbines noise estimation over water.   For the PDE based approach, the Helmholtz Equation will be employed in order to examine different meteorological conditions and phenomena occurring over a flat reflecting surface. The experiments with the PDE include the simulation of meteorological conditions with different levels of refraction and changing ground impedance in order to take into account the effect of a shoreline. In addition a meteorological phenomenon called the low-level jet is investigated which is characterised by strong winds at relatively low altitude.   Noise prediction tools used by developers need to be able to consider these effects in order to allow for thorough planning of wind energy projects. Nonetheless, relatively more complex models such as the Helmholtz Equation require experienced users and significant computing time. Further research and development needs to be made in order to promote the wider use of noise prediction methods like the Helmholtz Equation in the wind industry.

This thesis investigates the possibility of approximating stationary solutions of stochastic differential equations using numerical methods. We consider a particular class of stochastic differential equations, which are known to generate random dynamical systems. The existence of stochastic stationary solution is proved using global attractor approach. Euler's numerical method, applied to the stochastic differential equation, is proved to generate a discrete random dynamical system. The existence of stationary solution is proved again using global attractor approach. At last we prove that the approximate stationary point converges in mean-square sense to the exact one as the time step of the numerical scheme diminishes.

We develop and evaluate a novel host-vector model of West Nile virus (WNV) transmission that incorporates multiple avian host species and host stage-structure (juvenile and adult stages), with both species-specific and stage-specific biting rates of vectors on hosts. We use this model to explore WNV transmission dynamics that occur between vectors and multiple structured host populations as a result of heterogeneous biting rates. Our analysis shows that increased exposure of juvenile hosts results in earlier, more intense WNV transmission when compared to the effects of differential host species exposure, regardless of other parameter values. We also find that, in addition to competence, increased juvenile exposure is an important mechanism for determining the effect of species diversity on the disease risk of a community.

A thesis submitted to the Faculty of Science, University of the Witwatersrand in fulfillment of the requirements for the degree of Doctor of Philosophy. Johannesburg, 3 April 2017. / The effect of suction and blowing at the base on the horizontal spreading under gravity of a two-dimensional thin fluid film and an axisymmetric liquid drop is in- vestigated. The velocity vn which describes the suction/injection of fluid at the base is not specified initially. The height of the thin film satisfies a nonlinear diffusion equation with vn as a source term. The Lie group method for the solution of partial differential equations is used to reduce the partial differential equations to ordinary differential equations and to construct group invariant solutions. For a group invari- ant solution to exist, vn must satisfy a first order linear partial differential equation. The two-dimensional spreading of a thin fluid film is first investigated. Two models for vn which give analytical solutions are analysed. In the first model vn is propor- tional to the height of the thin film at that point. The constant of proportionality is β (−∞ < β < ∞). The half-width always increases to infinity as time increases even for suction at the base. The range of β for the thin fluid film approximation to be valid is determined. For all values of suction and a small range of blowing the maximum height of the film tends to zero as time t → ∞. There is a value of β corresponding to blowing for which the maximum height remains constant with the blowing balancing the effect of gravity. For stronger blowing the maximum height tends to infinity algebraically, there is a value of β for which the maximum height tends to infinity exponentially and for stronger blowing, still in the range for which the thin film approximation is valid, the maximum height tends to infinity in a finite time. For blowing the location of a stagnation point on the centre line is determined by solving a cubic equation approximately by a singular perturbation method and then exactly using a trigonometric solution. A dividing streamline passes through the stagnation point which separates the flow into two regions, an upper region consisting of fluid descending due to gravity and a lower region consisting of fluid rising due to blowing. For sufficiently strong blowing the lower region fills the whole of the film. In the second model vn is proportional to the spatial gradient of the height with constant of proportionality β∗ (−∞ < β∗ < ∞). The maximum height always decreases to zero as time increases even for blowing. The range of β∗ for the thin fluid film approximation to be valid is determined. The half-width tends to infinity algebraically for all blowing and a small range of weak suction. There is a value of β∗ corresponding to suction for which the half-width remains constant with the suction balancing the spreading due to gravity. For stronger suction the half-width tends to zero as t → ∞. For even stronger suction there is a value of β∗ for which the half-width tends to zero exponentially and a range of β∗ for which it tends to zero in a finite time but these values lie outside the range for which the thin fluid film approximation is valid. For blowing there is a stagnation point on the centre line at the base. Two dividing streamlines passes through the stagnation point which separate fluid descending due to gravity from fluid rising due to blowing. An approximate analytical solution is derived for the two dividing streamlines. A similar analysis is performed for the axisymmetric spreading of a liquid drop and the results are compared with the two-dimensional spreading of a thin fluid film. Since the two models for vn are still quite general it can be expected that general results found will apply to other models. These include the existence of a divid- ing streamline separating descending and rising fluid for blowing, the existence of a strength of blowing which balances the effect of gravity so the maximum height remains constant and the existence of a strength of suction which balances spreading due to gravity so that the half-width/radius remains constant. / MT 2017

O estudo da evolução da concentração de elementos de uma reação química, conhecida como Cinética Química, é de extrema importância para a compreensão das complexas interações em sistemas biológicos. Uma maneira de descrever a cinética de uma reação química é utilizando um sistema de equações diferenciais ordinárias (EDOs). Uma vez que para resolver um sistema de equações diferenciais ordinárias pode ser uma tarefa difícil (ou mesmo inviável), métodos numéricos são utilizados para realizar simulações, ou seja, para obter concentrações aproximadas das espécies químicas envolvidas durante um determinado período de tempo. No entanto, quanto maior for o sistema simulado de EDOs, mais os métodos numéricos estão sujeitos a erros. Além disso, o aumento do tamanho do sistema muitas vezes resulta em simulações que são mais exigentes do ponto de vista computacional. Assim, o objetivo deste projeto de mestrado é o desenvolvimento de regras para simplificar os sistemas de equações diferenciais ordinárias que modelam a cinética de reações químicas e, portanto, a obtenção de um algoritmo para executar simulações numéricas de um modo mais rápido e menos propenso a erros. Mais do que diminuir o erro e o tempo de execução, esta simplificação possibilita o biólogo escolher a solução mais factível do ponto de vista de medida. Isso porque, a identificação dos sistemas (i.e., inferência dos parâmetros) requer que a concentração de todas as espécies químicas seja conhecida, ao menos em um certo intervalo de tempo. Contudo, em muitos casos, não é possível medir a concentração de todas as espécies químicas consideradas. Esta simplificação gera sistemas equivalentes ao original, mas que dispensa a utilização de certas concentrações de espécies químicas. Um sistema de equações diferenciais ordinárias pode ser simplificado considerando as relações de conservação de massa, que são equações algébricas. Além disso, no caso de reações enzimáticas, o sistema de equações diferenciais ordinárias pode ser simplificado pelo pressuposto de que a concentração do complexo enzima-substrato mantém-se constante, o que permite a utilização da equação de Michaelis-Menten. De todas as combinações possíveis das equações algébricas com as equações diferenciais, uma família de sistemas simplificados de EDOs foi construída, permitindo a escolha do sistema mais simples. Esta escolha segue um critério guloso que favorece a minimização do número de equações diferenciais e do número total de termos. As regras em desenvolvimento de simplificação dos sistemas de equações diferenciais ordinárias foram utilizados para projetar um algoritmo, que foi implementado usando a linguagem de programação Python. O algoritmo concebido foi testado utilizando instâncias artificiais. / The study of the evolution of the concentration of species in a chemical reaction, known as Chemical Kinetics, is of paramount importance for the understanding of complex interactions in biological systems. One way to describe the kinetics of a chemical reaction is using a system of ordinary differential equations (ODEs). Once to solve a system of ODEs can be a difficult (or even unfeasible) task, numerical methods are employed to carry out simulations, that is, to obtain approximated concentrations of the involved chemical species for a certain time frame. However, the larger is the simulated system of ODEs, the more numerical methods are subject to error. Moreover, the increase of the system size often results in simulations that are more demanding from the computational point of view. Thus, the objective is the development of rules to simplify systems of ODEs that models the kinetics of chemical reactions, hence obtaining an algorithm to execute numerical simulations in a faster way and less prone to error. More than decrease error and run time, this simplification allows the biologist to choose the most feasible solution from the point of view of measurement. This is because the identification of systems (i.e., inferring parameters) requires that the concentration of all chemical species is known, at least in a certain time interval. However, in many cases it is not possible to measure the concentration of all chemical species considered. This simplification creates systems equivalent to the original, but that does not require the use of certain concentrations of chemical species. A system of ODEs can be simplified considering the relations of mass conservation, which are algebraic equations. Furthermore, in the case of enzymatic reactions, the system of ODEs can be simplified under the assumption that the concentration of enzyme-substrate complex remains constant, which allows us to use the Michaelis-Menten equation. From all possible combinations of the algebraic equations with differential equations, a family of simplified systems of ODEs will be built, allowing the choice of a simplest system. This choice will follow a greedy criterion which favors the minimization of number of differential equations and the total number of terms. The rules under development to simplify systems of ODEs will be used to design an algorithm, which will be implemented using Python programming language. The designed algorithm will be tested using synthetic data.

Este trabalho apresenta o desenvolvimento de um algoritmo baseado na modelagem do sistema de transmissão por meio de equações diferenciais, formuladas através dos parâmetros resistência e indutância da linha de transmissão a ser protegida. Nesta abordagem não é necessário que a entrada do algoritmo seja puramente senoidal, admitindo-se a presença de harmônicos e componentes CC presentes na falta como parte da solução do problema. Utilizou-se o software ATP para a modelagem do sistema elétrico estudado e a obtenção do conjunto de dados para análise e testes, permitindo-se a representação detalhada da linha de transmissão por meio das características dos condutores e suas respectivas disposições geométricas nas torres de transmissão, além da modelagem das diversas manobras e defeitos que os afetam, buscando-se uma aproximação com uma situação real. Com relação ao uso direto das equações diferenciais para a tarefa de proteção das linhas, constatou-se que sua aplicação não produz uma estimativa aceitável para ser usada em relés digitais por possuírem convergência em tempos normalmente superiores a dois ciclos. Assim, foi feita uma filtragem das respostas do algoritmo, proporcionando-se um diagnóstico mais rápido das estimativas. Para isso foi usado um filtro de mediana de 5ª ordem para o cálculo da localização da falta. Para todos os tipos de falta testados, a estimativa da localização da falta com o uso do referido filtro mostrou-se altamente satisfatória para a finalidade de proteção, convergindo em menos de um ciclo e meio de pós-falta, após a filtragem das estimativas, imprimindo maior velocidade de resposta para os relés digitais. / The present work shows the development of an algorithm based on the modeling of the transmission system utilizing differential equations. The differential equation for the line is solved having its resistance and inductance as parameters. In this approach there is no need for the algorithm inputs to be pure sinusoidal, allowing the presence of harmonic and DC components in the line as a part of the solution to the problem. The software ATP was utilized for the modeling of the electric system under study as well as data collection for analysis and tests. This representation allowed a detailed representation of the transmission line through the characteristics of the conductors and its geometrical disposition in the transmission towers, as well as the simulation of faults that usually affect the electric system, reproducing a realistic situation. The direct use of differential equations do not give an acceptable estimation as far as digital relays are concerned because they have convergence times over two cycles. In this sense, a 5th order median filter was utilized, providing faster diagnosis for the fault location estimation. The estimation of the fault location has proved to be a coherent criteria for the algorithm. For the fault types tested, the estimation for the fault location utilizing line parameters has shown itself highly satisfactory for protection purposes. This work has shown that the algorithm oulputs converge in less than 1 and a half cycles afler the fault occurrence, presenting a much faster response for digital relays.

L’imagerie par résonance magnétique de diffusion (IRMd) est une technique non-invasive permettant d’accéder à l'information structurelle des tissus biologiques à travers l’étude du mouvement de diffusion des molécules d’eau dans les tissus. Ses applications sont nombreuses en neurologie pour le diagnostic de certaines anomalies cérébrales. Cependant, en raison du mouvement cardiaque, l’utilisation de cette technique pour accéder à l’architecture du cœur in vivo représente un grand défi. Le mouvement cardiaque a été identifié comme une des sources majeures de perte du signal mesuré en IRM de diffusion. A cause de la sensibilité au mouvement, il est difficile d’évaluer dans quelle mesure les caractéristiques de diffusion obtenues à partir de l’IRM de diffusion reflètent les propriétés réelles des tissus cardiaques. Dans ce cadre, la modélisation et la simulation numérique du signal d’IRM de diffusion offrent une approche alternative pour aborder le problème. L’objectif de cette thèse est d’étudier numériquement l’influence du mouvement cardiaque sur les images de diffusion et de s’intéresser à la question d’atténuation de l’effet du mouvement cardiaque sur le signal d’IRM de diffusion. Le premier chapitre est consacré à l’introduction du principe physique de l'imagerie par résonance magnétique(IRM). Le deuxième chapitre présente le principe de l’IRM de diffusion et résume l’état de l’art des différents modèles proposés dans la littérature pour modéliser le signal d’IRM de diffusion. Dans le troisième chapitre un modèle modifié de l’équation de Bloch-Torrey dans un domaine qui se déforme au cours du temps est introduit et étudié. Ce modèle représente une généralisation de l’équation de Bloch-Torrey utilisée dans la modélisation du signal d’IRM de diffusion dans le cas sans mouvement. Dans le quatrième chapitre, l’influence du mouvement cardiaque sur le signal d’IRM de diffusion est étudiée numériquement en utilisant le modèle de Bloch-Torrey modifié et un champ de mouvement analytique imitant une déformation réaliste du cœur. L’étude numérique présentée, permet de quantifier l’effet du mouvement sur la mesure de diffusion en fonction du type de la séquence de codage de diffusion utilisée, de classer ces séquences en terme de sensibilité au mouvement cardiaque et d’identifier une fenêtre temporelle par rapport au cycle cardiaque où l’influence du mouvement est réduite. Enfin, dans le cinquième chapitre, une méthode de correction de mouvement est présentée afin de minimiser l’effet du mouvement cardiaque sur les images de diffusion. Cette méthode s’appuie sur un développement singulier du modèle de Bloch-Torrey modifié pour obtenir un modèle asymptotique qui permet de résoudre le problème inverse de récupération puis correction de la diffusion influencée par le mouvement cardiaque. / Diffusion magnetic resonance imaging (dMRI) is a non-invasive technique allowing access to the structural information of the biological tissues through the study of the diffusion motion of water molecules in tissues. Its applications are numerous in neurology, especially for the diagnosis of certain brain abnormalities, and for the study of the human cerebral white matter. However, due to the cardiac motion, the use of this technique to study the architecture of the in vivo human heart represents a great challenge. Cardiac motion has been identified as a major source of signal loss. Because of the sensitivity to motion, it is difficult to assess to what extent the diffusion characteristics obtained from diffusion MRI reflect the real properties of the cardiac tissue. In this context, modelling and numerical simulation of the diffusion MRI signal offer an alternative approach to address the problem. The objective of this thesis is to study numerically the influence of cardiac motion on the diffusion images and to focus on the issue of attenuation of the cardiac motion effect on the diffusion MRI signal. The first chapter of this thesis is devoted to the introduction of the physical principle of nuclear magnetic resonance (NMR) and image reconstruction techniques in MRI. The second chapter presents the principle of diffusion MRI and summarizes the state of the art of the various models proposed in the litera- ture to model the diffusion MRI signal. In the third chapter a modified model of the Bloch-Torrey equation in a domain that deforms over time is introduced and studied. This model represents a generalization of the Bloch-Torrey equation used to model the diffusion MRI signal in the case of static organs. In the fourth chapter, the influence of cardiac motion on the diffusion MRI signal is investigated numerically by using the modified Bloch-Torrey equation and an analytical motion model mimicking a realistic deformation of the heart. The numerical study reported here, can quantify the effect of motion on the diffusion measurement depending on the type of the diffusion coding sequence. The results obtained allow us to classify the diffusion encoding sequences in terms of sensitivity to the cardiac motion and identify for each sequence a temporal window in the cardiac cycle in which the influence of motion is reduced. Finally, in the fifth chapter, a motion correction method is presented to minimize the effect of cardiac motion on the diffusion images. This method is based on a singular development of the modified Bloch-Torrey model in order to obtain an asymptotic model of ordinary differential equation that gives a relationship between the true diffusion and the diffusion reconstructed in the presence of motion. This relationship is then used to solve the inverse problem of recovery and correction of the diffusion influenced by the cardiac motion.