>Magister Scientiae - MSc / We present various methods of pricing Asian options. The methods include Monte Carlo simulations designed using control and antithetic variates, numerical solution of partial differential equation and using lower bounds.The price of the Asian option is known to be a certain risk-neutral expectation. Using the Feynman-Kac theorem, we deduce that the problem of determining the expectation implies solving a linear parabolic partial differential equation. This partial differential equation does not admit explicit solutions due to the fact that the distribution of a sum of lognormal variables is not explicit. We then solve the partial differential equation numerically using finite difference and Monte Carlo methods.Our Monte Carlo approach is based on the pseudo random numbers and not deterministic sequence of numbers on which Quasi-Monte Carlo methods are designed. To make the Monte Carlo method more effective, two variance reduction techniques are discussed.Under the finite difference method, we consider explicit and the Crank-Nicholson’s schemes. We demonstrate that the explicit method gives rise to extraneous solutions because the stability conditions are difficult to satisfy. On the other hand, the Crank-Nicholson method is unconditionally stable and provides correct solutions. Finally, we apply the pricing methods to a similar problem of determining the price of a European-style arithmetic basket option under the Black-Scholes framework. We find the optimal lower bound, calculate it numerically and compare this with those obtained by the Monte Carlo and Moment Matching methods.Our presentation here includes some of the most recent advances on Asian options, and we contribute in particular by adding detail to the proofs and explanations. We also contribute some novel numerical methods. Most significantly, we include an original contribution on the use of very sharp lower bounds towards pricing European basket options.

In this thesis, a mathematical model for HIV and TB co-infection with TB treatment among populations of Ethiopia is developed and analyzed. The TB model includes an age of infection. We compute the basic reproduction numbers RTB and RH for TB and HIV respectively, and the overall repro- duction number R for the system. We find that if R < 1 and R > 1; then the disease-free and the endemic equilibria are locally asymptotically stable, respectively. Otherwise these equilibria are unstable. The TB-only endemic equilibrium is locally asymptotically stable if RTB > 1, and RH < 1. How- ever, the symmetric condition, RTB < 1 and RH > 1, does not necessarily guarantee the stability of the HIV-only equilibrium, but it is possible that TB can coexist with HIV when RH > 1: As a result, we assess the impact of TB treatment on the prevalence of TB and HIV co-infection. To derive and formulate the nonlinear differential equations models for HIV and TB co-infection that accounts for treatment, we formulate and analyze the HIV only sub models, the TB-only sub models and the full models of HIV and TB combined. The TB-only sub model includes both ODEs and PDEs in order to describe the variable infectiousness and e ect of TB treatment during the infectious period. To analyse and solve the three models, we construct robust methods, namely the numerical nonstandard definite difference methods (NSFDMs). Moreover, we improve the order of convergence of these methods in their applications to solve the model of HIV and TB co-infection with TB treatment at the population level in Ethiopia. The methods developed in this thesis work and show convergence, especially for individuals with small tolerance either to the disease free or the endemic equilibria for first order mixed ODE and PDE as we observed in our models. / Mathematical Sciences / Ph. D. (Applied Mathematics)

É bem sabido que métodos clássicos de elementos finitos e diferenças finitas para o problema de Helmholtz apresentam efeito de poluição, que pode deteriorar seriamente a qualidade da solução aproximada. Controlar o efeito de poluição é especialmente difícil quando são utilizadas malhas não uniformes. Para malhas uniformes com elementos quadrados são conhecidos métodos (p. e. o QSFEM, proposto por Babuska et al) que minimizam a poluição. Neste trabalho apresentamos inicialmente dois métodos de elementos finitos de Petrov-Galerkin com formulação relativamente simples, o RPPG e o QSPG, ambos com razoável robustez para certos tipos de distorções dos elementos. O QSPG apresenta ainda poluição mínima para elementos quadrados. Em seguida é formulado o QOFD, um método de diferenças finitas aplicável a malhas não estruturadas. O QOFD apresenta grande robustez em relação a distorções, mas requer trabalho extra para tratar problemas não homogêneos ou condições de contorno não essenciais. Finalmente é apresentado um novo método de elementos finitos de Petrov-Galerkin, o QOPG, que é formulado aplicando a mesma técnica usada para obter a estabilização do QOFD, obtendo assim a mesma robustez em relação a distorções da malha, com a vantagem de ser um método variacionalmente consistente. Resultados numéricos são apresentados ilustrando o comportamento de todos os métodos desenvolvidos em comparação com os métodos de Galerkin, GLS e QSFEM. / It is well known that classical finite elements or finite difference methods for Helmholtz problem present pollution effects that can severely deteriorate the quality of the approximate solution. To control pollution effects is especially difficult on non uniform meshes. For uniform meshes of square elements pollution effects can be minimized with the Quasi Stabilized Finite Element Method (QSFEM) proposed by Babusv ska el al, for example. In the present work we initially present two relatively simple Petrov-Galerkin finite element methods, referred here as RPPG (Reduced Pollution Petrov-Galerkin) and QSPG (Quasi Stabilized Petrov-Galerkin), with reasonable robustness to some type of mesh distortion. The QSPG also shows minimal pollution, identical to QSFEM, for uniform meshes with square elements. Next we formulate the QOFD (Quasi Stabilized Finite Difference) method, a finite difference method for unstructured meshes. The QOFD shows great robustness relative to element distortion, but requires extra work to consider non-essential boundary conditions and source terms. Finally we present a Quasi Optimal Petrov-Galerkin (QOPG) finite element method. To formulate the QOPG we use the same approach introduced for the QOFD, leading to the same accuracy and robustness on distorted meshes, but constructed based on consistent variational formulation. Numerical results are presented illustrating the behavior of all methods developed compared to Galerkin, GLS and QSFEM.

Philosophiae Doctor - PhD / In recent years, there has been a great interest towards the higher order numerical methods for singularly perturbed problems. As compared to their lower order counterparts, they provide better accuracy with fewer mesh points. Construction and/or implementation of direct higher order methods is usually very complicated. Thus a natural choice is to use some convergence acceleration techniques, e.g., Richardson extrapolation, defect correction, etc. In this thesis, we will consider various classes of problems described by singularly perturbed ordinary and partial differential equations. For these problems, we design some novel numerical methods and attempt to increase their accuracy as well as the order of convergence. We also do the same for existing numerical methods in some instances. We find that, even though the Richardson extrapolation technique always improves the accuracy, it does not perform equally well when applied to different methods for certain classes of problems. Moreover, while in some cases it improves the order of convergence, in other cases it does not. These issues are discussed in this thesis for linear and nonlinear singularly perturbed ODEs as well as PDEs. Extrapolation techniques are analyzed thoroughly in all the cases, whereas the limitations of the defect correction approach for certain problems is indicated at the end of the thesis. / South Africa

Infrasound is undetectable by the human ear and excessive exposure may be a substantial health risk. Low frequency sound propagates through walls with minimal attenuation, making it difficult to avoid. This study interprets the results from both analytical calculations and simulations of pressure waves propagating through a wall in one dimension. The wall is thin compared to the wavelength; the model implements properties of three materials commonly used in walls. The results indicate that the geometry of the wall, most importantly the small ratio between wall width and wavelength, is the prime reason for the low levels of attenuation observed in transmitted amplitudes of low frequency sounds, and that damping is negligible for infrasound. Furthermore, a one-dimensional homogeneous wall model gives rise to periodicity in the transmitted amplitude, which is not observed in experiments. Future studies should prioritize the introduction of at least one more dimension to the model, to allow for variable angles of incidence.

Efficient Implementation of 3D Finite Difference Schemes on Recent Processors Abstract In this paper a solver is introduced that solves a problem set modelled by the Burgers equation using the finite difference method: forward in time and central in space. The solver is parallelized and optimized for Intel Xeon Phi 7120P as well as Intel Xeon E5-2699v3 processors to investigate differences in terms of performance between the two architectures. Optimized data access and layout have been implemented to ensure good cache utilization. Loop tiling strategies are used to adjust data access with respect to the L2 cache size. Compiler hints describing aligned memory access are used to support vectorization on both processors. Additionally, prefetching strategies and streaming stores have been evaluated for the Intel Xeon Phi. Parallelization was done using OpenMP and MPI. The parallelisation for native execution on Xeon Phi is based on OpenMP and yielded a raw performance of nearly 100 GFLOP/s, reaching a speedup of almost 50 at a 83\% parallel efficiency. An OpenMP implementation on the E5-2699v3 (Haswell) processors produced up to 292 GFLOP/s, reaching a speedup of almost 31 at a 85\% parallel efficiency. For comparison a mixed implementation using interleaved communications with computations reached 267 GFLOP/s at a speedup of 28 with a 87\% parallel efficiency. Running a pure MPI implementation on the PDC's Beskow supercomputer with 16 nodes yielded a total performance of 1450 GFLOP/s and for a larger problem set it yielded a total of 2325 GFLOP/s, reaching a speedup and parallel efficiency at resp. 170 and 33,3\% and 290 and 56\%. An analysis based on the roofline performance model shows that the computations were memory bound to the L2 cache bandwidth, suggesting good L2 cache utilization for both the Haswell and the Xeon Phi's architectures. Xeon Phi performance can probably be improved by also using MPI. Keeping technological progress for computational cores in the Haswell processor in mind for the comparison, both processors perform well. Improving the stencil computations to a more compiler friendly form might improve performance more, as the compiler can possibly optimize more for the target platform. The experiments on the Cray system Beskow showed an increased efficiency from 33,3\% to 56\% for the larger problem, illustrating good weak scaling. This suggests that problem sizes should increase accordingly for larger number of nodes in order to achieve high efficiency. Frederick Ceder / Effektiv implementering av finita differensmetoder i 3D på moderna processorarkitekturer Sammanfattning Denna uppsats diskuterar implementationen av ett program som kan lösa problem modellerade efter Burgers ekvation numeriskt. Programmet är byggt ifrån grunden och använder sig av finita differensmetoder och applicerar FTCS metoden (Forward in Time Central in Space). Implementationen paralleliseras och optimeras på Intel Xeon Phi 7120P Coprocessor och Intel Xeon E5-2699v3 processorn för att undersöka skillnader i prestanda mellan de två modellerna. Vi optimerade programmet med omtanke på dataåtkomst och minneslayout för att få bra cacheutnyttjande. Loopblockningsstrategier används också för att dela upp arbetsminnet i mindre delar för att begränsa delarna i L2 cacheminnet. För att utnyttja vektorisering till fullo så används kompilatordirektiv som beskriver minnesåtkomsten, vilket ska hjälpa kompilatorn att förstå vilka dataaccesser som är alignade. Vi implementerade också prefetching strategier och streaming stores på Xeon Phi och disskuterar deras värde. Paralleliseringen gjordes med OpenMP och MPI. Parallelliseringen för Xeon Phi:en är baserad på bara OpenMP och exekverades direkt på chipet. Detta gav en rå prestanda på nästan 100 GFLOP/s och nådde en speedup på 50 med en 83% effektivitet. En OpenMP implementation på E5-2699v3 (Haswell) processorn fick upp till 292 GFLOP/s och nådde en speedup på 31 med en effektivitet på 85%. I jämnförelse fick en hybrid implementation 267 GFLOP/s och nådde en speedup på 28 med en effektivitet på 87%. En ren MPI implementation på PDC's Beskow superdator med 16 noder gav en total prestanda på 1450 GFLOP/s och för en större problemställning gav det totalt 2325 GFLOP/s, med speedup och effektivitet på respektive 170 och 33% och 290 och 56%. En analys baserad på roofline modellen visade att beräkningarna var minnesbudna till L2 cache bandbredden, vilket tyder på bra L2-cache användning för både Haswell och Xeon Phi:s arkitekturer. Xeon Phis prestanda kan förmodligen förbättras genom att även använda MPI. Håller man i åtanke de tekniska framstegen när det gäller beräkningskärnor på de senaste åren, så preseterar både arkitekturer bra. Beräkningskärnan av implementationen kan förmodligen anpassas till en mer kompilatorvänlig variant, vilket eventuellt kan leda till mer optimeringar av kompilatorn för respektive plattform. Experimenten på Cray-systemet Beskow visade en ökad effektivitet från 33,3% till 56% för större problemställningar, vilket visar tecken på bra weak scaling. Detta tyder på att effektivitet kan uppehållas om problemställningen växer med fler antal beräkningsnoder. Frederick Ceder

[ES] La forma más exacta de conocer el desplazamiento de los neutrones a través de un medio material se consigue resolviendo la Ecuación del Transporte Neutrónico. Tres diferentes aproximaciones de esta ecuación se han investigado en esta tesis: Ecuación del transporte neutrónico resuelta por el método de Ordenadas Discretas, Ecuación de la Difusión y Ecuación de Armónicos Esféricos Simplificados. Para resolver estás ecuaciones se estudian diferentes esquemas del Método de Diferencias Finitas. La solución a estas ecuaciones describe la población de neutrones y las reacciones ocasionadas dentro de un reactor nuclear. A su vez, estas variables están relacionadas con el flujo y la potencia, parámetros fundamentales para el Análisis de Seguridad Nuclear. La tesis introduce la definición de las ecuaciones mencionadas y en particular se detallan para el estado estacionario. Se plantea el Método Modal como solución a los problemas de autovalores definidos por dichas ecuaciones. Primero se desarrollan varios algoritmos para la resolución del estado estacionario de la Ecuación del Transporte de Neutrones con el Método de Ordenadas Discretas para la discretización angular y el Método de Diferencias Finitas para la discretización espacial. Se ha implementado una formulación capaz de resolver el problema de autovalores para cualquier número de grupos energéticos con upscattering y anisotropía. Varias cuadraturas utilizadas por este método en su resolución angular han sido estudiadas e implementadas para cualquier orden de aproximación de Ordenadas Discretas. Además, otra formulación se desarrolla para la solución del problema fuente de la ecuación del transporte neutrónico. A continuación, se lleva a cabo un algoritmo que permite resolver la Ecuación de la Difusión de Neutrones con dos variantes del método de diferencias Finitas, una centrada en celda y otra en vértice o nodo. Se utiliza también el Método Modal calculando cualquier número de autovalores para varios grupos de energía y con upscattering. También se implementan los dos esquemas del Método de Diferencias Finitas anteriormente mencionados en el desarrollo de diferentes algoritmos para resolver las Ecuaciones de Armónicos Esféricos Simplificados. Además, se ha realizado un análisis de diferentes aproximaciones de las condiciones de contorno. Finalmente, se han realizado cálculos de la constante de multiplicación, los modos subcríticos, el flujo neutrónico y la potencia para diferentes tipos de reactores nucleares. Estas variables resultan esenciales en Análisis de Seguridad Nuclear. Además, se han realizado diferentes estudios de sensibilidad de parámetros como tamaño de malla, orden utilizado en cuadraturas o tipo de cuadraturas. / [CA] La forma més exacta de conèixer el desplaçament dels neutrons a través d'un mitjà material s'aconsegueix resolent l'Equació del Transport Neutrònic. Tres diferents aproximacions d'esta equació s'han investigat en aquesta tesi: Equació del Transport Neutrònic resolta pel mètode d'Ordenades Discretes, Equació de la Difusió i Equació d'Ármonics Esfèrics Simplificats. Per a resoldre estes equacions s'estudien diferents esquemes del Mètode de Diferències Finites. La solució a estes equacions descriu la població de neutrons i les reaccions ocasionades dins d'un reactor nuclear. Al seu torn, estes variables estan relacionades amb el flux i la potència, paràmetres fonamentals per a l'Anàlisi de Seguretat Nuclear. La tesi introduïx la definició de les equacions mencionades i en particular es detallen per a l'estat estacionari. Es planteja el Mètode Modal com a solució als problemes d'autovalors definits per les dites equacions. Primer es desenvolupen diversos algoritmes per a la resolució de l'estat estacionari de l'Equació del Transport de Neutrons amb el Mètode d'Ordenades Discretes per a la discretiztació angular i el Mètode de Diferències Finites per a la discretització espacial. S'ha implementat una formulació capaç de resoldre el problema d'autovalors per a qualsevol nombre de grups energètics amb upscattering i anisotropia. Diverses quadratures utilitzades per este mètode en la seua resolució angular han sigut estudiades i implementades per a qualsevol orde d'aproximació d'Ordenades Discretes. A més, una altra formulació es desenvolupa per a la solució del problema font de l'Equació del Transport Neutrònic. A continuació, es du a terme un algoritme que permet resoldre l'Equació de la Difusió de Neutrons amb dos variants del mètode de Diferències Finites, una centrada en cel·la i una altra en vèrtex o node. S'utilitza també el Mètode Modal calculant qualsevol nombre d'autovalors per a diversos grups d'energia i amb upscattering. També s'implementen els dos esquemes del Mètode de Diferències Finites anteriorment mencionats en el desenvolupament de diferents algoritmes per a resoldre les Equacions d'Harmònics Esfèrics Simplificats. A més, s'ha realitzat una anàlisi de diferents aproximacions de les condicions de contorn. Finalment, s'han realitzat càlculs de la constant de multiplicació, els modes subcrítics, el flux neutrònic i la potència per a diferents tipus de reactors nuclears. Estes variables resulten essencials en Anàlisi de Seguretat Nuclear. A més, s'han realitzat diferents estudis de sensibilitat de paràmetres com la grandària de malla, orde utilitzat en quadratures o tipus de quadratures. / [EN] The most accurate way to know the movement of the neutrons through matter is achieved by solving the Neutron Transport Equation. Three different approaches to solve this equation have been investigated in this thesis: Discrete Ordinates Neutron Transport Equation, Neutron Diffusion Equation and Simplified Spherical Harmonics Equations. In order to solve the equations, different schemes of the Finite Differences Method were studied. The solution of these equations describes the population of neutrons and the occurred reactions inside a nuclear system. These variables are related with the flux and power, fundamental parameters for the Nuclear Safety Analysis. The thesis introduces the definition of the mentioned equations. In particular, they are detailed for the steady state case. The Modal Method is proposed as a solution to the eigenvalue problems determined by the equations. First, several algorithms for the solution of the steady state of the Neutron Transport Equation with the Discrete Ordinates Method for the angular discretization and Finite Difference Method for spatial discretization are developed. A formulation able to solve eigenvalue problems for any number of energy groups, with scattering and anisotropy has been developed. Several quadratures used by this method for the angular discretization have been studied and implemented for any order of approach of the discrete ordinates. Furthermore, an adapted formulation has been developed as a solution of the source problem for the Neutron Transport Equation. Next, an algorithm is carried out that allows to solve the Neutron Diffusion Equation with two variants of the Finite Difference Method, one with cell centered scheme and another edge entered. The Modal method is also used for calculating any number of eigenvalues for several energy groups and upscattering. Both Finite Difference schemes mentioned before are also implemented to solve the Simplified Spherical Harmonics Equations. Moreover, an analysis of different approaches of the boundary conditions is performed. Finally, calculations of the multiplication factor, subcritical modes, neutron flux and the power for different nuclear reactors were carried out. These variables result essential in Nuclear Safety Analysis. In addition, several sensitivity studies of parameters like mesh size, quadrature order or quadrature type were performed. / Me gustaría dar las gracias al Ministerio de Economía, Industria y Competitividad y a la Agencia Estatal de Investigación de España por la concesión de mi contrato predoctoral de formación de personal investigador con referencia BES-2016-076782. La ayuda económica proporcionada por este contrato fue esencial para el desarrollo de esta tesis, así como para el financiamiento de una estancia. / Morato Rafet, S. (2020). Contributions to solve the Multi-group Neutron Transport equation with different Angular Approaches [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/159271 / TESIS

Une modélisation numérique des ondes poroélastiques, décrites par le modèle de Biot, est proposée dans le domaine temporel. La dissipation visqueuse à l'intérieur des pores est décrite par le modèle de perméabilité dynamique de Johnson-Koplik-Dashen (JKD). Certains coefficients du modèle de Biot-JKD sont proportionnels à la racine carrée de la fréquence, introduisant dans le domaine temporel des dérivées fractionnaires décalées d'ordre 1/2, revenant à un produit de convolution. Basé sur une représentation diffusive, le produit de convolution est remplacé par un nombre fini de variables de mémoire satisfaisant une équation différentielle ordinaire locale en temps, menant au modèle de Biot-DA (diffusive approximation). Les propriétés des deux modèles sont analysées : hyperbolicité, décroissance de l'énergie, dispersion. On montre que la meilleure méthode de détermination des coefficients de l'approximation diffusive - quadratures de Gauss, optimisation linéaire ou non-linéaire sur la plage de fréquence d'intérêt - est l'optimisation non-linéaire. Une méthode de splitting est utilisée numériquement : la partie propagative est discrétisée par un schéma aux différences finies ADER d'ordre 4, et la partie diffusive est intégrée exactement. Les conditions de saut aux interfaces sont discrétisées avec une méthode d'interface immergée. Des simulations numériques sont présentées pour des milieux isotropes et isotropes transverses. Des comparaisons avec des solutions analytiques montrent l'efficacité et la précision de cette approche. Des simulations numériques en milieux complexes sont réalisées : influence de la porosité d'os spongieux, diffusion multiple en milieu aléatoire. / A time-domain numerical modeling of Biot poroelastic waves is proposed. The viscous dissipation in the pores is described using the dynamic permeability model of Johnson-Koplik-Dashen (JKD). Some of the coefficients in the Biot-JKD model are proportional to the square root of the frequency: in the time-domain, these coefficients introduce shifted fractional derivatives of order 1/2, involving a convolution product. Based on a diffusive representation, the convolution product is replaced by a finite number of memory variables that satisfy local-in-time ordinary differential equations, resulting in the Biot-DA (diffusive approximation). The properties of the two models are analyzed: hyperbolicity, decrease of energy, dispersion. To determine the coefficients of the diffusive approximation, different methods of quadrature are analyzed: Gaussian quadratures, linear or nonlinear optimization procedures in the frequency range of interest. The nonlinear optimization is shown to be the best way of determination. A splitting strategy is applied numerically: the propagative part is discretized using a fourth-order ADER scheme on a Cartesian grid, and the diffusive part is solved exactly. An immersed interface method is implemented to discretize the jump conditions at interfaces. Numerical experiments are presented for isotropic and transversely isotropic media. Comparisons with analytical solutions show the efficiency and the accuracy of this approach. Some numerical experiments are performed in complex media: influence of the porosity of a cancellous bone, multiple scattering across a set of random scatterers.

Philosophiae Doctor - PhD / Fitted operator finite difference methods (FOFDMs) for singularly perturbed problems have been explored for the last three decades. The construction of these numerical schemes is based on introducing a fitting factor along with the diffusion coefficient or by using principles of the non-standard finite difference methods. The FOFDMs based on the latter idea, are easy to construct and they are extendible to solve partial differential equations (PDEs) and their systems. Noting this flexible feature of the FOFDMs, this thesis deals with extension of these methods to solve interior layer problems, something that was still outstanding. The idea is then extended to solve singularly perturbed time-dependent PDEs whose solutions possess interior layers. The second aspect of this work is to improve accuracy of these approximation methods via methods like Richardson extrapolation. Having met these three objectives, we then extended our approach to solve singularly perturbed two-point boundary value problems with variable diffusion coefficients and analogous time-dependent PDEs. Careful analyses followed by extensive numerical simulations supporting theoretical findings are presented where necessary.

In this dissertation, we consider the Hull-White term structure problem with the boundary value condition given as the payoff of a European bond option. We restrict ourselves to the case where the parameters of the Hull-White model are strictly positive constants and from the risk neutral valuation formula, we first derive simple closed–form expression for pricing European bond option in the Hull-White extended Vasicek model framework. As the European option can be exercised only on the maturity date, we then examine the case of early exercise opportunity commonly called American option. With the analytic representation of American bond option being very hard to handle, we are forced to resort to numerical experiments. To do it excellently, we transform the Hull-White term structure equation into the diffusion equation and we first solve it through implicit, explicit and Crank-Nicolson (CN) difference methods. As these standard finite difference methods (FDMs) require truncation of the domain from infinite to finite one, which may deteriorate the computational efficiency for American bond option, we try to build a CN method over an unbounded domain. We introduce an exact artificial boundary condition in the pricing boundary value problem to reduce the original to an initial boundary problem. Then, the CN method is used to solve the reduced problem. We compare our performance with standard FDMs and the results through illustration show that our method is more efficient and accurate than standard FDMs when we price American bond option. / Mathematical Sciences / (M.Sc. (Mathematics))