The last two years have seen the most volatile financial markets for decades with steep losses in asset values and a deteriorating world economy. The insolvency of several banks and their negative impact on the economy has led to criticism of their risk management systems for not being adequate and lacking foresight. This thesis will study the performance of two broadly adopted portfolio risk measures before and during the current financial turbulence to examine their accuracy and reliability. The study will be carried out on a case portfolio consisting of American and European fixed income and equity. The portfolio uses a dynamic asset allocation scheme to maximize the ratio between expected return and portfolio risk. The market risk of the portfolio will be calculated on a daily basis using both Value-at-Risk (VaR) and expected shortfall (ES) in a Monte Carlo framework. These risk measures are then compared with prior measurements and the actual loss over the period. The results from the study indicate that the implemented risk model do not give totally reliable estimates, with more frequent and larger real losses than predicted. Nevertheless, the study sees a significant worsening in the performance of the risk measures during the current financial crisis from June 2007 to December 2008 compared with the previous years. This thesis argues that VaR and ES are useful risk measures, but that users should be well aware of the pitfalls in the underlying models and take appropriate precautions.

Properties of the empirical interpolation (EI) method is investigated by solving selected model problems. Also, a more challenging example with deformed geometry is solved within the online/offline computational framework of the reduced basis method.

It is often very difficult, particularly in higher dimensions, to find a good multivariate model that describes both marginal behavior and dependence structure of data efficiently. The copula approach to multivariate models has been found to fit this purpose particularly well, and since it is a relatively new concept in statistical modeling, it is under frequent development. In this thesis we focus on the decomposition of a multivariate model into pairwise copulas rather than the usual multivariate copula approach. We account for the theory behind the decomposition of a multivariate model into pairwise copulas, and apply the theory on both daily and intra day financial returns. The results are compared with the usual multivariate copula approach, and problems applying the theory are accounted for. The multivariate copula is rejected in favor of the pairwise decomposed model on daily returns with a level of significance less than 1%, while our decomposed models on intra day data does not lead to a rejection of the models with multivariate copulas. On daily returns a pairwise decomposition with Student copulas is preferable to multivariate copulas, while the decomposed models on intra day data need more development before outperforming multivariate copulas.

Shape analysis and recognition is a field ripe with creative solutions and innovative algorithms. We give a quick introduction to several different approaches, before basing our work on a representation introduced by Klassen et. al., considering shapes as equivalence classes of closed curves in the plane under reparametrization, and invariant under translation, rotation and scaling. We extend this to a definition for nonclosed curves, and prove a number of results, mostly concerning under which conditions on these curves the set of shapes become manifolds. We then motivate the study of geodesics on these manifolds as a means to compute a shape metric, and present two methods for computing such geodesics: the shooting method from Klassen et. al. and the ``direct'' method, new to this paper. Some numerical experiments are performed, which indicate that the direct method performs better for realistically chosen parameters, albeit not asymptotically.

In this thesis we study market risk in turbulent markets over different risk horizons. We construct portfolios which represent possible investments for a life assurance fund. The portfolios consist of equities, fixed income instruments, cash positions and interest rate derivatives. Today, the most commonly used metrics for market risk are Value-at-Risk (VaR) and Expected Shortfall (ES), and they will be central. We introduce necessary theory from quantitative finance related to asset price dynamics and security pricing. Further, interest rate related instruments are handled by the LIBOR Market Model (LMM), while equity prices are modeled as geometric Brownian motions. We use implied volatilities for instruments where they are available, and historical for the rest. We implement a risk model and make daily and quarterly market risk estimates between 2000-2008 for the portfolios. We choose some central events from the last quarter of 2008, a critical phase of the ongoing financial crisis, and analyze how the portfolios and the corresponding risk estimates are affected. Comparison of the portfolio losses against risk estimates allows us to evaluate the reliability of the broadly adopted model.

Compare the Normal Inverse Gaussian market model against empirical financial market data, and price exotic options using the numerical path integration approach.

The governing equations for waves propagating in water are derived by use of conservation laws. The equations are then cast onto dimensionless form and two important parameters are obtained. Approximations by use of asymptotic expansions in one or both of the parameters are then applied on the governing equations and we show that several different completely integrable equations, with different scaling transformations and at different order of approximations, can be derived. More precisely, the Korteweg-de Vries, Kadomtsev-Petviashvili and Boussinesq are obtained at first order, while the Camassa-Holm, Degasperis-Procesi, nonlinear Schrödinger and the Davey-Stewartson equations are obtained at second order. We discuss shortly some of the properties for each of the obtained equations.

In this text, I will look at some new approaches that may shed some light on the Kadison Singer problem, mainly one instigated by Vern Paulsen using dynamical systems in the Stone-Cech compactification of a discrete group. In order to do this, I will try to develop the theory in a crossed product setting, and look at some aspects of it that may hold interest of their own.

In this thesis an open-source software for permeability upscaling is developed. The software is based on DUNE, an open-source C++ framework for finding numerical solutions of partial differential equations (PDEs). It provides functionality used in finite elements, finite volumes and finite differences methods. Permeability is a measure of the ability of a material to transmit fluids, and determines the flow characteristics in reservoir models. Permeability upscaling is a technique to include fine-scale variations of the permeability field in a coarse-scale reservoir model. The upscaling technique used in this thesis involves solving an elliptic partial differential equation. This is solved with mixed and hybrid finite element methods. The mixed method transforms the original second order PDE into a system of two linear equations. The great advantage with these methods compared with standard finite element methods is continuity of the variable of interest in the upscaling problem. The hybrid method was introduced for being able to solve larger problems. The resulting system of equations from the hybrid method can be transformed into a symmetric positive definite system, which again can be solved with efficient iterative methods. Efficiency of the implementation is important, and as for most implementations of PDE solvers, the computational time is dominated by solving a system of linear equations. In this implementation it is used an algebraic multigrid (AMG) preconditioner provided with DUNE. This is known to be efficient on system arising from elliptic PDEs. The efficiency of the AMG preconditioner is compared with other alternatives, and is superior to the others. On the largest problem investigated, the AMG based solver is almost three times faster than the next best alternative. The performance of the implementation based on DUNE is also compared with an existing implementation by Sintef. Sintef's implementation is based on a mimetic finite difference method, but on the grid type investigated in this thesis, the methods are equivalent. Sintef's implementation uses the proprietary SAMG solver developed by Fraunhofer SCAI to solve the linear system of equations. SAMG is 58% faster than DUNE's solver on a test case consisting of 322 200 unknowns. The scalability of SAMG seem to be better than DUNE's AMG as the problem size increases. However, a great advantage with DUNE's solver is 50% lower memory usage measured on a problem consisting of approx. 3 million unknowns. Another advantage is the licensing of the software. Both DUNE and the upscaling software developed in this thesis is GPL licensed which means that anyone is free to improve or adjust the software.

We discuss nonholonomic systems in general and numerical methods for solving them. Two different approaches for obtaining numerical methods are considered; discretization of the Lagrange-d'Alembert equations on the one hand, and using the discrete Lagrange-d'Alembert principle to obtain nonholonomic integrators on the other. Among methods using the first approach, we focus on the super partitioned additive Runge-Kutta (SPARK) methods. Among nonholonomic integrators, we focus on a reversible second order method by McLachlan and Perlmutter. Through several numerical experiments the methods we present are compared by considering error-growth, conservation of energy, geometric properties of the solution and how well the constraints are satisfied. Of special interest is the comparison of the 2-stage SPARK Lobatto IIIA-B method and the nonholonomic integrator by McLachlan and Perlmutter, which both are reversible and of second order. We observe a clear connection between energy-conservation and the geometric properties of the numerical solution. To preserve energy in long-time integrations is seen to be important in order to get solutions with the correct qualitative properties. Our results indicate that the nonholonomic integrator by McLachlan and Perlmutter sometimes conserves energy better than the 2-stage SPARK Lobatto IIIA-B method. In a recent work by Jay, however, the same two methods are compared and are found to conserve energy equally well in long-time integrations.