GPU accelerated algorithms for computing matrix function vector products with applications to exponential integrators and fractional diffusion

The efficient computation of matrix function vector products has become an important area of research in recent times, driven in particular by two important applications: the numerical solution of fractional partial differential equations and the integration of large systems of ordinary differential equations. In this work we consider a problem that combines these two applications, in the form of a numerical solution algorithm for fractional reaction diffusion equations that after spatial discretisation, is advanced in time using the exponential Euler method. We focus on the efficient implementation of the algorithm on Graphics Processing Units (GPU), as we wish to make use of the increased computational power available with this hardware. We compute the matrix function vector products using the contour integration method in [N. Hale, N. Higham, and L. Trefethen. Computing Aα, log(A), and related matrix functions by contour integrals. SIAM J. Numer. Anal., 46(5):2505–2523, 2008]. Multiple levels of preconditioning are applied to reduce the GPU memory footprint and to further accelerate convergence. We also derive an error bound for the convergence of the contour integral method that allows us to pre-determine the appropriate number of quadrature points. Results are presented that demonstrate the effectiveness of the method for large two-dimensional problems, showing a speedup of more than an order of magnitude compared to a CPU-only implementation.

A stochastic exponential Euler scheme for simulation of stiff biochemical reaction systems

In order to simulate stiff biochemical reaction systems, an explicit exponential Euler scheme is derived for multidimensional, non-commutative stochastic differential equations with a semilinear drift term. The scheme is of strong order one half and A-stable in mean square. The combination with this and the projection method shows good performance in numerical experiments dealing with an alternative formulation of the chemical Langevin equation for a human ether a-go-go related gene ion channel mode

Universality in the fast orientational relaxation near isotropic-nematic transition

Detailed molecular dynamics simulations of Lennard-Jones ellipsoids have been carried out to investigate the emergence of criticality in the single-particle orientational relaxation near the isotropic-nematic (IN) phase transition. The simulations show a sudden appearance of a power-law behavior in the decay of the second-rank orientational relaxation as the IN transition is approached. The simulated value of the power-law exponent is 0.56, which is larger than the mean-field value (0.5) but less than the observed value (0.63) and may be due to the finite size of the simulated system. The decay of the first-rank orientational time correlation function, on the other hand, is nearly exponential but its decay becomes very slow near the isotropic-nematic transition, The zero-frequency rotational friction, calculated from the simulated angular Velocity correlation function, shows a marked increase near the IN transition.

Fast decay of the velocity autocorrelation function in dense shear flow of inelastic hard spheres

We find in complementary experiments and event-driven simulations of sheared inelastic hard spheres that the velocity autocorrelation function psi(t) decays much faster than t(-3/2) obtained for a fluid of elastic spheres at equilibrium. Particle displacements are measured in experiments inside a gravity-driven flow sheared by a rough wall. The average packing fraction obtained in the experiments is 0.59, and the packing fraction in the simulations is varied between 0.5 and 0.59. The motion is observed to be diffusive over long times except in experiments where there is layering of particles parallel to boundaries, and diffusion is inhibited between layers. Regardless, a rapid decay of psi(t) is observed, indicating that this is a feature of the sheared dissipative fluid, and is independent of the details of the relative particle arrangements. An important implication of our study is that the non-analytic contribution to the shear stress may not be present in a sheared inelastic fluid, leading to a wider range of applicability of kinetic theory approaches to dense granular matter.

Radiative decay of quarkonium into Higgs scalar

A relativistic bound-state formalism is used to calculate the branching ratio Γ(V→H+γ)/Γ(V→e+e-) where H is a Higgs scalar and significant relativistic effects have been obtained compared to the nonrelativistic calculation originally due to Wilczek

What was behind the bark? : An assessment of decay among urban Tilia, Betula and Acer trees felled as hazardous in the Helsinki City area

Old trees growing in urban environments are often felled due to symptoms of mechanical defects that could be hazardous to people and property. The decisions concerning these removals are justified by risk assessments carried out by tree care professionals. The major motivation for this study was to determine the most common profiles of potential hazard characteristics for the three most common urban tree genera in Helsinki City: Tilia, Betula and Acer, and in this way improve management practices and protection of old amenity trees. For this research, material from approximately 250 urban trees was collected in cooperation with the City of Helsinki Public Works Department during 2001 - 2004. From the total number of trees sampled, approximately 70% were defined as hazardous. The tree species had characteristic features as potential hazard profiles. For Tilia trees, hollowed heartwood with low fungal activity and advanced decay caused by Ganoderma lipsiense were the two most common profiles. In Betula spp., the primary reason for tree removal was usually lowered amenity value in terms of decline of the crown. Internal cracks, most often due to weak fork formation, were common causes of potential failure in Acer spp. Decay caused by Rigidoporus populinus often increased the risk of stem breakage in these Acer trees. Of the decay fungi observed, G. lipsiense was most often the reason for the increased risk of stem collapse. Other fungi that also caused extensive decay were R. populinus, Inonotus obliquus, Kretzschmaria deusta and Phellinus igniarius. The most common decay fungi in terms of incidence were Pholiota spp., but decay caused by these species did not have a high potential for causing stem breakage, because it rarely extended to the cambium. The various evaluations used in the study suggested contradictions in felling decisions based on trees displaying different stages of decay. For protection of old urban trees, it is crucial to develop monitoring methods so that tree care professionals could better analyse the rate of decay progression towards the sapwood and separate those trees with decreasing amounts of sound wood from those with decay that is restricted to the heartwood area.

Wave analysis of sound decay in rectangular rooms

The decay of sound in a rectangular room is analyzed for various boundary conditions on one of its walls. It is shown that the decay of the sound-intensity level is in general nonlinear. But for specific areas and impedances of the material it is possible to obtain a linear initial decay. It is also shown that the coefficients derived from the initial decay rates neither correspond to the predictions of Sabine's or Eyring's geometrical theories nor to the normal coefficients of Morse's wave theory. The dependence of the coefficients on the area of the material is discussed. The influence of the real and the imaginary parts of the specific acoustic impedance of the material on the coefficients is also discussed. Finally, the existence of a linear initial decay corresponding to the decay of a diffuse field in the case of a highly absorbing material partially covering a wall is explained on the basis of modal coupling.

Efficient simulation of unsaturated flow in heterogeneous soils using an exponential integrator

- Provided a practical variable-stepsize implementation of the exponential Euler method (EEM). - Introduced a new second-order variant of the scheme that enables the local error to be estimated at the cost of a single additional function evaluation. - New EEM implementation outperformed sophisticated implementations of the backward differentiation formulae (BDF) of order 2 and was competitive with BDF of order 5 for moderate to high tolerances.

Dirichlet problem at infinity for p-harmonic functions on negatively curved spaces

The object of this dissertation is to study globally defined bounded p-harmonic functions on Cartan-Hadamard manifolds and Gromov hyperbolic metric measure spaces. Such functions are constructed by solving the so called Dirichlet problem at infinity. This problem is to find a p-harmonic function on the space that extends continuously to the boundary at inifinity and obtains given boundary values there. The dissertation consists of an overview and three published research articles. In the first article the Dirichlet problem at infinity is considered for more general A-harmonic functions on Cartan-Hadamard manifolds. In the special case of two dimensions the Dirichlet problem at infinity is solved by only assuming that the sectional curvature has a certain upper bound. A sharpness result is proved for this upper bound. In the second article the Dirichlet problem at infinity is solved for p-harmonic functions on Cartan-Hadamard manifolds under the assumption that the sectional curvature is bounded outside a compact set from above and from below by functions that depend on the distance to a fixed point. The curvature bounds allow examples of quadratic decay and examples of exponential growth. In the final article a generalization of the Dirichlet problem at infinity for p-harmonic functions is considered on Gromov hyperbolic metric measure spaces. Existence and uniqueness results are proved and Cartan-Hadamard manifolds are considered as an application.

Wave analysis of sound decay in rectangular rooms

The decay of sound in a rectangular room is analyzed for various boundary conditions on one of its walls. It is shown that the decay of the sound-intensity level is in general nonlinear. But for specific areas and impedances of the material it is possible to obtain a linear initial decay. It is also shown that the coefficients derived from the initial decay rates neither correspond to the predictions of Sabine's or Eyring's geometrical theories nor to the normal coefficients of Morse's wave theory. The dependence of the coefficients on the area of the material is discussed. The influence of the real and the imaginary parts of the specific acoustic impedance of the material on the coefficients is also discussed. Finally, the existence of a linear initial decay corresponding to the decay of a diffuse field in the case of a highly absorbing material partially covering a wall is explained on the basis of modal coupling.

An exponential stability criterion for certain nonlinear systems

Improved sufficient conditions are derived for the exponential stability of a nonlinear time varying feedback system having a time invariant blockG in the forward path and a nonlinear time varying gain ϕ(.)k(t) in the feedback path. φ(.) being an odd monotone nondecreasing function. The resulting bound on is less restrictive than earlier criteria.

An exponential stability criterion for certain nonlinear systems

Improved sufficient conditions are derived for the exponential stability of a nonlinear time varying feedback system having a time invariant blockG in the forward path and a nonlinear time varying gain ϕ(.)k(t) in the feedback path. φ(.) being an odd monotone nondecreasing function. The resulting bound on $$\left( {{{\frac{{dk}}{{dt}}} \mathord{\left/ {\vphantom {{\frac{{dk}}{{dt}}} k}} \right. \kern-\nulldelimiterspace} k}} \right)$$ is less restrictive than earlier criteria.

The effect of lignin content and lignin modification on Norway spruce wood properties and decay resistance

Three different Norway spruce cutting clones growing in three environments with different soil and climatic conditions were studied. The purpose was to follow variation in the radial growth rate, wood properties and lignin content and to modify wood lignin with a natural monolignol, coniferyl alcohol, by making use of inherent wood peroxidases. In addition, the incorporation of chlorinated anilines into lignin was studied with synthetic model compounds and synthetic lignin preparations to show whether unnatural compounds originating from pesticides could be bound in the lignin polymer. The lignin content of heartwood, sapwood and earlywood was determined by applying Fourier transform infrared (FTIR) spectroscopy and a principal component regression (PCR) technique. Wood blocks were treated with coniferyl alcohol by using a vacuum impregnation method. The effect of impregnation was assessed by FTIR and by a fungal decay test. Trees from a fertile site showed the highest growth rate and sapwood lignin content and the lowest latewood proportion, weight density and modulus of rupture (MOR). Trees from a medium fertile site had the lowest growth rate and the highest latewood proportion, weight density, modulus of elasticity (MOE) and MOR. The most rapidly growing clone showed the lowest latewood proportion, weight density, MOE and MOR. The slowest growing clone had the lowest sapwood lignin content and the highest latewood proportion, weight density, MOE and MOR. Differences between the sites and clones were small, while fairly large variation was found between the individual trees and growing seasons. The cutting clones maintained clone-dependent wood properties in the different growing sites although variation between trees was high and climatic factors affected growth. The coniferyl alcohol impregnation increased the content of different lignin-type phenolic compounds in the wood as well as wood decay resistance against a white-rot fungus, Coriolus versicolor. During the synthetic lignin preparation 3,4-dichloroaniline became bound by a benzylamine bond to β-O-4 structures in the polymer and it could not be released by mild acid hydrolysis. The natural monolignol, coniferyl alcohol, and chlorinated anilines could be incorporated into the lignin polymer in vivo and in vitro, respectively.