No description available.

No description available.

Philosophiae Doctor - PhD / Numerical simulations play a crucial role in testing current cosmological models of the formation and evolution of the cosmic structure observed in the modern Universe. Simulations of the collapse of both baryonic and non-baryonic matter under the influence of gravity have yielded important results in our understanding of the large scale structure of the Universe. In addition to the underlying large scale structure, simulations which include gas dynamics can give us valuable insight into, and allow us to make testable predictions on, the nature and distribution of baryonic matter on a wide range of scales. In this work we give an overview of cosmological simulations and the methods employed in the solution of many body problems. We then present three projects focusing on scales ranging from individual galaxies to the cosmic web connecting clusters of galaxies thereby demonstrating the potential and diversity of numerical simulations in the fields of cosmology and astrophysics. We firstly investigate the environmental dependance of neutral hydrogen in the intergalactic medium by utilising high resolution cosmological hydrodynamic simulations in Chapter 3. We find that the extent of the neutral hydrogen radial profile is dependant on both the environment of the galaxy and its classification within the group ie. whether it is a central or satellite galaxy. We investigate whether this effect could arise from ram pressure forces exerted on the galaxies and find good agreement between galaxies experiencing high ram pressure forces and those with a low neutral hydrogen content. In Chapter 4 we investigate the velocity–shape alignment of clusters in a dark matter only simulation and the effect of such an alignment on measurements of the kinetic Sunyaev–Zeldovich (kSZ) effect. We find an alignment not only exists but can lead to an enhancement in the kSZ signal of up to 60% when the cluster is orientated along the line-of-sight. Finally we attempt to identify shocked gas in clusters and filaments using intermediate resolution cosmological hydrodynamic simulations in Chapter 5 with a view to predicting the synchrotron emission from these areas, something that may be detectable with the Square Kilometer Array.

We introduce an elastic constitutive model of gravity that enables the interpretation of cosmological observations in terms of established ideas from Solid Mechanics and multiscale modeling. The behavior of physical space is identified with that of a material-like medium called "cosmic fabric," which exhibits constitutive behavior. This cosmic fabric is a solid hyperplate that is broad in the three ordinary spatial dimensions and thin in a fourth hyperspatial dimension. Matter in space is treated as fabric inclusions that prescribe in-plane (three-dimensional) strain causing the transverse bending of the fabric into the fourth hyperspatial dimension. The linearized Einstein-Hilbert action, which governs the dynamics of physical space, is derived from postulating Hooke’s Law for the fabric, and the Schwarzschild metric is recovered from investigating matterabric interactions. At the continuum length scale, the Principle of Relativity is shown to apply for both moving and stationary observers alike, so that the fabric’s rest reference frame remains observationally indistinguishable at such a length scale. Within the Cosmic Fabric paradigm, the structural properties of space at different hierarchical length scales can be investigated using theoretical notions and computational tools from solid mechanics to address outstanding problems in cosmology and fundamental physics. For example, we propose and offer theoretical support for the "Inherent Structure Hypothesis", which states that the gravitational anomalies currently attributed to dark matter may in fact be manifestations of the inherent (undeformed) curvature of space. In addition, we develop a numerical framework wherein one can perform numerical "experiments" to investigate the implications of said hypothesis.

One of the most important aspects for all life on this planet is the act to keep their biological processes in a state where they do not reach equilibrium. One part in the upholding of this imbalanced state is the barrier between the cells and their surroundings, created by the cell membrane. Additionally, terrestrial animal life often requires a barrier that protects the organism's body from external hazards and water loss. As an alternative to experiments, the investigation of the processes occurring at these barriers can be performed by using molecular dynamics simulations. Through this method we can obtain an atomistic description of the dynamics associated with events that are not accessible to experimental setups.  In this thesis the first paper presents an improved particle-mesh Ewald method for the calculation of long-range Lennard-Jones interactions in molecular dynamics simulations, which solves the historical performance problem of the method. The second paper demonstrate an improved implementation, with a higher accuracy, that only incurs a performance loss of roughly 15% compared to conventional simulations using the Gromacs simulation package. Furthermore, the third paper presents a study of cholesterol's effect on the permeation of six different solutes across a variety of lipid bilayers. A laterally inhomogeneous permeability in cholesterol-containing membranes is proposed as an explanation for the large differences between experimental permeabilities and calculated partition coefficients in simulations. The fourth paper contains a coarse-grained simulation study of a proposed structural transformation in ceramide bilayer structures, during the formation of the stratum corneum. The simulations show that glycosylceramides are able to stabilize a three-dimensionally folded bilayer structure, while simulations with ceramides collapse into a lamellar bilayer structure. / <p>QC 20160308</p>

One of the most important aspects for all life on this planet is theact to keep their cellular processes in a state where they do notreach equilibrium. One part in the upholding of this imbalanced stateis the barrier between the cells and their surroundings, created bythe cell membrane. In addition to experiments, the investigation ofprocesses occuring in the cell membrane can be performed by usingmolecular dynamics simulations. Through this method we can obtain anatomistic description of the dynamics associated with events that arenot accessible to experimental setups. Molecular dynamics relies onthe integration of Newton's equations of motion in order to sample therelevant parts of phase-space for the system, and therefore it isdependent on a correct description of the interactions between all thesimulated particles. In this thesis I first present an improved methodfor the calculation of long-range interactions in molecular dynamicssimulations, followed by a study of cholesterol's impact on thepermeation of small solutes across a lipid bilayer. The first paper presents a previously derived modification to theparticle-mesh Ewald method, which makes it possible to apply thisto long-range Lennard-Jones interactions. Old implementations of themethod have been haunted by an extreme performance degradation andhere I propose a solution to this problem by applying a modifiedinteraction potential. I further show that the historical treatmentof long-range interactions in simulations of lipid bilayers hasnon-negligible effects on their structural properties.In the second paper, this modification is improved such that the smallerrors introduced by the modified interaction potential becomenegligible. Furthermore, I demonstrate that I have also improved theimplementation of the method so that it now only incurs a performanceloss of roughly 15% compared to conventional simulations using theGromacs simulation package.The third paper presents a simulation study of cholesterol's effect onthe permeation of six different solutes across a variety of lipidbilayers. I analyze the effect of different head groups, tail lengths,and tail saturation by performing simulations of the solutes in fourdifferent bilayers, with cholesterol contents between 0% and50%. Analysis of the simulations shows that the impact of the surfacearea per lipid on the partitioning of the solute could be lower thanpreviously thought. Furthermore, a model with a laterallyinhomogeneous permeability in cholesterol-containing membranes isproposed, which could explain the large differences betweenpermeabilities from experiments and calculated partition coefficientsin simulations. / <p>QC 20140609</p>

We consider fast computation methods for simulation of dynamics of a collection of particles dispersed in an unbounded Stokesian suspension. Stokesian suspensions are of great practical interest in the manufacturing and processing of various commercial products. The most popular dynamic simulation method for these kind of suspensions was developed by Brady and Bossis (Brady and Bossis [1988]). This method uses a truncated multipole expansion to represent the fluid traction on particle surfaces. The hydrodynamic interactions in Stoke-sian suspension are long ranged in nature, resulting in strong coupled motion of all particles. For an N particle system, this method imposes an O(N3) computational cost, thus posing limitations to the number of particles that may be simulated. More recent methods (Sierou and Brady [2001], Scintilla, Darve and Shaqfeh [2005]) have attempted to solve this problem using Particle Mesh Ewald summation techniques by distributing the moments on a grid and using Fast Fourier Transform algorithms, resulting in an O(N log N) computational cost. We review these methods and propose a version that we believe is some-what superior. In the course of this study, we have identified and corrected errors in previous studies that maybe of some importance in determining the bulk properties of suspensions. Finally, we show the utility of our method in determining certain properties of suspensions and compare them to existing analytical results for the same.

In this paper we describe a parallel algorithm for calculating nonequispaced fast Fourier transforms on massively parallel distributed memory architectures. These algorithms are implemented in an open source software library called PNFFT. Furthermore, we derive a parallel fast algorithm for the computation of the Coulomb potentials and forces in a charged particle system, which is based on the parallel nonequispaced fast Fourier transform. To prove the high scalability of our algorithms we provide performance results on a BlueGene/P system using up to 65536 cores.

The present thesis provides a modularized view on the structure of fast numerical methods for computing Coulomb interactions between charged particles in three-dimensional space. Thereby, the common structure is given in terms of three self-contained algorithmic frameworks that are built on top of each other, namely fast Fourier transform (FFT), nonequispaced fast Fourier transform (NFFT) and NFFT based particle-mesh methods (P²NFFT). For each of these frameworks algorithmic enhancement and parallel implementations are presented with special emphasis on scalability up to hundreds of thousands of parallel processes. In the context of FFT massively parallel algorithms are composed from hardware adaptive low level modules provided by the FFTW software library. The new algorithmic NFFT concepts include pruned NFFT, interlacing, analytic differentiation, and optimized deconvolution in Fourier space with respect to a mean square aliasing error. Enabled by these generalized concepts it is shown that NFFT provides a unified access to particle-mesh methods. Especially, mixed-periodic boundary conditions are handled in a consistent way and interlacing can be incorporated more efficiently. Heuristic approaches for parameter tuning are presented on the basis of thorough error estimates. / Die vorliegende Dissertation beschreibt einen modularisierten Blick auf die Struktur schneller numerischer Methoden für die Berechnung der Coulomb-Wechselwirkungen zwischen Ladungen im dreidimensionalen Raum. Die gemeinsame Struktur ist geprägt durch drei selbstständige und auf einander aufbauenden Algorithmen, nämlich der schnellen Fourier-Transformation (FFT), der nicht äquidistanten schnellen Fourier-Transformation (NFFT) und der NFFT-basierten Teilchen-Gitter-Methode (P²NFFT). Für jeden dieser Algorithmen werden Verbesserungen und parallele Implementierungen vorgestellt mit besonderem Augenmerk auf massiv paralleler Skalierbarkeit. Im Kontext der FFT werden parallele Algorithmen aus den Hardware adaptiven Modulen der FFTW Softwarebibliothek zusammengesetzt. Die neuen NFFT-Konzepte beinhalten abgeschnittene NFFT, Versatz, analytische Differentiation und optimierte Entfaltung im Fourier-Raum bezüglich des mittleren quadratischen Aliasfehlers. Mit Hilfe dieser Verallgemeinerungen bietet die NFFT einen vereinheitlichten Zugang zu Teilchen-Gitter-Methoden. Insbesondere gemischt periodische Randbedingungen werden einheitlich behandelt und Versatz wird effizienter umgesetzt. Heuristiken für die Parameterwahl werden auf Basis sorgfältiger Fehlerabschätzungen angegeben.

In this paper we describe a parallel algorithm for calculating nonequispaced fast Fourier transforms on massively parallel distributed memory architectures. These algorithms are implemented in an open source software library called PNFFT. Furthermore, we derive a parallel fast algorithm for the computation of the Coulomb potentials and forces in a charged particle system, which is based on the parallel nonequispaced fast Fourier transform. To prove the high scalability of our algorithms we provide performance results on a BlueGene/P system using up to 65536 cores.