In this thesis, we present a novel algorithm for computing the radiation field in astrophysical simulations. The algorithm is tree-based, similar to many gravity solvers, and allows computation of radiative transfer in O(Nsink logNsource) in cases without absorption, and O(Nsink logNsource logN) time with absorption. The algorithm scales well with the number of processors due to its tree-based nature, and is highly tunable in accuracy and speed. It is also only weakly dependent on both the energy band and the number of energy bands used. We provide a suite of tests of the code showing its ability to create accurate fluxes, ionization fronts, hydrodynamics coupling, and shadowing. We apply the algorithm in a set of simulations on an isolated spiral galaxy from the AGORA project. The algorithm is used to calculate FUV fields within the galaxy, which self-consistently sets the dominant photoelectric heating in the gas. This has never before been performed in galaxy simulations. We find, in agreement with Ostriker et al. [2010], that FUV can be a very important regulation mechanism for star formation in a galaxy. Depending on the assumed opacity, FUV can decrease the average star formation rate (SFR) anywhere from 15% to a factor of twenty. We compare this regulation mechanism to a highly effective model of supernovae (SNe) feedback, which reduces the SFR by a factor of twenty as well. However, SNe feedback destroys most of the gas structure in the process, whereas FUV has minimal impact on the gas structure. In the simulations with FUV radiation, we are also able to create a two phase medium that is a function of the mean FUV intensity the gas receives. Finally, we find that simulations with FUV agree well with observations of nearby spirals on the Kennicutt-Schmidt relation, at least at gas surface densities of 0.2 - 30 M⊙. At surface densities higher than 30 M⊙, we find that FUV is not an effective regulator which is consistent with arguments that SNe or other feedback mechanisms should become the primary regulator. / Thesis / Doctor of Philosophy (PhD)

Made available in DSpace on 2015-03-04T18:50:37Z (GMT). No. of bitstreams: 1 TeseRubem.pdf: 3958802 bytes, checksum: 559985052b43efb925ede65de47826ff (MD5) Previous issue date: 2011-08-26 / This thesis aims at presenting a systematic study of resolution of two-dimensional problems of electromagnetic wave scattering, by conducting and homogeneous or non-homogeneous dielectric media, involving singular sources, large wavenumbers and absorbing boundary conditions, governed by the scalar Helmholtz equation. The resolution of this class of problems through the Finite Element Method, raises several orders of difficulty. The first one comes from the presence of singular sources, having been overcome by adopting a technique of singularity removal. Another one is related to the undefined character of the Helmholtz operator for large wavenumbers, which results in the so-called pollution error of the approximated solution. This error is minimized through a new formulation of Petrov-Galerkin, using quasi optimal weighting functions, called Quasi Optimal Petrov-Galerkin (QOPG) method. The QOPG method, originally developed for homogeneous media, is extended to treating problems with non-homogeneous media as well. Finally, the effects of Bayliss-Turkel absorbing boundary conditions on the quality of the numerical solutions is also evaluated. / Esta tese tem por objetivo apresentar um estudo sistemático da resolução,em duas dimensões espaciais, de problemas de espalhamento de ondas eletromagnéticas por meios condutores e meios dielétricos homogêneos e heterogêneos, envolvendo fontes singulares, números de onda elevados e condições de contorno absorventes, modelados pela equação de Helmholtz. A resolução dessa classe de problemas pelo Método dos Elementos Finitos apresenta dificuldades de várias ordens. A primeira decorre da presença de fontes singulares, tendo sido superada pela adoção de uma técnica de remoção de singularidades. Uma outra decorre do caráter indefinido do operador de Helmholtz para números de onda elevados, o que resulta em um erro de aproximação denominado erro de poluição. A minimização desse erro foi obtida utilizando-se a nova formulação de Petrov-Galerkin, com funções peso quase ótimas, denominada método Quasi Optimal Petrov-Galerkin (QOPG). O método QOPG, originalmente desenvolvido para meios homogêneos, foi adaptado para tratar problemas em meios heterogêneos. Finalmente, consideram-se os efeitos das condições de contorno absorventes de Bayliss-Turkel sobre a qualidade das soluções obtidas.

Esta tese tem por objetivo apresentar um estudo sistemático da resolução,em duas dimensões espaciais, de problemas de espalhamento de ondas eletromagnéticas por meios condutores e meios dielétricos homogêneos e heterogêneos, envolvendo fontes singulares, números de onda elevados e condições de contorno absorventes, modelados pela equação de Helmholtz. A resolução dessa classe de problemas pelo Método dos Elementos Finitos apresenta dificuldades de várias ordens. A primeira decorre da presença de fontes singulares, tendo sido superada pela adoção de uma técnica de remoção de singularidades. Uma outra decorre do caráter indefinido do operador de Helmholtz para números de onda elevados, o que resulta em um erro de aproximação denominado erro de poluição. A minimização desse erro foi obtida utilizando-se a nova formulação de Petrov-Galerkin, com funções peso quase ótimas, denominada método Quasi Optimal Petrov-Galerkin (QOPG). O método QOPG, originalmente desenvolvido para meios homogêneos, foi adaptado para tratar problemas em meios heterogêneos. Finalmente, consideram-se os efeitos das condições de contorno absorventes de Bayliss-Turkel sobre a qualidade das soluções obtidas. / This thesis aims at presenting a systematic study of resolution of two-dimensional problems of electromagnetic wave scattering, by conducting and homogeneous or non-homogeneous dielectric media, involving singular sources, large wavenumbers and absorbing boundary conditions, governed by the scalar Helmholtz equation. The resolution of this class of problems through the Finite Element Method, raises several orders of difficulty. The first one comes from the presence of singular sources, having been overcome by adopting a technique of singularity removal. Another one is related to the undefined character of the Helmholtz operator for large wavenumbers, which results in the so-called pollution error of the approximated solution. This error is minimized through a new formulation of Petrov-Galerkin, using quasi optimal weighting functions, called Quasi Optimal Petrov-Galerkin (QOPG) method. The QOPG method, originally developed for homogeneous media, is extended to treating problems with non-homogeneous media as well. Finally, the effects of Bayliss-Turkel absorbing boundary conditions on the quality of the numerical solutions is also evaluated.

Utrahochenergetische kosmische Strahlung (ultra-high-energy cosmic rays -- UHECR) besteht aus ionisierten Atomkernen mit den höchsten Teilchenergien, die je gemessen wurden. Zwar wurden die Quellen von UHECRs noch nicht eindeutig identifiziert, doch gibt es deutliche Anzeichen, dass sie extragalaktisch sind. Um die Beobachtungen zu interpretieren, wird ein Modell der Wechselwirkungen mit Photofeldern sowohl in der Quelle als auch während der extragalaktischen Propagation benötigt. Bei diesen Wechselwirkungen werden sekundäre Neutrinos erzeugt. Diese Dissertation behandelt Modelle der Quellen von UHECRs und die damit verbundene Produktion von Neutrinos sowohl in den Quellen als auch während der Propagation. Dafür wurde ein neuer Code, PriNCe, für die Propagation von UHECRs entwickelt. Dieser Code wird in einem umfangreichen Parameterscan für ein generisches Quellenmodell angewendet, welches mit dem Spektralindex, der maximalen Rigidität, der kosmologischen Quellenverteilung und der chemischen Komposition als freie Parameter definiert ist. Dabei wird der Einfluss von verschiedenen Photodisintegrations- und Luftschauermodellen auf die erwarteten Eigenschaften der Quellen demonstriert. Der Fluss kosmogenischer Neutrinos, der sich daraus robust vorhersagen lässt, liegt außerhalb der Reichweite aller derzeit geplanten Neutrinodetektoren. GRBs als mögliche Quellen von UHECRs werden im Multi-Collision Internal-Shock Modell simuliert, welches die Abhängigkeit der Strahlungsprozesse von den verschiedenen Dissipationsradien im Plasmajet berücksichtigt. Für dieses Modell wird der Effekt demonstriert, den verschiedene Annahmen über die anfängliche Verteilung des Plasmajets und das hydrodynamische Modell auf die resultierende UHECR- und Neutrinosstrahlung haben. Für den Gammastrahlenblitz GRB170817A, welcher zusammen mit einem Gravitationswellensignal beobachtet wurde, werden Vorhersagen für den Neutrinofluss und ihre Abhängigkeit vom Beobachtungswinkel gemacht. / Ultra-high-energy cosmic rays (UHECRs) are the most energetic particles observed in the Universe. While the astrophysical sources of UHECRs have not yet been uniquely identified, there are strong indications for an extragalactic origin. The interpretation of the observations requires both simulations of UHECR acceleration and energy losses inside the source environment as well as interactions during extragalactic propagation. Due to their extreme energies, UHECR will interact with photons in these environments, producing a flux of secondary neutrinos. This dissertation deals with models of UHECR sources and the accompanying neutrino production in the source environment and during extragalactic propagation. We have developed a new, computationally efficient code, PriNCe, for the extragalactic propagation of UHECR nuclei. The PriNCe code is applied for an extensive parameter scan of a generic source model that is described by the spectral index, the maximal rigidity, the cosmological source evolution and the injected mass composition. In this scan, we demonstrate the impact of different disintegration and air-shower models on the inferred source properties. A prediction for the expected flux of cosmogenic neutrinos is also derived. GRBs are discussed as specific UHECR source candidates in the multi-collision internal-shock model. This model takes the radiation from different radii in the GRB outflow into account. We demonstrate how different assumptions about the initial setup of the jet and the hydrodynamic collision model impact the production of UHECRs and neutrinos. Motivated by the multi-messenger observation of GRB170817A, we discuss the expected neutrino production from this GRB and its dependence on the observation angle. We show that the neutrino flux for this event is at least four orders of magnitude below the detection limit for different geometries of the plasma jet.