Model stellar atmospheres are fundamental tools for understanding stellar observations from interferometry, microlensing, eclipsing binaries and planetary transits. However, the calculations also include assumptions, such as the geometry of the model. We use intensity profiles computed for both plane-parallel and spherically symmetric model atmospheres to determine fitting coefficients in the BVRIHK, CoRot and Kepler wavebands for limb darkening using several different fitting laws, for gravity-darkening and for interferometric angular diameter corrections. Comparing predicted variables for each geometry, we find that the spherically symmetric model geometry leads to different predictions for surface gravities log g < 3. In particular, the most commonly used limb-darkening laws produce poor fits to the intensity profiles of spherically symmetric model atmospheres, which indicates the need for more sophisticated laws. Angular diameter corrections for spherically symmetric models range from 0.67 to 1, compared to the much smaller range from 0.95 to 1 for plane-parallel models.

O tópico abordado nesta tese é a perda de massa através de ventos coronais magnetizados em estrelas de baixa massa. Ventos estelares têm sido estudados extensivamente há vários anos, tendo inicialmente como foco o vento solar. Atualmente, sabe-se que o campo magnético é essencial na aceleração e aquecimento dos ventos coronais. Apesar do conhecimento detalhado que temos da estrutura magnética do Sol, pouco se sabe sobre a configuração do campo magnético em outras estrelas. Nesta tese, é investigada a estrutura do campo magnético nas coroas de estrelas do tipo solar na Seqüência Principal e de suas predecessoras na pré Seqüência Principal através de simulações numéricas magneto-hidrodinâmicas tri-dimensionais. Aqui, consideramos de forma auto-consistente a interação entre o vento e o campo magnético e vice-versa. Dessa forma, pela interação entre forças magnéticas e forças do vento, consegue-se determinar a configuração do campo magnético e a estrutura dos ventos coronais. Realizamos um estudo de ventos de estrelas do tipo solar e a dependência dos mesmos com o parâmetro beta do plasma (a razão entre as densidades de energia térmica e magnética). Este é o primeiro estudo a realizar tal análise resolvendo as equações tri-dimensionais da magneto-hidrodinâmica ideal. Em nossas simulações, adotamos um parâmetro de aquecimento descrito por gamma, que é responsável pela aceleração térmica do vento. Então, nós analisamos ventos com intensidades de campo magnético nos pólos no intervalo de B0 = 1 a 20 G e mostramos que a estrutura do vento apresenta características que são similares à do vento coronal do Sol. No estado estacionário, a topologia do campo magnético obtida é similar para todos os casos estudados, apresentando uma configuração do tipo helmet streamer, com zonas de linhas fechadas e abertas de campo magnético co-existindo. Intensidades mais altas de campo levam a ventos mais acelerados e mais quentes. O aumento na intensidade do campo gera também uma zona morta maior no vento, i.e., os loops fechados que previnem que a matéria escape da coroa em latitudes menores que ~45 graus se estendem a maiores distâncias da estrela. Além disso, mostramos também que a força de Lorentz gera naturalmente um vento que é dependente da latitude. Ao aumentar a densidade da coroa mantendo B0 = 20 G, mostramos que o sistema volta a apresentar ventos menos acelerados e mais frios. Para um valor fixo de gamma, mostramos que o parâmetro essencial na determinação do perfil de velocidade do vento é o parâmetro beta calculado na base da coroa. Dessa forma, acredita-se que haja um grupo de ventos magnetizados que apresenta a mesma velocidade terminal independentemente das densidades de energia térmica ou magnética, desde que o parâmetro beta seja o mesmo. No entanto, essa degenerescência pode ser removida ao se comparar outros parâmetros físicos do vento, tal como a taxa de perda de massa. Nós também analisamos a influência do gamma nos nossos resultados e mostramos que ele é importante na determinação da estrutura do vento. Além disso, investigamos ventos magnetizados de estrelas de baixa massa da pré Seqüência Principal. Em particular, analisamos sob quais circunstâncias tais estrelas apresentam estruturas magnéticas alongadas (e.g., helmet streamers, proeminências do tipo slingshot, etc). Focamos especialmente em estrelas do tipo T Tauri fracas, uma vez que o tênue disco de acreção, quando presente ao redor de tais estrelas, não deve causar forte influência na estrutura do vento estelar e nem na do campo magnético coronal. Nós mostramos que o parâmetro beta do plasma é um fator decisivo na configuração do campo magnético do vento estelar. Usando parâmetros iniciais adequados ao que se é observado para tais estrelas, nós mostramos que a configuração do campo magnético pode variar entre uma configuração semelhante à de um dipolo e uma configuração com linhas fortemente colimadas em torno do eixo polar e streamers fechados ao redor do equador (configuração de multi-componentes para o campo magnético). Mostramos que as estruturas alongadas do campo magnético somente estão presentes se o parâmetro beta do plasma na base da coroa é beta0 << 1. Usando nossos modelos magneto-hidrodinâmicos, auto-consistentes, tri-dimensionais, estimamos para ventos de estrelas da pré Seqüência Principal a escala temporal de migração planetária devido a forças de arraste exercidas pelo vento em um planeta tipo hot-Jupiter (i.e., um planeta gigante que orbita muito próximo da estrela). Nosso modelo sugere que os ventos estelares de coroas com multi-componentes de campo magnético não têm influências significativas na migração de hot-Jupiters. / The subject of this thesis is the mass loss of low-mass stars through magnetized coronal winds. Stellar winds have been a topic of extensive research in Astrophysics for a long time, and their first investigations focused on the solar wind. Nowadays, we know that the magnetic field plays a crucial role in the acceleration and heating of coronal winds. Despite of the knowledge of the fine structure of the solar magnetic field, much less information is known regarding the configuration of the magnetic field in other stars. In this thesis, we investigate the structure of the magnetic field in the coronae of solar-like stars and young stars by means of three-dimensional magnetohydrodynamical numerical simulations. We self-consistently take into consideration the interaction of the outflowing wind with the magnetic field and vice versa. Hence, from the interplay between magnetic forces and wind forces, we are able to determine the configuration of the magnetic field and the structure of the coronal winds. We investigate solar-like stellar winds and their dependence on the plasma-beta parameter (the ratio between thermal and magnetic energy densities). This is the first study to perform such analysis solving the fully ideal three-dimensional magnetohydrodynamics equations. We adopt in our simulations a heating parameter described by gamma, which is responsible for the thermal acceleration of the wind. We analyze winds with polar magnetic field intensities ranging from B0 = 1 to 20 G and we show that the wind structure presents characteristics that are similar to the solar coronal wind. The steady-state magnetic field topology for all cases is similar, presenting a configuration of helmet streamer-type, with zones of closed field lines and open field lines coexisting. Higher magnetic field intensities lead to faster and hotter winds. The increase of the field intensity generates a larger ``dead zone\'\' in the wind, i.e., the closed loops that inhibit matter to escape from latitudes lower than 45 degrees extend farther away from the star. The Lorentz force leads naturally to a latitude-dependent wind. We show that by increasing the density and maintaining B0 = 20 G, the system recovers to slower and cooler winds. For a fixed gamma, we show that the key parameter in determining the wind velocity profile is the beta-parameter at the coronal base. Therefore, there is a group of magnetized flows that would present the same terminal velocity despite of its thermal and magnetic energy densities, as long as the plasma-beta parameter is the same. This degeneracy, however, can be removed if we compare other physical parameters of the wind, such as the mass-loss rate. We also analyze the influence of gamma in our results and we show that it is also important in determining the wind structure. We further investigate magnetized stellar winds of low-mass pre-main-sequence stars. In particular we analyze under which circumstances these stars present elongated magnetic features (e.g., helmet streamers, slingshot prominences, etc). We focus on weak-lined T Tauri stars, as the presence of the tenuous accretion disk is not expected to have strong influence on the structure of the stellar wind neither on the coronal magnetic field. We show that the plasma-beta parameter is a decisive factor in defining the magnetic configuration of the stellar wind. Using initial parameters within the observed range for these stars, we show that the coronal magnetic field configuration can vary between a dipole-like configuration and a configuration with strong collimated polar lines and closed streamers at the equator (multicomponent configuration for the magnetic field). We show that elongated magnetic features will only be present if the plasma-beta parameter at the coronal base is beta0 << 1. Using our self-consistent three-dimensional magnetohydrodynamical model, we estimate for the stellar winds of pre-main-sequence stars the timescale of planet migration due to drag forces exerted by the stellar wind on a hot-Jupiter (i.e., on a giant planet that orbits very close to the star). Our model suggests that the stellar wind of these multicomponent coronae are not expected to have significant influence on the migration of hot-Jupiters.

O tópico abordado nesta tese é a perda de massa através de ventos coronais magnetizados em estrelas de baixa massa. Ventos estelares têm sido estudados extensivamente há vários anos, tendo inicialmente como foco o vento solar. Atualmente, sabe-se que o campo magnético é essencial na aceleração e aquecimento dos ventos coronais. Apesar do conhecimento detalhado que temos da estrutura magnética do Sol, pouco se sabe sobre a configuração do campo magnético em outras estrelas. Nesta tese, é investigada a estrutura do campo magnético nas coroas de estrelas do tipo solar na Seqüência Principal e de suas predecessoras na pré Seqüência Principal através de simulações numéricas magneto-hidrodinâmicas tri-dimensionais. Aqui, consideramos de forma auto-consistente a interação entre o vento e o campo magnético e vice-versa. Dessa forma, pela interação entre forças magnéticas e forças do vento, consegue-se determinar a configuração do campo magnético e a estrutura dos ventos coronais. Realizamos um estudo de ventos de estrelas do tipo solar e a dependência dos mesmos com o parâmetro beta do plasma (a razão entre as densidades de energia térmica e magnética). Este é o primeiro estudo a realizar tal análise resolvendo as equações tri-dimensionais da magneto-hidrodinâmica ideal. Em nossas simulações, adotamos um parâmetro de aquecimento descrito por gamma, que é responsável pela aceleração térmica do vento. Então, nós analisamos ventos com intensidades de campo magnético nos pólos no intervalo de B0 = 1 a 20 G e mostramos que a estrutura do vento apresenta características que são similares à do vento coronal do Sol. No estado estacionário, a topologia do campo magnético obtida é similar para todos os casos estudados, apresentando uma configuração do tipo helmet streamer, com zonas de linhas fechadas e abertas de campo magnético co-existindo. Intensidades mais altas de campo levam a ventos mais acelerados e mais quentes. O aumento na intensidade do campo gera também uma zona morta maior no vento, i.e., os loops fechados que previnem que a matéria escape da coroa em latitudes menores que ~45 graus se estendem a maiores distâncias da estrela. Além disso, mostramos também que a força de Lorentz gera naturalmente um vento que é dependente da latitude. Ao aumentar a densidade da coroa mantendo B0 = 20 G, mostramos que o sistema volta a apresentar ventos menos acelerados e mais frios. Para um valor fixo de gamma, mostramos que o parâmetro essencial na determinação do perfil de velocidade do vento é o parâmetro beta calculado na base da coroa. Dessa forma, acredita-se que haja um grupo de ventos magnetizados que apresenta a mesma velocidade terminal independentemente das densidades de energia térmica ou magnética, desde que o parâmetro beta seja o mesmo. No entanto, essa degenerescência pode ser removida ao se comparar outros parâmetros físicos do vento, tal como a taxa de perda de massa. Nós também analisamos a influência do gamma nos nossos resultados e mostramos que ele é importante na determinação da estrutura do vento. Além disso, investigamos ventos magnetizados de estrelas de baixa massa da pré Seqüência Principal. Em particular, analisamos sob quais circunstâncias tais estrelas apresentam estruturas magnéticas alongadas (e.g., helmet streamers, proeminências do tipo slingshot, etc). Focamos especialmente em estrelas do tipo T Tauri fracas, uma vez que o tênue disco de acreção, quando presente ao redor de tais estrelas, não deve causar forte influência na estrutura do vento estelar e nem na do campo magnético coronal. Nós mostramos que o parâmetro beta do plasma é um fator decisivo na configuração do campo magnético do vento estelar. Usando parâmetros iniciais adequados ao que se é observado para tais estrelas, nós mostramos que a configuração do campo magnético pode variar entre uma configuração semelhante à de um dipolo e uma configuração com linhas fortemente colimadas em torno do eixo polar e streamers fechados ao redor do equador (configuração de multi-componentes para o campo magnético). Mostramos que as estruturas alongadas do campo magnético somente estão presentes se o parâmetro beta do plasma na base da coroa é beta0 << 1. Usando nossos modelos magneto-hidrodinâmicos, auto-consistentes, tri-dimensionais, estimamos para ventos de estrelas da pré Seqüência Principal a escala temporal de migração planetária devido a forças de arraste exercidas pelo vento em um planeta tipo hot-Jupiter (i.e., um planeta gigante que orbita muito próximo da estrela). Nosso modelo sugere que os ventos estelares de coroas com multi-componentes de campo magnético não têm influências significativas na migração de hot-Jupiters. / The subject of this thesis is the mass loss of low-mass stars through magnetized coronal winds. Stellar winds have been a topic of extensive research in Astrophysics for a long time, and their first investigations focused on the solar wind. Nowadays, we know that the magnetic field plays a crucial role in the acceleration and heating of coronal winds. Despite of the knowledge of the fine structure of the solar magnetic field, much less information is known regarding the configuration of the magnetic field in other stars. In this thesis, we investigate the structure of the magnetic field in the coronae of solar-like stars and young stars by means of three-dimensional magnetohydrodynamical numerical simulations. We self-consistently take into consideration the interaction of the outflowing wind with the magnetic field and vice versa. Hence, from the interplay between magnetic forces and wind forces, we are able to determine the configuration of the magnetic field and the structure of the coronal winds. We investigate solar-like stellar winds and their dependence on the plasma-beta parameter (the ratio between thermal and magnetic energy densities). This is the first study to perform such analysis solving the fully ideal three-dimensional magnetohydrodynamics equations. We adopt in our simulations a heating parameter described by gamma, which is responsible for the thermal acceleration of the wind. We analyze winds with polar magnetic field intensities ranging from B0 = 1 to 20 G and we show that the wind structure presents characteristics that are similar to the solar coronal wind. The steady-state magnetic field topology for all cases is similar, presenting a configuration of helmet streamer-type, with zones of closed field lines and open field lines coexisting. Higher magnetic field intensities lead to faster and hotter winds. The increase of the field intensity generates a larger ``dead zone\'\' in the wind, i.e., the closed loops that inhibit matter to escape from latitudes lower than 45 degrees extend farther away from the star. The Lorentz force leads naturally to a latitude-dependent wind. We show that by increasing the density and maintaining B0 = 20 G, the system recovers to slower and cooler winds. For a fixed gamma, we show that the key parameter in determining the wind velocity profile is the beta-parameter at the coronal base. Therefore, there is a group of magnetized flows that would present the same terminal velocity despite of its thermal and magnetic energy densities, as long as the plasma-beta parameter is the same. This degeneracy, however, can be removed if we compare other physical parameters of the wind, such as the mass-loss rate. We also analyze the influence of gamma in our results and we show that it is also important in determining the wind structure. We further investigate magnetized stellar winds of low-mass pre-main-sequence stars. In particular we analyze under which circumstances these stars present elongated magnetic features (e.g., helmet streamers, slingshot prominences, etc). We focus on weak-lined T Tauri stars, as the presence of the tenuous accretion disk is not expected to have strong influence on the structure of the stellar wind neither on the coronal magnetic field. We show that the plasma-beta parameter is a decisive factor in defining the magnetic configuration of the stellar wind. Using initial parameters within the observed range for these stars, we show that the coronal magnetic field configuration can vary between a dipole-like configuration and a configuration with strong collimated polar lines and closed streamers at the equator (multicomponent configuration for the magnetic field). We show that elongated magnetic features will only be present if the plasma-beta parameter at the coronal base is beta0 << 1. Using our self-consistent three-dimensional magnetohydrodynamical model, we estimate for the stellar winds of pre-main-sequence stars the timescale of planet migration due to drag forces exerted by the stellar wind on a hot-Jupiter (i.e., on a giant planet that orbits very close to the star). Our model suggests that the stellar wind of these multicomponent coronae are not expected to have significant influence on the migration of hot-Jupiters.

Classical Cepheids are key probes of both stellar astrophysics and cosmology as standard candles and pulsating variable stars. It is important to understand Cepheids in unprecedented detail in preparation for upcoming Gaia, James Webb Space Telescope (JWST) and extremely-large telescope observations. Cepheid eclipsing binary stars are ideal tools for achieving this goal, however there are currently only three known systems. One of those systems, OGLE-LMC-CEP1812, raises new questions about the evolution of classical Cepheids because of an apparent age discrepancy between the Cepheid and its red giant companion. We show that the Cepheid component is actually the product of a stellar merger of two main sequence stars that has since evolved across the Hertzsprung gap of the HR diagram. This post-merger product appears younger than the companion, hence the apparent age discrepancy is resolved. We discuss this idea and consequences for understanding Cepheid evolution.

<p>The information about the chemical compositions of stars is encoded in their spectra. Accurate determinations of these compositions are crucial for our understanding of stellar nucleosynthesis and Galactic chemical evolution. The determination of elemental abundances in stars requires models for the stellar atmospheres and the processes of line formation. Nearly all spectroscopic analyses of late-type stars carried out today are based on one-dimensional (1D), hydrostatic model atmospheres and on the assumption of local thermodynamic equilibrium (LTE). This approach can lead to large systematic errors in the predicted stellar atmospheric structures and line-strengths, and, hence, in the derived stellar abundances. In this thesis, examples of departures from LTE and from hydrostatic equilibrium are explored. The effects of background line opacities (line-blocking) due to atomic lines on the statistical equilibrium of Fe are investigated in late-type stars. Accounting for this line opacity is important at solar metallicity, where line-blocking significantly reduces the rates of radiatively induced ionizations of Fe. On the contrary, the effects of line-blocking in metal-poor stars are insignificant. In metal-poor stars, the dominant uncertainty in the statistical equilibrium of Fe is the treatment of inelastic H+Fe collisions. Substantial departures of Fe abundances from LTE are found at low metallicities: about 0.3 dex with efficient H+Fe collisions and about 0.5 dex without. The impact of three-dimensional (3D) hydrodynamical model atmospheres on line formation in red giant stars is also investigated. Inhomogeneities and correlated velocity fields in 3D models and differences between the mean 3D stratifications and corresponding 1D model atmospheres can significantly affect the predicted line strengths and derived abundances, in particular at very low metallicities. In LTE, the differences between 3D and 1D abundances of C, N, and O derived from CH, NH, and OH weak low-excitation lines are in the range -0.5 dex to -1.0 dex at [Fe/H]=-3. Large negative corrections (about -0.8 dex) are also found in LTE for weak low-excitation neutral Fe lines. We also investigate the impact of 3D hydrodynamical model stellar atmospheres on the determination of elemental abundances in the carbon-rich, hyper iron-poor stars HE 0107-5240 and HE 1327-2326. The lower temperatures of the line-forming regions of the 3D models compared with 1D models cause changes in the predicted spectral line strengths. In particular we find the 3D abundances of C, N, and O to be lower by about -0.8 dex (or more) than estimated from a 1D analysis. The 3D abundance of Fe is decreased but only by -0.2 dex. Departures from LTE for Fe might actually be very large for these stars and dominate over the effects due to granulation.</p>

Stellare Magnetfelder spielen eine wichtige Rolle bei der Entstehung und Entwicklung von Sternen. Leider entziehen sie sich aber, aufgrund ihrer großen Entfernung zur Erde, einer direkten Beobachtung. Dies gilt zumindest für derzeitige und in naher Zukunft zur Verfügung stehende Instrumente. Um aber beispielsweise zu verstehen, ob Magnetfelder durch einen Dynamoprozess generiert werden oder Überbleibsel der Sternentstehung sind, ist es zwingend erforderlich, die Oberflächenstruktur und die zeitliche Entwicklung von stellaren Feldern zu untersuchen. Glücklicherweise haben wir mit der Dopplerverschiebung sowie der Polarisation von Licht Mittel zur Verfügung, um indirekt die Magnetfeldtopologie entfernter Sternen zu rekonstruieren, wenn auch nur die schnell rotierender. Die auf den beiden genannten Effekten basierende Rekonstruktionsmethode ist unter dem Namen Zeeman-Doppler Imaging (ZDI) bekannt. Sie stellt eine leistungsfähige Methode dar, um aus rotationsverbreiterten Stokes Profilen schnell rotierender Sterne Oberflächenkartierungen der Temperatur und Magnetfeldverteilung zu erstellen. Durch das ZDI konnten in den vergangenen Jahren die Magnetfeldverteilungen zahlreicher Sterne rekonstruiert werden. Diese Methode stellt allerdings sehr hohe Anforderungen sowohl an die Instrumentierung als auch an die Rechenleistung und ist deshalb häufig mit zahlreichen Annahmen und Näherungen verbunden. Ziel dieser Arbeit war es, Methoden für ein ZDI zu entwickeln, das darauf ausgelegt ist, zeitaufgelöste spektropolarimetrische Daten von aktiven späten Sternen zu invertieren. Es sollte also insbesondere den komplexen und lokalen Magnetfeldstrukturen dieser Sterne Rechnung getragen werden. Um die Orientierung und Stärke solcher Felder zuverlässig rekonstruieren zu können, sollte die Inversion im Stande sein, alle vier Stokes-Komponenten einzubeziehen. Ferner war vorgesehen auf vollständigen polarisierten Strahlungstransportmodellierungen aufzubauen. Dies ermöglicht eine simultane und selbstkonsistente Temperatur- und Magnetfeld-Inversion, die damit dem komplexen Zusammenspiel zwischen Temperatur und Magnetfeld gerecht wird. Schließlich sollte die Anwendung eines neu zu entwickelnden ZDI Programms auf Stokes I und V Beobachtungen von II Pegasi (kurz: II Peg) erste Magnefeldkarten dieses sehr aktiven Sterns liefern. Um den hohen Rechenaufwand, der mit der Inversionsmethode einhergeht, besser bewältigen zu können, wurde zunächst eine schnelle Approximationsmethode für den polarisierten Strahlungstransport entwickelt. Sie basiert auf einer Hauptkomponentenanalyse (PCA) sowie auf künstlichen Neuronalen Netzen. Letztere approximieren den funktionalen Zusammenhang zwischen atmosphärischen Parametern und den zugehörigen lokalen Stokes Profilen. Inverse Probleme sind potentiell schlecht gestellt und erfordern in der Regel eine Regularisierung. Der entwickelte Ansatz verwendet eine lokale Entropie, die auf die Besonderheiten bei der Rekonstruktion lokalisierter Magnetfeder eingeht. Ein weiterer neuartiger Ansatz befasst sich mit der Rauschreduktion polarimetrischer Beobachtungsdaten. Er macht sich die Hauptkomponentenanalyse zu Nutze, um mit Hilfe einer Vielzahl beobachteter Spektrallinien, einzelne Linien mit drastisch vergrößertem Signal-zu-Rausch-Verhältnis wieder zu geben. Diese Methode hat gegenüber anderen Multi-Spektrallinien-Verfahren den Vorteil, nach wie vor eine Inversion auf der Basis einzelner Spektrallinien durchführen zu können. Schließlich wurde das Inversionsprogramm iMap entwickelt, das die zuvor genannten Methoden implementiert. Detaillierte Testrechnungen demonstrieren die Funktionsfähigkeit und Genauigkeit der schnellen Synthese-Methode und weisen einen Zeitgewinn von nahezu drei Größenordnungen gegenüber der konventionellen Strahlungstransportberechnung auf. Desweiteren untersuchen wir den Einfluss der verschiedenen Stokes Komponenten (IV bzw. IVQU) auf die Zuverlässigkeit, ein bekanntes Magnetfeld zu rekonstruieren. Damit belegen wir die Zuverlässigkeit unseres Inversionsprogrammes und zeigen darüber hinaus auch Einschränkungen von Magnetfeldinversionen im allgemeinen auf. Eine erste Inversion von Stokes I und V Profilen von II Peg liefert zum ersten Mal für diesen Stern simultan Temperatur- und Magnetfeldverteilungen. / Stellar magnetic fields, as a crucial component of star formation and evolution, evade direct observation at least with current and near future instruments. However investigating whether magnetic fields are generated by a dynamo process or represent relics from the formation process, or whether they show a behavior similar to the sun or something very different, it is essential to investigate their structure and temporal evolution. Fortunately nature provides us with the possibility to indirectly observe surface topologies on distant stars by means of Doppler shift and polarization of light, though not without its challenges. Based on the mentioned effects, the so called Zeeman-Doppler Imaging technique is a powerful method to retrieve magnetic fields from rapid rotating stars based on measurements of spectropolarimetric observations in terms of Stokes profiles. In recent years, a large number of stellar magnetic field distributions could be reconstructed by Zeeman-Doppler Imaging (ZDI). However, the implementation of this method often relies on many approximations because, as an inversion method, it entails enormous computational requirements. The aim of this thesis is to develop methods for a ZDI, designed to invert time-resolved spectropolarimetric data of active late type stars, and to account for the expected complex and small scale magnetic fields on these stars. In order to reliably reconstruct the detailed field orientation and strength, the inversion method is employed to be able to use of all four Stokes components. Furthermore it is based on fully polarized radiative transfer calculations to account for the intricate interplay between temperature and magnetic field. Finally, the application of a newly developed ZDI code to Stokes I and V observations of II Pegasi (short: II Peg) was supposed to deliver the first magnetic surface maps for this highly active star. To accomplish the high computational burden of a radiative transfer based ZDI, we developed a novel approximation method to speed up the inversion process. It is based on Principal Component Analysis and Artificial Neural Networks. The latter approximate the functional mapping between atmospheric parameters and the corresponding local Stokes profiles. Inverse problems, as we are dealing with, are potentially ill-posed and require a regularization method. We propose a new regularization scheme, which implements a local entropy function that accounts for the peculiarities of the reconstruction of localized magnetic fields. To deal with the relatively large noise that is always present in polarimetric data, we developed a multi-line denoising technique based on Principal Component Analysis. In contrast to other multi-line techniques that extract from a large number of spectral lines a sort of mean profile, this method allows to extract individual spectral lines and thus allows for an inversion on the basis of specific lines. All these methods are incorporated in our newly developed ZDI code iMap, which is based on a conjugated gradient method. An in depth validation of our new synthesis method demonstrates the reliability and accuracy of this approach as well as a gain in computation time by almost three orders of magnitude relative to the conventional radiative transfer calculations. We investigated the influence of the different Stokes components (IV / IVQU) on the ability to reconstruct a known synthetic field configuration. In doing so we validate the capability of our inversion code, and we also assess limitations of magnetic field inversions in general. In a first application to II Peg, a K2 IV subgiant, we derived temperature and magnetic field surface distributions from spectropolarimetric data obtained in 2004 and 2007. It gives for the first time simultaneously the temporal evolution of the surface temperature and magnetic field distribution on II Peg.

The information about the chemical compositions of stars is encoded in their spectra. Accurate determinations of these compositions are crucial for our understanding of stellar nucleosynthesis and Galactic chemical evolution. The determination of elemental abundances in stars requires models for the stellar atmospheres and the processes of line formation. Nearly all spectroscopic analyses of late-type stars carried out today are based on one-dimensional (1D), hydrostatic model atmospheres and on the assumption of local thermodynamic equilibrium (LTE). This approach can lead to large systematic errors in the predicted stellar atmospheric structures and line-strengths, and, hence, in the derived stellar abundances. In this thesis, examples of departures from LTE and from hydrostatic equilibrium are explored. The effects of background line opacities (line-blocking) due to atomic lines on the statistical equilibrium of Fe are investigated in late-type stars. Accounting for this line opacity is important at solar metallicity, where line-blocking significantly reduces the rates of radiatively induced ionizations of Fe. On the contrary, the effects of line-blocking in metal-poor stars are insignificant. In metal-poor stars, the dominant uncertainty in the statistical equilibrium of Fe is the treatment of inelastic H+Fe collisions. Substantial departures of Fe abundances from LTE are found at low metallicities: about 0.3 dex with efficient H+Fe collisions and about 0.5 dex without. The impact of three-dimensional (3D) hydrodynamical model atmospheres on line formation in red giant stars is also investigated. Inhomogeneities and correlated velocity fields in 3D models and differences between the mean 3D stratifications and corresponding 1D model atmospheres can significantly affect the predicted line strengths and derived abundances, in particular at very low metallicities. In LTE, the differences between 3D and 1D abundances of C, N, and O derived from CH, NH, and OH weak low-excitation lines are in the range -0.5 dex to -1.0 dex at [Fe/H]=-3. Large negative corrections (about -0.8 dex) are also found in LTE for weak low-excitation neutral Fe lines. We also investigate the impact of 3D hydrodynamical model stellar atmospheres on the determination of elemental abundances in the carbon-rich, hyper iron-poor stars HE 0107-5240 and HE 1327-2326. The lower temperatures of the line-forming regions of the 3D models compared with 1D models cause changes in the predicted spectral line strengths. In particular we find the 3D abundances of C, N, and O to be lower by about -0.8 dex (or more) than estimated from a 1D analysis. The 3D abundance of Fe is decreased but only by -0.2 dex. Departures from LTE for Fe might actually be very large for these stars and dominate over the effects due to granulation.

Mass loss, in the form of slow stellar winds, is a decisive factor for the evolution of cool luminous giants, eventually turning them into white dwarfs. These dense outflows are also a key factor in the enrichment of the interstellar medium with newly produced elements from the interior of these stars. There are strong indications that these winds are accelerated by radiation pressure on dust grains, but the actual grain species responsible for driving the outflows in M-type Asymptotic Giant Branch stars are still a matter of debate. Observations of dust features in the circumstellar environment of these stars suggest that magnesium-iron silicates are possible wind-drivers. However, the optical properties of these silicate grains are strongly influenced by the Fe-content. Fe-bearing condensates heat up strongly when interacting with the radiation field and therefore cannot form close enough to the star to trigger outflows. Fe-free condensates, on the other hand, have a low absorption cross-section at near-IR wavelengths where AGB stars emit most of their flux.  To solve this conundrum, it has been suggested that winds of M-type AGB stars may be driven by photon scattering on Fe-free silicate grains with sizes comparable to the wavelength of the flux maximum, rather than by true absorption. In this thesis we investigate dynamical models of M-type AGB stars, using Fe-free silicates as the wind-driving dust species. According to our findings these models produce both dynamic and photometric properties consistent with observations. Especially noteworthy are the large photometric variations in the visual band during a pulsation cycle, seen both in the observed and synthetic fluxes. A closer examination of the models reveals that these variations are caused by changes in the molecular layers, and not by changes in the dust. This is a strong indication that stellar winds of M-type AGB stars are driven by dust materials that are very transparent in the visual and near-infrared wavelength regions, otherwise these molecular effects would not be visible.

Dans cette thèse, j'analyse des observations spectropolarimétriques d'étoiles froides évoluées dans tous les paramètres de Stokes ($I$, $Q$, $U$ et $V$).L'étude des spectres polarisés circulairement de trois étoiles supergéantes rouges (RSG) m'a permis de détecter un faible champ magnétique (de l'ordre de 1 Gauss) à leur surface ainsi que de mettre en évidence l'influence de la forte polarisation linéaire de ces étoiles sur la mesure du champ magnétique.Cette forte polarisation linéaire est d'origine non magnétique et est reliée à la polarisation du continu dont l'observabilité résulte de la présence de structures brisants la symétrie cylindrique du disque stellaire.Après une analyse en profondeur de la polarisation du continu des étoiles RSG, j'utilise un modèle analytique spectropolarimétrique permettant de remonter à la position sur le disque de ces structures, en bon accord avec des observations interférométriques.Les étoiles variables pulsantes montrent elles aussi des signaux polarisés linéairement intenses dont l'origine est la polarisation intrinsèque (où de résonance) associée à certaines raies spectrales.Pour ces étoiles, les gradients de vitesses, introduits par les ondes de chocs se propageants dans leur atmosphère, amplifient la polarisation intrinsèque des raies, un mécanisme d'éclaircissement Doppler, bien connu dans le cas du Soleil, qui offre de nouvelles perspectives dans l'étude de ces objets. / In this thesis work, full Stokes ($I$, $Q$, $U$ et $V$) spectropolarimetric observations of cool evolved stars have been analysed.I have measured a weak magnetic field at the surface of three well known red supergiant stars. Furthermore, I have shown that the strong level of linear polarisation, measured for these stars, can lead to some ambiguity in the measurement of magnetic fields.This strong, unmagnetic, linear polarisation originate from depolarisation of the continuum which, in turn, is only detectable because of symmetry-breaking effects on the stellar disk.I study in some details the continuum polarisation of RSG stars and I present a way of mapping symmetry-breaking strucutres through the mean of a spectropolarimetric model. Good agreement with interferometric observations is reached.From their side, pulsating variable stars also show strong linearly polarised features in their spectra. These features are indeed due to resonant scattering polarisation associated to metallic lines, a phenomenon reminiscent of the second solar spectrum.For these stars, the velocity gradients, pertaining to the shock waves propagating through their atmosphere, enhance the intrinsic polarisation of the lines. This Doppler brightening effect, well known in the solar case, offer a new innovative method for the analysis of these stars.

In this paper, we present the analysis of incoherent non-thermal radio emission from a sample of hot magnetic stars, ranging from early-B to early-A spectral type. Spanning a wide range of stellar parameters and wind properties, these stars display a commonality in their radio emission which presents new challenges to the wind scenario as originally conceived. It was thought that relativistic electrons, responsible for the radio emission, originate in current sheets formed, where the wind opens the magnetic field lines. However, the true mass-loss rates from the cooler stars are too small to explain the observed non-thermal broad-band radio spectra. Instead, we suggest the existence of a radiation belt located inside the inner magnetosphere, similar to that of Jupiter. Such a structure explains the overall indifference of the broad-band radio emissions on wind mass-loss rates. Further, correlating the radio luminosities from a larger sample of magnetic stars with their stellar parameters, the combined roles of rotation and magnetic properties have been empirically determined. Finally, our sample of early-type magnetic stars suggests a scaling relationship between the non-thermal radio luminosity and the electric voltage induced by the magnetosphere's co-rotation, which appears to hold for a broader range of stellar types with dipole-dominated magnetospheres (like the cases of the planet Jupiter and the ultracool dwarf stars and brown dwarfs). We conclude that well-ordered and stable rotating magnetospheres share a common physical mechanism for supporting the generation of non-thermal electrons.