京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第18795号 / 理博第4053号 / 新制||理||1583(附属図書館) / 31746 / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)教授 嶺重 慎, 教授 太田 耕司, 准教授 樽家 篤史 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第20908号 / 理博第4360号 / 新制||理||1626(附属図書館) / 京都大学大学院理学研究科物理学・宇宙物理学専攻 / (主査)准教授 樽家 篤史, 教授 佐々木 節, 教授 川合 光 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

No description available.

To Understand how galaxies were assembled across the cosmic time remains one of the most outstanding questions in astronomy. The core of this question is how today's Hubble Sequence, namely the differentiation of galaxy morphology and its correlation to galaxy physical properties, is formed. In this thesis, we investigate the origin of the Hubble Sequence through galaxies at z~2, an epoch when the cosmic star formation activity reaches its peak and the properties of galaxies undergo dramatic transitions. Galaxies at z~2 have two important features that are distinct from nearby galaxies: much higher frequency of clumpy morphology in star-forming systems, and much compacter size. To understand the nature of the two features requires investigations on the sub-structure of galaxies in a multi-wavelength way. In this thesis, we study samples of galaxies that are selected from GOODS and HUDF, where ultra-deep and high-resolution optical and near-infrared images allow us to study the stellar populations of the sub-structures of galaxies at the rest-frame optical bands for the first time, to answer two questions: (1) the nature of kiloparsec-scale clumps in star-forming galaxies at z$\sim$2 and (2) the existence of color gradient and stellar population gradient in passively evolving galaxies at z~2, which may provide clues to the mechanisms of dramatic size evolution of this type of galaxies. We further design a set of color selection criteria to search for dusty star-forming galaxies and passively evolving galaxies at z~3 to explore the question: when today's Hubble Sequence has begun to appear.

Comprehensive and robust analysis of galaxies found throughout cosmic time provides the means to probe the underlying characteristics of our Universe. Coupling observations and theory, spectral energy distribution (SED) fitting provides a method to derive the intrinsic properties of distant galaxies which then aid in defining galaxy populations and constraining current galaxy formation and evolution scenarios. One such population are the sub-millimeter galaxies (SMGs) whose high infrared luminosities -- typically associated with dust-obscured star formation -- and redshift distribution places them as likely key components in galaxy evolution. To fully analyze these systems, however, requires a near complete sampling of the full SED, detailed models that encapsulate the variety of physical processes and sophisticated methods for comparing the data and models. In this dissertation, we present the general propose, Monte Carlo Markov Chain (MCMC) based SED fitting routine SED Analysis Through Markov Chains (SATMC) and the insight we have gained in modeling a sample of AzTEC 1.1mm-detected SMGs. The MCMC engine and Bayesian formalism used in the construction of SATMC offers a unique view at the constraints on model parameter space that are often grossly simplified in traditional SED fitting methods. We first present the motivation behind SATMC and its MCMC algorithm. We also highlight a series of test cases that verify not only its reliability but its versatility to various astrophysical applications, including the field of photometric redshift estimation. We then present the AzTEC SMG sample and preliminary results obtained through counterpart identification, X-ray spectral modeling and SED fitting with SATMC. Finally, we present the latest work in detailed SED analysis of SMGs and how these results influence our understanding of the SMG population.

No description available.

No description available.

This thesis explores the use of machine learning redshift estimation models trained on segmented galactic images. Segmentation of galaxies from the background is accomplished using a flood fill segmentation method which is novel to the field of galactic segmentation. Astronomy datasets are so large due to high volume modern surveys that automated analysis techniques are now required. Redshift is a prime example of an expensive measurement that is a candidate for automation. The Sloan Digital Sky Survey alone imaged more than 1 billion objects in 9 years, but only produced 4 million spectra over more than 20 years. Machine learning is an automation technology that promises to efficiently analyze imaging data alone such that redshift can be estimated with a high degree of accuracy. Ground truth redshift and multi-band galactic images were obtained for 200,000 galaxies from the Sloan Digital Sky Survey. Two model architectures were experimented with, a fully connected artificial neural network, and a convolutional neural network. Experiments were conducted on flood fill parameters, crop sizes, color spaces, and thresholding. We demonstrated that model performance on flood fill segments is higher than on unsegmented images across many crop sizes. The best achieved model performances for artificial neural networks, and convolutional neural networks are median absolute dispersions of 0.024 and 0.031, respectively. / Master of Science / Astronomy has progressed to the point where there is too much data for humans to analyze manually. Every year more data is created. Machine learning is a technology that can sort through all this data. We used machine learning to create a method to predict redshift of galaxies. Redshift is a measure of the object's speed and tells us about the history of the universe. We also created a novel way of separating galaxies in images from the background using flood fill. Machine learning creates a function that connects two datasets. Our first dataset is 200,000 galactic images from the Sloan Digital Sky Survey, and our second dataset is redshift measurements of these galaxies. We connected images to their redshift using a machine learning model. This lets us take images of galaxies and estimate their redshift. Our best models exceeded the performance of some other redshift estimation techniques.

The work contained within this thesis Is made up primarily of two pieces Both address active galactic nuclei And the galaxies that live nearby The obscured fraction of the population Is the topic of one publication And the type-II fraction in the optical regime In chapter four this is the theme I research the vicinity overdensity Around radio galaxies in chapter three, you’ll see I reduce some spectra at redshift one But not all observations in the end got done With the spectra I have I do what I can As if all target observations had actually ran In the end I conclude with results and the theme of research to be done further downstream. The works contained herein addresses two major topics in extragalactic astrophysics, namely the Type-II AGN fraction and the Overdensity-Radio power relation. Quantifying the Type-II AGN fraction has been attempted by many works in many different observational regimes, finding rather contrasting results. Accretion onto supermassive black holes contributes between 5 per cent and 20 per cent of the luminosity of the Universe, and seems to be closely linked to star formation processes. The large uncertainty on this value is due to the ill-determined contribution from obscured accretion, namely the Type-II fraction. In Chapters 3 and 4 I address this issue from a theoretical standpoint in the X-ray regime and an observational standpoint in the optical regime respectively. In Chapter 3 I show how crude X-ray spectroscopy of partially obscured AGN can lead to catastrophic underestimations of the intrinsic X-ray luminosity of these sources. Acting over an entire population, these partial obscurers can produce an obscured AGN fraction which decreases as a function of observed luminosity. The results are consistent with observations in the X-ray vs. IR luminosity of AGN classes. In Chapter 4 I select a statistically significant sample of AGN from an unbiased 250μm galaxy sample. After spectroscopic classification I find the optical Type- II AGN fraction to be consistent across several decades in [OIII] luminosity, a common proxy for intrinsic AGN luminosity. I also investigate the relation of AGN activity to host galaxy mass, as well as star formation activity and star formation history. Probing the environments of protoclusters will help to constrain the models of structure formation in the Universe. Until now, no dataset has been big enough to probe the environments of high redshift radio galaxies at a statistical level; While many believe that the feedback processes of high luminosity radio jets will have a direct impact on star formation in the surrounding medium it has not been tested. In Chapter 2 I investigate this on an statistical level, finding no meaningful correlation between radio galaxy radio power and source overdensity in the vicinities of these sources. In Chapter 5 I discuss the reduction of a 24μm sample at redshift z ∼ 1 for direct comparison with a local 12μm sample. With only a fraction of the target sample being observed, no statistically significant results could be derived, but the objects are spectroscopically classified and spectroscopic redshifts are measured where possible. Correlations in the data set are investigated and the limitations of the sample selection strategy are discussed.

Les supernovae de Type Ia sont devenues un outil essentiel dans la cosmologie observationnelle moderne. En étudiant la relation distance-redshift d'un grand nombre de supernovae, la nature de l'énergie noire peut être contrainte. Les distances au SNe de Type Ia sont néanmoins affectées par l'effet de lentilles gravitationnelles qui pourrait induire des effets systématiques dans les mesures de cosmologie. La plupart des supernovae sont faiblement demagnifiées et une petite fraction sont magnifiées de manière importante du fait de la distribution de masse dans la ligne de visée. Ceci induit naturellement une dispersion supplementaire dans les magnitudes observées. Il existe 2 façons d'estimer l'amplification des SNe Ia. Une première méthode consiste à comparer la luminosité de la supernova, qui est mesuré avec une précision typique de 15% , à la moyenne des luminosités de SNe au même redshift. Une autre estimation peut être obtenue en prédisant l'amplification induit par la densité de matière en avant-plan modelée en se basant sur les mesures de la luminosité des galaxies avec un à priori initial sur la relation de masse-luminosité des galaxies. La corrélation entre ces 2 estimateurs permet d'accorder la relation de masse-luminosité utilisée initialement pour obtenir une mesure indépendante fondée sur la luminosité des SNe Ia. Bien évidemment, cette mesure nécessite dans un premier temps la détection de cette corrélation et cette thèse a été dédiée à la mesure de la corrélation dans l'échantillon de SNLS 3 ans. Les supernovae de Type Ia sont devenues un outil essentiel dans la cosmologie observationnelle moderne. En étudiant la relation distance-redshift d'un grand nombre de supernovae, la nature de l'énergie noire peut être contrainte. Les distances au SNe de Type Ia sont néanmoins affectées par l'effet de lentilles gravitationnelles qui pourrait induire des effets systématiques dans les mesures de cosmologie. La plupart des supernovae sont faiblement demagnifiées et une petite fraction sont magnifiées de manière importante du fait de la distribution de masse dans la ligne de visée. Ceci induit naturellement une dispersion supplementaire dans les magnitudes observées. Il existe 2 façons d'estimer l'amplification des SNe Ia. Une première méthode consiste à comparer la luminosité de la supernova, qui est mesuré avec une précision typique de 15% , à la moyenne des luminosités de SNe au même redshift. Une autre estimation peut être obtenue en prédisant l'amplification induit par la densité de matière en avant-plan modelée en se basant sur les mesures de la luminosité des galaxies avec un à priori initial sur la relation de masse-luminosité des galaxies. La corrélation entre ces 2 estimateurs permet d'accorder la relation de masse-luminosité utilisée initialement pour obtenir une mesure indépendante fondée sur la luminosité des SNe Ia. Bien évidemment, cette mesure nécessite dans un premier temps la détection de cette corrélation et cette thèse a été dédiée à la mesure de la corrélation dans l'échantillon de SNLS 3 ans.