Research And Application Of Stellar Population Synthesis

Spectrum contains plentiful information about the properties of galaxy. Finding a way to analyze the spectra of observed galaxies and determine the parameters of large sample of galaxies is crucial for investigating galaxy formation and evolution. With the advent of large-area spectroscopic surveys, galaxy spectral data reached unprecedent-ed numbers, especially the Sloan Digital Sky Survey provide millions of high-quality spectra. With the increase of data availability, the development of fast and automated extraction methods are required. In this thesis, we use blind source separation meth-ods to analyze a large sample of galaxies spectra observed by large-area surveys, and use our spectral fitting method to explore the evolution of ULIRGs.Firstly, several blind source separation (BSS) techniques are introduced, including principal component analysis (PCA), independent component analysis (ICA) and prob-ability independent component analysis (PICA). In particularly, we describe the navie mean field ICA (EL-ICA) and advance mean field ICA (MF-ICA) in PICA. Compar-ing to other BSS methods, MF-ICA is more precise and efficient. Therefore we adopt MF-ICA algorithm to our spectral analysis method.In Chapter3, the evolutionary population synthesis (EPS) model are introduced, and the ingredients of EPS are summarized, including star formation rate, initial mass function, stellar evolution tracks and stellar spectral library. EPS models provide a large sample of simple stellar populations (SSPs), galaxy spectrum fitting and physical parameters determining can be achieved by invert a galaxy spectrum into a combination of several SSPs.In Chapter4, we apply MF-ICA method on galaxy spectral analysis and achieve galaxy spectrum fitting and stellar population analysis. Moreover, we analyze all spec-tral data obtained from SDSS DR8. The MF-ICA algorithm can compress the stellar spectral library into a few Independent Components (ICs), and galaxy spectrum can be reconstructed by these ICs. Comparing to method which decompose a galaxy spectrum into a combination of several simple stellar populations, MF-ICA approach offers a large improvement in the efficiency, for galaxy spectrum fitting and physical parame-ters determining. To check the reliability of this approach, three different methods are used:(1) simulations,(2) comparison with previous work based on a different tech- nique, and (3) analysis of the consistency of results obtained for a sample of galaxies from SDSS. Besides providing excellent fits to observed galaxy spectra, we find that the stellar population parameters, such as stellar contents, star formation histories, s-tarlight reddening and stellar velocity dispersion, can be accurately recovered by our MF-ICA method. To investigate the effects of changes in the spectral resolution and spectral signal-to-noise ratio, we also apply our fitting method to those spectroscopic data from DEEP2, and find it provides excellent fitting.Finally, we apply our spectral fitting method on ULIRGs sample, explore their evolution. We cross-matched Wide-field Infrared Survey Explorer (WISE) sources with the Sloan Digital Sky Survey (SDSS) galaxy spectroscopic catalog and present a large catalog of419ULIRGs. classify them into three subsamples, based on their e-mission line properties:HⅡ-ike ULIRGs, Seyfert2ULIRGs and composite ULIRGS. We apply our new efficient spectral synthesis technique, which based on mean field approach to Bayesian independent component analysis (MF-ICA) method, on the inte-grated spectra of these ULIRGs, and analyze the stellar population properties, including percentage contribution, stellar age and stellar mass, for these three types of ULIRGs, and explore the evolution among them. We find no significant difference between the properties of stellar populations in ULIRGs with or without active galactic nucleus components. Our results suggest that there is possibly no evolutionary link among these three type ULIRGs.