Author ORCID Identifier

Date Available


Year of Publication


Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation


Arts and Sciences


Physics and Astronomy

First Advisor

Dr. Renbin Yan


Empirical libraries of stellar spectra represent a key ingredient needed for modeling the integrated spectra of stellar populations, such as galaxies, through a process called stellar population synthesis (SPS). In order to make use of such libraries, accurate stellar atmospheric parameter estimates are required. Here, I present a methodology that was developed to build a stellar parameter catalog to accompany the MaNGA stellar library (MaStar), a comprehensive collection of empirical, medium-resolution stellar spectra. This parameter catalog was constructed using a multicomponent χ2 fitting approach to match the MaStar spectra to models generated by interpolating the ATLAS9-based BOSZ models. The total χ2 for a given model is defined as the sum of components constructed to characterize narrow-band features of observed spectra (e.g., absorption lines) and the broadband continuum shape separately. Extinction and systematics due to imperfect flux calibration are taken into account in the fitting of each spectrum. The χ2 distribution for a given region of model space is sampled using a Markov Chain Monte Carlo (MCMC) algorithm, and the resulting sample is used to extract atmospheric parameter estimates (Teff, log g, [Fe/H], and [α/Fe]), their corresponding uncertainties, and direct extinction measurements, given in the form of AV . Two methods are used to extract parameters and uncertainties, the results from which are referred to as the “BestFit” and “Bayesian” parameter sets. The former approach accepts the minimum-χ2 result found by the MCMC and uses the χ2 sample to compute error estimates using a simple likelihood-weighted Mean deviation-squared calculation. The latter approach uses Bayesian inference to compute likelihood-weighted mean parameter estimates and associated errors from the χ2 distribution sampled by the MCMC. The parameter distributions from these data sets are compared with each other, and the BestFit parameter set is deemed more reliable for external use in SPS. Both data sets are evaluated for internal consistency using repeated observations where available, and the BestFit parameter set is further evaluated through comparison with external data sets APOGEE-2 and Gaia DR2, as well as results obtained by other parameter-determination efforts from within the MaStar collaboration. This spectral-fitting exercise reveals possible deficiencies in current theoretical model spectra, illustrating the value of MaStar spectra for helping to improve the models.

Digital Object Identifier (DOI)

Funding Information

This study was supported by the National Science Foundation grant (no. AST-1715898) in 2017.