http://arxiv.org/abs/1403.1356
Optimised population synthesis provides an empirical method to extract the relative mix of stellar evolutionary stages and the distribution of atmospheric parameters within unresolved stellar systems, yet a robust validation of this method is still lacking. We here provide a calibration of population synthesis via non-linear bound-constrained optimisation of stellar populations based upon optical spectra of mock stellar systems and observed Galactic Globular Clusters (GGCs). The MILES stellar library is used as a basis for mock spectra as well as templates for the synthesis of deep GGC spectra from Schiavon et al. (2005). Optimised population synthesis applied to mock spectra recovers mean light-weighted stellar atmospheric parameters to within a mean uncertainty of 240 K, 0.04 dex, and 0.03 dex for T_eff, log(g), and [Fe/H], respectively. Decompositions of both mock and GGC spectra confirm the method’s ability to recover the expected mean light-weighted metallicity in dust-free conditions (E[B-V] < 0.15) with uncertainties comparable to evolutionary population synthesis methods. Dustier conditions require either appropriate dust-modelling when fitting to the full spectrum, or fitting only to select spectral features. We derive light-weighted fractions of stellar evolutionary stages from our population synthesis fits to GGCs, yielding on average a combined 25+/-6 per cent from main sequence and turnoff dwarfs, 64+/-7 per cent from subgiant, red giant and asymptotic giant branch stars, and 15+/-7 per cent from horizontal branch stars and blue stragglers. Excellent agreement is found between these fractions and those estimated from deep HST/ACS CMDs. Overall, optimised population synthesis remains a powerful tool for understanding the stellar populations within the integrated light of galaxies and globular clusters.
C. Barber, S. Courteau, J. Roediger, et. al.
Fri, 7 Mar 14
43/47