Uncertainties in Population Synthesis Models: Applications to Local and Distant Galaxies Gustavo Bruzual A. CIDA, Mérida, Venezuela

Motivation: Look at some problems or failures of current generation of population synthesis models and see what is the best we can do to understand them using available tools and data. What we would like to have in the years to come…

My view: There is not a perfect model or set of models. Depending on the application, a set of models built with a particular set of ingredients may be optimal. Most data sets are incomplete and may remain incomplete for a long time Work in collaboration with S. Charlot (soon to be published, CB10).

Typical Isochrone Marigo et al. (2008) Theoretical HRD

Typical Isochrone (detail) Marigo et al. (2008) Theoretical HRD

Stellar libraries: wavelength coverage

Stellar libraries: Teff coverage

IndoUS, Miles, and Stelib [Fe/H] Distribution Complete near Solar (more or less)

Increasing spectral resolution Coelho (< 1Å) IndoUS (~ 1Å) Miles (2.4 Å) Stelib (3 Å) HNGSL (~ 5Å) Pickles (5 Å) Kurucz (20 Å) Padova 94 tracks

Behaviour of line strength indices: Models vs SDSS bright galaxies IndoUS (new)Stelib (BC03)

Caution: Flux calibration problem in IndoUS library. Calibration and stellar parameters under revision by Prugniel and collaborators

STIS NGSL UV coverage NGSL (black) vs IUE (cyan) and Kurucz models (green) Solar metallicity 1 Gyr SSP Chabrier IMF Padova 1994 tracks STIS NGSL is being enlarged and recalibrated by Lindler & Heap and collaborators

STIS NGSL UV coverage NGSL vs Kurucz  models (green) Z = 0.2 x Solar (blue) 0.4 x Solar (cyan) 1.0 x Solar (red) 12 Gyr SSP Chabrier IMF Padova 1994 tracks

Problem 1: Incompleteness of stellar libraries U-UV spectral range. Excess of U flux in population models built using incomplete stellar libraries. Completeness matters

Walcher et al. (2008): VVDS data (grey shading) vs. models (contours) Conclude that the wavelength range from 3300 to 4050 Å is not correctly reproduced by models based mostly in the Miles library, which contains few hot stars.

Walcher et al. (2008): Excess U flux is clearly seen in fit to typical galaxy sed

Improving the U-UV spectral range Tlusty Models: Grid of NLTE plane parallel hydrostatic model atmospheres for: O-stars: Lanz & Hubeny (2003) B-stars: Lanz & Hubeny (2007) High spectral resolution from 54.8 Å to FIR (R = 50,000) 15,000 ≤ Teff ≤ 55,000 K; 0 ≤ Z ≤ 2 x Zo

Improving the U-UV spectral range Martins et al. (2005) Models: Grid of NLTE plane parallel hydrostatic model atmospheres Coverage: 3,000 ≤ Teff ≤ 27,500 K; 0.10 x Zo ≤ Z ≤ 2 x Zo 3000 to 7000 Å (R = 20,000)

Improving the U-UV spectral range UVBLUE and BLURED: Grid of model atmospheres based on Kurucz (1993) model atmospheres: UVBLUE: Rodriguez-Merino et al. (2005) Coverage: 3,000 to 50,000 K 850 to 4700 Å (R = 50,000) -2.0 ≤ [Fe/H] ≤ +0.5 BLUERED: Bertone et al. (2008) 3500 to 7000 Å (R = 500,000)

Walcher (2009): VVDS data (grey shading) vs. models (contours) Major improvement after including Tlusty, Martins et al., and UVBLUE stellar atmosphere models to complement MILES library in SSP modelling (work in progress).

Another look at the problem: Pure Miles (black) Extended Miles (red) In the pure Miles case intermediate Teff stars were represented by hotter stars. It is important to use a stellar library as complete as possible.

Ionizing photons: Depend on Tlusty models For SSP’s of Z = (green) (black) Improvement over BaSeL Atlas values (e.g. HeII)

UV spectral indices Defined by e.g. Fanelli et al., can be computed directly from the UV sed, the same as in the visible range. Z = 0.5 x Zo (blue) 1.0 x Zo (black) 2.5 x Zo (red) Recent work on UV indices: Maraston et al. (2008) Chavez et al. (2009)

Problem 2: Tracks should predict the right number of stars, at least in most relevan evolutionary phases (e.g. TP-AGB): NIR stellar sed’s Evolutionary tracks and the mass of galaxies

Improving the NIR spectral range IRTF library: Infrared Telescope Facility Spectral Library (Cool Stars) Rayner et al. (2009) Stellar Models for C-stars: Aringer et al. (2009) Both represent big improvements over previous data sets.

AGB candidates from SAGE LMC survey (Srinavasan et al. 2009)

AGB candidates from SAGE LMC survey Must take into account effects of dusty envelope and mass loss. Dusty code Poster by González-Lopezlira et al.

AGB candidates from SAGE LMC survey Must take into account effects of dusty envelope and mass loss. Dusty code Poster by González-Lopezlira et al.

Evolutionary tracks and the mass of galaxies Bertelli et al. (2008): Z = 0, , , 0.001, 0.002, 0.004, 0.008, 0.017, 0.040, TP-AGB evolutionary prescriptions by: Marigo & Girardi (2007): calibrated with MC clusters and star counts Bertelli et al (2008): extrapolate results from the Marigo and Girardi prescription to different chemical content of the stellar envelope. Hence they are uncalibrated.

olor Color evolution vs. MC cluster data

TP-AGB stars contribute roughly 60% of K light in the galaxy rest frame in the redshift range from 3 to 8

TP-AGB more numerous, brighter, and redder (240 vs 60 per 1 million Mo cluster)

Inferred mass in B and K

HUDF data CB07 SSP’s τ (V) = 0 t=0.6 – 1.3 Gyr M 11 =0.09–0.25

HUDF data CB07 SSP’s τ (V) = 1 t=0.7 – 1.3 Gyr M 11 = 0.1 – 0.3

HUDF data CB07 SSP’s τ (V) = 3 t=0.7 – 0.9 Gyr M 11 = 0.1 – 0.5

Problem 3: Why different codes produce different results? Show example

All these models are for the same Z = and IMF (Salpeter) LMC clusters Different tracks Different sed’s

Tracks Padova 2008, Different Z’s

Stochastic effects

Large amount of work by many groups on: Alpha-enhanced stellar and population models: Vazdekis and collaborators; Coelho and collaborators; Maraston and collaborators; Thomas and collaborators Variable light/heavy element ratio stellar and population models: Worthey and collaborators Star formation history in large galaxy samples: A legion of really bright and hard workers The role of binaries Vanbeveren, Zhanwen Han Propagation of uncertainties in population synthesis models and derived properties of galaxies: Series of papers by Conroy, Gunn, & White

The future:  The answers to distant galaxy problems will come from understanding nearby stars  We have collected more photons from distant galaxies than from nearby stars  Population synthesis models can get better only if ingredients get better. Most of the uncertainties come from critical stages in the evolutionary tracks or missing spectra of important stellar evolutionary phases.  This can happen both from the observational and the theoretical framework

Basic things that we do not know well enough:  Distance to star clusters needed for calibration of stellar evolution models to a higher precision than at present. These errors translate into errors in the age of galaxies or other stellar populations dated using synthesis models.  Physical properties of more stars distributted all over the HR diagram: mass, Teff, radius, chemical abundance, chemical anomalies…  Blue stragglers, origin and role in stellar evolution and relevance in galaxies as integrated populations  The role of binaries in the evolution of stellar populations of various ages and metallicities, and in different environments (stellar density)

Basic things that we do not know well enough:  Evolutionary tracks or evolutionary prescription for EHB stars. Frequency, lifetimes, dependence on metallicity  The role of mass loss.  Is the IMF universal?  Effects of dust  More complete stellar spectral libraries (HST ?) that fill the current gaps (TP-AGB, EHB, Blue stragglers) in as wide a wavelength range as possible and with a good flux calibration, or  Model stellar atmospheres that can be used instead of the observed spectra mentioned in the previous point, including complete line lists and the right geometry and kinematics

Basic things that we do not know well enough:  Can stellar rotation be neglected (as we always do)?  Up to what extent can multiple MS in “SSPs” be neglected?  Do light/heavy element ratios vary from star to star or galaxy to galaxy in a way we can understand? Are models a la Worthey really necessary?

IndoUS vs STELIB

HNGSL UV coverage HNGSL (black) vs IUE (cyan) and Kurucz models (green) Solar metallicity 12 Gyr SSP Chabrier IMF Padova 1994 tracks

HNGSL UV coverage HNGSL (black) vs IUE (cyan) and Kurucz models (green) Solar metallicity 500 Myr SSP Chabrier IMF Padova 1994 tracks

HNGSL UV coverage HNGSL vs Kurucz models (green) Z = 0.2 x Solar (blue) 0.4 x Solar (cyan) 1.0 x Solar (red) 1 Gyr SSP Chabrier IMF Padova 1994 tracks

HNGSL in the visible range HNGSL (black) vs STELIB (red), Kurucz models (green), and Pickles library (cyan) Solar metallicity 12 Gyr SSP Chabrier IMF Padova 1994 tracks

Flux calibration problem in IndoUS library

Flux calibration problem in IndoUS library