The Gaia-ESO Survey: Matching Chemo-Dynamical Simulations to Observations of the Milky Way

Thompson, B. B.; Few, C. G.; Bergemann, M.; Gibson, B. K.; MacFarlane, B. A.; Serenelli, A.; Gilmore, G.; Randich, S.; Vallenari, A.; Alfaro, E. J.; Bensby, T.; Francois, P.; Korn, A. J.; Bayo, A.; Carraro, G.; Casey, A. R.; Costado, M. T.; Donati, P; Franciosini, E.; Frasca, A.; Hourihane, A.; Jofre, P.; Hill, V.; Heiter, U.; Koposov, S. E.; Lanzafame, A.; Lardo, C.; de Laverny, P.; Lewis, J.; Magrini, L.; Marconi, G.; Masseron, T.; Monaco, L.; Morbidelli, L.; Pancino, E.; Prisinzano, L.; Recio-Blanco, A.; Sacco, G.; Sousa, S. G.; Tautvaisiene, G.; Worley, C. C.; Zaggia, S.
eprint arXiv:1709.01523


The typical methodology for comparing simulated galaxies with observational surveys is usually to apply a spatial selection to the simulation to mimic the region of interest covered by a comparable observational survey sample. In this work we compare this approach with a more sophisticated post-processing in which the observational uncertainties and selection effects (photometric, surface gravity and effective temperature) are taken into account. We compare a `solar neighbourhood analogue' region in a model Milky Way-like galaxy simulated with RAMSES-CH with fourth release Gaia-ESO survey data. We find that a simple spatial cut alone is insufficient and that observational uncertainties must be accounted for in the comparison. This is particularly true when the scale of uncertainty is large compared to the dynamic range of the data, e.g. in our comparison, the [Mg/Fe] distribution is affected much more than the more accurately determined [Fe/H] distribution. Despite clear differences in the underlying distributions of elemental abundances between simulation and observation, incorporating scatter to our simulation results to mimic observational uncertainty produces reasonable agreement. The quite complete nature of the Gaia-ESO survey means that the selection function has minimal impact on the distribution of observed age and metal abundances but this would become increasingly more important for surveys with narrower selection functions.