Can we really pick and choose? Benchmarking various selections of Gaia Enceladus/Sausage stars in observations with simulations

Large spectroscopic surveys plus Gaia astrometry have shown us that the inner stellar halo of the Galaxy is dominated by the debris of Gaia Enceladus/Sausage (GES). With the richness of data at hand, there are a myriad of ways these accreted stars have been selected. We investigate these GES selections and their effects on the inferred progenitor properties using data constructed from APOGEE and Gaia. We explore selections made in eccentricity, energy-angular momentum (E-Lz), radial action-angular momentum (Jr-Lz), action diamond, and [Mg/Mn]-[Al/Fe] in the observations, selecting between 144 and 1,279 GES stars with varying contamination from in-situ and other accreted stars. We also use the Auriga cosmological hydrodynamic simulations to benchmark the different GES dynamical selections. Applying the same observational GES cuts to nine Auriga galaxies with a GES, we find that the Jr-Lz method is best for sample purity and the eccentricity method for completeness. Given the average metallicity of GES (-1.28 < [Fe/H] < -1.18), we use the $z=0$ mass-metallicity relationship to find an average $\rm M_{\star}$ of $\sim 4 \times 10^{8}$ $\rm M_{\odot}$. We adopt a similar procedure and derive $\rm M_{\star}$ for the GES-like systems in Auriga and find that the eccentricity method overestimates the true $\rm M_{\star}$ by $\sim2.6\times$ while E-Lz underestimates by $\sim0.7\times$. Lastly, we estimate the total mass of GES to be $\rm 10^{10.5 - 11.1}~M_{\odot}$ using the relationship between the metallicity gradient and the GES-to-in-situ energy ratio. In the end, we cannot just `pick and choose' how we select GES stars, and instead should be motivated by the science question.