Interpreting Gaia data using high-resolution cosmological MHD simulations Rob Grand (MPA) in collaboration with Volker Springel (MPA), Rüdiger Pakmor (MPA), Facundo Gómez (La Serena), Federico Marinacci (Bologna), Christine Simpson (Chicago), Adrian Jenkins (Durham), Carlos Frenk (Durham), Simon White (MPA), Alis Deason (Durham), Vasily Belokurov (Cambridge) A brief history of MW cosmological simulations In the early days, only bulge-dominated galaxies could be produced… Hopkins+14 Scannapieco+11,12 Guedes+11 Aumer+13 …but recently, galaxies have become more like discy, late-type, star-forming spirals why? Strong stellar (AGN) feedback, better resolution, better codes… The Auriga simulations: cosmological “zoom” simulations for the formation of Milky Way mass galaxies (https://wwwmpa.mpa-garching.mpg.de/auriga/) Hi-res central galaxy Low-res (100 Mpc)3 box Hi-res halo region Simulated with gas (AREPO) and galaxy formation model: A large suite (40) of Milky Way-mass systems (Grand+17) Star-forming Range of spiral-bar morphologies Chemical dichotomy [alpha/Fe] Grand+18a [Fe/H] Rotationally supported Rotationally supported Thick & thin discs Thin discs Grand+17 Grand+18b Aurigaia: Gaia DR2 mock catalogues generated from Auriga (Grand+18b) based on the methods of Hunt+ 2015 (SNAPDRAGONS) and Lowing+ 2015 • Population synthesis using PARSEC isochrones and assuming star particle is SSP; • 3D dust extinction: 2D Schlegel + Sharma 3D model (available without extinction map); • Approx. Selection function: V<16 (everywhere); V<20 (|b| > 20 degrees) • Phase space interpolation for “synthetic star” generation; • Gaia-added error convolved with observables; Aurigaia: Gaia DR2 mock catalogues generated from Auriga (Grand+18b) based on the methods of Hunt+ 2015 (SNAPDRAGONS) and Lowing+ 2015 • Population synthesis using PARSEC isochrones and assuming star particle is SSP; • 3D dust extinction: 2D Schlegel + Sharma 3D model (available without extinction map); • Approx. Selection function: V<16 (everywhere); V<20 (|b| > 20 degrees) • Phase space interpolation for “synthetic star” generation; • Gaia-added error convolved with observables; Example application: RR Lyrae stars can trace halo rotation out to ~80 kpc 1 simulation Another simulation Coloured lines = various mocks Gaia’s big discoveries (to name a few) Disequilibrium dynamics Resonant planar dynamics Kawata+18 (Fragkoudi+18, Trick+18, Hunt+18, Sellwood+19…) Antoja+18 (Binney & Schönrich 18, Tian+ 18…) Gaia-Enceladus-Sausage + Stellar streams + substructure inner stellar halo Helmi+18 (Belokurov+18, Di Matteo+19, Ibata+19 (Helmi+17,Koppelman+18, Myeong+18 …) Gallart+19…) The halo has a prominent component with low-rotation/high radial velocity anisotropy (Belokurov+2018; Helmi+2018) Gaia-DR2 data reveals a dominant radially anisotropic component (Sausage, Belokurov+18) with beta~0.9 (Fattahi+2019) (see also Helmi+18, Di Matteo+19, Gallart+19) Origin of the Gaia-sausage according to Auriga (Fattahi+2019) A diverse range of velocity-space features, including a radially-anisotropic component —> we can look for origin in formation histories of this diversity Origin of the Gaia-sausage according to Auriga (Fattahi+2019) Extreme-deconvolution (Bovy+ 2011) - fit 2 or 3 Gaussian components —> derive beta (anisotropy) and fraction of accreted mass in “sausage” component Haloes with a prominent radial component tend to have: • Had earlier accretion histories • Had a peak stellar mass of M~109-1010 Msun • Had been accreted between 6-10 Gyr ago A big Splash (Belokurov… RG 19 subm.) Connection between GES, thick disc and inner halo Gaia DR2 (parallax-corrected a la Schönrich+ 19) cross-matched with spectroscopic data and Sanders+Das 2018 for ages • Evidence for a kicked out metal-rich “Splash” component • LAMOST K-giants illustrate extended distribution Constraining the GES merger time from Gaia If GES merger created splash, ages of counter-rotating stars betray merger time Data Simulations CR metal-rich stars almost all old Same story in Auriga simulations and AuriGaia mock catalogues Belokurov.. RG 19 Constraining the GES merger time from Gaia (and Auriga) If GES merger created splash, ages of counter-rotating stars betray merger time Confirmed: • 80th percentile of metal-rich, CR stars good proxy for GES merger time • Fraction of metal-rich CR stars correlates with GES prog. Mass GES, the splash and the thick disc (Grand in prep) GE merger: dynamical heating and in-situ star formation Stars formed in-situ during merger Bef After Pre-existing stars (or “Splash” population) Splash stars heated and more radially extended than starburst population Impact on chemodynamics of thick disc and halo (Grand in prep) Evolutionary scenario: • Before GES merger, there was a metal-rich proto-disc • GES merger “splashed” the proto-disc —> metal-rich inner halo • GES was gas-rich —> starburst —> “thick disc” created in short time • Thin disc formation starts from cold gas disc after afterwards Inferring MW mass from the high velocity tail (Deason+19): The effects of stellar streams Model: • The high-velocity tail of local accreted stars and the escape velocity radial profile follow a power law 3 parameters to constrain Assumptions: • The distribution function of the system is smooth, i.e. is well-mixed in phase space (Leoneard+Tremaine 90) • The velocities extend all the way up to the escape velocity Inferring MW mass from the high velocity tail (Deason+19): The effects of stellar streams Model input: radii and total velocities (Gaia DR2) of (counter-rotating) stars with radial velocity information Put prior on k, by calibrating to simulated stellar haloes with high velocity anisotropy (Gaia Sausage/Enceladus, Belokurov+18, Helmi+19, Fattahi+19) 240 CR stars with heliocentric Apply Bayesian likelihood distance < 3kpc technique to find best marginalising over poorly constrained The impact of streams on the escape speed measurement (Grand+ 19) 30 haloes; 4 solar positions (R=8 kpc, equidistant azimuth) per halo Mimicking the selection from Deason+19: • Select accreted star particles in each local volume • Take subsamples of 240 stars until star particles used up (i.e. 4 subsamples in a volume 3 kpc spheres containing 1000 star particles) —> 892 local accreted star samples Highly substructured phase space leads to under-/over-estimates vesc=601[605] vesc=650[608] vesc=589[608] vesc=583[606] Volume 1 Volume 2 Volume 3 Volume 4 Velocity distribution sporadically populated with bumps (differently at each solar pos) —> range of position-dependent escape velocities Example of Smooth phase space —> more accurate Escape velocities between positions …but sometimes high-velocity tail truncates below true escape speed Distribution of mass estimates across simulation suite Simulation ID Scatter: Bias: 75% comes from noise and variation of High velocity tail does not reach true dynamical substructure escape speed in most haloes 25% is spherical NFW dark halo assumption MW mass estimate revised to: Summary Cosmological zoom-in simulations are reaching the quality required to help interpret Galactic surveys • Realistic discs, high resolution with resolved stellar substructures • Mock Gaia catalogues Re: Gaia-Enceladus-Sausage (GES) gas-rich merger: • Stellar mass ~ 109 Msun, merger time ~ 10 Gyr • Kicked-up proto-disc stars (metal-rich inner halo) and created thick disc in starburst Stellar streams cause bias and increase uncertainty of MW mass estimate from high velocity tail of halo stars • New MW mass revised to: .