Testing the orbital histories of satellite galaxies of Milky Way-mass hosts using static, axisymmetric potentials against the FIRE cosmological simulations

The Milky Way (MW) hosts an interesting population of lower-mass galaxies, called satellite galaxies, that interact with each other and with their more massive host. In the current paradigm of galaxy formation, galaxies form hierarchically through numerous mergers of such lower-mass galaxies which can leave an imprint on the host galaxy’s stellar kinematics and elemental abundances. For the low-mass galaxies that survive to present-day, understanding the co-evolution of satellite galaxies of the MW requires understanding their full orbit histories across cosmic time. Nearly all works that model satellite orbits assume a gravitational potential for the MW halo that is fixed across time and spherically symmetric. However, given the mass growth of the halo (and galaxy), triaxiality, and substructure, neither of these assumptions is true. Fortunately, simulations today can model the evolution of both the MW potential and low-mass satellite galaxies at high resolution. I explore and test the limitations of static, axisymmetric orbit modeling against the FIRE-2 cosmological simulations of MW-mass systems. Specifically, I fit the gravitational potential of each host halo at z=0 and integrate back in time satellite orbits, comparing to the true orbit histories of the satellites within the cosmological simulations. When a typical surviving satellite fell in (3.4 - 9.7 Gyr ago), the host halo mass and radius were typically 33 - 86 per cent and 26 - 73 per cent of their values today. This cautions how much the MW-mass environment changes while satellite galaxies orbit in the fixed potential models. I derive and compare many satellite orbit properties, such as the most recent and minimum pericenter distances and velocities, the timing of the most recent pericenter, the most recent apocenter, and orbital periods and eccentricities between the model and simulations. The median fractional differences between the model and simulations of recent orbit properties, such as the timing and distance of the most recent pericenter, are smaller (median offset ~ 2-3 percent, 68th percentile width ~40 percent) than properties that occurred farther back in time, such as the minimum pericenter (median offset ~ 7 percent, 68th percentile width > 100 percent). However, for each orbit property the full distribution across our sample shows differences of up to ~75 per cent or more. These results thus highlight the uncertainty when using this commonly applied orbit modeling technique.