A comparison of hydrodynamic techniques for modelling collisions between main-sequence stars

Abstract

An Eulerian total variation diminishing (TVD) code and a Lagrangian smoothed particle hydrodynamics (SPH) code are used to simulate the off-axis collision of equal-mass main-sequence stars in order to address the question of whether stellar mergers can produce a remnant star where the interior has been replenished with hydrogen due to significant mixing. Each parent main-sequence star is chosen to be found near the turn-off, with hydrogen depleted in the core, and is modelled with a M = 0.8 M ⊙ realistic stellar model and as a n = 3 polytrope. An ideal fluid description with adiabatic index y = 5/3 is used for all hydrodynamic calculations. We found good agreement between the simulations for the polytropic case, with the remnant showing strong, non-local mixing throughout. In the interior quarter of the mass, ∼35 per cent is mixed in from larger radii and on average the remnant is ∼50 per cent fully mixed. For the realistic model, we found less mixing, particularly in the interior and in the SPH simulation. In the inner quarter, ∼20 per cent of the contained mass in the TVD case, but only ∼3 per cent in the SPH one is mixed in from outside. The simulations give consistent results for the overall profile of the merger remnant and the amount of mass-loss, but the differences in mixing suggests that the intrinsic difference between grid and particle based schemes remains a possible artefact. We conclude that both the TVD and SPH schemes can be used equally well for problems that are best suited to their strengths and that care should be taken in interpreting results about fluid mixing.