2006MNRAS.373.1013C

Using hydrodynamic simulations of disc-galaxy major mergers, we investigate the star formation history and remnant properties when various parametrizations of a simple stellar feedback model are implemented. The simulations include radiative cooling, a density-dependent star formation recipe and a model for feedback from massive stars. The feedback model stores supernova feedback energy within individual gas particles and dissipates this energy on a time-scale specified by two free parameters; τ_fb, which sets the dissipative time-scale, and n, which sets the effective equation of state in star-forming regions. Via this model, feedback energy can provide pressure support to regions of gas that are thermally cold. Using a self-consistent disc galaxy, modelled after a local Sbc spiral, in both isolated and major-merger simulations, we investigate parametrizations of the feedback model that are selected with respect to the quiescent disc stability. These models produce a range of star formation histories for discs evolved in isolation, or during a major merger, yet all are consistent with the star formation relation found by Kennicutt. We suggest that this result is produced by the adopted recipe for star formation and is not a byproduct of the feedback model. All major mergers produce a population of new stars that is highly centrally concentrated, demonstrating a distinct break in the r^1/4 surface density profile, consistent with previous findings. The half-mass radius and one-dimensional velocity dispersion are affected by the feedback model used. In tests with up to an order of magnitude higher resolution, the star formation history is nearly identical, suggesting that we have achieved a numerically converged star formation history. Finally, we compare our results to those of previous simulations of star formation in disc-galaxy major mergers, addressing the effects of star formation normalization, the version of smoothed particle hydrodynamics (sph) employed and assumptions about the interstellar medium. We conclude by suggesting several methods by which future studies may better constrain feedback models.