To understand the impact of star formation models on galaxy evolution, we perform cosmological zoom-in radiation-hydrodynamic simulations of a dwarf dark matter halo with virial mass 1e9 Msun at z=6. Two different star formation models are compared, a model based on a local gravo-thermo-turbulent condition and a model based on a sink particle algorithm. Using idealized tests of collapsing isothermal spheres and giant molecular clouds with different turbulent structures, we determine the optimal accretion radius to be twice the cell size and the resolution required to achieve reasonable convergence in star formation efficiency to 0.3 pc for the sink algorithm. As a first study in this series, we use cosmological zoom-in simulations with different spatial resolutions and find that star formation is more bursty in the runs with the sink algorithm, generating stronger outflows than in the runs with the gravo-thermo-turbulent model. The main reason for the increased burstiness is that the gas accretion rates on the sinks are high enough to form stars on very short timescales, leading to more clustered star formation. As a result, the star-forming clumps are disrupted more quickly in the sink run due to more coherent radiation and supernova feedback. The difference in burstiness between the two star formation models becomes even more pronounced when the supernova explosion energy is artificially increased. Our results suggest that improving the modelling of star formation on small, sub-molecular cloud scales can significantly impact the global properties of simulated galaxies.
The paper presents a new statistical method that enables the use of systematic errors in the maximum-likelihood regression of integer-count Poisson data to a parametric model. The method is primarily aimed at the characterization of the goodness-of-fit statistic in the presence of the over-dispersion that is induced by sources of systematic error, and is based on a quasi-maximum-likelihood method that retains the Poisson distribution of the data. We show that the Poisson deviance, which is the usual goodness-of-fit statistic and that is commonly referred to in astronomy as the Cash statistics, can be easily generalized in the presence of systematic errors, under rather general conditions. The method and the associated statistics are first developed theoretically, and then they are tested with the aid of numerical simulations and further illustrated with real-life data from astronomical observations. The statistical methods presented in this paper are intended as a simple general-purpose framework to include additional sources of uncertainty for the analysis of integer-count data in a variety of practical data analysis situations.