A modified star formation law as a solution to open problems in galaxy evolution

Abstract

In order to reproduce the low mass end of the stellar mass function, most\ncurrent models of galaxy evolution invoke very efficient supernova feedback.\nThis solution seems to suffer from several shortcomings however, like\npredicting too little star formation in low mass galaxies at z=0. In this work,\nwe explore modifications to the star formation (SF) law as an alternative\nsolution to achieve a match to the stellar mass function. This is done by\napplying semi-analytic models based on De Lucia & Blaizot, but with varying SF\nlaws, to the Millennium and Millennium-II simulations, within the formalism\ndeveloped by Neistein & Weinmann. Our best model includes lower SF efficiencies\nthan predicted by the Kennicutt-Schmidt law at low stellar masses, no sharp\nthreshold of cold gas mass for SF, and a SF law that is independent of cosmic\ntime. These simple modifications result in a model that is more successful than\ncurrent standard models in reproducing various properties of galaxies less\nmassive than 10^{10}Msun. The improvements include a good match to the observed\nauto-correlation function of galaxies, an evolution of the stellar mass\nfunction from z=3 to z=0 similar to observations, and a better agreement with\nobserved specific star formation rates. However, our modifications also lead to\na dramatic over-prediction of the cold mass content of galaxies. This shows\nthat finding a successful model may require fine-tuning of both star formation\nand supernovae feedback, as well as improvements on gas cooling, or perhaps the\ninclusion of a yet unknown process which efficiently heats or expels gas at\nhigh redshifts.\n