SIMBAD references

We compare the properties of clouds in simulated M33 galaxies to those observed in the real M33. We apply a friends of friends algorithm and CPROPS to identify clouds, as well as a pixel-by-pixel analysis. We obtain very good agreement between the number of clouds, and maximum mass of clouds. Both are lower than occurs for a Milky Way-type galaxy and thus are a function of the surface density, size, and galactic potential of M33. We reproduce the observed dependence of molecular cloud properties on radius in the simulations, and find this is due to the variation in gas surface density with radius. The cloud spectra also show good agreement between the simulations and observations, but the exact slope and shape of the spectra depend on the algorithm used to find clouds, and the range of cloud masses included when fitting the slope. Properties such as cloud angular momentum, velocity dispersions, and virial relation are also in good agreement between the simulations and observations, but do not necessarily distinguish between simulations of M33 and other galaxy simulations. Our results are not strongly dependent on the level of feedback used here (10 and 20 per cent) although they suggest that 15 per cent feedback efficiency may be optimal. Overall our results suggest that the molecular cloud properties are primarily dependent on the gas and mass surface density, and less dependent on the localized physics such as the details of stellar feedback, or the numerical code used.