One very fruitful class of applications of generating observables from simulations via radiative transfer is to test methods of inferring physical quantities from observations, which must necessarily rely on simplifying assumptions that do not necessarily hold for all galaxies. In Hayward et al. (2014b), we performed such a test. Specifically, we investigated how well the total infrared (IR) emission traces the true star formation rate (SFR). Because the true SFRs of our simulated galaxies are known, unlike those of real galaxies, our technique provide a unique method for directly addressing how well the IR luminosity traces the SFR. This SFR indicator relies on two assumptions: (1) young (<~100-Myr-old) stars dominate the dust heating. This requires other sources of dust heating, e.g. intermediate-age and old stars and AGN, to be sub-dominant. (2) The dust acts as a bolometer, i.e. the emission from young stars is completely absorbed by dust and re-radiated in the IR. If these two assumptions are completely true, then the IR luminosity of a galaxy should be a perfect tracer of its SFR. If assumption (1) does not hold, the IR luminosity will overestimate the SFR, whereas if assumption (2) is violated, it will underestimate the SFR.
In Hayward et al. (2014b), we performed dust radiative transfer on a suite of simulations of isolated and merging galaxies. We then converted the predicted IR luminosities to SFRs using the standard conversion. The results for the isolated galaxy simulations are shown in the figure below. The IR luminosity traces the SFR well for the two higher-mass simulated galaxies. However, for the lower-mass simulated galaxies, a significant fraction of the photons from the young stars escape without being absorbed by dust. Thus, the IR luminosity systematically underestimates the SFRs of these galaxies (the green and red points).
For the galaxy mergers (shown below; the points are connected to illustrate the time evolution), the situation is more complicated: prior to the starburst induced at merger coalescence, the IR luminosity traces the SFR very well. During the starbursts induced at merger coalescence (denoted by the large star symbols), it slightly overestimates the SFR because of non-negligible contributions of intermediate-age stars and AGN to the dust heating. As the star formation is ‘quenched’ (i.e. the instantaneous SFR decreases rapidly) because of a combination of gas consumption and AGN feedback, the overestimate becomes very severe because stars formed during and prior to the starburst are responsible for significantly dust heating even though the instantaneous SFR is very low. Thus, our simulations predict that for quenching and quenched galaxies, the IR luminosity can overestimate the true SFR by multiple orders of magnitude!
