Does the evolution of the radio luminosity function of star-forming galaxies match that of the star formation rate function?

The assessment of the relationship between radio continuum luminosity and star formation rate (SFR) is of crucial importance to make reliable predictions for the forthcoming ultra-deep radio surveys and to allow a full exploitation of their results to measure the cosmic star formation history. We have addressed this issue by matching recent accurate determinations of the SFR function up to high redshifts with literature estimates of the 1.4 GHz luminosity functions of star-forming galaxies (SFGs). This was done considering two options, proposed in the literature, for the relationship between the synchrotron emission (L-synch), that dominates at 1.4 GHz, and the SFR: a linear relation with a decline of the L-synch/SFR ratio at low luminosities or a mildly non-linear relation at all luminosities. In both cases, we get good agreement with the observed radio luminosity functions but, in the non-linear case, the deviation from linearity must be small. The luminosity function data are consistent with a moderate increase of the L-synch/SFR ratio with increasing redshift, indicated by other data sets, although a constant ratio cannot be ruled out. A stronger indication of such increase is provided by recent deep 1.4-GHz counts, down to mu Jy levels. This is in contradiction with models predicting a decrease of that ratio due to inverse Compton cooling of relativistic electrons at high redshifts. Synchrotron losses appear to dominate up to z similar or equal to 5. We have also updated the Massardi et al. evolutionary model for radio loud AGNs.