3.2. Summary of Results

These results, in particular the general correlation between radio and FIR luminosity, suggest that the radio emission arising in the faint sub-mJy and microJy radio source population is largely associated with massive star formation in distant star forming galaxies. However, a significant fraction of all the sub-mJy sources (~ 30%) are also identified with low-luminosity AGN. The remaining 10-20% of the faint radio source population are assoiciated with either extremely faint (R > 25) optical identifications, or remain unidentified altogether - even in the HDF-N itself (I > 28). Note that these conclusions are dominated by the faintest (and therefore more numerous) microJy radio sources - for example, the AGN fraction increases rapidly at higher (sub-mJy) flux density limits. In addition, simply labeling sources as pure "starbursts" or pure "AGN" is a little misleading; it is quite possible (even likely) that both phenomena co-exist in some of these faint systems. When we label sources in this way, we are only identifying the dominant phenomenon that gives rise to the bulk of the observed radio emission.

Studies of nearby (nuclear) starburst galaxies such as M82 and Arp 220 give us some idea of how the radio emission arises in these star forming galaxies. According to conventional theories, the chief ingredients are supernova (SN) events associated with massive star formation (see Marcaide this lecture series), and in particular, the global acceleration of cosmic ray electrons via shocks associated with these events. The total radio luminosity of a "normal" galaxy is therefore a direct measure of the SN event rate, and in turn, the star formation rate (SFR) of massive stars (e.g. Condon 1992). In this scenario, the tight correlation between the FIR and radio luminosity of star forming galaxies is explained by the FIR emission arising from the absorption and re-radiation (via dust) of the intense uv emission also associated with massive stars. By assuming an Initial Mass Function (e.g. a Salpeter IMF) for the stellar population, and some scaling factors based on local observations of the Milky Way, radio observations provide unbiased estimates of the SFR that are largely unaffected by extinction due to dust. The SFR inferred in this way for M82 and Arp 220 are ~ 10 and 100 M/yr respectively. The levels of radio and FIR emission observed for the more most distant radio sources in the HDF implies much higher star formation rates: ~ 1000 M/yr.