Although star formation activity has historically been defined by some absolute quantity such as IR luminosity, it would be preferable to use a relative measure which compares the total SFR to the the size of the system. In this way we can address whether star formation proceeds in an inherently different manner at these luminosities, or if it is a scaled up version of processes taking place at lower luminosities. Blue luminosity is frequently used as a substitute for the stellar mass (e.g. Fig. 1b), although this interpretation is wrought with uncertainties due to the presence of young stars and large amounts of dust. LIR / LK is a more useful measure, although LK is still affected by the presence of red AGB stars.
A popular normalization introduced by millimeter astronomers is the ratio of the IR luminosity to the molecular gas mass (LIR / MH2), where MH2 is estimated from measurements of the CO line. This ratio is termed the "star formation efficiency" (SFE) [20, 23, 24] as it represents in some global sense the number of massive stars formed per giant molecular cloud. Since LIR ~ MSFR , the SFE is inversely related to the gas depletion time, and is thus equivalent to the more classical definition of a starburst (i.e tburst << Ho-1).
Many studies have evaluated the SFE in merging galaxies concluding that ULIR mergers are forming stars up to an order of magnitude more efficiently than normal spirals (see Figure 2 and ). While there is some question as to whether the Galactic conversion factor between CO and H2 applies in merging galaxies [19, 25, 26], the expected variation leads to an overestimate of the molecular gas and therefore underestimates the actual SFEs. It therefore seems hard to escape the conclusion that massive stars are being formed at a higher rate for a given amount of cold gas than in quiescent systems, although the exact level of the enhancement is uncertain.
Figure 2. (a) total IR luminosity vs. derived H2 mass for all objects observed at OVRO. Normal spirals fall below the L/M = 10 line. (b) LIR vs. the molecular gas column density averaged over a 1 kpc diameter region (H2,nuclear) for the 12 objects observed with sufficient resolution. (c) SFE vs. H2,nuclear. From Yun & Hibbard, in preparation.