Appendix A, Figs. 3-5 are only available in electronic form at http://www.aanda.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

I.e., where the source extent was smaller than the largest angular scale which can be mapped by the VLA in the given configuration and frequency band.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 22.5 GHz

Note, however, that a steeper power-law fit to the high frequency radio data would not fit the $1.4~\rm GHz$ and $408~\rm MHz$ data points.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... $10^4~\rm yr$

Type 1b and type II SNe are thought to be the source of the synchrotron emission in ULIRGs, and occur by the core-collapse of stars more massive than $8~M_{\odot}$ . The difference in lifetime between a star of $8~M_{\odot}$ and a star of $100~M_{\odot}$ is $\sim$ 5 $\times$ $10^7~\rm yr$ . Therefore, for a normal initial mass function, even if the stars formed instantaneously, the SN rate cannot show large variations on timescales as short as $10^4~\rm yr$ .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.