Paper detail

Cosmic star formation probed via parametric stack-fitting of known sources to radio imaging

The promise of multi-wavelength astronomy has been tempered by the large disparity in sensitivity and resolution between different wavelength regimes. Here we present a statistical approach which attempts to overcome this by fitting parametric models directly to image data. Specifically, we fit a model for the radio luminosity function (LF) of star-forming galaxies to pixel intensity distributions at 1.4 GHz coincident with near-IR selected sources in COSMOS. Taking a mass-limited sample in redshift bins across the range $0<z<4$ we are able to fit the radio LF with ~0.2 dex precision in the key parameters (e.g. Phi*,L*). Good agreement is seen between our results and those using standard methods at radio and other wavelengths. Integrating our luminosity functions to get the star formation rate density we find that galaxies with a stellar mass greater than $10^{9.5}\,$M$_{\odot}$ contribute at least 50 per cent of cosmic star formation at since $z=4$. The scalability of our approach is empirically estimated, with the precision in LF parameter estimates found to scale with the number of sources in the stack as $\sqrt{N}$. This type of approach will be invaluable in the multi-wavelength analysis of upcoming surveys with the SKA pathfinder facilities; LOFAR, ASKAP and MeerKAT.