My research has focused on locating and measuring star formation and AGN activity in different environments with interferometric and single-dish radio observations. As my first PhD project, I studied the complex interaction between an intermediate redshift (z 0.3) starburst galaxy and a nearby ( 7 kpc separation) QSO using sub-arcsecond VLA observations. I found new evidence for jet-induced star formation activity in the companion galaxy, making the system a strong candidate for this rare, and potentially important process in the early Universe. In my second paper, I investigated the infrared-radio correlation (IRRC) of spheroid- and disc-dominated galaxies in the COSMOS field out to z 1.5. With 1.4 GHz data and Herschel photometry I found that the redshift evolution reported in recent works is due to an increasing radio excess emission associated with spheroid-dominated galaxies, compared to disc-dominated ones, i.e. the ‘purest' star-forming systems in our sample. I theorize that the extra radio power in spheroid-dominated systems is due to low-level AGN activity, even though these sources were not identified by most commonly-used diagnostics as AGN hosts. This finding will significantly increase the accuracy of future high-redshift radio surveys measuring star formation. In my third project I assembled and analysed the largest-to-date low-z IRRC sample of galaxies. I demonstrated the importance of selection effects influencing IRRC statistics, and carried out an improved IRRC analysis that yielded more accurate measures of the correlation's properties. With rich ancillary data it will provide insight into the physical processes that give rise to the IRRC. Finally, I adopted an MCMC-based model optimization to fit a radiative transfer model to ammonia line spectra of a binary molecular cloud core. I determined the physical structures and the masses of the cores and found they are gravitationally unbound.

Over the last three decades, the far-infrared emission from distant galaxies has been revealed to us. This far-infrared light is emitted by dust clouds heated by UV radiation from young stars. This reveals to us some of the most remarkable and highly star-forming galaxies in the Universe. The Herschel space observatory was able to capture this light. With this thesis I have attempted to get a better understanding of the underlying galaxy population. I have done this by observing the most extreme forms of star formation in the early Universe seen in maps obtained by the SPIRE instrument and using prior information from deep high resolution surveys. In particular I have examined the dependencies of dusty galaxy properties on their environment. I have confirmed that star formation is primarily dependent on both galaxy mass and whether a galaxy lies in the "blue cloud". Environment is the primary influence on the fraction of galaxies lying in the blue cloud and has a minor, but significant, affect on the average star formation rate of star forming galaxies. The highest redshift galaxies directly detected in the Herschel SPIRE maps are very rare, but due to the large area of the HerMES surveys we are able to find a statistical significant sample. With the addition of longer wavelength SCUBA-2 data I further confine the redshift of the dusty galaxies and find that the star formation rates of those sources are extremely high and exceed 1000 M_ a year. The observed number counts of these extremely bright sources have been a problem for galaxy evolution models. I am able to explain the observed number count of red SPIRE sources by adding correlated confusion noise and Gaussian instrumental noise to simulated galaxy catalogues. My results emphasise that it is crucial to correct for noise and selection effects for comparison with simulations. I exploit a novel way of fitting the full SPIRE maps using prior information from deep high resolution surveys, obtained from wavelengths ranging from optical to radio. In doing so I obtain the most accurate values of the cosmic infrared background (CIB) at the SPIRE wavelengths. With these results we have a better indication of which sources are producing the CIB, and therefore the bulk of star formation. My results indicate that future large area surveys like LSST are likely to resolve a substantial fraction of the population responsible for the CIB at 250 μm ≤ λ ≤ 500 μm.