Abstract
The Q/U-Imaging ExperimenT (QUIET) is a ground-based coherent polarimeter experiment, primarily targeting the detection of low-l B-mode polarization in the CMB. Such modes are a signature of primordial tensor perturbations, also known as gravitational waves. The existence of tensor perturbations is as yet unproven, but uniquely predicted by theories of cosmological inflation. The B-modes are thus frequently referred to as ``the smoking gun of inflation'', and efforts towards their detection are many and fierce. The first season of QUIET observations, split between the Q- and W- frequency bands, was carried out in 2008-2010. The analysis of the Q-band data has been completed, and was published in late 2010.
QUIET uses two independent data analysis pipelines. The Maximum Likelihood (M-L) pipeline is written and maintained by the QUIET participants at the University of Oslo. QUIET data analysis using the M-L pipeline is the topic for this thesis. I attempt to give a broad and comprehensible account of the process of data analysis for a non-ideal project in radio cosmology, investigating each step in the transformation from raw observational data to final estimates of power spectra and cosmological parameters. Following this narrative, the pipeline is applied in a reanalysis of the QUIET Q-band data set.
Since the completion of the original Q-band analysis, significant changes have been made to the M-L pipeline. A reanalysis of the Q-band data serves as a validation of the new features. We find signs of improved performance in some places, while at other points there is need of further assessment. Most importantly, we find signs of a bias in the polarization null test suite, which is thought to be caused by ground pickup, although this will have to be confirmed.
The reanalysis also has significant educational merit, as achieving a solid understanding of the data analysis process is among my main motives for undertaking this project. The modest size of the Q-band data set makes it well suited for demonstration.