Open Collections

Abstract

The Canadian Hydrogen Intensity Mapping Experiment (CHIME) is a radio telescope that we built to map the large-scale structure of the Universe between redshifts 0.8 and 2.5, when dark energy is expected to begin the transition from a decelerating to an accelerating phase in the expansion of the Universe. CHIME was designed to perform an intensity mapping survey using the 21 cm line of neutral hydrogen, a novel method that has the potential to enable enormous surveys of the distant Universe, but also significant observational challenges to overcome. In this thesis, I describe contributions I made to the CHIME data acquisition system and calibration effort, culminating in a detection of cosmological 21 cm emission in cross-correlation with measurements of the Lyman-𝜶 forest. The large data rate from the CHIME correlator is processed in real time by a high-performance digital pipeline, the development of which I participated in extensively. A few specific processing tasks where I led the design and implementation are highlighted in this work. In order to detect the 21 cm signal amidst the much brighter foreground emission from nearby sources, a very precise instrumental calibration is required. Calibrating the telescope's beam is a particular concern. One of many approaches being pursued for CHIME is the holographic observation of bright celestial sources in concert with a second radio telescope. I describe work I did to derive beam measurements from such observations and their analysis, including a scheme for calibrating the polarised beam response. I report the detection of 21 cm emission at an average redshift z = 2.3 in the cross-correlation of CHIME maps with measurements of the Lyman-𝜶 forest from the eBOSS. Data collected by CHIME over 88 days in the 400-500 MHz frequency band (1.8 < z < 2.5) were formed into maps of the sky and high-pass delay filtered to suppress the foreground power. Line-of-sight spectra to the eBOSS background quasar locations were extracted from the CHIME maps and combined with the Lyman-𝜶 forest flux transmission spectra to estimate the 21 cm-Lyman-𝜶 cross-correlation function. Fitting a simulation-derived template to this measurement results in a detection of 9-𝝈 significance.