Skip to Content

Wael Farah

Nearly a decade ago, a new class of radio transients emerged, dubbed Fast Radio Bursts (FRBs), holding the potential to provide additional insight to the workings of the ephemeral universe. FRBs are millisecond wide, intense bursts of radio emission, that together share a very exciting feature: the observed mean electron density along the line of sight to their sources substantially exceeds the maximum expected from our own Galaxy. In other words, models suggest that FRB emitters reside at cosmological distances, attributing the excess in electrons seen to the intergalactic medium. As such, FRBs are thought to be promising cosmological probes.

Although it is estimated that an FRB strikes earth every few minutes from a random direction on the sky, detecting them is not an easy task as our telescopes have limited fields-of-view and/or sensitivities. Moreover, the very radio spectrum in which they are discovered is increasingly polluted by our own communication technologies. All this, coupled with the fact that new generation telescopes will swamp researchers with data, means that gone are the days when astronomers had to inspect their data by eye. Machine learning provides new methods of data analysis without the need of explicit programming.

The aim of my Ph.D. thesis is threefold: to develop novel tools that will aid in the acceleration of Fast Radio Burst discoveries, to gather a large sample of these bursts for statistical analysis and to understand the completeness of previous/ongoing FRB surveys, such as the ones conducted by the Parkes radio telescope.

Building a real-time, machine learning-based, FRB detection pipeline to operate on the newly refurbished Molonglo Observatory Synthesis Telescope (UTMOST) was among the first projects completed during the initial stages of my Ph.D. A real time detection using UTMOST will allow a rapid multi-wavelength followup of an event, and a better localisation on sky. The pipeline aided in the discovery of the third FRB detected by the interferometer to date, and promises the detection of many more.

Office   AMDC812.06