Optimisation and parallelisation of pulsar searches in radio astronomy

Pulsars are collapsed stars that emit beams of radiation that are detected by radio telescopes on earth as regular pulses caused by the beam sweeping across the telescope. The timing of the pulses is regular to a very high degree of accuracy, making pulsars natural clocks that can be used to test the predictions of fundamental theories in physics, notably Einstein’s theory of general relativity. The Square Kilometre Array (SKA) is the world’s foremost project in radio astronomy to build a distributed radio telescope of unprecedented power that can in principle find the entire population of pulsars in our galaxy.

The data analysis challenges of the pulsar searching on the SKA are formidable. The data is gathered from at a rate of Terabits (10^12 bytes) a second. It is impossible to store this quantity of data and do the pulsar searching offline, therefore processing methods and algorithms that can process the data as fast as it it gathered are required. The School of Computer Science at the University of Manchester is working with the pulsar astronomers at Jodrell Bank Observatory to develop new methods of signal processing to address this problem. The PhD project will examine methods of optimising and parallelising this process using modern developments in processors, such as GPUs, to provide sufficient computational power.