Over the past few years, we have been conducting an all-Northern-sky survey for pulsars and fast transients using the LOw Frequency ARray (LOFAR), a next generation interferometric telescope that operates at low observing frequencies of ~30-270 MHz. The LOFAR Tied-Array All-Sky Survey (LOTAAS) utilises the simultaneous beamforming capability of LOFAR to form 222 beams for an observation, giving a large instantaneous field-of-view, allowing a large dwell time of 1 hour. LOTAAS has discovered 73 new pulsars and redetected a further 312 pulsars. Using the LOTAAS pipeline, I processed the data and searched for periodic pulsar candidates. Here, I describe the survey specifications and current status, as well as report on the pulsars discovered and redetected by the survey. Using the DMs of the discoveries, I have provided constraint to the Galactic electron density models, in which the YMW16 model underestimated the electron density off the Galactic plane by at least 10%. LOTAAS has so far generated an estimated 40 million periodic pulsar candidates, which require a Machine Learning classifier to reduce the number of candidates. The first ML classifier, LC1 was suspected to perform sub-optimally. Hence, I developed a new ensemble classifier of five decision trees, LC2 to address the issue. LC2 used a larger number of features compared to LC1 and a new training set that includes a large number of pulsars misclassified by LC1. A new class of RFI instances was also defined for LC2. LC2 performed better than LC1 across all metrics and is able to identify pulsars with wide integrated pulse profiles that were misclassified by LC1. I am responsible for the discovery of 32 of the 73 pulsars, including two binary pulsars and the slowest-spinning radio pulsar currently known, PSR J0250+5854. Here, I present a multi-frequency study (150, 350, 1500 MHz) of the properties of 22 of the LOTAAS discoveries, including PSR J0250+5854. The rotational properties of the sample of pulsars timed as part of my thesis appear to be of older pulsars of age > 10 Myr. The multifrequency study of the pulse profiles showed that PSRs J1643+1338 and J1749+5952 do not follow the radius-to-frequency mapping model. The average spectrum of the pulsars is found to be steeper than the general pulsar population. PSR J1658+3630 has a spin period of 33 ms and is in a 3.0 day orbit with a CO white dwarf companion of minimum mass of 0.87 Msun. My analysis of PSR J0250+5854 showed that the pulsar is located around the different models of pulsar death lines, has different emission properties at different observing frequencies, characterised by a stable emission with occasional nulling at 150MHz and sporadic strong pulses with a weak regular emission at 350 MHz. The discovery of PSR J0250+5854, spinning twice as slow as the next slowest radio pulsar, suggests that there are more similarly long-period pulsars waiting to be discovered.
|Date of Award
|1 Aug 2019
- The University of Manchester
|Michael Keith (Supervisor) & Benjamin Stappers (Supervisor)