PEACE: Pulsar evaluation algorithm for candidate extraction-a software package for post-analysis processing of pulsar survey candidates

Modern radio pulsar surveys produce a large volume of prospective candidates, the majorityof which are polluted by human-created radio frequency interference or other forms of noise. Typically, large numbers of candidates need to be visually inspected in order to determineif they are real pulsars. This process can be labour intensive. In this paper, we introducean algorithm called Pulsar Evaluation Algorithm for Candidate Extraction (PEACE) whichimproves the efficiency of identifying pulsar signals. The algorithm ranks the candidates basedon a score function. Unlike popular machine-learning-based algorithms, no prior training datasets are required. This algorithm has been applied to data from several large-scale radiopulsar surveys. Using the human-based ranking results generated by students in the AreciboRemote Command Center programme, the statistical performance of PEACE was evaluated. It was found that PEACE ranked 68 per cent of the student-identified pulsars within the top0.17 per cent of sorted candidates, 95 per cent within the top 0.34 per cent and 100 per centwithin the top 3.7 per cent. This clearly demonstrates that PEACE significantly increases thepulsar identification rate by a factor of about 50 to 1000. To date, PEACE has been directlyresponsible for the discovery of 47 new pulsars, 5 of which are millisecond pulsars that maybe useful for pulsar timing based gravitational-wave detection projects.