Effectively processing SKA-low precursor pulsar search data in China

Xue, Mengyao

As a sensitive low-frequency aperture array with a large field of view (FoV), the upcoming SKA1-LOW AA* will serve as an efficient pulsar search facility, especially for mid-to-high Galactic latitude which covers quite large sky regions with relatively low dispersion-measure (DM). Making use of the Murchison Widefield Array (MWA)——the low-frequency precursor of SKA, the Southern-sky MWA Rapid Two-metre (SMART) pulsar survey acts as an important demonstrator survey for SKA1-LOW, helping optimise survey strategies and offering valuable insights into pulsar population at low-frequency in the Southern sky. With a survey speed of ~600 deg²/hr, the SMART survey has covered the entire sky below declination +30° in just 94 hours, recording channelised complex voltages from each station (‘tile’). This voltage capture system (VCS) provides maximal flexibility with a practical data volume for offline data reduction, enabling the formation of over 8,000 tied-array beams to cover the entire field of view for each observation. However, this process demands intensive computational resources and poses significant I/O challenges. With the current processing software and available computing resources (e.g. OzSTAR/NT), it may take more than 10 years to complete SMART survey data processing and pulsar searching. To help address these challenges, and as part of our broader strategy for distributed search processing across multiple HPCs, we have undertaken processing SMART survey's voltage data using the computing resources in China. This collaboration with the Computer Network Information Center, Chinese Academy of Sciences (CNIC, CAS) demonstrates the feasibility of cross-continental raw voltage data transfer and processing for low-frequency radio telescope arrays. Within six months, we have made over 30,000 tied-array beams and conducted pulsar searches across a DM range of 0 to 500 pc cm-3. The processing has led to seven new pulsar discoveries to date. Our work establishes a practical model for processing intensive radio interferometer data, and the results validate our approach of intercontinental voltage data transfer and utilisation of remote computing facilities for efficient post hoc processing.