The Petabyte Project for Fast Radio Transients

We are in a golden era in the investigation of flashing, short-duration astronomical events detected by radio telescopes. Progress in understanding novel, highly luminous phenomena like fast radio bursts (FRBs) and highly intermittent pulsars is being enabled by the broad availability of archives, new observing systems on a host of telescopes worldwide, and extensive computing systems in place to provide throughput for huge amounts of data.

This field has rapidly evolved in the past decade. Searches for such transients have been done by constantly evolving analysis techniques, each improving on the last. However, new techniques are typically applied to only a small set of data.

A research group at West Virginia University (WVU) will undertake “The Petabyte Project,” which will harness many diverse archival and new data sets to perform an unprecedented uniform search across multiple observing systems and radio frequencies. This search for transient events has key analysis structures in place that were not available in past searches, including a systematic check of completeness of the constituent surveys, and a rigorously tested machine-learning-based event classifier.

To broaden the impact of their work, the researchers will work with two strong existing outreach programs at WVU, the Science Public Outreach Team (SPOT) to introduce area high-school students to machine learning applications, and the Pulsar Search Collaboratory (PSC) summer program, to enable high school students to participate in the research by helping to classify radio sources.

There will be several key scientific, software, and data outcomes from this search. The work will lead to the discovery of new FRBs and pulsars, particularly at high radio frequencies where few searches have been done. It will provide the first statements on the completeness of FRB searches as a function of dispersion measure (distance) and radio frequency, leading to new knowledge on the diversity and range of FRB properties.

The team will also provide the first broad, blind multi-frequency exploration of extreme intermittency in the radio pulsar population. All code generated by this project will be open-access and accompanied by in-depth documentation, allowing broad use of the unified transient data read/write tool, machine-learning suite, pipeline, and completeness assessment software.

Labelled search products will be made public on a permanent server for use in future interference, fast-transients, or machine-learning studies.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.