realfast: real-time, commensal fast transient searches at the Very Large Array
As the most sensitive cm-wavelength radio interferometer on earth, the VLA will revolutionize the study of fast radio transients with its ability to localize sources with arcsecond spatial resolution. A pioneering science case for
realfast is the mysterious new class of radio transients known as fast radio bursts. We have now used the VLA imaging to make the first localization and identification the host galaxy of an FRB. We have shown that FRBs are occur at cosmological distances and can be used to as novel probes of the intergalactic medium. Similarly, the interferometric discovery and localization of other classes of millisecond transient will address topics ranging from the missing baryon problem to (sub)stellar magnetism to the search for exotic binary pulsar systems.
The challenge to using the VLA for millisecond imaging is that it produces roughly 1 TB of data per hour and requires forming many thousands of images per second. This data rate is so large that it cannot be transferred via the internet for data analysis. The computing requirements are so severe that no single computer can manage the search. The question is: how can we manage a TB/hour data stream to find a millisecond transient in hudreds of hours of data?Our answer is
realfast, a system for real-time fast transient searches at the VLA. Real-time processing is critical, as it allows triggered data recording and opens access to "commensal" observing in conjunction with other VLA observations.
realfastis supported by the NSF ATI program starting in late 2016.
realfast is built from a few key technologies:
- Computing hardware at the VLA: Computing is the most significant missing component to the creation of a real-time, commensal VLA fast transient detection system. This proposal will support the purchase of servers, GPUs, and infrastructure to be dedicated to realfast.
- Commensal data stream: The commensal data stream is a high-speed duplicate of the primary observing stream.
- Transient search pipeline: The new hardware will run a transient search pipeline. Currently that pipeline has been demonstrated with CPUs, but ultimately it will be migrated to GPUs.
- Candidate data management system:
realfastwill serve data products and transient alerts to the public. Initially, team members will classify each candidate (e.g., candidate summary plot), but ultimately we will train a statistical classifier to automatically identify good candidates.
- Geoff Bower (ASIAA)
- Sarah Burke-Spolaor (WVU)
- Bryan Butler (NRAO; NRAO lead)
- Paul Demorest (NRAO)
- Shakeh Khudikyan (JPL)
- Casey Law (UC Berkeley; Principal Investigator)
- Joe Lazio (JPL)
- Martin Pokorny (NRAO)
- James Robnett (NRAO)
- Michael Rupen (DRAO)
The core software for the fast transient search is
rtpipe. We used it to make the first blind interferometric localization of a transient neutron star, searched a 200-hour, 200-TB VLA imaging survey for FRBs, and made the first interferometric localization and host identification of an FRB.
rtpipe was built as an open tool to support analysis of interferometric data for fast transients. For more information, see this blog post. You can also launch an interactive Jupyter notebook with
rtpipe pre-installed by visiting the github repo.