Daily Image
17-06-2024
A Needle in a Cosmic Haystack: A Review of FRB Search Techniques - II
| Submitter: | Joeri van Leeuwen and Kaustubh Rajwade |
| Description: | Since the discovery of Fast Radio Bursts (FRBs) in single-dish archival data in 2007, detection software has evolved tremendously. Pipelines now detect bursts in real time within a matter of seconds, operate on interferometers, buffer high-time and frequency resolution data, and issue real-time alerts to other observatories for rapid multi-wavelength follow-up. We recently published a review on FRB search pipelines (Rajwade & van Leeuwen, 2024), were we discuss the proven techniques that were adopted from pulsar searches, and highlight newer, more efficient techniques for detecting FRBs. A previous Daily Image quantified how challenging finding FRBs is. Today's image displays how this analysis and selection takes place. All real-life FRB detection pipelines produce many spurious candidates due to various terrestrial sources. These so-called ’false positives’ make an actual detection extremely challenging in environments with high radio frequency interference (RFI). Cellular network towers, radar stations and satellite constellations emitting intense radiation across the radio-band can generate millions of false FRB candidates in any telescope even after proper RFI excision. Hence, one has to use clever techniques to identify a real astrophysical source from the noise. Generally, these fall in two classes: First, sifting and clustering, and next: machine learning techniques. These are explained in the review, and shown in the today's image. In the image, we illustrate how candidate classification generally involves, from left to right, candidate clustering, extraction of best candidates per cluster as a dynamic spectrum, and analysing each of these using a convolutional neural network (CNN) that produces a classification metric. Each cluster of candidates is marked with a different colour. |
| Copyright: | JvL/KR |
| Tweet |  |