Using our SMARTs to find pulsars

Australia boasts a proud record of having telescopes and instruments that have historically surveyed the sky at radio frequencies, adding to a large swath of knowledge in our understanding of the Universe. Now, a global team of astronomers, led by researchers from the Curtin University node of the International Centre for Radio Astronomy Research (ICRAR) have released two new pre-print papers, describing the use of one of our most unusual-looking radio telescopes to survey the sky rapidly and efficiently for exotic objects known as pulsars. What makes the Southern-sky MWA Rapid Two-Metre (SMART) pulsar survey different from our other radio surveys is that it searches the entire Southern sky for lower-frequency emissions from these high-mass stellar remnant objects. 

Pulsars are rapidly rotating, highly magnetised neutron stars that emit beams of radio waves from their poles sweeping out across the cosmos like a lighthouse. If those beams happen to pass over the Earth and we’re looking at them with a radio telescope, we register a pulse of radio emission. They’re found all across our Milky Way Galaxy, though the majority of them reside in the central Galactic plane - where most objects live. 

Since their discovery in 1967, pulsars have played a vital role in testing astrophysical theories and models, as they are extreme objects - they have roughly 1 - 2 times the mass of our Sun, compressed into a ball roughly 20 kilometres in diameter, with some of them spinning hundreds of times per second on their axis. They also have some of the most powerful magnetic fields that we can measure in the Universe - about 10 billion times stronger than a typical fridge magnet. One consequence of these extreme characteristics is that pulsars spin with remarkable stability, in some cases even rivalling our best atomic clocks.

“The exciting thing about pulsars is that they allow us to study truly extreme physics that we have no way of recreating in a laboratory on Earth,” said PhD student Nick Swainston who led the processing and development of the SMART survey’s first pass - accounting for many of the discoveries. “Each pulsar we discover provides a new lab to test our theories and search for clues as to how pulsars work.”

These features have allowed astronomers to apply the regular, periodic ticking from pulsars in highly-precision timing experiments, which has given us our first-ever detections of planets outside the Solar System (1992) as well as the first indirect evidence of gravitational waves (1974). These days, these exotic objects are used in Galactic-scale arrays to search for low-frequency gravitational waves, generated by the mergers of supermassive black hole binaries (i.e., when galaxies collide), testing the extremes of nuclear physics, and for performing some of the most exquisite tests of the theories of gravity, e.g., Einstein’s General Relativity.  

Whilst pulsars were first discovered and studied in lower radio frequency bands, many pulsar surveys eventually moved to frequencies above 1 GHz in an effort to combat signal degradation experienced at low frequencies. - The higher sensitivity achievable by telescopes at 1-3 GHz, and the substantially decreased sky background noise, was ultimately a winning move.. Unfortunately, the lower frequency radio bands have to battle with the ‘noise’ from the background signal of the sky, which is larger at lower frequencies and can offset the fact that pulsars are brighter at lower frequencies. On the upside, low-frequency telescopes are much more efficient at surveying the sky because they have a much wider instantaneous field of view, which means they can rapidly cover the entire visible sky.

Regardless of whether you’re observing at 100 MHz or 2 GHz, before arriving at Earth, pulsar signals must also battle through the density of electrons that make up part of the Interstellar Medium (ISM - the stuff between the stars, which contrary to popular belief, is not a perfect vacuum). This electron column density causes the pulsar’s signals to spread out in time as a function of frequency, and so, different frequency portions of the signal arrive at different times, and unfortunately, at lower frequencies, this can disperse the signal significantly. The pulsar's signal can also be broadened in time due to the intervening ISM material through a process called multi-path propagation, which acts to scatter radio waves in random directions so that they don’t all arrive at the same time and so the signal appears to be smeared out. If the amount of broadening approaches, or is greater than, the pulsar’s spin period, this results in a loss of sensitivity. Again, this effect is stronger at lower frequencies. Lastly, there are also the scintillating effects that give rise to apparent modulations in the observed brightness or strength of the signal in time and frequency, making it harder to observe them. 

“From an observational perspective, pulsars tend to be inherently brighter at lower frequencies, and so we will very likely be able to detect pulsars that might be too faint to have been detected in previous surveys at higher frequencies,” said Dr Bradley Meyers, a co-author on the study. 

“Secondly, understanding how pulsars behave across wide ranges in frequency is important if we want to understand the physics of how they produce the radio emission we see in the first place! There’s also some evidence that we might be detecting older populations of pulsars at lower frequencies, which is important for understanding the Galactic population of pulsars.” 

So far, we’ve discovered about 3,300 pulsars in our Galaxy, with a handful in both the Small and Large Magellanic Clouds (satellite galaxies of the Milky Way). This is less than 10% of the total number of predicted detectable pulsars in the Galaxy - we are pretty sure we’re missing some! For example, a reason we might not detect a pulsar is that the radio beams are not pointed in our direction, which means we never see them “pulsing”. 

Here is where the SMART pulsar survey steps in. The objectives of this survey are to exploit the recent advances in instrumentation, technologies and methodologies to search for pulsars in a rather uncommon observing frequency band, which can allow astronomers to further refine and analyse the overall population, potentially uncovering new phenomena, and/or observing variability or transient behaviour in known pulsars. 

“We don’t really know enough about pulsar and neutron star populations - it’s a long-standing problem,” said Dr Meyers.

“There are so many unanswered questions. What is the total number of Galactic neutron stars? What is their birth rate? And how does it compare to the Galactic supernova rate? What are the physical processes that govern the emission of electromagnetic radiation from pulsars?”

“Continuing to find new pulsars in different parts of the sky with different telescopes around the world is one important way to make progress in answering these questions,” he said.  

The SMART pulsar survey uses the Murchison Widefield Array (MWA), an unusual radio telescope that is actually made up of many groups of smaller antennas. Unlike traditional dish antennas, which can operate as a single dish, or in multi-dish configurations, the MWA is made up of spider-like dipole antennas that all work together to look at the whole sky at these lower radio frequencies. 

Located at Inyarrimanha Ilgari Bundara, the CSIRO Murchison Radio-astronomy Observatory, the MWA features 256 “tiles”, each 4m x 4m in size, that are located at varying distances from the geographic centre of the instrument. The largest distance between the furthest tiles is approx. 5 kilometres apart, and this creates an interferometric baseline of this distance for the telescope. 

Sitting atop each of the 256 mesh tiles are 16 dipole antennas (the “spiders”) in a 4-by-4 configuration. It operates at a frequency range between 70 - 300 MHz, though the SMART pulsar survey is surveying the southern sky (everything south of +30-degrees declination) in the 140 - 170 MHz band, where the MWA is most sensitive for detecting pulsars. To find the pulsars the SMART team used Swinburne’s OzSTAR Supercomputer to crunch their data, with time allocated by the Astronomy Supercomputer Time Allocation Committee (ASTAC). ASTAC is formed by Astronomy Australia Limited (AAL) and managed by Astronomy Data and Computing Services (ADACS).

“The SMART survey perfectly complements (in frequency and sky coverage) similar large efforts around the world,” said Dr Ramesh Bhat, who is the project lead.

“It is an innovative approach using a next-generation radio telescope, which brings challenges, but also presents new opportunities; these can be exploited to enable faster confirmation and characterisation of new discoveries - which can potentially transform the way we find and follow-up new pulsars with the SKA telescope.”

“It explores a new parameter space (the low-frequency southern sky) for pulsar science and discoveries, and as such will be several times more sensitive than previous-generation low-frequency pulsar surveys in the southern sky,” he said. 

“In the coming years the team aspires to ramp up the efforts around the processing and analysis of several Petabytes of data collected as part of the survey, as they transition from the current early science phase to a deep-survey phase where new search algorithms will be integrated and the search parameters will be extended to target pulsars with very short spin periods and those with companion stars.” 

“Our long-term goals are to uncover new populations of pulsars, discover exotic objects, and exploit them for exciting science that is awaiting on the horizon.”

When the MWA is being used to conduct the SMART pulsar survey, it is configured in a compact layout, using only 128 tiles that are all located within 300-metres of each other - which allows for approx. 450 square degrees of the sky to be covered in an hour. Substantial parts of the sky have already been observed to progress the survey, but it is not yet complete. This staggered observing approach has translated to the successful surveying of 70% of the visible sky, though only a small fraction (5-10%) of this data has been processed and analysed to date.

“In the compact configuration, the total amount of processing required to sample the sky with maximum-sensitivity ‘pencil beams’ is reduced,” said Dr Meyers. “This is because the size of each pencil beam is larger if the tiles are closer together.”

“In the compact configuration, we need to create 100 times fewer pencil beams to achieve what we want. It is really a computational limit - we simply wouldn’t be able to achieve the survey in its current form using the extended configuration, even on supercomputers,” he said. 

From initial processing (sensitive to only the simplest type of pulsars) and the scrutiny of only a small fraction of the data, the team of astronomers have already made two new pulsar discoveries, an independent pulsar discovery (i.e., a pulsar also discovered with another telescope as part of a different survey), as well as the rediscovery of a known pulsar with corrected characteristics. These four objects were later followed up with Murriyang, the 64-m CSIRO Parkes radio telescope, as well as the Giant Metrewave Radio Telescope (GMRT, located in Pune, India) - which operate in the higher frequency bands, relative to the MWA. This allowed astronomers to observe these pulsars across a large range of radio frequencies (from 100 to 4000 MHz), enabling detailed characterisation of their spectral properties and the ISM propagation effects in those directions. Additionally, the SMART survey also redetected 120 known pulsars, bringing the total population of unique MWA pulsar detections to 180. 

“Of the four pulsar discoveries, two appear to be low-luminosity objects that are also brighter at low frequencies – a class of pulsars that belong to the lowermost 2 percentile of the currently known pulsar population,” said Dr Bhat. 

“Another exhibits the ‘sub-pulse drifting’ phenomenon, whereby the emission structure tends to progressively march in time as the pulsar rotates, but in a way that is truly remarkable, and can potentially prove to be an important test-bed for theories and models of pulsar emission physics.”

SMART also pulls down a lot of data (42 terabytes per observation) which has required the optimisation of observation and processing techniques, involving both humans and machines, in helping make (and confirm) any detections. The remaining 30% of the survey observations will be undertaken from early-2023 onwards, as the overall data set continues to be processed, ramping up in mid-2023 to bring the full sensitivity to bear. 

“Using machine learning as part of the processing pipeline is absolutely vital,” said Dr Sam McSweeney, another co-author of the paper.

“Without it, it would take a single person decades to sift through the millions of candidates that the pipeline generates and inspect them one by one. Fortunately, there exist robust machine learning algorithms that can decimate the pool of candidates by identifying the obvious non-detections, making it tractable to complete the task within months (of full-time equivalent effort) instead of years, and even less when multiple people join the effort.”

The SMART pulsar survey is one of many observing projects that are scanning the southern skies as a precursor to the upcoming activation of an even more ambitious telescope- the Square Kilometre Array (SKA). With the low-frequency component located in outback Western Australia (the high-frequency component is located in South Africa), Australia’s SKA instrument (known as SKA-Low) will observe the skies in the 50-350 MHz band - a similar region of the radio spectrum to the MWA. 

Therefore, surveys like SMART build a reference for future pulsar searches that are planned with the SKA, not to mention the ability to push science, instrumentation, data processing pipelines, and engineering towards SKA’s goals. In particular, the SKA-Low will have the same sky coverage as SMART, so this creates a unique advantage over northern hemisphere-based SKA precursor studies. 

“The SMART survey will serve as an important reference survey for future larger-scale surveys planned with the SKA-Low,” said Dr Bhat. “The important role of reference surveys is vividly demonstrated by two generations of multibeam pulsar surveys with Parkes, which benefited from their predecessor survey to confirm new discoveries and rapidly characterise new pulsars. At present, no such reference survey exists for SKA-Low; the SMART survey is designed to fill that void.”

One of the SKA’s key science goals is to undertake a deep look at the southern sky for pulsars - hoping to not only detect these objects in these low-frequency bands, but also observe any transient behaviour, or probe deeper through the ISM to find more to add into the population we already know about. 

“The SKA will provide an enormous boost to our ability to both find pulsars and study them in detail, across a wide range of frequencies with the world’s most sensitive telescope,” said Dr Meyers. “SKA-Low, in particular, will be great at finding new pulsars because of its wide field of view, as well as monitoring and characterising the effects that the ISM has on pulsar signals.” 

“The MWA and other low-frequency precursor telescopes have shown that pulsar emission sometimes does weird things at different frequencies, so having the ability to study these objects over wide frequency ranges, especially below 300 MHz, will unveil some curiosities, I’m sure.” 

Find out more about the SMART pulsar survey, by reading the accessible first pre-print paper about survey design and the processing pipeline, and also the second pre-print paper, which is about the survey stats, pulsar census and the first few discoveries. They are also available in the publications below. 

We acknowledge the Wajarri Yamaji people as the traditional owners of the Inyarrimanha Ilgari Bundara, the CSIRO Murchison Radio-astronomy Observatory, where the MWA and ASKAP instruments are located.

Video credit: SMART survey / MWA.