A Repeating FRB: Guest Post by Paul Scholz

This is a guest post by Paul Scholz, currently a graduate student working with Vicky Kaspi at McGill University. Paul specializes in radio and X-ray observations of pulsars. He was second author on this Nature Paper describing the discovery of a repeating(!!) fast radio burst, and was the first person to notice the repeated bursts it in the data from the follow-up observations. -dave

There's been a few posts already on this blog about FRBs  but a new result in a Nature paper last week, on which I'm co-author, adds a big twist to the story. As a reminder (or in case you haven't read Dave's and Dusty's posts) Fast Radio Bursts (FRBs) are short (~millisecond) bursts detected at radio frequencies, that originate from cosmological distances, or at least that's what's implied from their dispersion measures. The twist is that we've discovered repeating bursts from the same source as a previously discovered FRB. First I'll give a bit of background and then I'll get to why the result is exciting.

FRB 121102 was detected in the PALFA survey in 2013 by Laura Spitler and was the first FRB seen at a telescope other than Parkes. This was an important step in out understanding of this new phenomenon as it put to rest fears that FRBs were caused be some terrestrial or atmospheric phenomena local to Parkes. (since then the Green Bank Telescope has also found an FRB) 

The Parkes FRBs have so far never been seen to repeat despite quite a bit of telescope time dedicated to determining whether or not they do. This, along with their extreme distances, has led to many cataclysmic explanations for FRBs such as the collapse of a rapidly rotating neutron star into a black hole, and the merger of neutron stars or white dwarfs. 

For the PALFA FRB we set out to put a robust limit on the repeatability of bursts from the source of FRB121102. So, we performed a campaign of observations using Arecibo in May and June 2015. When we searched those observations for bursts we found some very bright signals. In total we found 10 new bursts that had the same dispersion measure as the original FRB 121102.

11 bursts from FRB 121102 including the original burst. They're corrected for the $\nu^2$ dispersion sweep and the shape of the instrumental bandpass.

It was immediately obvious that this was a big deal in the study of FRBs. Our discovery of repeating bursts from FRB 121102 shows that the source of these bursts cannot be cataclysmic, that is the source must survive the event. Also, six of the bursts were found within 10 minutes of each other, that says that it can't be caused by an event that occurs on longer timescales than that, such as giant flares from magnetars.

A model for FRBs that does fit with these bursts that can repeat on relatively short timescales is supergiant pulses from an extragalactic pulsar or magnetar. This also fits with the unusual and highly variable burst spectra that we see (see the right panels in the above plot), since similar spectral variability is seen in the Crab pulsar, the prime example of a young, energetic pulsar in a supernova remnant in our Galaxy. Another that might fit is radio counterparts to short (millisecond to second) X-ray bursts from a magnetar.

At any rate, this result shows that FRBs can repeat. But, it most definitely does not imply that all FRBs repeat and that some of them cannot be of the cosmological, cataclysmic flavor. The Keane et al. Nature paper that came out a week before ours suggested the merging of two neutron stars into a black hole (i.e. a cataclysmic event) as the origin of their bursts. But the host-galaxy association that that interpretation was based on has since been questioned (see Dave's post). Still, keep in mind that the Parkes-discovered FRBs have been followed up for many hours and no repeat bursts have been found. Perhaps they are all repeating and Parkes is missing the fainter bursts that we can see from FRB 121102 since we're using Arecibo. But this seems like especially bad luck, surely Parkes would have seen at least a second burst from one of the 15 FRBs discovered there by now.

This leads to the exciting prospect that there are two or more physical origins for FRBs. Time will tell: we'll keep characterizing FRB 121102, other FRBs will continue to be monitored for repeats, and in the next few years large field-of-view radio telescopes, such as the CHIME telescope in Canada, should give us a flood of new FRBs that will give us a much better picture of the FRB population.


EDIT - For posterity, here is a repeating Furby. -dave

The Case of the Fickle Found Furby

Fast Radio Bursts

Early last week, Dusty discussed Fast Radio Bursts (FRBs, or as I prefer to call them, Furbies). To recap, they are short (~millisecond) bursts in the radio (GHz) frequencies, that seem to come from cosmological distances, as determined from their dispersion measures, related to the amount of time between arrivals of the pulse at different radio frequencies. The more free electrons between us and a radio burst, the longer this delay will be, with the time delay scaling as the inverse of the frequency squared. The column density of electrons indicate that these bursts come from either dense plasma environments, or from cosmological distances (with the electrons coming from the intergalactic medium).

The first detected FRB, known as the Lorimer burst. The $\nu^{-2}$ dispersion, large dispersion measure, and short duration define the class.

The other kind of Furbies, also radio emitters, but more easily localized, with varying optical colors.  

Since these bursts are so fast (it's in the name!), it is very difficult to figure out precisely where they are coming from. Until last week we didn't have any clear identification of a true distance or redshift. We know that they seem to be isotropically distributed in the sky (thus supporting the hypothesis that they are cosmological). Most have been discovered by the Parkes radio observatory in Australia, mostly due to survey design reasons, but several have also been seen by the Arecibo and Green Bank Radio Telescopes. One detected by GBT also showed signs of circular polarization, seeming to indicate that it was associated with a strongly magnetized/rotating object. So far about 17 have been detected, though when convolving this number with the beam size and observational strategies of the various radio surveys, this gives a rate of roughly thousands per sky per day!

A Furby Located?

Last Thursday, an extremely interesting Nature paper came out discussing the apparent localization of a Fast Radio Burst by an afterglow and the identification of a host galaxy. Keane et al. report that using their extremely impressive real time identification of FRB 150418. (most FRBs have been detected by an archival search of the data), a large number of followup telescopes were triggered with an afterglow in the radio (5 GHz and 7.5 GHz) detected by ATCA that faded over about 6 days.

From Keane et al. 2016, on the purported radio afterglow associated with FRB 150418. The detection are the two purple points and one green point on the left, with low level emission and upper limits after. The FRB was detected at time 0. 

If true this is a HUGE deal. It would be the first identification of an FRB host galaxy, providing a distance and redshift (z = 0.49). The combination of dispersion measure and redshift provide an estimate of the "missing" intergalactic baryon density, providing an estimate of the mass in the intergalactic medium (IGM) that agrees well with the cosmological results from WMAP.

The host environment, however is at odds with a magnetar being the source of FRBs. Having FRBs be associated with giant magnetar flares is, I feel, the leading candidate for their astrophysical source. However, magnetars exist in young stellar populations, while the galaxy in question is "red and dead", meaning it has an old stellar population with no star formation). Additionally the energetics of the afterglow (which contains much more energy than the initial FRB) suggest that it must come from something much more powerful than a magnetar flare. The authors even suggest that these factors indicate a NS merger as a source for the burst and afterglow. This, however, is problematic, as the rate of neutron star mergers is not nearly high enough to account for the inferred FRB rate, which is close to the supernova rate! If the NS-mergers accounted for a large fraction of the inferred rate of FRBs this would mean that 1) Short GRBs are extremely beamed 2) Initial LIGO should have probably seen a merger if they happen so often. The alternative many suggest is that there may be multiple progenitor classes (though the rates must still be comparable for it to have been picked up with so small a sample).

Keane et al. also examine the possibility of a false positive, by looking at the rate of radio transients, and conclude that there is a < 6% chance of a coincident transient detection. However, in the last few days it has emerged that only considering the radio transient rate may have been a crucial, but subtle mistake.

Not So Fast, Radio Burst!

Over the weekend Peter Williams and Edo Berger of Harvard CfA posted a preprint questioning the Keane et al. results. (They actually posted a link to their preprint on facebook on Friday morning, which is impressive turn around time indeed!) They argue that it is the rate of variable radio sources that should be considered, and show that there should be order unity of these expected in the beam of Parkes associated with the FRB. They argue that the "afterglow" then, is likely associated with AGN variability. Most compellingly (to me), they show how the evolving ratio of the 5.5 GHz to 7.5 GHz flux from the source is inconsistent with a relativistic fireball that one expects to be powering a true afterglow of a burst event. (This also jives with how it requires a lot more energy to power such an afterglow than most think would be available for a single compact object).

If the "afterglow" of FRB 150418 is indeed due to AGN variability, this would be easy enough to check with a campaign of follow-up observations, which of course Edo and Peter carried out. The result was this Astronomer's Telegram:

ATel 8752 from Williams and Berger reporting on the VLA follow up of the purported FRB host galaxy.

Using the VLA, they reported a 157 microJansky detection of the radio source, which suggests a re-brightening at the 3 sigma level. That's pretty damning of the afterglow interpretation, though there is no direct detection of variability in their two 1.5 hour observations. 

It is worth noting that radio transient/variability people that I know and respect think this closes the case, but some of my colleagues who study AGN for a living have told me that such variability from a radio quiescent AGN isn't actually expected (they said they expect such variability from a blazar source, though the optical observations make it clear it is not a blazar in this case). [EDIT - Edo writes to mention that since the source is only a few degrees from the Galactic plane, the variability seen is consistent with the level of ISM scintillation you'd expect from an AGN with no intrinsic variability. Interesting! Though pulsar people should not have missed this as they deal with scintillation all the time!] As an outsider to both fields I can't really provide a solid interpretation, but my general impression is that the radio sky (and particular quiescent radio galaxies) at these frequencies and timescales are not well studied. Regardless, the re-brightening, if held up, combined with the strange spectral evolution, does make it hard to believe the afterglow interpretation, and subsequently the association with a host galaxy and distance. 

A Furby, disassociated. 

Too bad!