## Hard X-Ray Burst

I just learned something cool: 0.4 seconds after LIGO saw those gravitational waves on 14 September 2015, a satellite named Fermi detected a burst of X-rays!

• V. Connaughton et al, Fermi GBM observations of LIGO gravitational wave event GW150914.

It lasted one second. It was rather weak (for such things). The photons emitted ranged from 50 keV to 10 MeV in energy, with a peak around 3.5 MeV. The paper calls this event a ‘hard X-ray source’. Wikipedia says photons with an energy over 100 keV deserve the name gamma rays, while those between 10 keV and 100 keV are ‘hard X-rays’. So, maybe this event deserves to be maybe a gamma ray burst. I suppose it’s all just a matter of semantics: it’s not as if there’s any sharp difference between a highly energetic X-ray and a low-energy gamma ray.

Whatever you call it, this event does not appear connected with other previously known objects. It’s hard to tell exactly where it happened. But its location is consistent with what little we know about the source of the gravitational waves.

If this X-ray burst was caused by the same event that created the gravitational waves, that would be surprising. Everyone assumed the gravitational waves were formed by two large black holes that had been orbiting each other for millions or billions of years, slowly spiraling down. In this scenario we don’t expect much electromagnetic radiation when the black holes finally collide.

Perhaps those expectations are wrong. Or maybe—just maybe—both the gravitational waves and X-rays were formed during the collapse of a single very large star! That’s what typically causes gamma ray bursts—we think. But it’s not at all typical—as far as we know—for a large star to form two black holes when it collapses! And that’s what we’d need to get that gravitational wave event: two black holes, which then spiral down and merge into one!

Here’s an analysis of the issue:

As he notes, the collapsing star would need to have an insane amount of angular momentum to collapse into a dumb-bell shape and form two black holes, each roughly 30 times the mass of our Sun, which then quickly spiral down and collide.

Furthermore, as Tony Wells pointed to me, the lack of neutrinos argues against the idea that this event involved a large collapsing star:

• ANTARES collaboration, High-energy neutrino follow-up search of Gravitational wave event GW150914 with ANTARES and IceCube.

To add to the muddle, another satellite devoted to observing gamma rays, called INTEGRAL, did not see anything:

It will take a while to sort this out.

But luckily, the first gravitational wave burst seen by LIGO was not the only one! Dennis Overbye of the New York Times writes:

Shortly after the September event, LIGO recorded another, weaker signal that was probably also from black holes, the team said. According to Dr. Weiss, there were at least four detections during the first LIGO observing run, which ended in January. The second run will begin this summer. In the fall, another detector, Advanced Virgo, operated by the European Gravitational Observatory in Italy, will start up. There are hopes for more in the future, in India and Japan.

So we will know more soon!

For more on Fermi:

### 29 Responses to Hard X-Ray Burst

1. jessemckeown says:

Well, if there were two black holes orbiting each other, there might have been three or more stars there before; and with ~3M☉ converted to radiant geometric disturbance, any neighbours may well have felt something of a nudge.

2. WebHubTelescope says:

As a way of trying to understand this, there seem to be two theories proposed.
1. Two black holes that have been in mutual orbit for millions or billions of years, are caught in the moment that they finally collapse into each other.
2. A massive star collapses which then spontaneously creates two black holes and these are not far enough apart so they quickly collapse into each other.

And they think the latter is a preferable theory because gamma rays need a source of mass to emit, and any mass in theory #1 is well in the past.

• John Baez says:

Yes, that’s right.

And here’s the dilemma: theory 1 is implausible because in this scenario we don’t see how a gamma ray burst would occur, and theory 2 is implausible because a star would need to be rotating at an insane rate, stretched out into a kind of dumbbell, to collapse into two separate black holes rather than one.

However, theorists are inventive, and we’re just beginning to see people start trying to explain this event. Jesse’s idea is nice: take theory 1 and put another star nearby, which gets pulverized by gravitational radiation and emits some gamma rays. One needs to get quantitative to see how plausible this might be.

3. domenico says:

I am thinking that a couple of positive (or negative) charged black holes could generate electromagnetic radiation in the relativistic merger, but there is the problem of the 0.4 sec arrival delay of the electromagnetic radiation.
I am thinking that if the emission mechanism is similar (a jet of gravitational wave and a jet of electromagnetic wave with the same directions), then there is only a zone of the space (along the line between the source and the Earth) where the signal is observable.

• Nix says:

The problem there is that charging a black hole like that (and having it retain its charge for any notable amount of time) is probably very hard. Most stuff in space is net-uncharged, after all, and if a hole gets charged it will preferentially attract oppositely-charged particles until it is uncharged again.

• John Baez says:

Yes, I see no mechanism for two black holes to become highly charged. And if they were charged and orbiting each other 250 times per second (as the ones that collided were, at the end), they would produce radiation with a frequency of 500 hertz. Why would they produce gamma rays? Okay, I guess in the merger there would be a kind of ‘spark’: that would be interesting to analyze. But a much more promising method of getting gamma rays is to get some ordinary matter into the act.

• domenico says:

I am thinking that there are simple mechanisms to charge a black hole; for example, some galaxies with different total charge, so that exist zone of space where the charge is positive (or negative) over long distance, or charged stars before the black hole birth because of solar winds, or …
I try a simple solution of the problem: if the gamma-ray burst are localized in jets, how it is possible to observe the arrival of the gamma-ray burst, and gravitational waves, if the two processes have not similar mechanism of production, and similar waves?
If there is not jets, then the processes can be different; but I don’t know if GW150914-GBM is a jet emission.

4. arch1 says:

Can anyone explain the logarithmic term in the formula for the false alarm probability in section 2.2 of Connaughton et al? They say that this term “accounts for the search window trials”. Thanks!

• John Baez says:

I’m not a statistician, so let me just quote the passage you mention and see if someone understands it. Or maybe if I think about it a while it’ll make sense:

We determine the significance of a GBM [gamma-ray burst monitor] counterpart candidate by considering both its frequency of occurrence, and its proximity to the GW [gravitational wave] trigger time. The candidates are assigned a false-alarm probability of $2 \lambda \Delta t$ where $\lambda$ is the candidate’s false-alarm rate in the GBM data, and $\Delta t$ is its absolute time-difference to the GW time. Our method, described in Blackburn (2015) allows us to account for all the search windows in the interval over which we performed our search, while assigning larger significance to those events found closest to the time of interest. This two-parameter ranking method frees us from having to choose a fixed search interval. We can also limit the length of the search interval to a value that is computationally reasonable without fear of truncating our probability distribution.

With a false alarm rate of $4.79 \times 10^{-4}$ Hz for GW150914-GBM, which begins 0.4 s after the time of the GW event, we calculate a false alarm probability for GW150914-GBM, $P = 9.58 10^{-4}$ Hz $\times 0.4 s \times (1 + \ln(30 s / 0.256 s)) = 0.0022,$ where the logarithmic term accounts for the search window trials.

“Hz” means “hertz” which means “per second”. I’m suspicious of anyone who gives three decimals of precision for a false alarm rate: it’s generally hard to know exactly how often you screw up. But that’s not very important.

A more important thing is to understand the role of their “search windows”. Elsewhere these are called “bins”, and they’re said to be .256 seconds long. I guess the idea is that in each such time interval they count the number of photons detected—see Figure 2 in the paper. I don’t see where the figure of 30 seconds is coming from.

I think it would help to read the paper by Blackburn:

• Blackburn et al, Significance of two-parameter coincidence.

5. amarashiki says:

Extreme Astronomy, Astrophysics and Cosmology: gamma rays, X-rays, neutrino astronomy, gravitational wave astronomy, cosmic rays, and much more are pushing ahead our high energy limits…and phenomenology…In time, theory and new theories will follow them…

6. davetweed says:

I wonder if Greg Egan could sue the universe for copyright infringement of a scene from Diaspora? (OK, they were neutron stars with lethal levels of radiation emitted, but plagiarism is bad, mkay?)

• arch1 says:

(chuckle) I think he’d have trouble establishing priority.

7. arch1 says:

Another beginner Q: The calculated probability of false alarm seems to be based on an assumed background rate which is the upper end of a 90% confidence interval.

How can they conclude a ~0.22% probability of false alarm, if there is a 5% chance that their statistically inferred background rate is too small? Shouldn’t they decrease the 5% (which would cause the 0.22% to increase) until the two are at least in the same ballpark?

• John Baez says:

Good question. Again, it takes someone with more understanding of statistics than I have to give a good answer. But I think it’s not necessarily a paradox that the 5% greatly exceeds the ~0.22%. Say there’s a 95% chance that an event occurs once in a millennium (on average) and a 5% chance that it occurs once in a century (on average). You can work out the chance that it occurs on a given day, and it’s a lot less than 5%.

The problem is that there could be a ‘long tail’. E.g. suppose there’s a 95% chance that an event occurs once in a millennium (on average) and a 4.9% chance that it occurs once in a century (on average) and a 0.1% chance that it occurs once an hour. This changes everything.

• arch1 says:

Thanks to your example, John, I see that my thinking was incorrect – the two probabilities involved need not be even roughly comparable.

I now think the “right” way to compute the false alarm rate, given that it depends on an assumed background rate b which is itself uncertain, would be as the integral of
pr{false alarm | b=r} * p(r) dr,
where p is b’s assumed probability density function.

I guess that the people writing and reading such papers have a good sense of when such considerations might make a difference (due e.g. to long tails or sensitive dependencies), and just ignore them otherwise. (At least, I hope they don’t ignore them just because they are inconvenient and messy!)

8. Crippa75 says:

Do they know what the effect would be if a extremely strong gravity wave would create distortion in dark matter or dark energy?

Even if all normal matter is gone there should still be some dark matter clusted nearby.

• John says:

Dark matter, being uncharged, couldn’t be the origin of this flash. I don’t think we really know enough about dark energy to say for sure, but a simple cosmological constant is a pure geometric term and shouldn’t produce any EM flash either.

You might also think about normal matter like a nearby star being disrupted by the GWs, but for this event the GWs would not disrupt a material body even very close to the black holes. The source was ~400 Mpc ~= 1e22 kilometres away, and the peak strain at Earth was 1e-21. The strain falls off like 1/r, so a totally naive estimate would say you need to be only 1000 km from the “source” for the strain to approach 1%, which is well inside the near field where the waves are not linear anyway. I doubt there was any material, dark or not, left hanging about in that area after it was swept clean by a couple of huge black holes zooming about at a large fraction of light speed.

9. Māris Ozols says:

Maybe it was a ring-shaped black hole rather than a collapsing star:
http://www.cam.ac.uk/research/news/five-dimensional-black-hole-could-break-general-relativity
However, that would require 5 dimensions… :)

• John Baez says:

Heh, yes — those ring-shaped 5d black holes have been in the news lately.

By the way, it’s not really known if ring-shaped event horizons are possible in our universe. At one point it was thought that as two black holes collided, their event horizons might momentarily be connected by two bridges, forming an event horizon with the topology of a torus. But so far, detailed simulations have instead given results like this:

So, the idea of a ring-shaped horizon is no longer so popular… but nobody has proved it can never happen.

For details, see:

• Michael I. Cohen and Jeffrey D. Kaplan and Mark A. Scheel, On toroidal horizons in binary black hole inspirals, Phys. Rev. D 85 (2012), 024031.

• arch1 says:

Is there some reason a black hole can’t merge with two others at once?

If not, it hand-wavingly seems one could almost force a torus given enough black holes and enough ingenuity (two carefully oriented and timed colliding necklaces, or something).

• John Baez says:

arch1 wrote:

Is there some reason a black hole can’t merge with two others at once?

No, it’s just incredibly unlikely here in the actual Universe.

If not, it hand-wavingly seems one could almost force a torus given enough black holes and enough ingenuity (two carefully oriented and timed colliding necklaces, or something).

Yes, a ring of small black holes all moving towards each other seem like they could form a torus as they meet. I doubt anyone has tried simulating that yet.

10. […] Baez wrote a bit on it https://johncarlosbaez.wordpress.com/2016/02/25/gamma-ray-burst/   Greg Bernhardt, Feb 26, 2016 at 9:34 […]

11. Jonathan Scott says:

The reference to the energy being “about” 50keV seems wrong; the Fermi team paper says “above” 50keV and the detailed energy spectrum results (section 2.5) show a spread of energies starting above 50keV which peak around 3.5MeV and which appear to continue up to at least 10MeV and possibly even up to 50MeV.

• John Baez says:

Thanks! That’s interesting.

I was going to correct the title of this blog article again, restoring its original title “Gamma ray burst”, but the beginning of this section calls this event “a weak but significant hard X-ray source with a spectrum that extends into the MeV range”—so they seem to think “hard X-ray source” is a suitable term even though Wikipedia calls photons with energy more than 100 keV “gamma rays”. They also say:

For a deconvolution assuming a source position at the northeastern tip of the southern lobe (entry 10 in Table 2), the Comptonized model converges to find a best fit $E_{peak}$ of $3.5^{+2:3}_{-1.1}$ MeV.

where earlier they’d defined $E_{peak}$ to be the “peak energy in the spectral energy distribution”. And they say:

The fit parameter values are typical for short GRBs [gamma-ray bursts].

12. […] Bernhardt said: John Baez wrote a bit on it at https://johncarlosbaez.wordpress.com/2016/02/25/gamma-ray-burst/

Perhaps those expectations are wrong. Or maybe—just maybe […]

13. Peter Bloem says:

Wouldn’t proponents of the single-star-collapsing-hypothesis also need to account for the 0.4 second gap? I know the speed of gravitational waves hasn’t been fully established, but it’s unlikely to be faster than x-rays, right?

So if the collapse of the star causes the gamma rays and the black holes, and the black holes merging causes the gravitational waves, I don’t see how the gravitational waves can possibly arrive first.

• John Baez says:

Peter wrote:

Wouldn’t proponents of the single-star-collapsing-hypothesis also need to account for the 0.4 second gap?

Good point!

I know the speed of gravitational waves hasn’t been fully established, but it’s unlikely to be faster than x-rays, right?

According to general relativity the speed of gravitational waves is exactly the speed of light. Anyone who doubts general relativity can’t trust the usual interpretation of anything we see with LIGO. So, we should assume general relativity is right about the speed of gravitational waves when trying to understand this event: otherwise we are tying one hand behind our back.

In an ordinary type II supernova, formed by a collapsing star, the neutrinos come out before the light. The reason is that neutrinos move at almost the speed of light in vacuum, while the light needs to penetrate through the body of the collapsing star, repeatedly getting absorbed and re-radiated.

A good example is supernova 1987a:

Approximately two to three hours before the visible light from SN 1987A reached Earth, a burst of neutrinos was observed at three separate neutrino observatories. This is likely due to neutrino emission, which occurs simultaneously with core collapse, but preceding the emission of visible light. Transmission of visible light is a slower process that occurs only after the shock wave reaches the stellar surface. At 07:35 UT, Kamiokande II detected 11 antineutrinos; IMB, 8 antineutrinos; and Baksan, 5 antineutrinos; in a burst lasting less than 13 seconds. Approximately three hours earlier, the Mont Blanc liquid scintillator detected a five-neutrino burst, but this is generally not believed to be associated with SN 1987A.

Supernova 1987A is considered a ‘peculiar’ type II supernova: for one thing, nobody has yet seen the neutron star that was expected to form from the collapse! There are various possible explanations: it might have formed a black hole… or it might even have formed a quark star.

14. George Dishman says:

The idea that a “nearby” star might be influenced into some cataclysmic response by the energy released in the merger sounds interesting, the question that springs to mind is how stable would its orbit be if it were at a location away from the pair but the same distance from us, or more generally on a hyperbola with 0.4s path difference. These aren’t SMBH so perhaps a binary BH system within a dense globular cluster could produce such configuration.

Of course the idea is crucially dependent on how much energy the star could extract from the wave burst, the luminosity would be enormous at that range but I would guess that any energy absorption would depend on “tidal heating” of the star.

• John Baez says:

Yes, testing the plausibility of this idea requires some calculations and estimates. Unfortunately I’m not in a position to do them. I sure hope people figure out what happened! And if they do, I hope someone adds another comment on this blog article, to let me know.