Thread by @dcastelvecchi

So, @naturenews, why did you hype up the possibility that today's event was a neutron star–black hole merger?
I'm glad you asked.
[Warning: monster thread ahead]
[Image credit: A. Tonita, L. Rezzolla, F. Pannarale]
nature.com/articles/d4158…

After all, @LIGO and @ego_virgo’s own estimate said it only had a 13% chance to be that type of event.
[2/]
gracedb.ligo.org/superevents/S1…

And I’m glad I asked LIGO’s own Chad Hanna and B.S. Sathyaprakash. Both are physicists at Penn State, and here I will try to explain what they told me, hopefully without butchering it too much
[3/]

Those percentages are produced by an automated algorithm. It’s like when Spock says “Computer, what is the probability that…?” But I would trust Spock’s estimation over the computer’s.
[4/]

First of all, these alerts are produced by the “low-latency pipelines”, algorithms that constantly monitor the three interferometrs in search of signals. Their estimates of the parameters of the event are based on a library of “templates”...
[5/]

..., which are relatively rough predictions of what two objects should sound like (in the gravitational-wave sense) when the spiral into each other. The libraries contain hundreds of thousands of templates, to cover a broad range of scenarios.
[6/]

Still, the templates are only good approximations for vanilla binaries. They will inevitably be poor matches for outliers — such as systems with eccentric orbits, or where one object is much more massive than the other, or where the spin axes are weirdly oriented, and so on.
[7/]

The low-latency algorithms then merely produce their best guesses based on the templates. To have an accurate estimation of the parameters of a specific event — the masses of the objects, the distance of the system, etc — you need to run very intensive computer simulations.
[8/]

So it will be a while before LIGO and Virgo figure this out, and when they do, they will probably not reveal their results until they have written a paper and had it peer-reviewed.
[9/]

Meanwhile, though, why is it that experts can eyeball this event and guess that it’s much more than 13% likely to be a neutron star–black hole merger?
[10/]

First of all, there is the distance. Based on the waveform and amplitude, it seems likely to have occurred 375 megaparsecs away. That is very far for a NS-NS merger. Nominally, LIGO’s most sensitive detector (La.) can currently see these only out to a distance of 140 Mpc...
[11/]

I say nominally because this 140 Mpc is averaged over the whole sky; an interferometer is not equally sensitive in all directions (and it even has some blind spots)
[Fig. credit: Living Rev. Relativity, 12, (2009), 2]
[12/]

Also, the loudness of the waves depends on the orientation of the orbital plane as seen from Earth: the waves are at their maximum loudness in the face-on direction, and half as loud from an edge-on point of view.
[13/]

Moreover, the sensitivity distance is conventionally based on an event with a signal-to-noise ratio of 8, but events could be lower-SNR and still be detectable. And finally, the sensitivity distance is also increased when you combine the power of multiple interferometers.
[14/]

For today's event to be a NS–NS merger, it would have to have been extreme in a number of ways:
1. The neutron stars had to be unusually massive.
2. They had to be almost exactly face-on.
3. They had to be in the best possible direction in the sky.
[15/]

Remove any one of these lucky coincidences, and it becomes pretty much impossible for the detectors to have seen a NS-NS event as far as 375 megaparsecs.
[16/]

Sathyaprakash told me, “it’s very unlikely that it was in an optimal position. From that, you can guess it can’t be a binary neutron star.” On the other hand, he warns, the distance could be overestimated. If it was substantially closer, NS–NS could again be favored
[17/]

But one of the objects being a black hole seems more plausible: unlike neutron stars, black holes are not limited in how massive they can be. “In that situation, we can see such systems at further distance, because of the heavy black hole being part of it,” Hanna told me.
[end]

Update 27 April: In today's updated sky map, the region to search has shrunk quite a bit, but the distance has not changed significantly...

https://twitter.com/cplberry/status/1122144943190028288