What kind of thing is GW190814?

There’s been a lot of interest in the past day or two over an event that occurred in the LIGO detectors last August, entitled GW190814. A paper has appeared declaring this to be “the observation of a compact binary coalescence involving a 22.2–24.3 M  black hole and a compact object with a mass of 2.50–2.67 M “. That would be interesting of course because the smaller object is smaller than the black holes involved in previous detections and its mass suggests the possibility that it may be a neutron star, although no electromagnetic counterpart has yet been detected.  It’s a mystery.

I was quite excited when I saw the announcement about this yesterday but my enthusiasm was dampened a bit when I saw the data from the two LIGO detectors at Hanford and Livingston in the USA and the Virgo detector in Italy.

Visually, the Livingston detection seems reasonably firm, but the paper notes that there were thunderstorms in the area at the time of  GW190814 which affected the low-frequency data. There doesn’t look like anything at all but noise in the Virgo channel. The Hanford data may show something but, according to the paper, the detector was “not in nominal observing mode at the time of GW190814” so the data from this detector require special treatment. What you see in the Hanford channel looks rather similar to the two (presumably noise) features seen to the left in the Livingston plot.

I know that – not for the first time – I’m probably going to incur the wrath of my colleagues in the gravitational waves community but I have to sound a note of caution. Before asking whether the event involves a black hole or a neutron star you have to be convinced that the event is an event at all.  Fortunately, at least some of the data relating to this have been released and will no doubt be subjected to independent scrutiny.

Now I’m going to retreat into my bunker and hide from the inevitable comments…

11 Responses to “What kind of thing is GW190814?”

  1. Thaddeus Gutierrez Says:

    Of August, 2019 LIGO-Virgo suprevents (N=7), 4/7=57% were retracted https://gracedb.ligo.org/superevents/public/O3/
    Geophysical noise associated with storm activity is likely to have been dominated by broadband electromagnetic saturation around Schumann resonance modes, which LIGO has admitted are problematic in themselves, in addition to impulsive charging, particle showers, radio disturbances, and secondary magnetization from [often distant] TLE-generating thunderstorms.

    We would be led to believe that LIGO-Virgo can distinguish a true GW signal from a false trigger during periods of time when false triggers are most likely. My credulity had already been tested with GW190425 (magnetospheric mode, global-coherent CG lightning behavior, single-detector, very low confidence EM transient in elevated background at energetic particle-enhanced spacecraft altitude), and GW190412 (arrived during magnetospheric mode, golbal-coherent CG lightning, reconnection footprint, multipole moments indistinguishable from EM).

    Compare arrival order https://photos.app.goo.gl/RL6xSnr55zdjiABb8

    O2: GW170729 GW170809 GW170814 GW170817, GW170818, GW170823

    O3: S190727h S190728q S190808ae S190814bv S190816i S190822c

    For August 2019:

    retracted BNS trig. S190822c >99% BNS
    rRetracted NS-BH S190816i nearly identical luminosity dist. as putative NS-BH S190814bv. Notice confidence in a mass-gap assessment was 100% for S190814bv, then 100% NS-BH. https://gracedb.ligo.org/superevents/public/O3/

    DL ~40 Mpc #GW170817 BNS

    DL ~40 Mpc #S190822c BNS retracted

    DL ~267 Mpc #S190814bv NSBH

    DL ~261 Mpc #S190816i NSBH retracted

    See https://fulguritics.blogspot.com/2018/11/ligo-single-detector-trigger-rate-for.html
    and lightning, space weather for LIGO-Virgo events (including O3 retractions)

    • Phillip Helbig Says:

      Since weather information is available essentially in real time, if what you say is true (not just in the comment above but elsewhere), then you should be able to predict when LIGO will make an announcement. I’m sure that there are bookies in England who would be willing to bet on this. So why aren’t you rich?

      • Thaddeus Gutierrez Says:

        Space weather follows cycles as well, some of which are orbit-bound, while others are more complex and involve topological properties of the heliospheric current sheet responding to shock modulation of the interplanetary plasma within the long-range gravitational potential of the sun (and beyond).

        Well-behaved mutual order between discrete O3 samples [e.g. retractions[60, 129]≈non-retractions[62, 130]≈0.43|𝐾,
        𝑁=56 non-retracted, 𝑁=24 retracted
        1/(56/24)≈0.43]: https://photos.app.goo.gl/jg91PkCVe2ttypHR6

        We are discussing a kind of electromagnetic intermittency that is linked to cloud-ground lightning activity as a response function from a larger scale phase-locked special substorm mode affecting global lightning, ground polarity, and point discharges (which are much more common – 5:1 – than CG/GC discharges, and which are assumed to become coherent during the dipolarization intervals surrounding LIGO-Virgo events), and local coupling to storm noise involved much more than the acoustical energy from thunder (which is not expected to demonstrate a conveniently sharp strain-coupling cutoff at 30 Hz). Such a spurious transparency stunt masks the actual severity of the lightning/TLE/particle shower coupling issue associated with quasiperiodic reconnection phases and global-coherent bursting CG lightning associated with magnetospheric sawtooth injections.

        One can’t bet on LVC choice from “network SNR” given an almost meaningless conditional technique for estimating FAR for the purposes of dismissing morphologically-valid signals (which arrive on the order of 100/day), where multiple comparisons are evident from prior models anchored to self-confirmation of updated event lists. If all LVC signals are local and related to ordinary scale-invariant co-moving terrestrial coupling modes I have identified, then the diurnal/annual cyclicity of signal arrivals can be readily explained. Automorphic systematics are such that we would be foolish to ignore the statistical analysis of the output of LIGO’s own adaptive methodology, where non-retractions and retractions show geometric properties, which can be dismissed as orbit-linked to differential critical points between stationary horizons and continental footprint, with amphidromic/transient localization radial correlations inherited from thunderstorm domains.

        What does LIGO find? It shouldn’t have any non-random content after three experimental sampling periods, with all triggers restricted to rigid intervals with secular geographic oscillation for magnetic reconnection in all its splendor and subtlety.

      • Phillip Helbig Says:

        The LIGO announcements are weeks or months after the event. You claim that the event could be bogus, due to the weather effects you mention. But if that is the case, why don’t you do an analysis of the weather every day, as soon as possible, and not only on the day of an even announced (weeks or months later) by LIGO? If you are sure that the events are bogus, based on your analysis, then you should be able to predict when LIGO will make such a bogus announcement. So if you publish your predictions and are right much more often than could be due to chance, people will believe you, otherwise they won’t. Since you manage to do an analysis within a day or so of a LIGO announcement, the analysis can’t take that long, and you could do one every day.

      • telescoper Says:

        Actually candidate signals are announced almost immediately…

      • Thaddeus Gutierrez Says:

        Solar-IMF-ground connectivity:LVC trigger correspondence is further supported in consecutive event correlation to geometry of CME-ICME joint velocity field and its front/sheath boundary dispersive arrival times (deviations identical to conserved discrete bow shock-magnetotail propagation lags). GW190521g is another LVC trigger that has been published as a “confirmed” event concomitant with GW190814, but has received less attention. For consecutive O3 LIGO-Virgo triggers S190524q [retracted], S190521r, S190521g/GW190521g/GW190521, S190519bj, CACTus-LASCO CME, GOES X-ray flux, and ACE L1 interplanetary magnetic field and solar wind data clearly support corotating-coalescing CME/X-ray flux ToF|ToA for this segment of N=4 consecutive O3 LIGO-Virgo triggers during a period of enhanced Earth-sun connectivity – showing the emergence of complimentary lengths in field-restricted gyrofrequency:ToF, reproducing LVC arrival t0 (correlated CME and May 21, 2019 LIGO-Virgo S190521r-GW190521g trigger intervals of 4:48:00 and 4:41:33 respectively).

        Calculations and plots for this series of events: https://photos.app.goo.gl/DoNXJ1hgej6zWQo99

        Near-periodic Earth-solar wind-interplanetary magnetic field interaction periods follow daily and annual cycles that have been shown to also correlate to the arrival times LIGO-Virgo triggers as a population, which – paradoxically – never arrive during geosphysically quiet periods free of nonstationary transverse spectral contents. IMF-magnetosphere-ground reconnection of the extended scale occurring during all waveform-conformal LVC triggers is presaged by ubiquitous quasiperiodic-correlated noise floor intervals centered on LVC trigger ToA – especially as Earth sun connectivity to heliopause is polarity-restrictive undergoes discrete variation (crossings and interference) in localized coupling strength to enhanced solar wind/IMF with profound symmetry transformations. In other words, LIGO-Virgo cannot detect gravitational wave signals during quiet periods with Gaussian noise and low Earth-sun connectivity https://photos.app.goo.gl/jg91PkCVe2ttypHR6.

        Bursting global synchronized (dipolarization-injection front-triggered) CG lightning around N=4 consecutive LIGO-Virgo events including putative GW190521g, demonstrates the reality of global correlated ground state/noise during reconnecting quasiperiodic magnetospheric mode. Magnetic IMF/magnetospheric field structural disorder and field line reconnection phases are remarkable during LIGO-Virgo events. Particle precipitation in field-aligned currents and quasi-perpendicular boundary bifurcations (charged layer propagation) and magnetopause distortion also peak for phase during LIGO-Virgo triggers:

        1) S190524q (May 24, 2019 04:52:06 UTC)
        BATSRUS magnetic field lines
        BATSRUS magnetosphere (magnetopause) to 15 RE
        CG lightning around trigger

        2) S190521r (May 21, 2019 07:43:59 UTC)
        BATSRUS magnetic field lines
        BATSRUS magnetosphere (magnetopause) to 15 RE
        CG lightning around trigger

        3) S190521g/GW190521g/GW190521 (May 21, 2019 03:02:29 UTC)
        BATSRUS magnetic field lines
        BATSRUS magnetosphere (magnetopause) to 15 RE
        CG lightning around trigger

        4) S190519bj May 19, 2019 15:35:44 UTC
        BATSRUS magnetic field lines
        BATSRUS magnetosphere (magnetopause) to 15 RE
        CG lightning around trigger

        LIGO-Virgo N=24 O3 events up to S190521r synchronized with critical geoelectric field variation coupled by Earth-magnetosphere feedback to solar wind during quasi-coherent crossing events. Trailing shocks associated with recent long flare enhancements and ICME activity (303 km/s) during multiple connectivity phase map to S190521g-S190521r; a proton flare occurs immediately preceding GW190521g/S190521g arrival within 5-minute lag for Larmour-bound relativistic proton penetration to ground given intermittent magnetospheric configuration
        GW190521g/GW190521/S190521g high false alarm rate of 1 per 8.3367 years is being model-associated with a possible detection of non-coincident flaring from the assumed neighborhood of the model-dependent sky localization. The claimed finding is decorated by a supportive prediction of a second flaring event at ~1.6 years from initial observation at Palomar of the assumed delayed [non-coincident] bursting, which was also not found in archived data for at least a month after the initial LIGO-Virgo trigger. https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.124.251102
        Graham et al. 2020 is an extremely weak claim for other preclusionary reasons, such as the actual low SNR data involved and the highly uncertain distance of the modeled source, which maximizes systematic uncertainty. GW190521g/GW190521/S190521g, being the second “most distant” LVC trigger (luminosity distance 3931±953 Mpc), could only be constrained to within a very large 765 deg2 sky area.

        CACTus CME properties are automatically extracted from LASCO data http://sidc.be/cactus/. Heliospheric models utilized are avaliable at https://iswa.ccmc.gsfc.nasa.gov/. Lightning data obtained at http://en.blitzortung.org/archive_data.php. LIGO-Virgo event data available at https://gracedb.ligo.org/superevents/public/O3/.

  2. Must have been an elusive quark star than, which is supposed to have a predicted mass somewhere in between a neutron star and a black hole.

    • Jonivar Skullerud Says:

      If this is real then it is unlikely to be a quark star, since very few quark matter equations of state give maximum masses as high as this. If something in this mass range is confirmed not to be a black hole (eg, associated with electromagnetic or neutrino radiation) it would be quite a discovery and give QCD practitioners a lot of work to do!

  3. Jonathan Thornburg Says:

    Some thoughts on this event:

    The Ap.J.Letters paper is fully open-access, yea! (I have no idea of how much this might have cost the LIGO-Virgo collaboration (LVC).)

    The LVC Webinar outlining the observations and their interpretation is quite interesting. They ran out of space for the live zoom presentation (initially capped I think at 500 people), but they’ve posted a recording on youtube:

    As noted in the paper, although the LIGO Hanford detector was doing engineering work around the time of the event, “Within a 5 minute window around GW190814, this procedure was not taking place; therefore, LIGO Hanford data for GW190814 are usable in the nominal range of analyzed frequencies.”

    Also as noted in the paper, thunderstorms near LIGO Livingston “affects frequencies up to 30Hz”, so (for this detector) only data above 30Hz was used in the analysis.

    The fact that you (and I) don’t see much in the LIGO Hanford time-frequency plot, or in the Virgo plot, doesn’t actually argue against this being a compact-binary-coalescence gravitational-wave (GW) signal. If you had a signal strong enough to see by eye in a time-frequency plot, that would be nice, but if you think about a universe with sources randomly distributed, and a 1/r falloff in signal amplitude with (luminosity) distance, it’s easy to see that there will be many more weak signals than strong signals. (That is, there’s a much larger volume of the universe in which a given “absolute magnitude” source would be a weak signal here on the Earth, than there is where that same source would be a strong signal here on the Earth.)

    *If* all we know about the source was that it emitted some energy, then I suspect we couldn’t do much better than looking for high-energy regions in a time-frequency plot.
    (And in fact, the LVC does run this sort of analysis continuously, so they’re sensitive to “unmodelled signals”, as opposed to the things that depend on source models which I’m about to describe.)

    But we know know that binary compact-objects exist (whether these objects are black holes or neutron stars isn’t important to the point I’m making here), so it’s reasonable to also search the detector data stream for GW signals of the sort that binary compact-objects would emit. If general relativity (GR) is correct, a compact-binary coalescence (CBC) has a very distinctive GW pattern (which we first estimated from approximate analytical of the Einstein equations, and then since 2005 have been able to calculate by numerically solving the Einstein equations): CBC GWs are a “chirp”, a roughly sinusoidal waveform (recall that the actual detector output is phase-coherent, i.e., they’re measuring something analogous to cosine and sine components as a function of time) which gradually sweeps upwards in frequency and amplitude as the binary decays, until (if the compact objects are black holes) the signal cuts off and then decays at a much higher frequency as the post-merger binary relaxes to a stationary state.

    The actual detection of this event was done by matched filtering, looking for patterns in the data set which might be consistent with such a chirp. Matched filtering is much better than the human eye at spotting such signals, because it can look at the whole data pattern (sine & cosine components from each of 3 detectors, as a function of time) to see whether (in a Bayesian sense) this is more consistent with a precomputed “template” or with noise. (This relies on having an appropriate precomputed GW template. Since we don’t know the actual source parameters ahead of time, actual CBC matched-filtering searches repeat this test against each member of a “template bank” of some millions of templates, each corresponding to different values of the two object masses and other source parameters. A “detection” corresponds to one or more of the templates being a good match to the data stream.)

    Obviously there’s a multiple-comparison problem here (http://xkcd.com/882/) which has to be (and is) taken into account when calculating the statistical significance of such a detection.

    For this event the detection is quite robust: LIGO Livingston alone had a matched-filtering signal/noise of 21.6, LIGO Hanford 10.6, and Virgo 4.5. The overall network signal/noise ratio (combining data from all three detectors) was about 25:1. In other words, at the time of this event, all 3 detectors saw something which looked much more like a CBC template (convolved with the known detector response) than it looked like noise.

    The detector noise is highly non-Gaussian, non-white, and non-stationary, so calculating the “statistical signifcance” of such a detection is quite tricky. One way to describe this is the false-alarm rate, the estimated time it would take for the detector noise to produce a false-alarm signal this strong or stronger. As discussed in section 3.3 of the paper, there are several different ways to do this (e.g., do you include this event itself in “the detector noise”?), but their best estimates (page 9 of the paper, left column, end of the 2nd-to-last paper, describing the estimate “that yields a mean-unbiased estimate of the distribution of noise events”) are less than 1 false alarm in 10^5 years for one detection pipeline, and less than 1 in 4.2*10^4 years for another pipeline.

    So, the analysis suggests that this was very likely an actual CBC GW signal.

    To me, the most convincing argument that this is indeed a black-hole/{either black hole or neutron star} coalescence signal is section 4.2 of the paper, especially figure 7, which shows a clear detection of a (phase-coherent) higher-multipole signal at 1.5 times the dominant (quadrupole) frequency. (For this detection they estimate a classical p-value of p < 2.2*10^-4.) (There's also a weaker detection at 2.0 times the dominant frequency.) This is as predicted from GR, and doesn't seem like something that you'd expect to see in detector noise.

  4. Jonathan Thornburg Says:

    There’s some discussion of possible formation scenarios for the secondary in the LVC colloquium slides, which can be downloaded (pdf) at

  5. Thaddeus Gutierrez Says:

    First, we would have to be completely conditioned to terminate meta-experimental inquiry at the most superficial point to ignore what is happening surrounding Earth during LIGO-Virgo triggers, which is exactly the kind of environment (both socially and physically) to produce false positive claims. The choices in hypotheses people have been given is narrative-bound in almost all respects to the degree that is expected where almost pure induction-retroduction . Our specializations are no excuse for these lapses, as foreground data/models are widely available and show that LIGO-Virgo events are not independent of the properties of magnetospheric mode. Population statistics utilized to evaluate FAR and other probabilities are derived from sampling from this window-biased kind of fishing, and as such are either equivocal or circular at best.

    For example, for so-called GW190814 (Aug. 14, 2019
    21:10:39 UTC), the interplanetary environment was highly-unstable and undergoing multiscaled reconnection events within corotating interaction regions that affect the phase lengths and disorder of the local geomagnetic vacuum – itself nonstationary.
    BATSRUS magnetopause/bowshock/nightside reconnection-injection region
    BATSRUS IMF-magnetospheric-ground magnetic field line connectivity
    ENLIL heliospheric density-magnetic field line connectivity

    INTEGRAL and Fermi orbits can be found to have crossed enhanced nightside regions during all events that have become bulwarks for LVC multimessenger appeals. These conditions are also difficult to reject as correlated to network error. Terrestrial [relativistic] coherent particle bursts (proton flares, TGFs) seem to be ignored.
    For GW170817, BATSRUS IMF-magnetospheric-ground magnetic field line connectivity https://photos.app.goo.gl/MVK2L8yr8pP8dYgLA, BATSRUS magnetopause/bowshock/nightside reconnection-injection region
    For GW190425, BATSRUS IMF-magnetospheric-ground magnetic field line connectivity https://photos.app.goo.gl/7H8HM2QNrnBAw6Rp8,
    BATSRUS magnetopause/bowshock/nightside reconnection-injection region https://photos.app.goo.gl/ckoMPDe8R2yHwAdP8.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

<span>%d</span> bloggers like this: