GWSkyNet-Multi: A Machine-learning Multiclass Classifier for LIGO–Virgo Public Alerts

Abstract

Compact object mergers which produce both detectable gravitational waves and electromagnetic (EM) emission can provide valuable insights into the neutron star equation of state, the tension in the Hubble constant, and the origin of the r-process elements. However, EM follow-up of gravitational wave sources is complicated by false-positive detections, and the transient nature of the associated EM emission. GWSkyNet-Multi is a machine learning model that attempts facilitate EM follow-up by providing real-time predictions of the source of a gravitational wave detection. The model uses information from Open Public Alerts (OPAs) released by LIGO–Virgo within minutes of a gravitational wave detection. GWSkyNet was introduced in Cabero et al. as a binary classifier and uses the OPA skymaps to classify sources as either astrophysical or as glitches. In this paper, we introduce GWSkyNet-Multi, an extension of GWSkyNet which further distinguishes sources as binary black hole mergers, mergers involving a neutron star, or non-astrophysical glitches. GWSkyNet-Multi is a sequence of three one-versus-all classifiers trained using a class-balanced and physically motivated source mass distribution. Training on this data set, we obtain test set accuracies of 93.7% for binary black hole-versus-all, 94.4% for neutron star-versus-all, and 95.1% for glitch-versus-all. We obtain an overall accuracy of 93.4% using a hierarchical classification scheme. Furthermore, we correctly identify 36 of the 40 gravitational wave detections from the first half of LIGO–Virgo’s third observing run (O3a) and present predictions for O3b sources. As gravitational wave detections increase in number and frequency, GWSkyNet-Multi will be a powerful tool for prioritizing successful EM follow-up.

Bayesian analysis of LIGO-Virgo mergers: Primordial vs. astrophysical black hole populations

“…Finally, it is interesting to consider how our results might change with the inclusion of recent new detections in the third observing run of LIGO-Virgo (O3). This has so far yielded a system, GW190814, with low mass ratio q ≈ 0.1 having one component in the lower mass gap with m ≈ 2.6 M [45], the black hole binary merger GW190412 with low mass ratio q ≈ 0.3 [44], and roughly 50 new binary black hole detections. A repeat of our analysis on the full sample of O3 events is forthcoming, so for now we focus on these two unusual systems 11 .…”

“…The lower mass cut-off is motivated by observations of X-ray binaries [66] and appears to be roughly 5 M , with a mass gap expected between this and the predicted upper limit for a neutron star of roughly 2 M (see however Ref. [45] for a recent detection of a compact object in this mass gap). The upper limit is more uncertain, and is partly motivated a posteriori from the GWTC-1 catalogue.…”

“…The S/N threshold roughly translates into suppression in the sensitivity to component masses below about 5 M relative to the maximum sensitivity at that redshift. The Figure makes it clear that LIGO in the O1O2 observing runs was sensitive to large mass ratios, with this increasingly true for the more sensitive O3 run which has yielded objects with q ≈ 0.3 [44] and q ≈ 0.1 [45]. Note that the sensitivity falls to zero at very high (a few hundred solar) masses where the waveforms have no support above the minimum frequency f low .…”

“…We close this section by noting that the weak constraints on the Model B parameter β q reflect the weak constraints on mass ratios in the GWTC-1 catalogue (see Figure 1). Recently the LIGO-Virgo Collaboration reported detections of BBH mergers with significantly asymmetric masses having q ≈ 0.3 [44] and q ≈ 0.1 [45]. Consequently the constraints on β q tighten significantly when these sources are included.…”

Minimally modified gravity with auxiliary constraints formalism

“…So far, GR stands strongly against tests such as consistency checks [4], merger remnants [5] and the properties of the generation and propagation (e.g. the propagating speed, dispersion relations and polarization states) of GWs [6][7][8][9][10][11][12][13] (see [14] for reviews).…”