A Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814
K. D. Alexander, G. Schroeder, K. Paterson, W. Fong, P. Cowperthwaite, S. Gomez, B. Margalit, R. Margutti, E. Berger, P. Blanchard, R. Chornock, T. Eftekhari, T. Laskar, B. D. Metzger, M. Nicholl, V. A. Villar, P. K. G. Williams
DDraft version February 19, 2021
Typeset using L A TEX twocolumn style in AASTeX63
A Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814
K. D. Alexander, ∗ G. Schroeder, K. Paterson, W. Fong, P. Cowperthwaite, † S. Gomez, B. Margalit, ∗ R. Margutti, E. Berger, P. Blanchard, R. Chornock, T. Eftekhari, T. Laskar, B. D. Metzger,
6, 7
M. Nicholl, V. A. Villar, ‡ and P. K. G. Williams
3, 10 Center for Interdisciplinary Exploration and Research in Astrophysics (CIERA) and Department of Physics and Astronomy,Northwestern University, Evanston, IL 60208, USA The Observatories of the Carnegie Institution for Science, 813 Santa Barbara Street, Pasadena, CA 91101, USA Center for Astrophysics | Harvard & Smithsonian, 60 Garden St., Cambridge, MA 02138, USA Astronomy Department and Theoretical Astrophysics Center, University of California, Berkeley, Berkeley, CA 94720, USA Department of Physics, University of Bath, Claverton Down, Bath BA2 7AY, UK Department of Physics and Columbia Astrophysics Laboratory, Columbia University, New York, NY 10025, USA Center for Computational Astrophysics, Flatiron Institute, 162 5th Ave, NY 10011, USA Birmingham Institute for Gravitational Wave Astronomy and School of Physics and Astronomy, University of Birmingham,Birmingham B15 2TT, UK Department of Astronomy, Columbia University, New York, NY 10027-6601, USA American Astronomical Society, 1667 K St. NW Ste. 800, Washington, DC 20006, USA
ABSTRACTGW190814 was a compact object binary coalescence detected in gravitational waves by AdvancedLIGO and Advanced Virgo that garnered exceptional community interest due to its excellent local-ization and the uncertain nature of the binary’s lighter-mass component (either the heaviest knownneutron star, or the lightest known black hole). Despite extensive follow up observations, no elec-tromagnetic counterpart has been identified. Here we present new radio observations of 75 galaxieswithin the localization volume at ∆ t ≈ −
266 days post-merger. Our observations cover ∼ ∼ ◦ (the best-fit binary inclinationderived from the gravitational wave signal) and assumed electron and magnetic field energy fractionsof (cid:15) e = 0 . (cid:15) B = 0 .
01, we can rule out a typical short gamma-ray burst-like Gaussian jet withisotropic-equivalent kinetic energy 2 × erg propagating into a constant density medium n (cid:38) . − . These are the first limits resulting from a galaxy-targeted search for a radio counterpart to agravitational wave event, and we discuss the challenges, and possible advantages, of applying similarsearch strategies to future events using current and upcoming radio facilities. Keywords: radio sources (1358) – radio transient sources (2008) — gravitational waves (678) INTRODUCTIONRecent detections of gravitational waves (GW) haverevolutionized our understanding of the populationof compact object binaries, impacting many areas ofphysics and astrophysics (Abbott et al. 2016, 2017a,2019a,b, 2020b,c; LIGO Scientific Collaboration et al.2020). While much can be learned from the GW sig- ∗ NHFP Einstein Fellow † NHFP Hubble Fellow ‡ Simons Junior Fellow nals alone, understanding the full astrophysical con-text of the merger event, including the association toits host galaxy, requires discovery of an electromag-netic (EM) counterpart. Mergers of two neutron starsare thus of particular interest, as they are predicted toproduce radiation across the EM spectrum and havebeen long-theorized to be the origin of short gamma-ray bursts (SGRBs; e.g. Narayan et al. 1992; Eichleret al. 1989; Fong & Berger 2013; Berger 2014). Thiswas spectacularly confirmed by the discovery of the bi-nary neutron star (BNS) merger GW170817 (Abbottet al. 2017a,b), which not only had associated gamma- a r X i v : . [ a s t r o - ph . H E ] F e b Alexander et al. ray emission (Goldstein et al. 2017; Abbott et al. 2017c;Savchenko et al. 2017), but also a bright kilonova de-tected in the ultraviolet, optical, and IR bands (An-dreoni et al. 2017; Arcavi et al. 2017a; Chornock et al.2017; Cowperthwaite et al. 2017; Coulter et al. 2017;D´ıaz et al. 2017; Drout et al. 2017; Kasliwal et al. 2017;Nicholl et al. 2017; Lipunov et al. 2017; Pian et al. 2017;Pozanenko et al. 2017; Smartt et al. 2017; Tanvir et al.2017; Utsumi et al. 2017; Valenti et al. 2017; Villar et al.2017, 2018) and a long-lasting synchrotron afterglow de-tected from radio through X-ray wavelengths (Alexan-der et al. 2017, 2018; Haggard et al. 2017; Hallinan et al.2017; Margutti et al. 2017, 2018; Troja et al. 2017, 2018,2019, 2020; D’Avanzo et al. 2018; Dobie et al. 2018; Ly-man et al. 2018; Mooley et al. 2018a,b,c; Nynka et al.2018; Ruan et al. 2018; Fong et al. 2019; Ghirlanda et al.2019; Hajela et al. 2019; Lamb et al. 2019; Piro et al.2019).Mergers between a neutron star and a black hole arealso predicted to result in detectable EM emission insome cases, particularly if the mass ratio of the bi-nary is not too extreme (Kawaguchi et al. 2016; Metzger2019). It is however an open question whether neutronstar-black hole (NSBH) mergers also produce SGRBs(Murguia-Berthier et al. 2017; Gompertz et al. 2020),and how those SGRBs would compare to the cosmologi-cal SGRB population. The prompt gamma-ray emissionfrom off-axis relativistic jets is highly suppressed due torelativistic beaming, and would likely be undetectableat the larger distances where most GW mergers will oc-cur (Abbott et al. 2017c; Goldstein et al. 2017; Margutti& Chornock 2020). Thus, except for the small fractionof on-axis mergers, the best opportunity to determinewhether NSBH mergers produce relativistic jets or out-flows of sub-relativistic material is to search for syn-chrotron emission in the radio or X-ray, as was done forGW170817.On 2019 August 14, Advanced LIGO/Virgo reportedthe detection of a new compact object merger candidateGW190814, with a preliminary false alarm rate of one in10 years (GCN 25324; LIGO Scientific Collaboration& Virgo Collaboration 2019a). It was initially classifiedas a MassGap event (meaning that the lighter member ofthe binary had a mass between 3 − M (cid:12) ), but the classi-fication was revised to a NSBH merger approximately 12hours later (GCN 25333; LIGO Scientific Collaboration& Virgo Collaboration 2019b). This classification, to-gether with the excellent localization (23 deg with 90%confidence in the skymap provided by LIGO ScientificCollaboration & Virgo Collaboration 2019b 13.5 hourspost-merger), generated considerable interest and tele-scope investment from the astronomy community. Nu- merous follow up efforts across the EM spectrum re-vealed no evidence for any counterpart (Dobie et al.2019; Gomez et al. 2019; Ackley et al. 2020; Andreoniet al. 2020; Antier et al. 2020; Gompertz et al. 2020;Morgan et al. 2020; Page et al. 2020; Thakur et al. 2020;Vieira et al. 2020; Watson et al. 2020), broadly consis-tent with the highly unequal binary mass ratio revealedby the full gravitational wave analysis (Abbott et al.2020b) and the small NS radius inferred from observa-tions of GW170817 (Capano et al. 2020). The natureof the lighter 2 . +0 . − . M (cid:12) component – neutron star orblack hole – remains unclear (LIGO Scientific Collabo-ration et al. 2020; Essick & Landry 2020; Tsokaros et al.2020; Tews et al. 2021). Nevertheless, GW190814 pro-vided an excellent test-bed for various multi-wavelengthobserving strategies, as its precise localization (tight-ened to 18.5 deg in the final analysis by Abbott et al.2020b) and large distance (241 +41 − Mpc) will likely betypical of GW events discovered in O4 and beyond.In particular, GW190814 prompted several indepen-dent searches for a radio counterpart. Unlike the opti-cal sky, the variable radio sky is not well-characterizedon timescales of months at the typical flux densitiesof plausible gravitational-wave transients (although thebackground rate of extragalactic radio transients is ex-pected to be low, e.g. Metzger et al. 2015). Severalwide-field radio searches for GW merger counterpartshave been previously employed even in cases when nobright radio counterpart is expected, to better under-stand the likely background rates of potential contami-nating sources (e.g. Mooley et al. 2018d; Bhakta et al.2020). However, GW190814 was the first event for whicha significant fraction of the localization area could becovered to any significant depth by current radio facil-ities. A wide-field single-frequency radio search cover-ing 89% of the LIGO Scientific Collaboration & VirgoCollaboration (2019b) localization region was conductedwith ASKAP at early times (2 −
33 days post-merger),ruling out the presence of an on-axis relativistic jetwith isotropic-equivalent kinetic energy E iso = 10 ergwithin the observed region under standard assumptionsabout the jet microphysics (Dobie et al. 2019). This isconsistent with the lack of bright X-ray or gamma-rayemission observed at early times, which would have beenexpected if such a jet were present (Palmer et al. 2019;Page et al. 2020; Watson et al. 2020).Here, we present targeted late-time radio observationsof 75 galaxies within GW190814’s localization volume,spanning 1 − Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 +17 − deg; Abbott et al. 2020b). The timescale of ourobservations also allows for some limited constraints onthe presence of slower-moving kilonova (KN) ejecta. Asecondary goal is to characterize the background of vari-able and transient radio sources likely to be encounteredin future radio searches for gravitational wave coun-terparts, with a focus on the implications for galaxy-targeted strategies. We present our observations in Sec-tion 2 and discuss our counterpart search and additionaltargeted observations of our most promising candidatein Section 3, ultimately concluding that this object ismore likely to be an unrelated background source. InSection 4, we place limits on the existence of a radiocounterpart to GW190814 and discuss the nature of theunrelated variable radio sources uncovered in our search.We conclude in Section 5 with some implications of ourwork for future radio searches for GW counterparts. OBSERVATIONS2.1.
Galaxy selection criteria
To maximize the likelihood of observing the counter-part location with a minimal expenditure of telescopetime, we selected a galaxy-targeted search strategy forGW190814. Galaxy-targeted searches are particularlyappropriate for telescopes with smaller instantaneousfields of view (like the VLA). The galaxies we targetinclude many of the most optically luminous galaxieswithin the localization volume, the same galaxies thathave been prioritized for counterpart searches at opti-cal and X-ray wavelengths. Such galaxies are attrac-tive targets because they contain much of the stellarmass within the localization volume (and thus have thehighest probability of actually containing the merger).However, they are also more likely to have detectableradio emission from unrelated sources (e.g. star forma-tion, AGN activity), making searches for radio tran-sients more challenging.To select our targets, we generated a list of all galax-ies from the Galaxy List for the Advanced Detector Era(GLADE) catalog (D´alya et al. 2018) with B -band lu-minosities L (cid:38) . L ∗ in the LALINFERENCE
90% local-ization volume that was circulated by the LIGO/Virgocollaboration 13.5 hr post-discovery (LIGO ScientificCollaboration & Virgo Collaboration 2019b). We thenranked this list based on a weighting of the galaxy’sspatial position within the localization volume and thegalaxy’s B -band luminosity (a proxy for stellar mass),using the same procedure followed in Gomez et al. (2019)and Hosseinzadeh et al. (2019). We observed the top 75galaxies on this list, out of a total of 723 galaxies. The observed galaxies contained ∼
21% of the cataloged in-tegrated stellar luminosity in the region. However, theGLADE catalog is known to be incomplete at the dis-tance of GW190814 (D´alya et al. 2018; Abbott et al.2020b). We therefore estimate the completeness of thecatalog by integrating a Schechter B-band galaxy lumi-nosity function down to 0 . L ∗ to approximate the truetotal number of galaxies and the corresponding total in-tegrated stellar luminosity contained in the region (seeGomez et al. 2019 for the exact function used). We findthat the GLADE catalog is ∼
50% complete down to0 . L ∗ at the distance of GW190814, in terms of numberof galaxies. We estimate that our 75 galaxies contain14% of the total integrated stellar luminosity within the LALINFERENCE localization volume.The final localization volume presented in Ab-bott et al. (2020b) shifted slightly compared to the
LALINFERENCE localization volume (Figure 1, left panel)and shrank significantly, from 1 . × Mpc to 3 . × Mpc . We therefore repeat the above calculations forthe final localization volume. We find that 65 of our tar-get galaxies remain within the final 90% localization vol-ume presented in Abbott et al. (2020b), comprising 32%of the total integrated stellar luminosity within this re-gion. Thus, assuming that the merger probability tracksstellar light, we have a roughly 32% chance that our ob-servations covered the true position of the merger.2.2. VLA observations
We observed the top 75 galaxies in our ranked list inSeptember 2019 ( ∼ ∼ − clean . Alexander et al. h m m m h m -25°-30° Right Ascension D e c li n a t i o n LALINFERENCE SkymapPublished Skymap
ESO 474-035
30 arcsec
NECandidate 1 (variable) dR = 98 kpc Source 2 (constant)
Figure 1.
Left: The positions of our 75 galaxies (purple circles) in relation to the GW skymaps (dotted lines show the
LALINFERENCE map used to create our ranked list; solid lines show the revised final map from the full GW analysis, Abbott et al.2020b). The size of each circle corresponds to the area searched around each galaxy (100 kpc physical offset, 61 − (cid:48)(cid:48) dependingon the galaxy distance). The position of the galaxy ESO 474-035 is highlighted in white. Right: The field containing ESO474-035, as seen in the optical (grayscale, i -band image from Magellan taken 1.48 days post-merger) and the radio (red contours,6 GHz image from our VLA program taken 213 days post-merger). We detect two significant point-like radio sources within oursearch region (dashed light blue line): “Candidate 1” (variable, not detected at 35 days post-merger), and “Source 2” (consistentwith constant flux density). We also observe weak extended emission near the nucleus of ESO 474-035. At the distance of ESO474-035 ( d L = 271 Mpc), Candidate 1 would be at a projected offset of 98.0 kpc (blue line), at the upper end of the distributionmeasured from short GRBs. Unlike Source 2, which has a clear optical counterpart, Candidate 1 has no coincident opticalemission to i (cid:38) .
23 mag at 1.48 days after the merger, and no underlying, static host galaxy to r = 24 .
92 mag.
The first epoch of observations was taken when theVLA was in its most extended A configuration (beam-size ∼ . (cid:48)(cid:48) at 6 GHz) and the second epoch when it wasin its more compact C configuration (beamsize ∼ . (cid:48)(cid:48) at 6 GHz). As any physically-plausible radio afterglowat the distance implied by the GW signal ( d L = 241 +41 − Mpc; Abbott et al. 2020b) is predicted to be an un-resolved point source on the timescale of our observa-tions, the differing resolution of the radio observationsshould not impact the flux density recovered for any(sufficiently isolated) bona fide counterpart. Neverthe-less, the C configuration data is more sensitive to diffuseemission from the merger host galaxy or other sourcesin the field, which caused additional challenges with thedata imaging for a subset of our targets. This manifestsas an elevated rms noise level in a small fraction of ourimages. In epoch 1, we achieved a median image rms of12.2 µ Jy beam − and in epoch 2, we achieved a medianimage rms of 17.6 µ Jy beam − . In both epochs, ourtypical time on source was ∼
6m 20s for each galaxy. COUNTERPART SEARCH3.1.
Candidate Selection Criteria
We next searched each galaxy field for a radio counter-part to GW190814. To identify radio sources, we usedSource Extractor (Bertin & Arnouts 1996) in combina-tion with the distance from the GLADE catalog to lo-cate all radio sources detected with > σ significance ineach image located within 100 kpc of each of our targetgalaxies. We chose this search radius because (cid:38) imtool fitsrc com-mand within the pwkit package (Williams et al. 2017).All flux densities were extracted assuming a point source Distances are calculated using an assumed flat ΛCDM cosmologywith H = 70 km s − Mpc − , Ω M = 0 .
27, and Ω Λ = 0 . Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 V s m >1 mJy0.1-1 mJy<0.1 mJysingle epoch detectioncoincident with knownoptical source (PS1)variable (this work)Candidate 1GW170817off-axis tophat jet model Figure 2.
The variability statistic ( V s ) versus modulation index ( m ) for the population of radio sources detected in ourobservations (circles are sources detected in both epochs, diamonds are detected in only one epoch). The sources outlined inblack are coincident with optical sources in archival data; we consider them less likely to be possible counterparts to GW190814.Sources within the darker gray shaded region would have been defined as significantly variable by Radcliffe et al. (2019), whilethe lighter gray region indicates the definition used by Bhakta et al. (2020). However, these criteria may miss genuine GWcounterparts: for example, a model off-axis tophat jet with E K,iso = 5 × erg, n = 0.09 cm − and viewing angle 45 ◦ would appear at the position of the blue star if it were observed with the cadence and sensitivity of our observations, andGW170817’s radio counterpart is not obviously highly variable on the timescale of our observations (magenta star). We utilizea less-restrictive variability definition in this work, resulting in a larger sample of possibly variable objects (black crosses). Weselect one of these variables for additional multi-frequency follow up, based on its large modulation index and its lack of anoptical counterpart (Candidate 1, red star). fit. We found that due to the differing rms noise levelin the epoch 1 and 2 images, some sources that wereonly recovered by Source Extractor with > σ signifi-cance in one of the two epochs may nevertheless be con-sistent with constant flux density. For each source de-tected in only one epoch, we therefore ran fitsrc at thesource position in the other epoch to search for lower-significance emission. We then used the recovered fluxdensity (or 3 times the image rms at the source positionif no emission was detected) to measure the variabilityof each object.We assess the variability of our radio sources in severalways. Several previous wide-field searches for radio tran-sients have characterized variability between two epochsin terms of the modulation index, m , and the variabilityt-statistic, V s (e.g. Mooley et al. 2016; Radcliffe et al.2019; Bhakta et al. 2020). We define: m = ∆ S (cid:104) S (cid:105) = 2 S − S S + S (1) where S is the flux density of each source, as determinedfrom a point source fit with fitsrc , and V s = (cid:12)(cid:12)(cid:12)(cid:12) ∆ Sσ (cid:12)(cid:12)(cid:12)(cid:12) = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) S − S (cid:112) σ + σ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) (2)where σ is the total measurement error on the flux den-sity. We include both the uncertainty derived from thepoint source fit and an additional error term of 5% corre-sponding to the accuracy of the absolute flux calibrationscale of the VLA (Perley & Butler 2017) in this quan-tity. We plot the distribution of our radio sources in V s and m in Figure 2.Previous work has often focused on maximizing thepurity of constructed samples of radio variables andtransients, and thus has imposed fairly strict cutoffs forvariability: Mooley et al. (2016) and Radcliffe et al.(2019) require V s > . | m | > .
26 (darker grayshaded region in Figure 2), while Bhakta et al. (2020)require V s > | m | > .
18 (light gray shaded re-
Alexander et al. gion). Applying these criteria, we find 8 and 13 variablesources respectively in our sample, corresponding to avariability fraction of 12% and 19%. This is significantlyhigher than previous blind untargeted searches, whichhave found that only a few percent of radio sources overlarge sky areas exhibit this level of variability (e.g. Carilliet al. 2003; Frail et al. 2012; Mooley et al. 2016; Bhan-dari et al. 2018; Radcliffe et al. 2019; Dobie et al. 2019;Bhakta et al. 2020; Sarbadhicary et al. 2020). We notethat all 13 variables are coincident with bright galaxiesin archival optical imaging, and 12 of them are coinci-dent with the nuclei of our target galaxies to within as-trometric errors. One natural explanation for the higherprevalence of variability in our sample compared to pre-vious work is thus that the centers of (relatively) nearbygalaxies selected as likely hosts for GW mergers are morelikely than average regions of space to contain sources ofdetectable radio emission (e.g. weak AGN). We explorethis further in Section 4.3.One downside of optimizing for sample purity in ourdata is that GW radio counterparts are expected overa broad range of timescales: weeks to months for rela-tivistic jet afterglows, or months to years or even decadesfor kilonova afterglows (e.g. Nakar & Piran 2011). Thus,with only two epochs of data, it is worth exploring ad-ditional, conservative methods to ensure that transientswith variability timescales not well-aligned to our ob-serving cadence are also discovered. For example, suchtechniques would be necessary to recover a GW170817-like radio transient in our data: if we compute m and V s for GW170817 using the 6 GHz observations collectedclosest to the timescale of our observations (at 39 d and217 d respectively; Alexander et al. 2017, 2018), it wouldnot be classified as a variable source by the methodsoutlined in either Radcliffe et al. (2019) or Bhakta et al.(2020) (Figure 2, magenta star). Even an off-axis rel-ativistic jet that peaks on the timescale of our secondepoch may not satisfy the criterion for V s given the sen-sitivities achieved in our two epochs (e.g. sample modelin Figure 2, blue star).We are thus motivated to explore a less stringent cri-terion for variability, to emphasize completeness of oursample rather than purity. We therefore create a list ofall radio sources inconsistent with constant flux densityto within the > σ measurement uncertainties calcu-lated by fitsrc (using 3 times the image rms at thesource position as an upper limit on the flux density for This is equivalent to demonstrating variability at the > σ or > σ confidence level, respectively, in the case of Gaussian noise.More generally, V s ≤ . t -statistic. sources only detected in one epoch). We identified 30potential radio counterparts using this criterion (Figure2, black crosses). 28 of our 30 variable sources (includingall 13 of the “highly significant variables” satisfying theBhakta et al. 2020 criteria discussed above) increasedin brightness between epochs 1 and 2, while only twosources decreased in brightness. We do not expect thisimbalance to result from differences in absolute flux cal-ibration between the two epochs, as the variables werenot preferentially observed in any single scheduling block(and as mentioned above, we already include a conserva-tive additional 5% uncertainty term in the measurementerrors used to compute V s , based on the known accuracyof the VLA flux calibration scale). Instead, the apparentincrease in flux density of many of our sources may bepartially explained by the reconfiguration of the VLAbetween epochs 1 and 2: the C configuration epoch 2data are more sensitive to diffuse low surface brightnessemission (from e.g. ongoing star formation) than the Aconfiguration epoch 1 data. (This is also consistent withthe fact that a number of our radio sources appear point-like in epoch 1 and slightly extended in epoch 2, despitethe epoch 2 data having ∼ × lower resolution.)The centers of galaxies may be particularly prone tosuch spurious detections of variability, as they may con-tain a superposition of extended and compact emissioncomponents from star formation and/or AGN activity.Indeed, the positions of 17 of our variable sources areconsistent with the centers of their target GLADE galax-ies to within astrometric uncertainties. While somemodels predict an enhanced rate of compact objectmergers near the supermassive black holes at the centersof galaxies (Antonini & Perets 2012; McKernan et al.2020; Perna et al. 2021; Zhu et al. 2021), this resultsuggests that GW counterparts near galactic nuclei willbe particularly challenging to identify in the radio. Werule out these 17 sources as likely radio counterparts toGW190814.Finally, we searched the VLA Sky Survey (VLASS;Lacy et al. 2020) Quick Look images and thePanSTARRS-1 data archive (PS1; Chambers et al. 2016)at the positions of the remaining candidates to provideadditional insight into their nature. Three sources aremarginally detected in VLASS epoch 1 data that pre-dates GW190814, suggesting that they are unrelated tothe merger. Furthermore, we found that all but sixsources had a spatially coincident optical counterpartin the PS1 catalog, suggesting that these radio sourcesare also likely not plausible counterparts to GW190814.While the exact peak timescale of a hypothetical radiocounterpart to GW190814 is uncertain, models of off-axis relativistic jets with θ obs ∼ ◦ (GW190814’s bi- Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 Figure 3.
The 6.0 GHz light curve (left) and spectral energy distribution at 266 d (right) of Candidate 1 (black points, errorbars are 1 σ ). The triangle indicates a 3 σ upper limit. The red lines represent one allowed model from our kilonova ejectamodeling ( n = 10 − cm − , E ej = 3 . × erg, M ej = 0 . M (cid:12) , p = 3 . (cid:15) e = 0 . (cid:15) B = 0 . nary inclination) and parameters typical of short GRBsare expected to show significant variability between thetimescales of our two epochs (Figure 2, blue star). Ofour remaining six candidates, the source that shows thelargest | m | is located at R.A. = 00 h m s .418, Dec = -25 ◦ (cid:48) (cid:48)(cid:48) .16 (J2000), within the field of the galaxy ESO474-035, which was ranked sixth in our catalog of 75galaxies (Figure 1). We discuss this source (hereafter“Candidate 1”) and its evolution in more detail in thenext section.3.2. Modeling of Candidate 1 in the Field of ESO474-035
We now focus on Candidate 1 to assess its viability asa counterpart to GW190814. Previous Magellan obser-vations at 1.48 d after the GW trigger cover both thecatalogued GLADE galaxy ESO 474-035 and the posi-tion of Candidate 1, and placed a limit of i > .
23 magon any transient optical emission at that time (Figure1; see also Gomez et al. 2019). Moreover, deeper, pre-merger limits at the candidate position from the LegacySurvey (Dey et al. 2019) of r = 24 .
92 mag, derivedfrom the 5 σ depth of co-added DECam images (BrickID 0131m257), place stringent constraints on an under-lying source. For a constraint on a satellite galaxy atthe same distance as ESO 474-035, this translates to a L (cid:46) . × L (cid:12) . This is roughly 4 orders of magnitudebelow the luminosity of the Milky Way, ruling out allexcept the faintest dwarf galaxy regime (Simon 2019).Moreover, the lowest host galaxy luminosities derivedfor short GRBs are ≈ L (cid:12) (Berger 2014), well abovethe limit derived here. If instead there is a backgroundgalaxy at a similar luminosity to the Milky Way at theposition, it would need to be at z (cid:38) (cid:38) × ergs − for the radio transient, comparable to radio-loudquasars (Kellermann et al. 2016). Thus, we find thatwhile a background quasar cannot be ruled out, if theorigin of Candidate 1 is from a stellar progenitor, thenit likely originated from ESO 474-035.We triggered additional multi-frequency radio obser-vations of Candidate 1 at 1 −
12 GHz, which were carriedout on 2020 May 5 ( t = 266 d, “epoch 3”), with the VLAin C configuration. Our observations reveal a continuedincrease in the flux density at 6 GHz (Figure 3, left),confirming that the change in flux density is intrinsic tothe source, rather than an artifact of the VLA configu-ration change between epochs 1 and 2. The broadbandspectrum is optically-thin, consistent with a single powerlaw F ν ∝ ν β , where β = − . ± .
06 (Figure 3, right).At the highest frequencies observed (8 −
12 GHz), wenote that the emission appears partially resolved intotwo components, with centroids separated by ∼ (cid:48)(cid:48) . Atthe distance of ESO 474-035 (271 Mpc) this would cor-respond to a physical separation of ∼ Alexander et al. high-resolution data are not available. We therefore in-vest some additional effort in fitting Candidate 1’s radioemission with models appropriate for GW counterparts,to see if we can distinguish Candidate 1 from a typicalradio GW counterpart based on the physical propertiesrequired to fit the flux density alone. For this analysis,we use the combined flux density of the two resolvedcomponents when modeling the high-frequency epoch 3data and we assume that Candidate 1 is at the distanceof ESO 474-035.We consider two classes of radio GW counterpart mod-els for Candidate 1: radio emission from collimated fastejecta (i.e. an initially relativistic jet, possibly withsome velocity structure) and from slower ejecta (thesame material in which r-process nucleosynthesis occursat early times, producing the kilonova optical transient;assumed to be quasi-spherical and moving at up to afew tens of percent of c ). In both cases, the radio emis-sion is synchrotron radiation arising from a populationof electrons accelerated into a power-law distribution ofenergies, N ( γ ) ∝ γ − p for γ > γ m , as the merger ejectashocks and interacts with the surrounding interstellarmedium. The allowed parameter space is highly degen-erate, as we observe only the rising portion of the lightcurve and a single power-law segment of the synchrotronspectral energy distribution. We find p = 3 . ± .
11 usingthe spectral slope computed from our multi-frequencyobservations ( β = − . ± . ν m (the synchrotron frequency cor-responding to γ m ) and below the cooling frequency ( ν c )(Granot & Sari 2002). This value is in line with some su-pernovae (see e.g. Van Dyk et al. 1994; Chevalier 1998),although it disagrees with the precise value p = 2 . +0 . − . calculated for GW170817 (Hajela et al. 2019) and it ishigher than previously observed in many cosmologicalSGRB afterglows, where typically 2 < p < p = 3 . θ obs = 46 ◦ (to put the jet in alignmentwith the best-fit binary inclination as derived from theGW signal; Abbott et al. 2020b), and we assume thatthe fractions of energy carried by electrons and mag-netic fields in the shock are (cid:15) e = 0 . (cid:15) B = 0 . p, (cid:15) e and (cid:15) B are poorly constrained andmust be assumed. The full GW170817 dataset sug-gests that (cid:15) B may be much lower than 0.01 in at leastsome jets (Margutti & Chornock 2020); if this is truefor GW190814, then the constraints derived below are Figure 4.
The allowed E ej vs n phase space for Candidate 1,under the assumption that the radio emission is produced bythe shock between 0 . M (cid:12) of quasi-spherical ejecta and theambient medium (with p = 3 . (cid:15) e = 0 .
1, and (cid:15) B = 0 . − are ruled out by the 6 GHzlight curve. also affected. We consider both tophat jets (in whichall of the jet energy is contained within a narrow conewith opening angle 15 ◦ ) and jets with a relativistic coresurrounded by Gaussian wings of slower-moving mate-rial. We find that only the tophat jet models can re-produce the steep rise seen in Candidate 1’s 6 GHz lightcurve, and furthermore that we require a large isotropic-equivalent jet energy ( E K,iso ∼ × erg) and a highdensity ( n ∼ . − ); comparable to the largest en-ergy values and in the top 40% of density values inferredfor SGRBs (Fong et al. 2015). While we cannot entirelyrule out the possibility that Candidate 1 is an off-axisrelativistic jet launched by GW190814, we disfavor thispossibility due to the large energy required and the highvalue of p . The high density would also be unexpectedfor such a highly-offset transient, particularly due to thelack of any optical emission at the transient position tosuggest a satellite galaxy or globular cluster environ-ment for the transient.We next compare the radio behavior to slow kilo-nova (KN) ejecta models to determine whether theyare consistent with our observations. Following the pre-scriptions of Schroeder et al. (2020), we modeled the 6GHz lightcurve and multi-frequency spectrum of Can-didate 1 with a KN ejecta interaction model. We againfix p = 3 . (cid:15) e = 0 .
1, and (cid:15) B = 0 .
01 in our mod-eling. Optical and near-infrared follow-up studies ofGW190814 have placed constraints on the ejecta mass
Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 M ej (cid:46) . − . M (cid:12) (Gomez et al. 2019; Kawaguchiet al. 2020; Andreoni et al. 2020; Morgan et al. 2020;Ackley et al. 2020; Thakur et al. 2020; Vieira et al. 2020).We set the ejecta mass M ej to 0 . M (cid:12) , as lower valuesof M (cid:12) would cause the time of observation of epoch 3to approach the deceleration time, t dec , of the ejecta.If t dec ≈ t obs , , the time of observation of epoch 3, theKN model light curve would start to decline, whereas weobserve the 6 GHz light curve still rising through epoch3. Even after making these assumptions, several param-eter degeneracies remain in our modeling. We thereforecreate a grid of 250 light-curve models exploring a rangeof values for the density n and ejecta energy E ej . Wefind a broad range of combinations that are consistentwith our observations at the times of the two 6 GHzdetections (Figure 4, solid line). Densities above n ∼ − are ruled out, as these models would also requirethe light curve to begin declining by the time of ourepoch 3 observation. As with the relativistic jet models,the ejecta energy required to match the observations islarge, particularly for the lower densities expected giventhe large offset of Candidate 1. These models also re-quire very high ejecta velocities, β ∼ . − . c for n ∼ . − − cm − (Figure 5, red dashed line). Whilethe distributions of ejecta velocities in some publishedKN models have tails to high velocities ∼ . c (e.g. Fig-ure 5, gray lines), the bulk of the energy must be car-ried by slower-moving ejecta ( ≈ . − . c ), to matchthe optical and infrared properties of the KN emission(Bauswein et al. 2013; Hotokezaka et al. 2013; Sekiguchiet al. 2016; Ciolfi et al. 2017; Sekiguchi et al. 2016; Moo-ley et al. 2018a). It is therefore difficult to explain thehigh ejecta energy required to fit our radio light curvewith existing KN models.In summary, both relativistic jet models and quasi-spherical KN ejecta models ultimately require very highenergies and fast ejecta velocities to match the radio evo-lution of Candidate 1. A comparison to ejecta modelsconsistent with GW170817 and cosmological SGRBs canbe seen in Figure 5. It is clear that either the KN modelsused to explain Candidate 1 are probing a new regime ofparameter space, one with higher energies and velocitiesthan the other models that have been found to be con-sistent with the population of cosmological SGRBs andGW170817, or that some parameters (e.g. (cid:15) e and (cid:15) B ) dif-fer from the standard values we assumed. Nevertheless,even for lower (cid:15) B values it is difficult to construct a plau-sible physical scenario that accelerates sufficient ejectato such high velocities, particularly for compact objectmergers with highly unequal mass ratios like GW190814.Ultimately, we conclude that this analysis disfavors Can- Figure 5.
The E ej vs Specific Momentum (Γ β ) phase spacefor KN models consistent with Candidate 1’s radio evolution(red dashed line) in comparison to other ejecta models pro-posed for compact object mergers. Both quasi-spherical KNejecta models and relativistic ejecta models struggle to repro-duce the high energy required to match Candidate 1’s radioproperties. The orange circles show the energy of the red,blue, and purple KN components associated with GW170817from Villar et al. 2017, while the gray lines show two differentmodels for the velocity distribution of quasi-spherical ejectain this event (Mooley et al. 2018a). The purple shaded re-gion is a representative range of maximum energies found forSGRB slow ejecta derived from late-time radio observations(Schroeder et al. 2020). The green line is the structured jetmodel for GW170817 from Margutti et al. (2018), while theblue shaded region is the beaming-corrected energy of the jetcomponent in SGRBs (Fong et al. 2015). didate 1 as a radio counterpart to GW190814 on physicalgrounds. We suggest that similar analyses can be ap-plied in future GW counterpart searches to discriminatebetween genuine radio GW counterparts and unrelatedbackground AGN. DISCUSSION4.1.
Limits on a Relativistic Jet Launched byGW190814
Apart from Candidate 1, we detect no other con-vincing radio counterpart to GW190814 in our obser-vations. We therefore next investigate the limits we canplace on the existence of a relativistic jet (assuming thatthe merger occurred within one of the galaxies we tar-geted). We generate a grid of light curves at 6 GHzusing afterglowpy (Ryan et al. 2020) for 2 relativisticjets: 1) a tophat jet with an opening angle θ jet = 15 ◦ ,and 2) a Gaussian jet with a 15 ◦ jet core ( θ jet ) andwings extending out to 6 θ jet . We compute the lightcurves for isotropic-equivalent kinetic energies, E K,iso =2 × , 5 × , and 5 × ergs (representing a typi-cal SGRB energy, as well as two more optimistic cases),0 Alexander et al.
Figure 6.
Portions of parameter space ruled out by the non-detection in our two epochs at ∼
38 and ∼
208 days for a 15 ◦ tophat (left) and Gaussian (right) jet, with E K,iso = (0.2, 0.5, and 5) × erg, p = 2 . , (cid:15) e = 0 .
1, and (cid:15) B = 0 .
01. The shadedregion shows the uncertainty from the distance. All space above and to the left of these limits is ruled out by our non-detections.Under these assumptions, we can rule out an energetic jet similar to those seen in cosmological short GRBs for a range of viewingangles and densities. The binary inclination angle and its associated uncertainty calculated from the full GW analysis (Abbottet al. 2020b) are shown by the dashed lines. over 0 . − n = 10 − − cm − , and for viewing angles ranging for0 ◦ (on-axis) to 90 ◦ . We fix p = 2.2, (cid:15) e = 0 . (cid:15) B = 0 . d L = 241 +41 − Mpc. We then compare these lightcurves to the limits placed by our VLA observations at38 and 208 days, using 3 × the typical RMS of our imagesin each epoch as an upper limit on the flux density ofany counterpart at that time (Figure 6). These limits al-low us to rule out the higher density parameter space, ashigher densities results in a brighter source which wouldhave been detected in our observations.Compared to the best-fit viewing angle of θ obs = 46 ◦ calculated by LIGO/Virgo, we find our observations ruleout densities n (cid:38) − assuming atop-hat jet model for E K,iso = 2 × , 5 × , and5 × erg respectively. Compared to the circumburstdensities found for the SGRB population (assuming (cid:15) e = 0.1 and (cid:15) B = 0.01 for consistency; Fong et al. 2015),these constraints are comparable to the median of thepopulation (at the ∼ −
65% level compared to SGRBdensities). These limits are a strong function of theviewing angle; for the jet with E K,iso = 2 × erg,the lower limit on the density ranges from 0 . − . − for the full range of viewing angles 35 ◦ − ◦ al-lowed by the GW analysis. For Gaussian jet models with the same energies, we find more constraining densitiesof n (cid:38) .
01, 0.003, 6 × − cm − (at the ∼ − θ obs = 46 ◦ ( n (cid:38) − − . − for θ obs = 35 ◦ − ◦ at d L = 241 Mpc). In comparison,the search conducted by Dobie et al. (2019) can rule outcomparable densities for a jet with E K,iso = 10 erg, θ jet = 10 ◦ , (cid:15) e = 0 . (cid:15) B = 0 .
01, and p = 2 . (cid:46) ◦ off-axis; for a jet that is 46 ◦ off-axis, the density is not constrained by their data. Thisemphasizes the importance of continuing radio transientsearches to late times ( (cid:38) Properties of our most variable sources: extremeAGN flares?
The differing resolution and sensitivity of our datamay impact the interpretation of sources with both smalland large apparent flux density changes on the timescaleof our observations, as discussed in Section 3.1. Never-theless, five of of our 13 “highly-variable sources” (se-lected using the definition of Bhakta et al. 2020) havestrong detections in VLASS data predating 2019 August18 and three have marginal detections, confirming thatthey are likely unrelated to GW190814. The brightest ofthese radio sources is coincident with ESO 474-026 (the
Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 . ± .
03 mJy in epoch 1 to 1 . ± .
14 mJyin epoch 2. This factor ∼ ∼ (cid:46)
2; Hovatta et al. 2008). The large-amplitude variabil-ity of this source may hint that it has recently restartedjet activity, and it may even be related to the popula-tion of quasars recently discovered to have increased inbrightness by 100% − > (cid:46) Considerations for Future Galaxy-Targeted RadioCounterpart Searches
Finally, we briefly expand upon some implicationsof the high variability fraction of our sample of ra-dio sources. Compared to untargeted searches, nearbygalaxies are likely to have a higher surface density of de-tectable radio transients with unresolved, compact emis-sion, including radio supernovae, tidal disruption events(TDEs), and AGN flares (Romero-Ca˜nizales et al. 2011;Metzger et al. 2015; Alexander et al. 2015; Irwin et al.2015; Anderson et al. 2019), in addition to more dif-fuse emission associated with star formation. Given thesmall number of galaxies we observed, their fairly proxi-mal distances, and the rates of transients per L ∗ galaxy,we expect to find (cid:46) − Alexander et al. in our data, complicating our efforts to determine howmuch of the variability is intrinsic for these sources. Thisreinforces the importance of using variability selectioncriteria tailored for galaxy-targeted searches, and takingextra care near the nuclei of target galaxies where suchextended emission is most likely to be detected. Deeptemplate images of each galaxy in the relevant configu-ration(s) would be necessary to attempt to deconvolve apotential near-nuclear GW counterpart from the back-ground variability of its host. CONCLUSIONSWe carried out the first galaxy-targeted search for aradio counterpart to a gravitational wave merger event,GW190814. Although we detected several transient orvariable sources, all are consistent with AGN variabilityor are artificial variables created by the different uv cov-erage of our two epochs of data, and are thus unlikely tobe genuine radio counterparts to GW190814. Via addi-tional monitoring of one initially promising candidate,we demonstrate that multi-frequency radio observationscan help distinguish background AGN flares from bonafide radio GW counterparts, as they may have differ-ent spectral indices and/or require unphysical parame-ter values to fit the shape of the radio light curve in thecontext of relativistic jet or kilonova afterglow models.For the 75 galaxies that we observed, comprising 32% ofthe stellar luminosity in the final localization volume, wecan rule out a relativistic jet at the best-fit LIGO/Virgoviewing angle of ∼ ◦ with isotropic-equivalent energies E K = 2 × , 5 × , and 5 × erg propagat-ing in an ISM-like constant density medium of n (cid:38) − for a tophat jet model, or n (cid:38) × − cm − for a Gaussian jet model (assuming p = 2 . (cid:15) e = 0 .
1, and (cid:15) B = 0 . (cid:38) (cid:46)
50. Such eventswill remain rare (we expect at most one such BNS orNSBH merger in O4; Abbott et al. 2020a), but are nev-ertheless likely to be among the best-studied mergers.The increasing availability of deep template maps of theradio sky will also improve our ability to interpret theresults of future radio counterpart searches, just as wefound VLASS observations of our target galaxies to beuseful in this work. Radio searches will remain the onlyway to discover electromagnetic counterparts to com-pact object mergers that occur in the daytime sky, whoseoptical and X-ray emission cannot be studied, and willthus remain an important tool for multi-messenger stud-ies, despite their challenges.
Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814
Alexander et al.
Table 1 . Galaxies targeted in our observations.
Galaxy Name Date observed (UT) RA (J2000) DEC (J2000) Distance z M B Epoch 1 Epoch 2 (deg) (deg) (Mpc)ESO 474-026 2019 Sep 22.31362 2020 Feb 29.79077 11.781363 -24.370647 244.25 0.0263 -21.98IC 1587 2019 Sep 22.34542 2020 Feb 29.82260 12.180364 -23.561686 252.33 0.0442 -21.83PGC 198197 2019 Sep 21.34325 2020 Mar 14.78605 12.091079 -25.126814 297.78 0.0661 -21.39PGC 198196 2019 Sep 21.33787 2020 Mar 14.78064 11.870618 -25.440655 267.39 0.0594 -21.32PGC 2864 2019 Sep 22.35535 2020 Feb 29.83249 12.25617 -23.811317 236.13 0.0525 -21.14ESO 474-035 a Table 1 continued
Late-Time Galaxy-Targeted Search for the Radio Counterpart of GW190814 Table 1 (continued)
Galaxy Name Date observed (UT) RA (J2000) DEC (J2000) Distance z M B Epoch 1 Epoch 2 (deg) (deg) (Mpc)PGC 198221 2019 Sep 18.29289 2020 Mar 13.82147 12.385612 -26.538588 301.73 0.0670 -20.89PGC 198201 2019 Sep 22.37976 2020 Feb 29.85692 12.637869 -23.295488 239.17 0.0532 -20.73PGC 774472 2019 Sep 21.32782 2020 Mar 14.77060 11.9691 -25.8414 246.68 0.0550 -20.08PGC 142558 2019 Sep 18.27759 2020 Mar 13.80616 12.332034 -26.476397 254.34 0.0567 -20.41PGC 133700 2019 Sep 21.41394 2020 Mar 14.85671 13.477119 -24.077032 210.96 0.0471 -20.72ESO 474-036 2019 Sep 22.39506 2020 Feb 29.87222 13.192388 -22.975018 236.97 0.0480 -21.94PGC 766121 2019 Sep 18.37785 2020 Mar 13.90642 13.358978 -26.599817 278.84 0.0624 -20.48PGC 771842 2019 Sep 18.32821 2020 Mar 13.85680 12.881038 -26.077213 295.12 0.0656 -20.05PGC 198252 2019 Sep 21.32326 2020 Mar 14.76606 11.7816 -25.66073 268.67 0.0598 -20.122MASX 00511861-2620430 2019 Sep 18.31373 2020 Mar 13.84227 12.827568 -26.345284 247.73 0.0551 -19.92PGC 198242 2019 Sep 22.30363 2020 Feb 29.78082 11.264906 -25.019766 275.32 0.0614 -20.58PGC 773323 2019 Sep 21.36318 2020 Mar 14.80596 12.4675 -25.94625 306.20 0.0679 -20.11PGC 3235862 2019 Sep 18.35799 2020 Mar 13.88655 13.3536 -25.8268 260.30 0.0579 -19.51PGC 198247 2019 Sep 21.36777 2020 Mar 14.81056 12.525684 -25.957939 338.83 0.0748 -20.93PGC 2993 2019 Sep 18.30737 2020 Mar 13.83597 12.808345 -26.461119 202.90 0.0449 -20.49PGC 777373 2019 Sep 21.38416 2020 Mar 14.82694 12.7184 -25.57706 226.34 0.0507 -19.46PGC 200164 2019 Sep 22.29907 2020 Feb 29.77624 11.054956 -24.327574 288.19 0.0641 -20.75PGC 3083 2019 Sep 18.37329 2020 Mar 13.90188 13.149672 -26.750933 211.01 0.0469 -20.51PGC 773284 2019 Sep 21.31789 2020 Mar 14.76068 11.555073 -25.950188 268.84 0.0599 -20.92PGC 3235913 2019 Sep 18.34351 2020 Mar 13.87208 12.9027 -25.94219 261.52 0.0582 -19.36PGC 3196 2019 Sep 22.40954 2020 Feb 29.88672 13.566429 -23.535162 262.21 0.0583 -20.83PGC 3093 2019 Sep 18.34807 2020 Mar 13.87665 13.194542 -25.671635 169.14 0.0391 -20.70PGC 773198 2019 Sep 21.37957 2020 Mar 14.82234 12.6371 -25.95719 293.77 0.0653 -19.66PGC 100480 2019 Sep 18.39395 2020 Mar 13.92271 23.594837 -32.835316 286.65 0.0638 -22.18PGC 198243 2019 Sep 22.32367 2020 Feb 29.80084 11.3958 -24.24854 231.52 0.0516 -19.82PGC 198225 2019 Sep 22.34006 2020 Feb 29.81722 12.174922 -23.368631 330.03 0.0729 -21.14PGC 3198 2019 Sep 22.40498 2020 Feb 29.88215 13.570972 -23.552662 213.56 0.0480 -20.85PGC 2939 2019 Sep 22.38969 2020 Feb 29.86687 12.639195 -23.016602 236.27 0.0526 -20.66PGC 198202 2019 Sep 22.39962 2020 Feb 29.87677 13.522819 -23.194635 266.81 0.0593 -21.13PGC 133698 2019 Sep 22.41953 2020 Feb 29.89667 14.254662 -23.837297 220.63 0.0499 -21.21PGC 773149 2019 Sep 18.33358 2020 Mar 13.86215 12.8157 -25.9609 300.61 0.0667 -19.53IC 1581 2019 Sep 21.31334 2020 Mar 14.75611 11.442945 -25.920193 222.59 0.0500 -21.10
Note — a Galaxy selected for additional multi-frequency follow up, based on the presence of a variable radio source within 100 kpc discoveredin our observations.
Facilities:
VLA
Software:
CASA (McMullin et al. 2007), emcee,Numpy, pwkitREFERENCES
Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2016,PhRvL, 116, 061102,doi: 10.1103/PhysRevLett.116.061102—. 2017a, Physical Review Letters, 119, 161101,doi: 10.1103/PhysRevLett.119.161101—. 2017b, ApJL, 848, L12, doi: 10.3847/2041-8213/aa91c9—. 2017c, ApJL, 848, L13, doi: 10.3847/2041-8213/aa920c —. 2019a, Physical Review X, 9, 031040,doi: 10.1103/PhysRevX.9.031040—. 2019b, ApJL, 882, L24, doi: 10.3847/2041-8213/ab3800—. 2020a, Living Reviews in Relativity, 23, 3,doi: 10.1007/s41114-020-00026-9Abbott, R., Abbott, T. D., Abraham, S., et al. 2020b,ApJL, 896, L44, doi: 10.3847/2041-8213/ab960f Alexander et al. —. 2020c, arXiv e-prints, arXiv:2010.14527.https://arxiv.org/abs/2010.14527Ackley, K., Amati, L., Barbieri, C., et al. 2020, A&A, 643,A113, doi: 10.1051/0004-6361/202037669Alexander, K. D., Soderberg, A. M., & Chomiuk, L. B.2015, ApJ, 806, 106, doi: 10.1088/0004-637X/806/1/106Alexander, K. D., Berger, E., Fong, W., et al. 2017, ApJL,848, L21, doi: 10.3847/2041-8213/aa905dAlexander, K. D., Margutti, R., Blanchard, P. K., et al.2018, ApJL, 863, L18, doi: 10.3847/2041-8213/aad637Anderson, G. E., Miller-Jones, J. C. A., Middleton, M. J.,et al. 2019, MNRAS, 489, 1181,doi: 10.1093/mnras/stz1303Andreoni, I., Ackley, K., Cooke, J., et al. 2017, PASA, 34,e069, doi: 10.1017/pasa.2017.65Andreoni, I., Goldstein, D. A., Kasliwal, M. M., et al. 2020,ApJ, 890, 131, doi: 10.3847/1538-4357/ab6a1bAntier, S., Agayeva, S., Aivazyan, V., et al. 2020, MNRAS,492, 3904, doi: 10.1093/mnras/stz3142Antonini, F., & Perets, H. B. 2012, ApJ, 757, 27,doi: 10.1088/0004-637X/757/1/27Arcavi, I., Hosseinzadeh, G., Howell, D. A., et al. 2017a,Nature, 551, 64, doi: 10.1038/nature24291Arcavi, I., McCully, C., Hosseinzadeh, G., et al. 2017b,ApJL, 848, L33, doi: 10.3847/2041-8213/aa910fBauswein, A., Goriely, S., & Janka, H. T. 2013, ApJ, 773,78, doi: 10.1088/0004-637X/773/1/78Berger, E. 2014, ARA&A, 52, 43,doi: 10.1146/annurev-astro-081913-035926Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393,doi: 10.1051/aas:1996164Bhakta, D., Mooley, K. P., Corsi, A., et al. 2020, arXive-prints, arXiv:2010.15042.https://arxiv.org/abs/2010.15042Bhandari, S., Bannister, K. W., Murphy, T., et al. 2018,MNRAS, 478, 1784, doi: 10.1093/mnras/sty1157Capano, C. D., Tews, I., Brown, S. M., et al. 2020, NatureAstronomy, 4, 625, doi: 10.1038/s41550-020-1014-6Carilli, C. L., Ivison, R. J., & Frail, D. A. 2003, ApJ, 590,192, doi: 10.1086/375005Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016,arXiv e-prints, arXiv:1612.05560.https://arxiv.org/abs/1612.05560Chevalier, R. A. 1998, ApJ, 499, 810, doi: 10.1086/305676Chornock, R., Berger, E., Kasen, D., et al. 2017, ApJL,848, L19, doi: 10.3847/2041-8213/aa905cCiolfi, R., Kastaun, W., Giacomazzo, B., et al. 2017,PhRvD, 95, 063016, doi: 10.1103/PhysRevD.95.063016Coulter, D. A., Foley, R. J., Kilpatrick, C. D., et al. 2017,Science, 358, 1556, doi: 10.1126/science.aap9811 Cowperthwaite, P. S., Berger, E., Villar, V. A., et al. 2017,ApJL, 848, L17, doi: 10.3847/2041-8213/aa8fc7D´alya, G., Galg´oczi, G., Dobos, L., et al. 2018, MNRAS,479, 2374, doi: 10.1093/mnras/sty1703D’Avanzo, P., Campana, S., Salafia, O. S., et al. 2018,A&A, 613, L1, doi: 10.1051/0004-6361/201832664Dey, A., Schlegel, D. J., Lang, D., et al. 2019, AJ, 157, 168,doi: 10.3847/1538-3881/ab089dD´ıaz, M. C., Macri, L. M., Garcia Lambas, D., et al. 2017,ApJL, 848, L29, doi: 10.3847/2041-8213/aa9060Dobie, D., Kaplan, D. L., Murphy, T., et al. 2018, ApJL,858, L15, doi: 10.3847/2041-8213/aac105Dobie, D., Stewart, A., Murphy, T., et al. 2019, ApJL, 887,L13, doi: 10.3847/2041-8213/ab59dbDrout, M. R., Piro, A. L., Shappee, B. J., et al. 2017,Science, 358, 1570, doi: 10.1126/science.aaq0049Ducoin, J. G., Corre, D., Leroy, N., & Le Floch, E. 2020,MNRAS, 492, 4768, doi: 10.1093/mnras/staa114Eichler, D., Livio, M., Piran, T., & Schramm, D. N. 1989,Nature, 340, 126, doi: 10.1038/340126a0Essick, R., & Landry, P. 2020, ApJ, 904, 80,doi: 10.3847/1538-4357/abbd3bFong, W., & Berger, E. 2013, ApJ, 776, 18,doi: 10.1088/0004-637X/776/1/18Fong, W., Berger, E., Margutti, R., & Zauderer, B. A.2015, ApJ, 815, 102, doi: 10.1088/0004-637X/815/2/102Fong, W., Blanchard, P. K., Alexander, K. D., et al. 2019,ApJL, 883, L1, doi: 10.3847/2041-8213/ab3d9eFrail, D. A., Kulkarni, S. R., Ofek, E. O., Bower, G. C., &Nakar, E. 2012, ApJ, 747, 70,doi: 10.1088/0004-637X/747/1/70Gehrels, N., Cannizzo, J. K., Kanner, J., et al. 2016, ApJ,820, 136, doi: 10.3847/0004-637X/820/2/136Ghirlanda, G., Salafia, O. S., Paragi, Z., et al. 2019,Science, 363, 968, doi: 10.1126/science.aau8815Goldstein, A., Veres, P., Burns, E., et al. 2017, ApJL, 848,L14, doi: 10.3847/2041-8213/aa8f41Gomez, S., Hosseinzadeh, G., Cowperthwaite, P. S., et al.2019, ApJL, 884, L55, doi: 10.3847/2041-8213/ab4ad5Gompertz, B. P., Levan, A. J., & Tanvir, N. R. 2020, ApJ,895, 58, doi: 10.3847/1538-4357/ab8d24Granot, J., & Sari, R. 2002, ApJ, 568, 820,doi: 10.1086/338966Haggard, D., Nynka, M., Ruan, J. J., et al. 2017, ApJL,848, L25, doi: 10.3847/2041-8213/aa8edeHajela, A., Margutti, R., Alexander, K. D., et al. 2019,ApJL, 886, L17, doi: 10.3847/2041-8213/ab5226Hallinan, G., Corsi, A., Mooley, K. P., et al. 2017, Science,358, 1579, doi: 10.1126/science.aap9855