INTRODUCTION

On 14 August 2019 the LIGO and Virgo collaborations detected the compact binary merger S190814bv 1 with the LIGO Livingston (L1), LIGO Hanford (H1) and Virgo (V1) gravitational wave detectors (LIGO Scientific Collaboration and Virgo Collaboration et al. 2019a). The event was classified as a neutron star-black hole (NSBH) merger, where the lighter component has a mass < 3 M ⊙ , and the heavier component has a mass > 5 M ⊙ , (LIGO Scientific Collaboration and Virgo Collaboration et al. 2019b). The accuracy of this classification is dependent on the physical upper-limit for neutron star mass which is not well constrained, but may be less than the above definition (Zhang et al. 2019;Cromartie et al. 2019). The probability of there being matter outside the remnant object is < 1% (LIGO Scientific Collaboration and Virgo Collaboration et al. 2019a), therefore the expected nature of any electromagnetic radiation from the merger (if any) is unclear.

The preferred skymap (LALInference.v1.fits.gz) has a 90% localisation region of 23 degfoot_0 and a skyaveraged distance estimate of 267±52 Mpc. High-energy observations (Molkov et al. 2019;Kocevski et al. 2019;Pilia et al. 2019;Sugizaki et al. 2019;Palmer et al. 2019) find no evidence for a coincident short gamma-ray burst (GRB). Optical observations found numerous candidate counterparts that have since been ruled out with further photometric and spectroscopic observations (Andreoni et al. in prep.).

While the low probability of remnant matter (LIGO Scientific Collaboration and Virgo Collaboration et al. 2019b) may suggest that the merger produced no electromagnetic counterpart, the lack of optical counterparts may also be explained by intrinsic factors such as inclination angle, mass ratio, remnant lifetime or a lack of polar ejecta (Kasen et al. 2017), or extrinsic factors like dust-obscuration. In this case, radio emission may be the only way to localise this event.

We performed follow-up of S190814bv with the Australian Square Kilometre Array Pathfinder (ASKAP; Johnston et al. 2008). In Section 3 we discuss our untargeted radio transients search. In Section 4 we summarise multi-wavelength follow-up of candidate counterpart AT2019osy that was initially detected in this search.

OBSERVATIONS & DATA REDUCTION

We observed a target field centred on (J2000) coordinates α = 00 h 50 m 37. s 5, δ = -25 • 16 ′ 57. s 37 at ∆T = 2, 9 and 33 days post-merger with ASKAP. This target field, shown in Figure 1 at ∆T = 2 days, covers 89% of the skymap probability.

Table 1 gives a summary of our ASKAP observations. Data were observed using 36 beams arranged in a closep-ack36 footprint 2 with beam spacing of 0.9 degrees. The field was tracked for a nominal time of 10.5 hrs and 288 MHz of bandwidth was recorded with a center frequency of 944 MHz. Typical sensitivity was ∼ 39 µJy with a beam size of ∼ 12 ′′ .

We imaged the data with the ASKAPsoft pipeline version 0.24.4 (Whiting et al. 2017), using a set of parameters optimised for deep continuum fields. Each beam was imaged independently and then combined using a linear mosaic. Multi-frequency synthesis with two Taylor terms was used, along with Multi-scale CLEAN using scales up to 27 pixels in size. Visibilities were weighted using Wiener preconditioning with a robustness parameter of zero. Two major cycles of self-calibration were used to refine the antenna gain solutions derived from observations of PKS B1934-638 in each beam (see Mc-Connell et al. 2016, for a description of the ASKAP beamforming and calibration process). We also used pre-release data from the 888 MHz Rapid ASKAP Continuum Survey (RACSfoot_1 ) as a reference epoch.

The astrometric accuracy and flux scaling of each epoch is consistent with every other epoch. The median flux ratio of compact sources for any two of the ASKAP observations is consistent with 1 within uncertainties. The median RA offset is 0.09-0.36 ′′ and the median declination offset is 0.02-0.2 ′′ (smaller than the pixel size) with a typical standard deviation of 0.7 ′′ and 0.6 ′′ respectively.

UNTARGETED SEARCH FOR RADIO TRANSIENTS AND VARIABLES

To search for a radio counterpart to S190814bv, we performed an untargeted search for transients and highly variable sources using the LOFAR Transients Pipeline (TraP; Swinbank et al. 2015). We ran TraP with source detection and analysis thresholds of 5σ and 3σ respectively and used the 'force beam' option to con- 1. Sources that showed a decline between epochs 1 and 2, followed by a rise between epochs 2 and 3. 41 sources were excluded.

2. Sources detected in RACS epoch 0 where epochs 1 and 2 had lower integrated flux values than epoch 0. 3 sources were excluded.

We then searched the GLADE catalogue (GLADE; Dálya et al. 2018) for galaxies in the localisation volume within 20 ′′ (or ∼ 20 kpc at the estimated distance of S190814bv LIGO Scientific Collaboration and Virgo Collaboration et al. 2019b) of a variable source. We found one candidate (ASKAP J005547.4-270433) that is near 2dFGRS TGS211Z177, a catalogued galaxy with z = 0.0738 (Colless et al. 2001). This source was the only strong candidate after epoch 2 and prior to the acquisition of epoch 3 we performed multi-wavelength follow-up which we discuss in Section 4. We excluded two candidates that matched with a GLADE galaxy > 3σ beyond the estimated distance to S190814bv (267 ± 52 Mpc LIGO Scientific Collaboration and Virgo Collaboration et al. 2019b).

We crossmatched the 42 remaining variable candidates with the Photometric Redshifts for the Legacy Surveys (PRLS) catalogue (Zhou et al. in prep.), which is based on Data Release 8 of DESI Legacy Imaging Surveys (Dey et al. 2019). We excluded 22 variable sources that had all optical matches at distances differing by > 3σ from the estimated distance to S190814bv. This left 7 sources with at least one crossmatch within the localisation volume and 13 sources with no reliable distance estimate (see Table 2).

FOLLOW-UP OF ASKAP J005547.4-270433

Radio Observations

We carried out follow-up observations of ASKAP J005547.4-270433 (hereafter AT2019osy) with the ATCA (C3278, PI: Dobie) using two 2 GHz bands centered on 5.5 and 9 GHz at 14, 22 and 34 days postmerger. We reduced the data using the same method as Dobie et al. ( 2018) using PKS B1934-638 and B0118-272 as flux and phase calibrators respectively.

We also carried out VLA observations (VLA 18B-320, PI: Frail) on 2019 Aug 28 and Sep 09. Standard 2 bit WIDAR correlator setups were used for L and S bands, and 3 bit setups for C and X bands to obtain a contiguous frequency coverage between 1 -12 GHz. 3C48 and J0118-2141 were used as the flux and phase calibrators respectively. The data were processed using the NRAO CASA pipeline and imaged using the clean task in CASA.

A summary of our observations is given in Table 3. We find a flux density offsetfoot_2 of ∼ 40% between the initial ATCA and VLA observations, however later observations with both telescopes are self-consistent. We therefore find no evidence for radio variability beyond the initial rise observed with ASKAP.

Optical Observations

We conducted optical imaging of AT2019osy with the Dark Energy Camera (DECam, Flaugher et al. 2015) on the 4m Blanco telescope under NOAO program ID 2019B-0372 (PI: Soares-Santos). Images including the location of AT2019osy were taken in i and z bands nightly from 2019-08-15 to 2019-08-18 and on 2019-08-21 (UT) and reduced in real-time (Goldstein et al. 2019). A detailed offline analysis of the subtraction images zooming in on the location around AT2019osy, reveals no robust point source at this location to a depth of i > 21.2mag and z > 20.0mag on UT 2019-08-15 (the night of the merger) increasing linearly in limiting magnitude to i > 23.5mag and z > 23.5mag on UT 2019-08-21 (consistent with independent analysis by Herner et al. 2019). We also analyzed the DECam images using The Tractor image modeling software (Lang et al. 2016) and found that a model with an exponential galaxy profile with a point source at the galaxy nucleus is required to fit the data, both before and after S190814bv. This suggests that there is no optical transient temporally coincident with S190814bv but possibly some underlying nuclear variability. On 2019-08-22 UT, we observed AT2019osy in the near infrared using the Wide-field Infrared Camera (WIRC, Wilson et al. 2003) with the 200-inch Hale telescope at Palomar Observatory for a total of 10 minutes exposure time (De et al. 2019). The WIRC data were reduced and stacked using a custom pipeline (De et al., in preparation). No counterpart to AT2019osy was detected down to an AB limiting magnitude of J > 21.5 (5σ).

We also obtained a spectrum of the host galaxy of AT2019osy using the Double Beam Spectrograph (Oke & Gunn 1982) on the Palomar 200-inch Hale Telescope (P200), which we reduced using pyraf-dbsp (Bellm & Sesar 2016). The spectrum is dominated by red continuum that is likely primarily associated with the host galaxy; no obvious broad features are evident. We identify several narrow emission lines (Hα; [NII]λλ6548,6583, [SII]λλ6716,6731, and marginal [OII]λ3727) at a common redshift of 0.0733, consistent within 2-sigma of the LVC distance constraint. Hβ and [OIII]λ5007 are not detected in the spectrum. We measure a flux ratio of log[NIIλ6583/Hα]=0.2, indicating at least partial contribution by an AGN (Kauffmann et al. 2003).

X-ray observations

We observed the field of AT2019osy, starting at 2019-09-23 10:30:48 UT for 20 ks with the Chandra ACIS-S instrument (S3 chip) and very faint data mode. The data were analyzed with CIAO (v 4.11;Fruscione et al. 2006) and calibration was carried out with CALDBv4.8.4.1. We reprocessed the primary and secondary data using the repro script, created X-ray images for the 0.3-8 keV range. No sources were visible near AT2019osy (verified with both wavdetect and celldetect), with a maximum count rate of 2.85×10 -4 s -1 . Assuming a neutral hydrogen column density N H = 1.8 × 10 20 cm -2 and a power-law model with index n = 1.66 (corresponding to the observed radio spectral index of -0.4), this count rate yields a 0.3-8 keV unabsorbed flux upper limit of 3.2×10 -15 erg cm -2 s -1 (as reported in Jaodand et al. 2019) or an unabsorbed luminosity of 4.2×10 40 erg s -1 .

Source classification

AT2019osy exhibits no significant radio variability beyond the initial rise and there is no evidence for a coincident optical transient. The coincident galaxy is edge-on, likely with significant dust obscuration towards the nucleus, and therefore the optical spectrum is consistent only rule out a small part of the parameter space around θ obs = 10 • and n = 1 cm -3 .

In comparison, if we scale the non-thermal lightcurve of GW170817 to 943 MHz based on a spectral index of α = -0.575 (Mooley et al. 2018;Hajela et al. 2019) and place it at a distance comparable to S190814bv, we find a peak flux density of ∼ 5 µJy, well below our detection threshold. We note that the non-thermal emission from GW170817 did not peak until ∼ 150 d post-merger (Dobie et al. 2018). Further observations on timescales of months-years post-merger will enable us to place tighter constraints on the circum-merger density and inclination angle, which may be useful in improving the gravitational wave localisation (Corley et al. 2019).

CONCLUSIONS

We have performed widefield radio follow-up of the NS-BH merger S190814bv with the Australian Square Kilometre Array Pathfinder. We cover 89% of the sky localisation with a single 30 deg 2 pointing centered on the localisation maxima. We found 21 candidate counterparts and performed comprehensive multi-wavelength follow-up of one, AT2019osy. The number of candidates is consistent with the expected rate of AGN variability. Most exhibit variability that is consistent with that expected from interstellar scintillation and are therefore unlikely to be related to S190814bv

The non-detection of a radio counterpart allows us to place constraints on the circum-merger density, n, and inclination angle of the merger, θ obs . Under the assumption of E iso = 10 51 erg, we constrain θ obs > 10 • for all n at the extreme of the probability distribution of distance to the event. We will be able to place tighter constraints on these merger parameters once inclination angle estimates from gravitational wave strain data are released publicly.

As well as probing different parameters to optical searches, radio observations of future events may detect a gravitational wave counterpart where optical follow-up is inhibited by observing constraints, or intrinsic properties of the merger. We have demonstrated that it is possible to perform comprehensive follow-up of gravitational wave events with ASKAP, due to its large field of view that enables a survey speed significantly faster than comparable radio facilities.