AGN flares as counterparts to the mergers detected by LIGO and Virgo: a novel spatial correlation analysis
Abstract
The primary formation channel for the stellar-mass Binary Black Holes which have been detected merging by the LIGO-Virgo-KAGRA (LVK) collaboration is yet to be discerned. One of the main obstacles is that such Gravitational Wave (GW) events are not in general expected to produce an Electromagnetic (EM) counterpart. This might not be the case if the mergers happen in gaseous environments, such as the accretion discs of Active Galactic Nuclei (AGN). Recently, 20 AGN flares detected by the Zwicky Transient Facility have been investigated as potential counterparts of GW events by Graham et al. (2023). We present a new spatial correlation analysis involving such events that uses the up-to-date posterior samples of 78 mergers detected during the third observing run of the LVK collaboration. We apply a likelihood method which takes into account the exact position of the EM signal within the 3D sky map of the GW events. We find that current data favour the hypothesis of no causal connection between the detected mergers and the AGN flares. We place an upper limit of 0.155 at a 90 per cent credibility level on the fraction of coalescences that are physically related to a flare. Moreover, we show that the mass distribution of the merging binaries that appear in coincidence with AGN flares, characterised by values typically larger than the ones of the full population of GW events, is also consistent with the no-connection hypothesis. This is because of a positive correlation between the binary mass and the reconstruction volume.
keywords:
Gravitational Waves - Active Galactic Nuclei - Localisation1 Introduction
During the third observing run (O3) of the LIGO-Virgo-KAGRA (LVK) collaboration, Gravitational Waves (GWs) coming from 80 confirmed mergers of binaries of compact objects were detected (Abbott et al., 2023b, 2024). Several channels for the formation of the merging systems have been proposed (see Mapelli, 2021, for a recent review). These pathways can be divided into two main categories: the evolution of isolated stellar binary systems, and the formation inside dense environments, such as Nuclear Star Clusters or accretion discs of Active Galactic Nuclei (AGN). The formation of merging Binary Black Holes (BBHs) in AGN discs is expected to be facilitated not only by the high density of compact objects, but also by their dynamical interaction with the gas (Stone et al., 2017; Tagawa et al., 2020; Li & Lai, 2022, 2023a, 2023b; DeLaurentiis et al., 2023; Qian et al., 2024; Rowan et al., 2023b, a; Rodríguez-Ramírez et al., 2024). For this reason, the so-called “AGN-channel” for the formation of merging binaries has been recently studied and modelled extensively.
The component masses of binaries coalescing in accretion discs around Massive Black Holes are expected to populate the high end of the astrophysical stellar-mass Black Hole mass spectrum presented in Abbott et al. (2023a). This is mainly because Type I migration caused by the interaction between a gaseous disk and the compact objects within it makes mass segregation very efficient, since the migration time-scale is inversely proportional to the mass of the compact object (Armitage, 2007; McKernan et al., 2011; Secunda et al., 2019). Moreover, the vicinity to the central Massive Black Hole implies that mergers of compact objects inside an AGN accretion disc happen in the presence of a deep gravitational potential. For this reason, the speed at which the remnant objects get kicked from the location of the merger, typically of the order of hundreds of kilometres per second, is likely lower than the escape speed of the environment. Therefore there is the possibility for it to be retained and to take part to another BBH merger. In this scenario of hierarchical mergers, the components of the binaries are expected to have high masses and dimensionless spins close to , which is the value that the remnant of a previous merger is expected to have, due to total angular momentum conservation (Gerosa & Berti, 2017; Gerosa & Fishbach, 2021). In addition, binaries of compact objects merging inside an AGN disc have been modelled as able to develop and maintain a measurable level of eccentricity also in the LVK frequency range (Samsing et al., 2022; Calcino et al., 2023; Fabj & Samsing, 2024). The detection of non-zero eccentricity in one or more GW signals would therefore represent a strong hint regarding the efficiency of disc-assisted BBH formation. Finally, the symmetry of an astrophysical environment such as a rotating disc has been claimed to be sufficient to explain the anti-correlation between the binary effective spin parameter and the mass ratio that has been observed in the population of mergers detected by the LVK collaboration (McKernan et al., 2022; Santini et al., 2023). The comparison of the binary parameters’ distributions predicted for the AGN channel with the ones of detected GW events is a viable approach for estimating the fractional contribution of this formation path to the total BBH merger rate. Nevertheless, such approach is not the only one. It is in fact also possible to put constraints on the this channel’s efficiency by analysing the spatial correlation between the sky maps of the events detected by the LVK interferometers and the positions of observed AGN (Bartos et al., 2017; Veronesi et al., 2022, 2023). In particular, in Veronesi et al. (2023) was found that the fraction of detected BBH mergers that took place in an AGN brighter than () is lower than () at 95 per cent credibility level.
A third way to identify the host environment of the merging binaries is the direct detection of a transient Electromagnetic (EM) counterpart. The vast majority of the detected GW events are coalescences of BBHs, which are not generally expected to produce a detectable associated EM signal. However, such counterparts are expected to be produced, even if not necessarily detectable, whenever mergers take place in gaseous environments, like AGN accretion discs (Bartos, 2016; McKernan et al., 2019). A possible origin of these counterparts is Bondi accretion on the merging objects at a hyper-Eddington rate, that triggers a Bondi explosion (Wang et al., 2021). Another potential source of a detectable EM transient is a jet coming from the accretion of magnetized medium onto the remnant object that traverses the gaseous disc after receiving a recoil kick from the merger event (Chen & Dai, 2024). Recently Graham et al. (2023, G23, hereafter) have identified AGN 20 unusual flaring activities observed by the Zwicky Transient Facility (ZTF; Bellm et al., 2019; Graham et al., 2019) as potential EM counterparts of the GW events detected during O3. These transients have been labelled as not caused by Supernovae (SNe), Tidal Disruption Events (TDEs), or regular AGN variability. A possible physical cause of such flares is Bondi drag accretion and shock of the gas that interacts gravitationally with the kicked remnant of a merger event that happened inside the disc. The flare is expected to be manifest when such remnant reaches the optical depth surface of the disc. In Tagawa et al. (2024) is presented an emission model based on the presence of a Blandford-Znajek jet produced from BHs in AGN discs (Tagawa et al., 2022, 2023), and it is applied to the flares reported in G23, finding that such model can be consistent with the observed events after a number of assumptions were made, mainly regarding the accretion rate onto the stellar-mass BH, and the fraction of jet power that ends up in radiation.
In this work we present the results of a spatial and temporal correlation study for the identification of the host environment of the mergers detected by the LVK collaboration. In particular, this analysis focuses on the connection between the mergers detected during O3 and the 20 EM transients presented and examined in G23.
We calculate the posterior probability distribution of , the fraction of the LVK events detected during O3 that are causally connected to an AGN flare. We do this by using an adapted version of the statistical approach presented in Veronesi et al. (2023). This is a model-independent approach, we only take into account spatial and temporal correlation, but otherwise remain agnostic about which AGN and GW events are favorable counterparts.
In Section 2 we present the properties of the data that have been used in the analysis, which include the GW detections and the AGN flares considered as potential EM counterparts. The description of the statistical method used to check the significance of the GW-AGN flares connection is in Section 3, while the results of its application to the observed data are presented in Section 4. Finally, in Section 5 we draw final conclusions from our results and discuss what are their implications concerning the physical relation between BBH mergers and AGN flares.
We adopt the cosmological parameters of the Cosmic Microwave Background observations by Planck (Planck Collaboration et al., 2016): , , and .
2 Datasets
In this section we first present the main properties of the AGN flares selected in G23 as potential EM counterparts to O3 detections. We then describe and list the GW events we use in our analysis. Finally we show all the spatial and temporal associations that exist between the two catalogues, both in the case of a 3D cross-match, and in the case of a 2D sky-projected one.
2.1 AGN flares
All the AGN flares we use in our analysis have been observed by ZTF. This facility uses a 47 field-of-view camera to cover the majority of the sky above a declination of every two or three nights in the g-band and in the r-band. To select potential EM counterparts for the GW events detected during O3, in G23 the data from the fifth release (DR5) 111https://www.ztf.caltech.edu/page/dr5 were used, and the search has been limited to AGN at a redshift , given the sensitivity the interferometers of LIGO and Virgo had during O3. The light curves of the observed AGN flares have been fitted with a Gaussian rise - exponential decay form (see Equation 13 in G23). From the original full sample were then removed all the flares that have been considered as originated from SNe, TDEs, or regular AGN variability. This selection has been done based on the timescales of the transients, their g-r colour, their rate of colour evolution, and their total observed energy. The resulting selected sample consists of 20 AGN flares that are considered as potential EM counterparts for the GW events detected during O3. All these flares have peaked during the time window that can allow a causal connection with at least one merger. This temporal match is considered possible if the AGN has not flared before the first GW detection or more than 200 days after the last one. Any merger happening after that time is not considered as a potential counterpart, since the estimated time required for the remnant of the merger to reach the edge of the accretion disc is of the order of tens of days.
In Table 1 we list the main properties of the 20 potential counterparts. In particular we report the name of the AGN, the Right Ascension (RA) and Declination (Dec), the redshift, the Modified Julian Day (MJD) of the peak, and the Gaussian rise time of the light curve, . The sky distribution of these transients is shown in Figure 1.
AGN name | RA | Dec | Redshift | ||
---|---|---|---|---|---|
[deg] | [deg] | [days] | |||
J053408.42+085450.7 | 83.535 | 8.914 | 0.5* | 58890 | 17 |
J120437.98+500024.0 | 181.158 | 50.007 | 0.389 | 58894 | 17 |
J124942.30+344928.9 | 192.426 | 34.825 | 0.438 | 58671 | 16 |
J140941.88+552928.1 | 212.425 | 55.491 | 0.074 | 58616 | 11 |
J143157.51+451544.0 | 217.990 | 45.262 | 0.693 | 58859 | 12 |
J143536.15+173755.4 | 218.901 | 17.362 | 0.095 | 58673 | 9 |
J145500.22+321637.1 | 223.751 | 32.277 | 0.177 | 58590 | 14 |
J150748.68+723506.1 | 226.953 | 72.585 | — | 58971 | 20 |
J152433.35+274311.6 | 231.139 | 27.720 | 0.069 | 58611 | 8 |
J154342.46+461233.4 | 235.927 | 46.209 | 0.599 | 58975 | 21 |
J154806.31+291216.3 | 237.026 | 29.205 | 1.090 | 58997 | 55 |
J160822.16+401217.8 | 242.092 | 40.205 | 0.627 | 58921 | 27 |
J161833.77+263226.0 | 244.641 | 26.541 | 0.126 | 58600 | 11 |
J163641.61+092459.2 | 249.173 | 9.416 | 1.155 | 58694 | 9 |
J181719.95+541910.0 | 274.333 | 54.319 | 0.234 | 58784 | 17 |
J183412.42+365655.2 | 278.552 | 36.949 | 0.419 | 58689 | 13 |
J224333.95+760619.2 | 340.891 | 76.105 | 0.353 | 58772 | 11 |
J233252.05+034559.7 | 353.217 | 3.767 | 1.119 | 58855 | 40 |
J233746.08-013116.3 | 354.442 | -1.521 | 0.115 | 58703 | 8 |
J234420.76+471828.9 | 356.087 | 47.308 | — | 58669 | 11 |
2.2 GW events
We make use of 78 GW events detected during O3, which are contained in the Gravitational Wave Transient Catalog (GWTC) 2.1 (Abbott et al., 2024) and in GWTC 3 (Abbott et al., 2023b). The former includes all the events detected up to the first half of O3, O3a, started on April 1st, 2019 and ended on October 1st of the same year, while the latter includes all the events detected during the second half of O3, O3b, that started on November 1st, 2019, and ended on March 27th, 2020. We only had to exclude GW200308_173609 and GW200322_091133 because for these two poorly-localized events the posterior samples do not allow the evaluation of the size of the 90 per cent Credibility Level localisation volume (V90).
We work with the latest (at the time in which this paper is written) versions of the posterior samples published by the Gravitational Wave Open Science Center (GWOSC) (Abbott et al., 2023c). Such samples were published on May 13th, 2022 for GWTC-2.1, and on June 23rd, 2023 for GWTC-3. Since G23 has been published in January 2023, we make use of a different, updated version of the posterior distributions for the events of GWTC-3, with respect to such work. The same is true for all the GW events for which in G23 GWTC-2.0 has been used. For such mergers, we use the posterior distributions of GWTC-2.1. The posterior samples coming from the IMRPhenomXPHM (Pratten et al., 2021) waveform model have been used for every GW event a part from GW190425_081805, GW191219_163120, GW200105_162426, and GW200115_042309, for which the Mixed posterior samples were used.
To reproduce the analysis of GW23, we have to estimate the fraction of the 90 per cent credibility level localisation area (A90) that is outside the galactic plane and has been observed by ZTF at least 20 times in both the g-band and r-band in the 200 days following the LVK detection. This threshold number of observations is used to be consistent with G23. We call this parameter . To calculate it we use the ZTF forced photometry service (Masci et al., 2023) to obtain the number and the MJD of the observations at each sky location. The values of for all the sky maps we use in this work are listed in Table LABEL:tab:gws, together with the ID of the corresponding GW event, the catalogue it is contained in, the sizes of A90 and V90, and the MJD of the detection. The sizes of V90 we report refer to comoving localisation volumes, and not to Euclidean volumes in luminosity distance. This choice implies a cosmological model to be assumed during the cross-matching process, but leads to no difference in its results.
2.3 Matching events
The positions of the 20 flares listed in Table 1 are cross-matched with the sky maps of the GW events listed in Table LABEL:tab:gws. This is done using the postprocess.crossmatch function of the ligo.skymap package, searching space by descending probability density per unit comoving volume. The cross-matching is performed both in the 3D comoving space, and in the 2D sky-projected space. For the 3D case, the two AGN flares without any redshift information are excluded. This is consistent with the approach used in G23, since these two potential counterparts are also not contained in their sample of 7 matching events.
To be considered as matching with a merger event, an AGN flare has to be found within the 90 per cent Credibility Level localisation volume (or area) of the sky map. Following GW23, we also require that the difference between the flare’s peak time and the GW detection time is smaller than 200 days and larger than the flare’s Gaussian rise time. The 3D (2D) cross-matching process leads to a total of 12 (32) matches. These are listed in Table 2 and Table 3. For each match we list the ID of the corresponding GW event, the name of the flaring AGN, and the credibility level corresponding to its position within the sky map of the merger.
GW ID | AGN name | Credibility level |
---|---|---|
GW190514_065416 | J224333.95+760619.2 | 0.175 |
GW190521_030229 | J124942.30+344928.9 | 0.360 |
GW190620_030421 | J124942.30+344928.9 | 0.778 |
GW190706_222641 | J183412.42+365655.2 | 0.181 |
GW190708_232457 | J233746.08-013116.3 | 0.839 |
GW190719_215514 | J181719.95+541910.0 | 0.185 |
GW190731_140936 | J053408.42+085450.7 | 0.479 |
GW190803_022701 | J053408.42+085450.7 | 0.644 |
GW190803_022701 | J120437.98+500024.0 | 0.543 |
GW200128_022011 | J120437.98+500024.0 | 0.879 |
GW200216_220804 | J154342.46+461233.4 | 0.727 |
GW200220_124850 | J154342.46+461233.4 | 0.137 |
GW ID | AGN position | Credibility level |
---|---|---|
GW190425_081805 | J124942.30+344928.9 | 0.750 |
GW190425_081805 | J140941.88+552928.1 | 0.555 |
GW190425_081805 | J152433.35+274311.6 | 0.281 |
GW190425_081805 | J163641.61+092459.2 | 0.120 |
GW190514_065416 | J224333.95+760619.2 | 0.479 |
GW190519_153144 | J234420.76+471828.9 | 0.206 |
GW190521_030229 | J124942.30+344928.9 | 0.640 |
GW190620_030421 | J124942.30+344928.9 | 0.747 |
GW190620_030421 | J143536.15+173755.4 | 0.893 |
GW190620_030421 | J163641.61+092459.2 | 0.597 |
GW190706_222641 | J183412.42+365655.2 | 0.685 |
GW190708_232457 | J233252.05+034559.7 | 0.713 |
GW190708_232457 | J233746.08-013116.3 | 0.626 |
GW190719_215514 | J181719.95+541910.0 | 0.727 |
GW190731_140936 | J053408.42+085450.7 | 0.379 |
GW190803_022701 | J053408.42+085450.7 | 0.777 |
GW190803_022701 | J120437.98+500024.0 | 0.544 |
GW190403_051529 | J124942.30+344928.9 | 0.660 |
GW190403_051529 | J140941.88+552928.1 | 0.856 |
GW191126_115259 | J150748.68+723506.1 | 0.888 |
GW191129_134029 | J154342.46+461233.4 | 0.895 |
GW191219_163120 | J154342.46+461233.4 | 0.409 |
GW200105_162426 | J053408.42+085450.7 | 0.790 |
GW200105_162426 | J154342.46+461233.4 | 0.552 |
GW200105_162426 | J154806.31+291216.3 | 0.369 |
GW200105_162426 | J160822.16+401217.8 | 0.819 |
GW200112_155838 | J154342.46+461233.4 | 0.839 |
GW200112_155838 | J160822.16+401217.8 | 0.341 |
GW200210_092255 | J154342.46+461233.4 | 0.836 |
GW200216_220804 | J154342.46+461233.4 | 0.820 |
GW200216_220804 | J154806.31+291216.3 | 0.890 |
GW200220_124850 | J154342.46+461233.4 | 0.145 |
3 Method
The goal of this work is to establish whether the matches between GW events and AGN flares that are listed in Table 2 and Table 3 are due to a causal connection or to random chance association. To do this, we calculate the posterior distribution of the parameter , the fraction of detected mergers that have a connection with a flaring event. We use an adapted version of the likelihood function presented in Veronesi et al. (2023). The general form of this function is the following:
(1) |
where is the total number of GW detections. The value of is used to weight the contribution of the -th single-event likelihood function. For example, a GW event which sky map is characterised by cannot contain any information regarding the connection with AGN flares observed by ZTF. Using as weight in the likelihood calculation takes this into account.
The signal probability density for the -th GW event is calculated as follows:
(2) |
where is the number of spatial and temporal matches the -th GW event has with flares, is the probability density associated to the position of the -th matching EM transient, and is the effective spatial number density correspondent to it. This number density is assumed to be uniform within V90.
In the analysis that involves the 2D sky-projected cross-match, is used instead of , and both and are in units of , and not in units of . We calculate the effective number density as follows:
(3) |
where 20 is the total number of AGN flares selected from ZTF DR5 as potential counterparts of GW events, is the total effective surveyed comoving volume, and is the total observing time of the search, between the start of O3 and 200 days after its end. We calculate the total effective comoving volume in order for it to be compatible with how has been calculated for each GW event, and to take into account the non uniformity of the source density in the ZTF footprint. We first map the sky into a HealPix grid with a resolution of NSIDE=32. To each pixel correspondent to a galactic latitude lower than or greater than is then associated a number between 0 and 1. This value is the fraction of the total number of GW events for which that specific pixel was observed at least 20 times in both the ZTF g-band and r-band in the 200 days following the merger detection. These values are then all summed together and the result is multiplied by the angular size of each pixel. This sum corresponds to the total effective area observed by the survey during , , and is used in the calculation of , instead of , in the case of the 2D sky-projected analysis. The value of is finally calculated multiplying the total comoving volume enclosed within the redshift limit adopted in the search presented in G23, , by the ratio between and the total area of the sky. Finally, the numerator of the second fraction in Equation 3 takes into account the width of the time window within which a specific flare has to peak in order to allow the match with a GW event. We calculate this quantity by subtracting the Gaussian rise time of the flare from 200 days, the maximum allowed delay time between the peak of such EM transient and the GW detection.
The background probability density function is calculated in the same way as in Veronesi et al. (2023):
(4) |
where the 0.9 factor is used so that and are normalised to the same value. Just like the signal density function (Equation 2) is used instead of for the 2D cross-matching analysis.
After normalising using a flat prior on in the interval, we obtain the posterior distribution of this parameter.
4 Results
In this Section we present the posterior distribution of the fraction of detected GW events that have a connection with an AGN flare, . Then we show the results of two more tests to check if this posterior is indeed compatible with the hypothesis. The first of the two tests focuses the total number of matches between the catalogue of AGN flares and the one of GW events, the second on the binary mass distribution one should expect for the mergers characterised by at least one match.
4.1 Posterior distributions
The posterior probability distribution on is estimated using the data presented in Section 2 and the method described in Section 3. The result is shown in Figure 2. For both the 3D and the 2D cross-matching analyses, the posterior distribution peaks at . In the case of the 3D analysis, the upper limits of the 68, 90, and 95 per cent Credibility Intervals (CIs) correspond to , , and , respectively. The upper limits for the same CIs in the 2D sky-projected analysis are , , and .
4.2 Background Monte Carlo realisations
The posterior distributions presented in Section 4.1 peak at , which corresponds to the hypothesis of no causal connection between mergers detected by the LVK interferometers and the AGN flares selected as potential EM counterparts for such events. In other words, the positioning of the flares within the GW sky maps, and the observed number of matches between GW events and AGN flares appear consistent with chance associations only, without any physical relation between the two events.
To further test this hypothesis, we perform 500 Monte Carlo (MC) realisations of this background scenario. Each of them is constructed as follows:
-
•
To generate a catalog of AGN flares that follows the ZTF real sampling of the sky, which is not fully uniform within the footprint, we sample 20 sky-positions from the catalogue of extra-galactic transient events presented in van Velzen et al. (2024);
-
•
For each simulated flare, we select its peak time by drawing from the times its sky position was observed by ZTF To match the peak time distribution of the 20 flares from G23, we require this peak time to be later than the start of O3 and earlier than the date that corresponds to 200 days after the end of the same observing run;
-
•
To each of the 20 sampled positions we associate a value of redshift. This is obtained by inverse-sampling a linear interpolation of the Cumulative Distribution Function of the redshifts of the potential EM counterparts to GW events listed in Table 1;
-
•
To each flare we associate a value for its rise time , drawn randomly from the ones listed in Table 1;
-
•
Once the catalogue of simulated background ZTF flares is constructed, it is cross-matched with the 3D sky maps of the GW events detected during O3 and listed in Table LABEL:tab:gws. As for the cross-match performed with observed data, in order for a match to be considered as valid, the flare has to peak not more than 200 days after the GW detection, and not before that a number of days equal to the corresponding have passed.
The histogram in Figure 3 shows the distribution of the number of spatial and temporal matches obtained in the 500 MC realisations of the background scenario. The average value of the sample is and its standard deviation is . The number of matches obtained from observed data (12) is less than one standard deviation away from the mean of the distribution. It is reasonable then to assess that 12 matches can be expected even only due to random chance association.
For each MC realisation, the posterior distribution of is also obtained. In 150 realisations this function peaks at , and the distribution of the value of extends to . This is similar to the posterior of obtained from observed data, again confirming that the outcome of our likelihood method (Eq.1) is consistent with the background hypothesis.
4.3 Binary Mass distributions
Due primarily to mass segregation and the possibility of undergoing subsequent hierarchical mergers, BBHs that coalesce inside dynamically dense environments like the accretion discs of an AGN are expected to have on average a higher mass with respect to the ones that formed through the evolution of an isolated binary stellar system. In Figure 4 it is shown the size of V90 for each GW event detected during O3, weighted by , as a function of the source-frame total mass of the merging binary. The horizontal and the vertical dashed lines indicate the median weighted size of V90 and the median total mass, respectively. The pink non-round markers indicate the 11 GW events that have at least one 3D match with a potential EM counterpart.
We indeed notice that most of matching GW events have a total mass that is higher than the median. A KS test for the hypothesis that the mass distributions of the matching and the non-matching merging binaries are not related yields . However, this apparent difference in mass might be because the matching events are also associated with localisation uncertainties larger than the median. To test this hypothesis we follow the procedure presented hereunder:
- •
-
•
For all the GW events we evaluate the expected number of matches due to chance association, , multiplying the size of their localisation volume by ;
-
•
We extract a number of matches for each GW event, drawing from a Poisson distribution with as expectation value;
-
•
We finally perform a two-sample Kolmogorov-Smirnov test between the distribution of the source-frame total mass of the GW events that have at least one match in this random sample and the distribution of the source-frame mass of the ones that don’t have any match. For each repetition we therefore obtain a value of the KS statistic.
The whole process is repeated 10,000 times. The histogram in Figure 5 shows the distribution of the KS statistic for all the repetitions of the random sampling. We see that the KS statistic of the observed 11 matching GW events is consistent with the distribution obtained from the random sampling process. This means that the fact that the observed matching events correspond on average to higher binary masses with respect to the non-matching ones should be expected due to random chance association between GW sky maps and the AGN flares.
5 Discussion and conclusion
We present a statistical investigation on the connection between the GW events detected during O3 and the AGN flares selected in G23 as potential EM counterparts. We do this using the most updated version of the posterior samples released by the LVK collaboration. We make use of the statistical method presented in Veronesi et al. (2023) to estimate the posterior distribution over the fraction of GW events that have an observed AGN flare as an EM counterpart, . We repeat the same analysis both performing a 3D cross-match between the GW sky maps and the positions of the flares, and a 2D sky-projected one. In both cases the posterior distribution peaks at .
The upper limit of the 90 per cent CI is in the case of the 3D cross-matching analysis and for the 2D sky-projected case.
Moreover, we perform 500 MC realisations of the background hypothesis, according to which the matches between GW events and AGN flares are due to random chance association. We find that the total number of matches that exist in the case of the real observed data (12) is compatible with this scenario.
Finally we find through a random sampling process, that the observed distribution of source-frame binary masses of the mergers that have a spatial and temporal match with at least one AGN flare is compatible with the same no-connection hypothesis.
We conclude that the hypothesis of no causal connection between detected GW events and AGN flares cannot be rejected. Moreover, it appears this hypothesis is preferred by currently available data, since the posterior on we obtain peaks at zero.
In G23 Poisson statistic is used to calculate the probability of obtaining a number of matches equal or bigger than the observed one under the no causal connection hypothesis. They estimate this probability to be . The estimate for the number of expected random matches used in G23 (2.83) is calculated assuming a spatial number density uniform in comoving volume up to and a temporal distribution of the flares over the whole duration of the data collection of ZTF DR5. This estimate is not compatible with the average number of matches we obtain in our Monte Carlo realisations of the background hypothesis (14.9). A key difference is that we take into account non-uniform distributions of the sky-position (by sampling from ZTF sources to obtain a realistic sky distributions) and redshift of the AGN flares (by sampling from the redshifts of the observed flares), and we allow these transient events to peak only in the time window that includes all the 20 selected potential counterparts. Below we explain this difference in more detail.
The main distinction between this work and G23 is that we measure a higher effective AGN flare number density, which leads to a higher expected number of matches under the background hypothesis. This difference is mainly driven by two factors. First, in G23 the fraction of matching temporal windows is calculated assuming a mean flare lifetime of 100 days and dividing it by 1000 days, approximately the duration of the observations of ZTF DR5. However, all the 20 AGN flares considered in the analysis peaked between the start of O3 and 200 days after its end, implying the effective duration of the ZTF flare search is 562 days. As such, we calculate for each flare the fraction of matching temporal windows by dividing the width of this window () by 562 days. This leads to a significant difference with respect to the value assumed in G23, since the average value of the rise times, , is about 17 days. The second factor is a different estimate of the effective total comoving volume probed in the ZTF search. Our estimate () takes into account the non-uniformity of the sky distribution of the ZTF extra-galactic sources and is approximately 1.5 times smaller compared to the ZTF volume used in G23. Taken together these two effects yield an effective source density that is approximately 4.6 times larger than the one used in G23. We find a value of the total localisation volume weighted by similar to the ones used in G23, even using updated sky maps (, that corresponds approximately to 68 per cent of the value used in G23).
It is evident that our analysis only constraints the fraction of GW events in AGN that yield a detectable flare. Several factor can make this counterpart challenging to detect. In order to produce a potentially observable flare not later than 200 days after the GW event, the recoil velocity has to be greater than a value that depends on the physical and geometrical characteristics of the AGN disc (see Equation 5 of G23). At the same time, an increase in the recoil velocity corresponds to a decrease in the Bondi-Hoyle-Lyttleton luminosity (see Equations 3 and 6 of G23). Therefore if the merger remnant travels too fast through the accretion disc, the hypothetical resulting flare will not be possible to detect as it will not significantly exceed the luminosity of the host AGN. On top of these physical factors regarding the observability of the flare caused by a kicked remnant exiting the accretion disc, there is a geometrical one. To be visible from instruments like ZTF, the flare has in fact to happen on the side of the disc that faces Earth, and on average this is expected to be true only in half of the cases. Finally, in order to be confidently identified as a potential EM counterpart to a GW signal, the flare has to happen in a position in the sky that allows the light-curve to be scanned with several observations in different bands over its lifetime.
Current and future time-domain surveys such as ZTF and the Vera C. Rubin Observatory (Ivezić et al., 2019) are promising tools for the identification of transient EM counterparts of GW events detected by the LVK collaboration. The low number density of events not compatible with regular AGN variability makes them ideal for spatial correlation analyses like the one presented in this work. Current data suggest that there is no correlation with events detected by the LVK interferometers, but updating the posterior distribution of with more data, or data more sensitive to lower amplitude flares, could yield better constraints on the relation between mergers of compact objects and AGN flares.
Acknowledgements
The authors thank Matthew Graham for making the data of the 20 AGN flares mentioned in G23 available. EMR acknowledges support from ERC Grant “VEGA P.", number 101002511. This research has made use of data or software obtained from the Gravitational Wave Open Science Center (gwosc.org), a service of LIGO Laboratory, the LIGO Scientific Collaboration, the Virgo Collaboration, and KAGRA. LIGO Laboratory and Advanced LIGO are funded by the United States National Science Foundation (NSF) as well as the Science and Technology Facilities Council (STFC) of the United Kingdom, the Max-Planck-Society (MPS), and the State of Niedersachsen/Germany for support of the construction of Advanced LIGO and construction and operation of the GEO600 detector. Additional support for Advanced LIGO was provided by the Australian Research Council. Virgo is funded, through the European Gravitational Observatory (EGO), by the French Centre National de Recherche Scientifique (CNRS), the Italian Istituto Nazionale di Fisica Nucleare (INFN) and the Dutch Nikhef, with contributions by institutions from Belgium, Germany, Greece, Hungary, Ireland, Japan, Monaco, Poland, Portugal, Spain. KAGRA is supported by Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan Society for the Promotion of Science (JSPS) in Japan; National Research Foundation (NRF) and Ministry of Science and ICT (MSIT) in Korea; Academia Sinica (AS) and National Science and Technology Council (NSTC) in Taiwan. Software: Numpy (Harris et al., 2020); Matplotlib (Hunter, 2007); SciPy (Virtanen et al., 2020); Astropy (Astropy Collaboration et al., 2013, 2018); BAYESTAR (Singer & Price, 2016).
Data Availabilty
The data underlying this article will be shared on reasonable request to the corresponding author.
References
- Abbott et al. (2023a) Abbott R., et al., 2023a, Physical Review X, 13, 011048
- Abbott et al. (2023b) Abbott R., et al., 2023b, Physical Review X, 13, 041039
- Abbott et al. (2023c) Abbott R., et al., 2023c, ApJS, 267, 29
- Abbott et al. (2024) Abbott R., et al., 2024, Phys. Rev. D, 109, 022001
- Armitage (2007) Armitage P. J., 2007, arXiv e-prints, pp astro–ph/0701485
- Astropy Collaboration et al. (2013) Astropy Collaboration et al., 2013, A&A, 558, A33
- Astropy Collaboration et al. (2018) Astropy Collaboration et al., 2018, AJ, 156, 123
- Bartos (2016) Bartos I., 2016, in American Astronomical Society Meeting Abstracts #228. p. 208.03
- Bartos et al. (2017) Bartos I., Haiman Z., Marka Z., Metzger B. D., Stone N. C., Marka S., 2017, Nature Communications, 8, 831
- Bellm et al. (2019) Bellm E. C., et al., 2019, PASP, 131, 018002
- Calcino et al. (2023) Calcino J., Dempsey A. M., Dittmann A. J., Li H., 2023, arXiv e-prints, p. arXiv:2311.13727
- Chen & Dai (2024) Chen K., Dai Z.-G., 2024, ApJ, 961, 206
- DeLaurentiis et al. (2023) DeLaurentiis S., Epstein-Martin M., Haiman Z., 2023, in AAS/High Energy Astrophysics Division. p. 100.30
- Fabj & Samsing (2024) Fabj G., Samsing J., 2024, arXiv e-prints, p. arXiv:2402.16948
- Gerosa & Berti (2017) Gerosa D., Berti E., 2017, Phys. Rev. D, 95, 124046
- Gerosa & Fishbach (2021) Gerosa D., Fishbach M., 2021, Nature Astronomy, 5, 749
- Graham et al. (2019) Graham M. J., et al., 2019, PASP, 131, 078001
- Graham et al. (2023) Graham M. J., et al., 2023, ApJ, 942, 99
- Harris et al. (2020) Harris C. R., et al., 2020, Nature, 585, 357
- Hunter (2007) Hunter J. D., 2007, Computing in Science and Engineering, 9, 90
- Ivezić et al. (2019) Ivezić Ž., et al., 2019, ApJ, 873, 111
- Li & Lai (2022) Li R., Lai D., 2022, MNRAS, 517, 1602
- Li & Lai (2023a) Li R., Lai D., 2023a, arXiv e-prints, p. arXiv:2303.12207
- Li & Lai (2023b) Li R., Lai D., 2023b, MNRAS, 522, 1881
- Mapelli (2021) Mapelli M., 2021, arXiv e-prints, p. arXiv:2106.00699
- Masci et al. (2023) Masci F. J., et al., 2023, arXiv e-prints, p. arXiv:2305.16279
- McKernan et al. (2011) McKernan B., Ford K. E. S., Lyra W., Perets H. B., Winter L. M., Yaqoob T., 2011, MNRAS, 417, L103
- McKernan et al. (2019) McKernan B., et al., 2019, ApJ, 884, L50
- McKernan et al. (2022) McKernan B., Ford K. E. S., Callister T., Farr W. M., O’Shaughnessy R., Smith R., Thrane E., Vajpeyi A., 2022, MNRAS, 514, 3886
- Planck Collaboration et al. (2016) Planck Collaboration et al., 2016, A&A, 594, A13
- Pratten et al. (2021) Pratten G., et al., 2021, Phys. Rev. D, 103, 104056
- Qian et al. (2024) Qian K., Li J., Lai D., 2024, ApJ, 962, 143
- Rodríguez-Ramírez et al. (2024) Rodríguez-Ramírez J. C., Bom C. R., Fraga B., Nemmen R., 2024, MNRAS, 527, 6076
- Rowan et al. (2023a) Rowan C., Whitehead H., Boekholt T., Kocsis B., Haiman Z., 2023a, arXiv e-prints, p. arXiv:2309.14433
- Rowan et al. (2023b) Rowan C., Boekholt T., Kocsis B., Haiman Z., 2023b, MNRAS, 524, 2770
- Samsing et al. (2022) Samsing J., et al., 2022, Nature, 603, 237
- Santini et al. (2023) Santini A., Gerosa D., Cotesta R., Berti E., 2023, Phys. Rev. D, 108, 083033
- Secunda et al. (2019) Secunda A., Bellovary J., Mac Low M.-M., Ford K. E. S., McKernan B., Leigh N. W. C., Lyra W., Sándor Z., 2019, ApJ, 878, 85
- Singer & Price (2016) Singer L. P., Price L. R., 2016, Phys. Rev. D, 93, 024013
- Stone et al. (2017) Stone N. C., Metzger B. D., Haiman Z., 2017, MNRAS, 464, 946
- Tagawa et al. (2020) Tagawa H., Haiman Z., Kocsis B., 2020, ApJ, 898, 25
- Tagawa et al. (2022) Tagawa H., Kimura S. S., Haiman Z., Perna R., Tanaka H., Bartos I., 2022, ApJ, 927, 41
- Tagawa et al. (2023) Tagawa H., Kimura S. S., Haiman Z., Perna R., Bartos I., 2023, ApJ, 950, 13
- Tagawa et al. (2024) Tagawa H., Kimura S. S., Haiman Z., Perna R., Bartos I., 2024, ApJ, 966, 21
- Veronesi et al. (2022) Veronesi N., Rossi E. M., van Velzen S., Buscicchio R., 2022, MNRAS, 514, 2092
- Veronesi et al. (2023) Veronesi N., Rossi E. M., van Velzen S., 2023, MNRAS, 526, 6031
- Virtanen et al. (2020) Virtanen P., et al., 2020, Nature Methods, 17, 261
- Wang et al. (2021) Wang J.-M., Liu J.-R., Ho L. C., Li Y.-R., Du P., 2021, ApJ, 916, L17
- van Velzen et al. (2024) van Velzen S., et al., 2024, MNRAS, 529, 2559
Appendix A List of GW events
In Table LABEL:tab:gws we list the properties of the GW events detected during O3 we used in the analysis presented in this work.
GW ID | Catalogue | A90 | V90 | MJD | |
---|---|---|---|---|---|
GW190403_051519 | GWTC-2.1 | 2731 | 3.872 | 58576.2 | |
GW190408_181802 | GWTC-2.1 | 271 | 8.922 | 58581.8 | |
GW190412_053044 | GWTC-2.1 | 25 | 1.112 | 58585.2 | |
GW190413_052954 | GWTC-2.1 | 668 | 2.017 | 58586.2 | |
GW190413_134308 | GWTC-2.1 | 562 | 2.100 | 58586.6 | |
GW190421_213856 | GWTC-2.1 | 1237 | 1.729 | 58594.9 | |
GW190425_081805 | GWTC-2.1 | 8728 | 7.772 | 58598.3 | |
GW190426_190642 | GWTC-2.1 | 4559 | 2.527 | 58599.8 | |
GW190503_185404 | GWTC-2.1 | 103 | 4.716 | 58606.8 | |
GW190512_180714 | GWTC-2.1 | 274 | 9.283 | 58615.8 | |
GW190513_205428 | GWTC-2.1 | 448 | 3.805 | 58616.9 | |
GW190514_065416 | GWTC-2.1 | 3186 | 1.063 | 58617.3 | |
GW190517_055101 | GWTC-2.1 | 365 | 3.042 | 58620.2 | |
GW190519_153544 | GWTC-2.1 | 672 | 1.180 | 58622.6 | |
GW190521_030229 | GWTC-2.1 | 1021 | 3.434 | 58624.1 | |
GW190521_074359 | GWTC-2.1 | 469 | 5.611 | 58624.3 | |
GW190527_092055 | GWTC-2.1 | 3640 | 7.788 | 58630.4 | |
GW190602_175927 | GWTC-2.1 | 739 | 1.412 | 58636.7 | |
GW190620_030421 | GWTC-2.1 | 6443 | 1.219 | 58654.1 | |
GW190630_185205 | GWTC-2.1 | 960 | 1.104 | 58664.8 | |
GW190701_203306 | GWTC-2.1 | 43 | 3.494 | 58665.9 | |
GW190706_222641 | GWTC-2.1 | 2596 | 7.799 | 58670.9 | |
GW190707_093326 | GWTC-2.1 | 893 | 6.443 | 58671.4 | |
GW190708_232457 | GWTC-2.1 | 11032 | 9.846 | 58673.0 | |
GW190719_215514 | GWTC-2.1 | 3564 | 1.317 | 58683.9 | |
GW190720_000836 | GWTC-2.1 | 35 | 2.303 | 58684.0 | |
GW190725_174728 | GWTC-2.1 | 2142 | 3.780 | 58689.7 | |
GW190727_060333 | GWTC-2.1 | 100 | 1.963 | 58691.3 | |
GW190728_064510 | GWTC-2.1 | 321 | 2.974 | 58692.3 | |
GW190731_140936 | GWTC-2.1 | 3532 | 8.919 | 58695.6 | |
GW190803_022701 | GWTC-2.1 | 1012 | 2.227 | 58698.1 | |
GW190805_211137 | GWTC-2.1 | 1538 | 1.342 | 58700.9 | |
GW190814_211039 | GWTC-2.1 | 22 | 3.590 | 58709.9 | |
GW190828_063405 | GWTC-2.1 | 340 | 2.502 | 58723.3 | |
GW190828_065509 | GWTC-2.1 | 593 | 2.700 | 58723.3 | |
GW190910_112807 | GWTC-2.1 | 8305 | 5.158 | 58736.5 | |
GW190915_235702 | GWTC-2.1 | 432 | 2.417 | 58742.0 | |
GW190916_200658 | GWTC-2.1 | 2368 | 1.537 | 58742.8 | |
GW190917_114630 | GWTC-2.1 | 1687 | 1.096 | 58743.5 | |
GW190924_021846 | GWTC-2.1 | 376 | 1.209 | 58750.1 | |
GW190925_232845 | GWTC-2.1 | 876 | 9.957 | 58752.0 | |
GW190926_050336 | GWTC-2.1 | 2015 | 7.945 | 58752.2 | |
GW190929_012149 | GWTC-2.1 | 1651 | 4.851 | 58755.1 | |
GW190930_133541 | GWTC-2.1 | 1493 | 1.223 | 58756.6 | |
GW191103_012549 | GWTC-3 | 2171 | 2.663 | 58790.1 | |
GW191105_143521 | GWTC-3 | 641 | 1.250 | 58792.6 | |
GW191109_010717 | GWTC-3 | 1649 | 4.863 | 58796.0 | |
GW191113_071753 | GWTC-3 | 2484 | 1.159 | 58800.3 | |
GW191126_115259 | GWTC-3 | 1378 | 5.990 | 58813.5 | |
GW191127_050227 | GWTC-3 | 983 | 3.588 | 58814.2 | |
GW191129_134029 | GWTC-3 | 856 | 5.496 | 58816.6 | |
GW191204_110529 | GWTC-3 | 3380 | 3.436 | 58821.5 | |
GW191204_171526 | GWTC-3 | 256 | 7.520 | 58821.7 | |
GW191215_223052 | GWTC-3 | 586 | 4.535 | 58832.9 | |
GW191216_213338 | GWTC-3 | 206 | 1.280 | 58833.9 | |
GW191219_163120 | GWTC-3 | 2232 | 7.504 | 58836.7 | |
GW191222_033537 | GWTC-3 | 2168 | 3.687 | 58839.1 | |
GW191230_180458 | GWTC-3 | 1086 | 4.376 | 58847.8 | |
GW200105_162426 | GWTC-3 | 7882 | 3.345 | 58853.7 | |
GW200112_155838 | GWTC-3 | 3200 | 5.599 | 58860.7 | |
GW200115_042309 | GWTC-3 | 512 | 3.792 | 58863.2 | |
GW200128_022011 | GWTC-3 | 2415 | 5.729 | 58876.1 | |
GW200129_065458 | GWTC-3 | 31 | 2.617 | 58877.3 | |
GW200202_154313 | GWTC-3 | 150 | 2.155 | 58881.7 | |
GW200208_130117 | GWTC-3 | 30 | 3.075 | 58887.5 | |
GW200208_222617 | GWTC-3 | 2040 | 1.214 | 58887.9 | |
GW200209_085452 | GWTC-3 | 877 | 2.325 | 58888.4 | |
GW200210_092255 | GWTC-3 | 1387 | 1.595 | 58889.4 | |
GW200216_220804 | GWTC-3 | 2924 | 1.113 | 58895.9 | |
GW200219_094415 | GWTC-3 | 781 | 1.902 | 58898.4 | |
GW200220_061928 | GWTC-3 | 4477 | 4.333 | 58899.3 | |
GW200220_124850 | GWTC-3 | 3129 | 1.185 | 58899.5 | |
GW200224_222234 | GWTC-3 | 42 | 1.947 | 58903.9 | |
GW200225_060421 | GWTC-3 | 498 | 8.177 | 58904.3 | |
GW200302_015811 | GWTC-3 | 6016 | 2.778 | 58910.1 | |
GW200306_093714 | GWTC-3 | 3907 | 4.302 | 58914.4 | |
GW200311_115853 | GWTC-3 | 35 | 5.799 | 58919.5 | |
GW200316_215756 | GWTC-3 | 187 | 3.634 | 58924.9 |