Tidal disruption flares are coming

The first tidal disruption flare in ZTF: from photometric selection to multi-wavelength characterization


We report the independent discovery of the transient source AT2018zr during commissioning observations of the Zwicky Transient Facility (ZTF), the first tidal disruption event (TDE) found in this survey. The ZTF light curve of the TDE samples the rise-to-peak exceptionally well, with 50 days of and -band detections before the time of maximum light. We also present our multi-wavelength follow-up observations that were triggered by the discovery of this flare: the detection of a weak X-ray source ( erg s) and a stringent upper limit to the radio emission. The X-ray emission shows a thermal spectrum ( eV) and is two orders of magnitude fainter than the contemporaneous optical/UV blackbody luminosity. We use observations of 128 known active galactic nuclei (AGN) to assess the quality of the ZTF astrometry, finding a median host-flare distance of 02 for genuine nuclear flares. Using ZTF observations of variability from known AGN and SNe we show how these sources can be separated from TDEs. A combination of light curve shape, color, and location in the host galaxy can be used to select a clean TDE sample from multi-band optical surveys such as ZTF or LSST.

galaxies: nuclei, accretion, accretion disks, surveys



Sjoert van Velzen

0000-0002-3859-8074]Sjoert van Velzen

0000-0003-3703-5154]Suvi Gezari

0000-0003-1673-970X]S. Bradley Cenko

0000-0003-3124-2814]James C. A. Miller-Jones

0000-0002-6485-2259]Nathaniel Roth

0000-0003-0901-1606]Nadejda Blagorodnova

0000-0003-1710-9339]Lin Yan

0000-0001-9676-730X]Sara Frederick

0000-0002-3168-0139]Matthew J. Graham

0000-0001-8018-5348]Eric C. Bellm

0000-0002-6540-1484]Thomas Kupfer

0000-0001-5390-8563]Shrinivas R. Kulkarni

0000-0003-2242-0244]Ashish Mahabal

0000-0002-8532-9395]Frank J. Masci

0000-0001-9515-478X]Adam A. Miller

0000-0001-7648-4142]Ben Rusholme

0000-0001-6753-1488]Maayane T. Soumagnac

0000-0001-6584-6945]Yutaro Tachibana (優太朗橘)

1 Introduction

Stars that pass within the tidal radius of a supermassive black hole are disrupted and a sizable fraction of the resulting stellar debris gets accreted onto the black hole. When this disruption occurs outside the black hole event horizon (Hills, 1975), the result is a luminous flare of thermal emission (Rees, 1988). These stellar tidal disruption flares provide a unique tool to study black hole accretion and jet formation (e.g., Giannios & Metzger, 2011; van Velzen et al., 2011; Tchekhovskoy et al., 2014; Coughlin & Begelman, 2014; Piran et al., 2015; Pasham & van Velzen, 2018).

Optical transient surveys currently dominate the discovery of tidal disruption events (TDEs); about a dozen candidates have been found to date (for a recent compilation see van Velzen, 2018). All-sky X-ray surveys provide a second avenue for discovery (e.g., Saxton et al., 2017). By late 2019, the eROSITA mission (Merloni et al., 2012) should significantly increase the number of TDEs discovered via their X-ray emission (Khabibullin et al., 2014).

The observed blackbody radii of known TDEs (Gezari et al., 2009) suggest the soft X-ray photons of these flares originate from the inner part of a newly-formed accretion disk ( cm), while the optical photons are produced at much larger radii,  cm. Only a handful of optically selected TDEs have received sensitive X-ray follow-up observations within the first few months of discovery. So far, every case has been different; optically selected TDEs can be X-ray faint (Gezari et al., 2012), or have (Holoien et al., 2016b), or even show a decreasing optical to X-ray ratio (Gezari et al., 2017b).

Unification of optical and X-ray properties of TDEs is possible if the optical emission is powered by shocks from intersecting stellar debris streams (Piran et al., 2015) and the X-ray photons are produced when parts of the stream get deflected towards a few gravitational radii and accreted (Shiokawa et al., 2015; Krolik et al., 2016). In this scenario, TDEs with low X-ray luminosities can be explained as inefficiencies in this circularization process. If instead most of the stellar debris is able to rapidly form an accretion disk (Bonnerot et al., 2016; Hayasaki et al., 2016), the X-rays from the disk have to be reprocessed at larger radii (e.g., Loeb & Ulmer, 1997; Bogdanović et al., 2004; Strubbe & Quataert, 2009; Guillochon et al., 2014) to yield the observed optical emission. When the reprocessing layer is optically thick to X-rays, TDE unification is established via orientation (e.g., Metzger & Stone, 2016; Auchettl et al., 2017; Dai et al., 2018).

Discriminating between the different unification proposals might be possible by comparing the X-ray and optical TDE rate (e.g., if the reprocessing picture is correct, the ratio of the X-ray/optical rate yields the covering factor of the region that absorbs the X-ray photons). However this requires a larger body of TDEs since the current rate estimates are dominated by Poisson uncertainties (van Velzen & Farrar, 2014; Holoien et al., 2016b; Hung et al., 2018). Moreover, rates from different X-ray surveys (Donley et al., 2002; Esquej et al., 2007) are discrepant by a factor —although this discrepancy can be fixed by a more uniform analysis of the X-ray TDEs (Saxton et al. 2018 in prep).

Insight into the emission mechanism of TDEs can also be gained from detailed observations of individual events. A lag of the X-ray emission in a cross correlation analysis (Pasham et al., 2017) and decrease of with time (Gezari et al., 2017a) have both been interpreted as evidence against a reprocessing scenario. However these conclusions are not definitive since the X-ray diversity of TDEs has not yet been mapped out.

Clearly more TDEs with multi-wavelength observations are needed to make progress, which brings us to the topic of this paper: a new TDE detected in commissioning data from the Zwicky Transient Facility (ZTF). As discussed in Hung et al. (2018), ZTF has the potential to significantly increase the TDE detection rate; this first TDE candidate is encouraging evidence that this survey will reach the anticipated performance.

Like the Palomar Transient Factory (PTF; Law et al., 2009; Rau et al., 2009), ZTF uses the Samuel Oschin 48″ Schmidt telescope at Palomar Observatory. The biggest improvement over PTF is the 47 deg field-of-view of the ZTF camera (Bellm et al., 2018). The public Northern Sky Survey of ZTF (Graham et al., 2018) began on 2018 March 17, and covers the entire visible sky from Palomar in both the and band every three nights (the Galactic Plane, , is covered with a one night cadence) to a typical depth of 20.5 magnitude. Using ZTF images a stream of alerts (Patterson et al., 2018) containing transients and variable sources is generated by IPAC (Masci et al., 2018). Besides the essential photometric information, this stream contains value-added products such as the quality of the subtraction (Mahabal et al., 2018) and probability that the alert is associated with a star versus a galaxy (Tachibana & Miller, 2018).

This paper is organized as follows. In Section 2 we present our observations of AT2018zr, the first TDE with ZTF observations. This source was also detected in Pan-STARRS and is known as PS18kh (Chambers et al., 2016); both ATLAS (Tonry et al., 2018) and ASAS-SN (Shappee et al., 2014) also obtained detections of AT2018zr, see Holoien et al. (2018). Here we present several new observations of this latest TDE: the ZTF light curve, XMM-Newton X-ray spectra, and VLA/AMI radio observations. The results from our HST campaign of UV spectroscopy and ground-based optical spectroscopic monitoring will be presented separately (Hung et al. in prep.). In Section 3 we compare AT2018zr to previous TDEs, SNe, and AGN. In Section 4 we show the astrometric quality of ZTF data for nuclear transients. In Section 5 we discuss the results.

Figure 1: ZTF and Swift/UVOT light curve. The dashed line indicates the time of the first SEDM spectrum, while the dotted lines label the time of XMM X-ray observations. Triangles denote 5 upper limits to the flux.

We adopt a flat cosmology with and . All magnitudes are reported in the AB system (Oke, 1974).

Figure 2: Spectral energy distribution. Grey open symbols indicate the host galaxy photometry (from SDSS and GALEX, obtained before the flare) and the best-fit synthetic galaxy spectrum is shown by the thin grey line. The circles show a subset of the UVOT monitoring observations and the corresponding best-fit blackbody spectrum. The unfolded X-ray spectra obtained from the two epochs of XMM-Newton observations are also shown.

2 Observations

2.1 Selection of nuclear flares in ZTF data

During the commissioning phase of the ZTF camera and IPAC alerts pipeline we assessed the quality of the alert stream, focusing on nuclear transients. A transient was considered nuclear when it had at least one detection with a distance between the location of the source in the reference frame and the location of the transient that was smaller than 06. We also required a match within 1″ of a known Pan-STARRS (Chambers et al., 2016) galaxy, selected using a star-galaxy score (Tachibana & Miller, 2018) of gcore . To remove sources with a very small flux increase, we also required that the point spread function (PSF) magnitude in the difference image (agpsf , see \citealt{tp_Masci:18:ZTFDataSystem) and the PSF magnitude of the source in the reference image (agnr ) obey the relation: \verbagpsf agnr $<$ 1.5ag. To apply these filters and to obtain visual confirmation of the alerts we used the GROWTH Marshal (Kasliwal et al., 2018).

The objective of our commissioning effort was to understand the quality of the astrometry of nuclear transients; these results are presented in Section 4. We also obtained spectroscopic follow-up for a subset of nuclear transients, which led to the identification of the transient AT2018zr as a TDE candidate.

2.2 Brief history of AT2018zr

On 2018 March 6, the source ZTF18aabtxvd2 was identified as a nuclear transient by our alert pipeline. Spectroscopic follow-up observations using SEDM (Blagorodnova et al., 2018) were obtained 3 days later; we measured a blue continuum without significant absorption or emission; a redshift and TDE classification was established using additional spectroscopic observations (Hung et al. 2018, in prep). Upon further investigation we noticed the reference frame was contaminated with light from the transient, which prohibited an earlier detection; after rebuilding the reference images and applying an image subtraction algorithm (Zackay et al., 2016) that is similar to the one used in the IPAC pipeline, we found the first ZTF detection was on 2018 February 7 (Fig. 1).

On March 24, Tucker et al. (2018, ATel 11473) reported spectroscopic follow-up and a TDE candidate classification for this source. This follow-up was based on the PS1 photometric observations reported to the Transient Name Server (TNS) on 2018 March 2.

2.3 Optical/UV observations

Observations with the Neil Gehrels Swift Observatory (Swift; Gehrels et al., 2004) started on 2018 March 27. We extracted the UVOT (Roming et al., 2005; Poole et al., 2008) flux with the help of the votsorce task, using an aperture radius of 5 arcsec.

The flux of the host galaxy in the UVOT bands was estimated by fitting a synthetic galaxy spectrum (Conroy et al., 2009; Conroy & Gunn, 2010) to the SDSS model magnitudes (Stoughton et al., 2002; Ahn et al., 2014), see Table 1. To construct the synthetic galaxy spectrum we adopt the default assumptions for the stellar parameters: a Kroupa 2001 initial mass function with stellar masses ; Padova isochrones and MILES spectral library Vazdekis et al. 2010. We assume an exponentially declining star formation rate (, with and as free parameters). We account for Galactic extinction by applying the Cardelli et al. (1989) extinction law with to the model spectrum. We also allow for extinction in the TDE host galaxy by modifying the model spectrum using a Calzetti et al. (2000) extinction law. The best-fit parameters for the formation history are  Gyr and  Gyr; the total stellar mass of the galaxy inferred for this model is .

Besides the Swift/UVOT and ZTF photometry, our light curve also includes P60/SEDM photometric data, host-subtracted using SDSS reference images (Fremling et al., 2016).

We correct the difference magnitude in each band for Galactic extinction,  mag (Schlegel et al., 1998), again assuming a Cardelli et al. (1989) extinction law with . The resulting light curve is shown in Fig. 1 and the photometry is available in Table 4.

By adding a blackbody spectrum to the synthetic host galaxy spectrum we find the best-fit temperature of the flare (Fig. 2). During the first 40 days of Swift monitoring the mean temperature was  K, and appears constant with an rms of only  K. In the last 20 days of monitoring (before the source moved out of the Swift visibility window), the temperature increased by a factor of (see also Holoien et al., 2018).

We searched for time lags between the optical and UV measurements taken with Swift using cross-correlations, following the procedure in Peterson et al. (1998). We first correct the light curves with a simple linear trend using a maximum-likelihood approach, and then use linear interpolation between data points in order to sample both UV and optical light curves on the same grid. We find no significant time lags between any combination of the the UV and optical bands at the 95% confidence level.

To search for outbursts in the years prior to the ZTF detection of AT2018zr, we applied a forced photometry method to the difference images from PTF and iPTF (Masci et al., 2017). We obtained 61 images, clustered at 6, 4, and 1 year before the current peak of the light curve. No prior variability was detected to a typical -band magnitude (5).

Using our ZTF observations of AT2018zr, we measure a mean angular distance between the host galaxy center and the flare of 012, or 162 pc. The rms of the offset, combining 46 offset measurements in Right Ascension and Declination and both - and -band detections, is 025. We thus conclude that the position of the flare is consistent with originating from the center of its host galaxy. Indeed, as we will see in Section 4, flares from AGN—which are expected to originate from the center of their galaxy—have a similar mean host-flare distance.

2.4 X-ray spectrum of AT2018zr

X-ray follow-up observations of AT2018zr were obtained using XMM-Newton (program 082204, PI: Gezari). Two epochs of XMM-Newton observations of the source were taken on 2018 April 11, and 2018 May 3 (see Table 2 for details). We reduced the XMM-Newton/pn data using the XMM-Newton Science Analysis System (SAS) and the newest calibration files. We started with the observation data files (ODFs) and followed standard procedures. Events were filtered with the conditions

ATTERN<= 4 and \verb FLAG==0 . We checked for high background flares (of which there were none). The source and background extraction regions are circular regions of radius 35". The response matrices were produced using {\sc rmfgen} and {\sc arfgen} in SAS. Spectral fitting was performed using
{\sc XS
EC v12.10 (Arnaud, 1996) with the c-statistic.

Figure 3: Top: The XMM-Newton spectra for the two epochs. The source spectrum is shown in blue and red points, and the background is shown as the shaded regions. Both epochs are well described by a single blackbody component with Galactic absorption (solid lines). Bottom: The ratio of the spectra to the best fit model.

Both spectra are well-described by a single blackbody component ( eV) and Galactic absorption ( cm; Kalberla et al. 2005). No additional absorption (at the redshift of the source) was required. The 0.3-1 keV luminosity of the first epoch is . The temperature of the thermal component remained constant between the two observations, though the flux decreased by a factor of 2. In Table 2 we list the full details of the spectral parameters.

The results from our XMM analysis are in mild conflict with the non-thermal spectrum (photon index ) reported by Holoien et al. (2018) using the Swift/XRT data. The XRT measurements overlap with our XMM-Newton epochs, but have a lower signal-to-noise ratio and less spectral coverage at low energies. Given these limitations, a thermal spectrum is easily mistaken for a steep power-law. Indeed, the XRT flux reported by Holoien et al. (2018) is consistent with the XMM-Newton flux. Given the superior quality of the XMM observations, we conclude that the thermal nature of the X-ray spectrum of AT2018zr is firmly established.

2.5 Radio upper limits of AT2018zr

Radio observations of AT2018zr were obtained using the Arcminute Microkelvin Imager Large Array (AMI-LA; Zwart et al. 2008; Hickish et al. 2017) and the Karl G. Jansky Very Large Array (VLA; program 18A-373, PI: van Velzen). AMI observed on 2018 March 28, followed by the VLA on 2018 March 30 and April 28. The AMI data reduction was performed using the custom calibration pipeline reduce_dc (e.g. Perrott et al. 2013) with 3C 286 used as the primary calibrator, and J0745+3142 as the secondary calibrator. The same calibrators were used for both VLA observations. For the VLA data analysis, we made use of the NRAO pipeline products and flagged a few additional spectral channels after manual inspection for radio frequency interference (RFI). The calibrated visibilites were imaged using the Common Astronomy Software Application (CASA; McMullin et al., 2007) task clean, with natural weighting.

The source was not detected in any of our radio observations. Our most sensitive observation was the first VLA epoch, which yields a 3 upper limit to the 10 GHz radio luminosity (defined as ) of  erg s. Full details are listed in Table 3.

Figure 4: Light curves of the five optical TDEs that have a resolved rise to peak. We show both the rest-frame -band luminosity (top) and the blackbody luminosity (bottom). Note the difference of the vertical axis scale between the two figures due to the bolometric correction from the optical luminosity to the blackbody luminosity.
Figure 5: Tidal disruption flares compared to other nuclear flares and transients detected by ZTF. Top: Rise timescale versus the fade timescale. Known TDEs have a longer rise and fade timescale compared to (most) SNe. Bottom: the mean - color versus its slope (Eq. 2). Unlike known SNe, all known TDEs have a near-constant color. Most AGN flares also have a constant color, but their mean colors (as measured in the difference image) show a much larger dispersion.

3 Comparison to known TDEs, supernovae, and AGN flares

To date, only four optical TDEs have a resolved rise to peak: PS1-10jh (Gezari et al., 2012), PS1-11af (Chornock et al., 2014), PTF-09ge3(Arcavi et al., 2014) and iPTF-16fnl (Blagorodnova et al., 2017). The earliest ZTF detection of AT2018zr is 50 days before the peak of the light curve. Measurements of the rise time to peak are important since this parameter is expected to scale with the mass of the black hole that disrupted the star (Rees, 1988; Lodato et al., 2009). In Fig. 4 we show the light curves of the five TDEs with pre-peak detections. We compare both the rest-frame -band luminosity and the blackbody luminosity. The -correction and bolometric correction were estimated using the mean blackbody temperature of the post-peak observations (van Velzen et al., 2018), except for AT2018zr we used the temperature estimated from the nearest Swift/UVOT observation.

Our sample of nuclear flares from ZTF data is large enough to provide a meaningful comparison of the photometric properties of TDEs, SNe, and AGN flares. Selecting alerts that were discovered between May 1 and August 8, we obtain 840 sources. Of these, 331 can be classified as AGN using the Million quasar catalog (Flesch, 2015, v5.2.). Our sample contains 81 spectroscopically confirmed SNe (of which 62 SNe Type Ia) and 3 cataclysmic variables (CVs). The spectroscopic observations for SN/CV classifications were obtained by the ZTF collaboration, with SEDM serving as the main instrument; SNe typing was established using SNID (Blondin & Tonry, 2007).

To be able to compare TDEs, SNe, CVs, and AGN flares, we use a single light curve model to describe the observations of these transients. We compute the fading timescale of the light curve with respect to the rise time to the peak of the light curve using an exponential decay, while rise to peak is modelled using a Gaussian function:


This simplistic light curve model is a compromise between using specific models for each type of object (e.g., SN Ia templates) and using a completely non-parametric approach (e.g., interpolating the light curve to measure the FWHM). Using all available ZTF data (i.e., including upper limits prior to the first detection of the alert), we fit both the -band and -band simultaneously. To model the SED, we use a constant color for the observation before the peak of the light curve, and a linear change of color with time for the post-peak observations:


To summarize, our light curve model has six free parameters: rise timescale (), fade timescale (), the time of peak (), flux at peak (), mean pre-peak color (), and rate of color change (, with units time).

For AT2018zr, we have no ZTF photometry post peak (because the flux of the transient is contained in the reference image of the public survey, which is not available for reprocessing until the first ZTF data release; Graham et al. 2018) and we instead use the Swift/UVOT photometry. For the other four TDEs with a resolved peak of the light curve, we include the published photometry4 up to 100 days post-peak (when a longer temporal baseline is used, an exponential decay no longer provides a good description of the TDE light curve).

In Fig. 5 we show the result of applying our light curve model to AT2018zr, other known TDEs, as well as the AGN, SNe and CVs in our sample of nuclear flares. We discuss these results in Section 5.4.

Figure 6: ZTF astrometric accuracy for nuclear flares. Shown is the rms of the angular offset of AGN flares as a function of the magnitude in the difference image. We show both the offset to the centroid of the reference image (squares) and the angular distance to the median location of the source in the difference images (circles). We used 11476 offset measurements (both R.A. and Decl.) of 128 AGN.

4 Host-flare astrometry in ZTF data

In the previous section we found that our sample of nuclear flares contains about 10% spectroscopically confirmed SNe. As explained in Section 2.1, the sample of nuclear flares was constructed from alerts with at least one detection with a host-flare distance smaller than 06. However we expect that the mean host-flare distance can be measured with a precision that is better than 06, thus facilitating a better separation of nuclear flares (AGN/TDEs) and SNe.

To understand how the measurement of the offset scales with the signal-to-noise ratio of the detection, we collected ZTF measurements for a sample of known AGN. To obtain a good measurement of the mean and rms of the offset, we required at least 7 detections and a median host-flare distance , leaving 128 AGN. Under the assumption that the variability of these sources originates from the photometric center of their host galaxy, the observed rms of the offset yields the uncertainty, .

In Fig. 6 we show binned by the PSF magnitude of the flare in the difference image (). We find the following dependence between these two parameters


We can use this relation to compute the inverse-variance weighted mean of the offset. In Fig. 7 we show the weighted mean host-flare distance for the SNe, AGN, and unclassified flares in our nuclear flare sample (again using only sources with at least 7 detections). Multiple observations of the same flare lead to an increased accuracy on the mean host-flare distance, yielding a typical uncertainty of 02 of nuclear flares with a few tens of detections (cf. the peak of the AGN distribution in Fig. 7).

Figure 7: Stacked histogram of the weighted (Eq. 3) mean host-flare distance for sources in our sample of nuclear flares, selected from 3 months of ZTF observations.

The measurements of the host-flare offset are not independent because each measurement depends on the same reference frame to yield the position of the host. To show the astrometric accuracy without the contribution of the reference frame, we also report the rms of the offset with respect to the median position of the flare in the difference image (Fig. 6).

5 Discussion

5.1 Origin of the thermal emission mechanism

AT2018zr is the fifth optical/UV-selected TDE with an X-ray detection, and only the third source that was detected within a few months of discovery. Its optical to X-ray flux ratio is similar to ASASSN-15oi (Fig. 8), yet in contrast to that TDE, increased between the two epochs of XMM-Newton observations. A longer temporal baseline is needed to confirm whether this high flux ratio persists or decreases to , as observed for most other optical TDEs with X-ray detections (Fig. 8).

The similarity of to ASASSN-15oi may lead one to consider that the low X-ray luminosity of AT2018zr is explained by a delay in the formation of the accretion disk (Gezari et al., 2017a). If instead disk formation is efficient and the optical emission can be explained by reprocessing (e.g. by bound-free absorption) of emission from the inner disk, the low X-ray luminosity can be explained by obscuration of this disk. If the observed X-ray plus optical emission is similar to the bolometric disk luminosity, the observed value of implies our line of sight to the disk lets through at most 1% of the X-ray photons.

The blackbody radius corresponding to the single temperature model of the X-ray spectrum is  cm for the first epoch of X-ray observations. This corresponds to the Schwarzschild radius () of a black hole with a mass of  . Such a low-mass black hole is not expected given the properties of the host galaxy of AT2018zr. If half of the stellar mass of the galaxy is in the bulge, the predicted black hole mass (Gültekin et al., 2009) is   (consistent with the black hole mass estimate of Holoien et al. 2018). If the observed X-ray photons originated from the inner part of an accretion disk, the intrinsic X-ray luminosity must be times higher to match the blackbody radius to the expected size of the inner disk. Part of this tension can be alleviated if the observed X-ray temperature is higher than the true temperature due to obscuration or a contribution of inverse Compton emission to the 0.3-1 keV spectrum. For example, if we adopt the upper limit to the intrinsic absorption from the X-ray spectral fit of , then the inferred blackbody temperature decreases by a factor of 2 and the unabsorbed X-ray luminosity increases by a factor of 1800, yielding a blackbody radius that is a factor 170 larger and of the same of order as the expected size of the inner disk.

Even after accounting for the potential effect of absorption on the X-ray spectrum, the X-ray blackbody radius from our XMM-Newton observations is two orders of magnitude smaller than the inner radius of of the elliptical disk model proposed by Holoien et al. (2018). While this extended elliptical disk could explain the properties of the optical emission lines, our X-ray observations suggest a small, compact accretion disk is present as well.

Figure 8: The ratio of the optical/UV blackbody luminosity to the X-ray luminosity (0.3-10 keV) as as function of time. At the time of our XMM-Newton observations AT2018zr was X-ray dim and its X-ray-to-optical ratio was remarkably similar to ASASSN-15oi. Data on previous TDEs taken from Gezari et al. (2017a).
Figure 9: Radio luminosity of optical TDEs, normalized to the peak of the optical light curves ( in the rest-frame -band; see Fig. 4). We only show radio observations obtained within one year of the first optical detection.

5.2 Interpretation of radio non-detection

The upper limit to the radio luminosity of AT2018zr is one order of magnitude lower than the observed radio emission of the TDE ASASSN-14li (van Velzen et al., 2016; Alexander et al., 2016; Bright et al., 2018). Currently, only the TDE iPTF-16fnl (Blagorodnova et al., 2017) has received radio follow-up observations close to the peak of the flare with a similar sensitivity. This source was also not detected at radio frequencies. However, this flare was exceptional, being the faintest and fastest fading TDE to date (see Fig. 4). The optical properties of AT2018zr, on the other hand, are similar to the mean properties of the current TDE sample (see Figs. 4 & 5 and Hung et al. 2017).

Our radio non-detection rules out the hypothesis that TDEs with a typical optical luminosity and fade timescale produce radio emission similar to ASASSN-14li (Fig. 9). However the X-ray luminosity of AT2018zr is two orders of magnitude lower than ASASSN-14li (Holoien et al., 2016b). If the radio luminosity scales linearly with the power of the accretion disk, as observed for ensembles of radio-loud quasars (Rawlings & Saunders, 1991; Falcke et al., 1999; van Velzen et al., 2015) and for ASASSN-14li (Pasham & van Velzen, 2018), the expected radio flux of AT2018zr would be too faint to be detectable.

Free-free absorption is unlikely to affect the 10 GHz flux of AT2018zr since it would require an unrealistically high electron density. For an electron temperature of  K, we require an emission measure (EM) of at least  cm pc (e.g., a mean electron density of  cm within one parsec) to yield a significant optical depth () for free-free absorption at 10 GHz (e.g., Condon, 1992). For higher electron temperatures, as expected for galaxy centers (Lazio et al., 1999), the lower limit on the EM would increase even further.

5.3 Rise timescale and black hole mass

While our current sample of TDEs with a resolved rise to peak is still small, we can start to search for the anticipated correlation between black hole mass and rise timescale (Rees, 1988; Lodato et al., 2009). To estimate the black hole mass we use the velocity dispersion measurements from Wevers et al. (2017) and Wevers et al. (2018, in prep) and the Gültekin et al. (2009) - relation. For the host galaxy of AT2018zr, a velocity dispersion measurement is not yet available and we adopt the black hole mass from the bulge mass (, see Section 5.1). To provide a more uniform comparison we also consider the relation between rise time and total stellar mass. The results are shown in Fig. 10.

Our measurement of the rise time uses a Gaussian function (Eq.1), which has no defined start time. To be able to compare our measurement to the predicted fallback timescale () we assume the disruption happened at (i.e., when the flux in the model light curve is just 1% of the flux at peak). In Fig. 10 we show the predicted rise time from the theoretical fallback time of Stone et al. (2013), for the disruption of a star with a mass of 1  and an impact parameter of unity,


We find no correlation between the rise time and black hole mass or total galaxy mass. This could be considered surprising given that a correlation between the fade timescale and the black hole mass has been reported (Blagorodnova et al., 2017; Wevers et al., 2017). However, our results also show that the rise and fade timescales themselves appear to be uncorrelated (Fig. 5). It could be possible that the post-peak light curve provides a better tracer for the fallback rate—and thus a better mass estimate (Mockler et al., 2018)—compared to the rise to peak.

Figure 10: The rest-frame rise timescale of the light curves, measured using a Gaussian function (Eq. 1), and the black hole mass (top) or the total stellar mass of the host galaxy (bottom). The black hole mass is estimated using the velocity dispersion of the host galaxy, with the exception of AT2018zr. The dashed line shows the expected scaling (Eq. 4) between the rise time to peak and black hole mass.

5.4 Photometric selection of TDEs in optical surveys

Using only 3 months of ZTF data, we confirm the conclusions from earlier TDE population studies (Gezari et al., 2009; van Velzen et al., 2011; Holoien et al., 2016a; Hung et al., 2017), showing that TDEs are a class of flares with a shared set of photometric properties.

Our work is the first to quantify the distribution of rise and fade timescales of TDEs. We find that most TDEs have both a longer rise time and fade time compared to SNe (Fig. 5, top panel). The TDE iPTF-16fnl is an interesting exception, displaying a rise timescale at the edge of the SN Ia distribution and a fade timescale that is faster compared to most SNe.

Using only pre-peak observations, effective removal of SNe Ia is possible by restricting to flares with a rise time () that is longer than  days. While AGN flares can have a wide range of rise/fade timescales, very few rise and fade within a few months. Only 5% of AGN in our sample of nuclear flares have rise and fade timescales that fall within the range spanned by known TDEs.

The largest contrast between TDEs and SNe is found when we consider the mean color and color change (Fig. 5). We see that TDEs cluster in the region of blue and constant colors (van Velzen et al., 2011). Near their peak, some SNe can be as blue as TDEs, yet these SNe also cool very fast.

We can conclude that a clean sample of TDEs can be selected from using a combination of photometric properties: the rise/fade timescale (Fig. 5, top panel), the flare’s color and its evolution (Fig. 5, bottom panel), and the location of the flare in the host galaxy (Fig. 7). While each of these metrics has exceptions (e.g., TDE from faint galaxies have large astrometric uncertainties on the host-flare offset, some TDE rise and fade rapidly), photometric selection will be unavoidable in the era of the Large Synoptic Survey Telescope (LSST). The TDEs detected by LSST will be too faint and too numerous ( per year; van Velzen et al., 2011) to use spectroscopic follow-up observations for classification.

Acknowledgments — Based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project, a scientific collaboration among the California Institute of Technology, the Oskar Klein Centre, the Weizmann Institute of Science, the University of Maryland, the University of Washington, Deutsches Elektronen-Synchrotron, the University of Wisconsin-Milwaukee, and the TANGO Program of the University System of Taiwan. Further support is provided by the U.S. National Science Foundation under Grant No. AST-1440341. We thank the National Radio Astronomy Observatory (NRAO) staff for the rapid scheduling of the VLA observations. NRAO is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. We thank the staff of the Mullard Radio Astronomy Observatory for their assistance in the operation of AMI. We acknowledge the use of public data from the Swift data archive. This research made use of Astropy, a community-developed core Python package for Astronomy (The Astropy Collaboration et al., 2018). S. Gezari is supported in part by NSF CAREER grant 1454816 and NSF AAG grant 1616566. M. M. Kasliwal acknowledges support by the GROWTH (Global Relay of Observatories Watching Transients Happen) project funded by the National Science Foundation PIRE (Partnership in International Research and Education) program under Grant No 1545949. N.R. acknowledges the support of a Joint Space-Science Institute prize postdoctoral fellowship. J.C.A.M.-J. is supported by an Australian Research Council Future Fellowship (FT140101082). This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 759194 - USNAC).
     Filter      Magnitude
V 18.49
B 19.40
U 20.81
UVW1 22.48
UVM2 23.61
UVW2 23.91

Note. – Obtained by convolving the best-fit galaxy model (Fig. 2) with the Swift/UVOT filter throughput. Not corrected for Galactic extinction.

Table 1: Synthetic host magnitudes
Start Int.5 Counts Flux6
(MJD) (ks) (cm) (eV) ( erg s cm)
58219.98 20 259
58241.92 25 190

Note. – Errors correspond to a 90% confidence level.

Table 2: XMM X-ray observations
Instrument Start Int.7 rms Flux
(MJD) (min) (Jy/beam) (Jy)
AMI (16 GHz) 58205.8 237.6 40.0
VLA X-band (10 GHz) 58207.16 6.0 9.1
VLA X-band (10 GHz) 58236.14 6.1 12.5
Table 3: Radio observations
MJD Instrument Filter Mag
58099.480 ZTF/P48 r
58100.470 ZTF/P48 r
58101.360 ZTF/P48 r
58105.310 ZTF/P48 r
58154.220 ZTF/P48 r
58156.230 ZTF/P48 r 21.06 0.16
58158.230 ZTF/P48 r 20.85 0.12
58160.230 ZTF/P48 r 20.47 0.16
58182.190 ZTF/P48 r 18.16 0.07
58183.180 ZTF/P48 r 17.98 0.02
58190.154 ZTF/P48 r 17.59 0.01
58190.155 ZTF/P48 r 17.59 0.02
58183.340 SEDM/P60 r 17.87 0.13
58222.136 SEDM/P60 r 18.23 0.02
58222.142 SEDM/P60 r 18.26 0.02
58222.148 SEDM/P60 r 18.25 0.01
58222.155 SEDM/P60 r 18.26 0.01
58222.161 SEDM/P60 r 18.24 0.01
58227.162 SEDM/P60 r 18.37 0.03
58227.168 SEDM/P60 r 18.40 0.03
58227.174 SEDM/P60 r 18.32 0.03
58213.031 Swift/UVOT V 17.86 0.42
58204.533 Swift/UVOT V 17.86 0.30
58237.078 Swift/UVOT V 17.97 0.36
58166.260 ZTF/P48 g 18.85 0.02
58167.180 ZTF/P48 g 18.84 0.03
58168.164 ZTF/P48 g 18.77 0.04
58168.176 ZTF/P48 g 18.64 0.03
58170.392 ZTF/P48 g 18.64 0.01
58222.138 SEDM/P60 g 18.21 0.03
58222.144 SEDM/P60 g 18.20 0.02
58222.150 SEDM/P60 g 18.24 0.02
58222.157 SEDM/P60 g 18.25 0.02
58222.163 SEDM/P60 g 18.19 0.02
58227.164 SEDM/P60 g 18.29 0.03
58227.170 SEDM/P60 g 18.35 0.02
58212.637 Swift/UVOT B 17.89 0.26
58208.840 Swift/UVOT B 17.58 0.22
58210.375 Swift/UVOT B 17.52 0.19
58213.029 Swift/UVOT B 18.10 0.28
58228.507 Swift/UVOT B 18.44 0.44
58231.961 Swift/UVOT B 18.15 0.34
58264.035 Swift/UVOT B 18.86 0.48
58223.266 Swift/UVOT B 17.94 0.21
58204.527 Swift/UVOT B 17.51 0.12
58240.132 Swift/UVOT B 18.38 0.24
58215.023 Swift/UVOT B 17.50 0.21
58249.364 Swift/UVOT B 19.09 0.49
58252.146 Swift/UVOT B 18.37 0.32
58221.212 Swift/UVOT B 18.24 0.22
58237.073 Swift/UVOT B 18.21 0.24
58212.637 Swift/UVOT U 17.68 0.15
58208.839 Swift/UVOT U 17.90 0.17
58210.375 Swift/UVOT U 17.98 0.17
58213.028 Swift/UVOT U 17.79 0.14
58228.507 Swift/UVOT U 18.33 0.23
58231.961 Swift/UVOT U 18.36 0.23
58264.034 Swift/UVOT U 18.68 0.22
58261.116 Swift/UVOT U 18.92 0.26
58234.018 Swift/UVOT U 18.22 0.18
58223.266 Swift/UVOT U 18.21 0.16
58204.526 Swift/UVOT U 17.66 0.09
58240.131 Swift/UVOT U 18.65 0.16
58215.023 Swift/UVOT U 18.14 0.20
58249.364 Swift/UVOT U 18.80 0.20
58255.732 Swift/UVOT U 19.04 0.34
58252.145 Swift/UVOT U 18.69 0.23
58242.317 Swift/UVOT U 18.45 0.32
58221.212 Swift/UVOT U 18.00 0.12
58237.072 Swift/UVOT U 18.53 0.17
58212.636 Swift/UVOT UVW1 18.14 0.13
58208.838 Swift/UVOT UVW1 18.39 0.14
58210.374 Swift/UVOT UVW1 18.12 0.13
58213.027 Swift/UVOT UVW1 18.12 0.11
58228.506 Swift/UVOT UVW1 18.40 0.15
58231.960 Swift/UVOT UVW1 18.42 0.15
58264.032 Swift/UVOT UVW1 18.69 0.12
58261.114 Swift/UVOT UVW1 18.79 0.13
58234.017 Swift/UVOT UVW1 18.63 0.14
58223.265 Swift/UVOT UVW1 18.41 0.12
58204.525 Swift/UVOT UVW1 17.89 0.08
58240.129 Swift/UVOT UVW1 18.79 0.11
58215.022 Swift/UVOT UVW1 18.39 0.15
58249.362 Swift/UVOT UVW1 18.68 0.12
58255.731 Swift/UVOT UVW1 18.59 0.14
58252.144 Swift/UVOT UVW1 18.87 0.14
58242.316 Swift/UVOT UVW1 18.40 0.19
58221.210 Swift/UVOT UVW1 18.51 0.10
58237.071 Swift/UVOT UVW1 18.80 0.12
58212.895 Swift/UVOT UVM2 19.04 0.15
58208.845 Swift/UVOT UVM2 18.53 0.09
58210.379 Swift/UVOT UVM2 18.38 0.10
58213.034 Swift/UVOT UVM2 18.37 0.09
58228.511 Swift/UVOT UVM2 18.91 0.12
58231.965 Swift/UVOT UVM2 18.79 0.12
58264.045 Swift/UVOT UVM2 18.77 0.08
58261.124 Swift/UVOT UVM2 18.76 0.09
58234.024 Swift/UVOT UVM2 18.82 0.10
58223.272 Swift/UVOT UVM2 18.66 0.09
58204.537 Swift/UVOT UVM2 18.06 0.06
58240.142 Swift/UVOT UVM2 19.01 0.09
58215.027 Swift/UVOT UVM2 18.50 0.11
58249.371 Swift/UVOT UVM2 19.14 0.15
58255.739 Swift/UVOT UVM2 18.94 0.10
58252.154 Swift/UVOT UVM2 18.94 0.09
58242.320 Swift/UVOT UVM2 19.11 0.17
58221.218 Swift/UVOT UVM2 18.58 0.16
58237.083 Swift/UVOT UVM2 19.02 0.09
58212.638 Swift/UVOT UVW2 18.75 0.16
58208.841 Swift/UVOT UVW2 18.74 0.11
58210.376 Swift/UVOT UVW2 18.49 0.11
58213.030 Swift/UVOT UVW2 18.57 0.10
58228.508 Swift/UVOT UVW2 18.90 0.14
58231.962 Swift/UVOT UVW2 19.09 0.15
58264.038 Swift/UVOT UVW2 19.00 0.09
58261.119 Swift/UVOT UVW2 18.79 0.09
58234.020 Swift/UVOT UVW2 19.16 0.13
58223.268 Swift/UVOT UVW2 18.87 0.11
58204.530 Swift/UVOT UVW2 18.22 0.07
58240.135 Swift/UVOT UVW2 19.23 0.10
58215.024 Swift/UVOT UVW2 18.55 0.12
58249.367 Swift/UVOT UVW2 19.10 0.10
58255.734 Swift/UVOT UVW2 19.07 0.12
58252.148 Swift/UVOT UVW2 19.16 0.11
58242.318 Swift/UVOT UVW2 19.23 0.19
58221.215 Swift/UVOT UVW2 18.78 0.09
58237.076 Swift/UVOT UVW2 19.13 0.10

Note. – Reported magnitudes have the host flux subtracted (see Table 1) and are corrected for Galactic extinction. Upper limits are reported at the 5 level.

Table 4: Optical/UV photometry


  1. journal: ApJ
  2. We internally nicknamed this source ZTF-NedStark.
  3. PTF-09djl and PTF-09axc (Arcavi et al., 2014) also have pre-peak detections, but no post-peak detections.
  4. Obtained using the Open TDE Catalog, http://TDE.space.
  5. The time on source.
  6. Flux at 0.3-1 keV.
  7. The time on source.
  8. footnotetext: The 3 upper limit to the flux.


  1. Ahn, C. P., Alexandroff, R., Allende Prieto, C., et al. 2014, ApJS, 211, 17
  2. Alexander, K. D., Berger, E., Guillochon, J., Zauderer, B. A., & Williams, P. K. G. 2016, ApJ, 819, L25
  3. Arcavi, I., Gal-Yam, A., Sullivan, M., et al. 2014, ApJ, 793, 38
  4. Arnaud, K. A. 1996, in Astronomical Society of the Pacific Conference Series, Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby & J. Barnes, 17
  5. Auchettl, K., Guillochon, J., & Ramirez-Ruiz, E. 2017, ApJ, 838, 149
  6. Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2018, submitted to PASP
  7. Blagorodnova, N., Gezari, S., Hung, T., et al. 2017, ApJ, 844, 46
  8. Blagorodnova, N., Neill, J. D., Walters, R., et al. 2018, PASP, 130, 035003
  9. Blondin, S., & Tonry, J. L. 2007, ApJ, 666, 1024
  10. Bogdanović, T., Eracleous, M., Mahadevan, S., Sigurdsson, S., & Laguna, P. 2004, ApJ, 610, 707
  11. Bonnerot, C., Rossi, E. M., Lodato, G., & Price, D. J. 2016, MNRAS, 455, 2253
  12. Bright, J. S., Fender, R. P., Motta, S. E., et al. 2018, MNRAS, 475, 4011
  13. Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ, 533, 682
  14. Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
  15. Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016, ArXiv e-prints, arXiv:1612.05560
  16. Chornock, R., Berger, E., Gezari, S., et al. 2014, ApJ, 780, 44
  17. Condon, J. J. 1992, ARA&A, 30, 575
  18. Conroy, C., & Gunn, J. E. 2010, ApJ, 712, 833
  19. Conroy, C., Gunn, J. E., & White, M. 2009, ApJ, 699, 486
  20. Coughlin, E. R., & Begelman, M. C. 2014, ApJ, 781, 82
  21. Dai, L., McKinney, J. C., Roth, N., Ramirez-Ruiz, E., & Miller, M. C. 2018, ArXiv e-prints, arXiv:1803.03265
  22. Donley, J. L., Brandt, W. N., Eracleous, M., & Boller, T. 2002, AJ, 124, 1308
  23. Esquej, P., Saxton, R. D., Freyberg, M. J., et al. 2007, A&A, 462, L49
  24. Falcke, H., Bower, G. C., Lobanov, A. P., et al. 1999, ApJ, 514, L17
  25. Flesch, E. W. 2015, PASA, 32, e010
  26. Fremling, C., Sollerman, J., Taddia, F., et al. 2016, A&A, 593, A68
  27. Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, 1005
  28. Gezari, S., Cenko, S. B., & Arcavi, I. 2017a, ApJ, 851, L47
  29. Gezari, S., Heckman, T., Cenko, S. B., et al. 2009, ApJ, 698, 1367
  30. Gezari, S., Chornock, R., Rest, A., et al. 2012, Nature, 485, 217
  31. Gezari, S., Hung, T., Cenko, S. B., et al. 2017b, ApJ, 835, 144
  32. Giannios, D., & Metzger, B. D. 2011, MNRAS, 416, 2102
  33. Graham, M., et al. 2018, pASP submitted
  34. Guillochon, J., Manukian, H., & Ramirez-Ruiz, E. 2014, ApJ, 783, 23
  35. Gültekin, K., Richstone, D. O., Gebhardt, K., et al. 2009, ApJ, 698, 198
  36. Hayasaki, K., Stone, N., & Loeb, A. 2016, MNRAS, 461, 3760
  37. Hickish, J., Razavi-Ghods, N., Perrott, Y. C., et al. 2017, ArXiv e-prints, arXiv:1707.04237
  38. Hills, J. G. 1975, Nature, 254, 295
  39. Holoien, T. W.-S., Kochanek, C. S., Prieto, J. L., et al. 2016a, MNRAS, 463, 3813
  40. —. 2016b, MNRAS, 455, 2918
  41. Holoien, T. W.-S., Huber, M. E., Shappee, B. J., et al. 2018, ArXiv e-prints, arXiv:1808.02890
  42. Hung, T., Gezari, S., Blagorodnova, N., et al. 2017, ApJ, 842, 29
  43. Hung, T., Gezari, S., Cenko, S. B., et al. 2018, ApJ, in press, arXiv:1712.04936
  44. Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005, A&A, 440, 775
  45. Kasliwal, M., et al. 2018, in preparation
  46. Khabibullin, I., Sazonov, S., & Sunyaev, R. 2014, MNRAS, 437, 327
  47. Krolik, J., Piran, T., Svirski, G., & Cheng, R. M. 2016, ApJ, 827, 127
  48. Kroupa, P. 2001, MNRAS, 322, 231
  49. Law, N. M., Kulkarni, S. R., Dekany, R. G., et al. 2009, PASP, 121, 1395
  50. Lazio, T. J. W., Cordes, J. M., Anantharamaiah, K. R., Goss, W. M., & Kassim, N. E. 1999, in Astronomical Society of the Pacific Conference Series, Vol. 186, The Central Parsecs of the Galaxy, ed. H. Falcke, A. Cotera, W. J. Duschl, F. Melia, & M. J. Rieke, 441
  51. Lodato, G., King, A. R., & Pringle, J. E. 2009, MNRAS, 392, 332
  52. Loeb, A., & Ulmer, A. 1997, ApJ, 489, 573
  53. Mahabal, A. A., et al. 2018, in preparation
  54. Masci, F. J., Laher, R. R., Rusholme, B., & Shupe, D. L. 2018, pASP submitted
  55. Masci, F. J., Laher, R. R., Rebbapragada, U. D., et al. 2017, PASP, 129, 014002
  56. McMullin, J. P., Waters, B., Schiebel, D., Young, W., & Golap, K. 2007, in Astronomical Society of the Pacific Conference Series, Vol. 376, Astronomical Data Analysis Software and Systems XVI, ed. R. A. Shaw, F. Hill, & D. J. Bell, 127
  57. Merloni, A., Predehl, P., Becker, W., et al. 2012, ArXiv e-prints, arXiv:1209.3114
  58. Metzger, B. D., & Stone, N. C. 2016, MNRAS, 461, 948
  59. Mockler, B., Guillochon, J., & Ramirez-Ruiz, E. 2018, ArXiv e-prints, arXiv:1801.08221
  60. Oke, J. B. 1974, ApJS, 27, 21
  61. Pasham, D. R., Cenko, S. B., Sadowski, A., et al. 2017, ApJ, 837, L30
  62. Pasham, D. R., & van Velzen, S. 2018, ApJ, 856, 14
  63. Patterson, M. T., et al. 2018, in preparation
  64. Perrott, Y. C., Scaife, A. M. M., Green, D. A., et al. 2013, MNRAS, 429, 3330
  65. Peterson, B. M., Wanders, I., Horne, K., et al. 1998, PASP, 110, 660
  66. Piran, T., Svirski, G., Krolik, J., Cheng, R. M., & Shiokawa, H. 2015, ApJ, 806, 164
  67. Poole, T. S., Breeveld, A. A., Page, M. J., et al. 2008, MNRAS, 383, 627
  68. Rau, A., Kulkarni, S. R., Law, N. M., et al. 2009, PASP, 121, 1334
  69. Rawlings, S., & Saunders, R. 1991, Nature, 349, 138
  70. Rees, M. J. 1988, Nature, 333, 523
  71. Roming, P. W. A., Kennedy, T. E., Mason, K. O., et al. 2005, Space Sci. Rev., 120, 95
  72. Saxton, R. D., Read, A. M., Komossa, S., et al. 2017, A&A, 598, A29
  73. Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
  74. Shappee, B. J., Prieto, J. L., Grupe, D., et al. 2014, ApJ, 788, 48
  75. Shiokawa, H., Krolik, J. H., Cheng, R. M., Piran, T., & Noble, S. C. 2015, ApJ, 804, 85
  76. Stone, N., Sari, R., & Loeb, A. 2013, MNRAS, 435, 1809
  77. Stoughton, C., Lupton, R. H., Bernardi, M., et al. 2002, AJ, 123, 485
  78. Strubbe, L. E., & Quataert, E. 2009, MNRAS, 400, 2070
  79. Tachibana, Y., & Miller, A. 2018, pASP submitted
  80. Tchekhovskoy, A., Metzger, B. D., Giannios, D., & Kelley, L. Z. 2014, MNRAS, 437, 2744
  81. The Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., et al. 2018, ArXiv e-prints, arXiv:1801.02634
  82. Tonry, J. L., Denneau, L., Heinze, A. N., et al. 2018, PASP, 130, 064505
  83. Tucker, M. A., Huber, M., Shappee, B. J., et al. 2018, The Astronomer’s Telegram, 11473
  84. van Velzen, S. 2018, ApJ, 852, 72
  85. van Velzen, S., Falcke, H., & Körding, E. 2015, MNRAS, 446, 2985
  86. van Velzen, S., & Farrar, G. R. 2014, ApJ, 792, 53
  87. van Velzen, S., Körding, E., & Falcke, H. 2011, MNRAS, 417, L51
  88. van Velzen, S., Mendez, A. J., Krolik, J. H., & Gorjian, V. 2016, ApJ, 829, 19
  89. van Velzen, S., Stone, N. C., Metzger, B. D., et al. 2018, ArXiv e-prints, arXiv:1809.00003
  90. van Velzen, S., Farrar, G. R., Gezari, S., et al. 2011, ApJ, 741, 73
  91. Vazdekis, A., Sánchez-Blázquez, P., Falcón-Barroso, J., et al. 2010, MNRAS, 404, 1639
  92. Wevers, T., van Velzen, S., Jonker, P. G., et al. 2017, MNRAS, 471, 1694
  93. Zackay, B., Ofek, E. O., & Gal-Yam, A. 2016, ApJ, 830, 27
  94. Zwart, J. T. L., Barker, R. W., Biddulph, P., et al. 2008, MNRAS, 391, 1545
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minumum 40 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description