Draft version January 11, 2021 Typeset using LATEX twocolumn style in AASTeX63

The NANOGrav 11 yr Data Set: Limits on Binaries in within 500 Mpc

Zaven Arzoumanian,1 Paul T. Baker ,2 Adam Brazier,3 Paul R. Brook ,4, 5 Sarah Burke-Spolaor ,4, 5 Bence Becsy,6 Maria Charisi§ ,7, 8, ∗ Shami Chatterjee ,3 James M. Cordes ,3 Neil J. Cornish ,6 Fronefield Crawford ,9 H. Thankful Cromartie ,10 Megan E. DeCesar ,11 Paul B. Demorest ,12 Timothy Dolch ,13 Rodney D. Elliott,14 Justin A. Ellis,15 Elizabeth C. Ferrara,16 Emmanuel Fonseca ,17 Nathan Garver-Daniels ,4, 5 Peter A. Gentile ,4, 5 Deborah C. Good ,18 Jeffrey S. Hazboun ,19 Kristina Islo,20 Ross J. Jennings ,3 Megan L. Jones ,20 Andrew R. Kaiser ,4, 5 David L. Kaplan ,20 Luke Zoltan Kelley ,21 Joey Shapiro Key ,19 Michael T. Lam ,22, 23 T. Joseph W. Lazio,24, 7 Jing Luo,25 Ryan S. Lynch ,26 Chung-Pei Ma,27 Dustin R. Madison ,4, 5, ∗ Maura A. McLaughlin ,4, 5 Chiara M. F. Mingarelli ,28, 29 Cherry Ng ,30 David J. Nice ,11 Timothy T. Pennucci ,31, 32, ∗ Nihan S. Pol ,4, 8, 5 Scott M. Ransom ,31 Paul S. Ray ,33 Brent J. Shapiro-Albert ,4, 5 Xavier Siemens ,34, 20 Joseph Simon ,24, 7 Renee´ Spiewak ,35 Ingrid H. Stairs ,18 Daniel R. Stinebring ,36 Kevin Stovall ,12 Joseph K. Swiggum ,11, ∗ Stephen R. Taylor ,8 Michele Vallisneri ,24, 7 Sarah J. Vigeland ,20 Caitlin A. Witt ,4, 5 The NANOGrav Collaboration

1X-Ray Astrophysics Laboratory, NASA Goddard Space Flight Center, Code 662, Greenbelt, MD 20771, USA 2Department of Physics and Astronomy, Widener University, One University Place, Chester, PA 19013, USA 3Cornell Center for Astrophysics and Planetary Science and Department of Astronomy, Cornell University, Ithaca, NY 14853, USA 4Department of Physics and Astronomy, West Virginia University, P.O. Box 6315, Morgantown, WV 26506, USA 5Center for Gravitational Waves and Cosmology, West Virginia University, Chestnut Ridge Research Building, Morgantown, WV 26505, USA 6Department of Physics, Montana State University, Bozeman, MT 59717, USA 7Theoretical AstroPhysics Including Relativity (TAPIR), MC 350-17, California Institute of Technology, Pasadena, California 91125, USA 8Department of Physics and Astronomy, Vanderbilt University, 2301 Vanderbilt Place, Nashville, TN 37235, USA 9Department of Physics and Astronomy, Franklin & Marshall College, P.O. Box 3003, Lancaster, PA 17604, USA 10University of Virginia, Department of Astronomy, P.O. Box 400325, Charlottesville, VA 22904, USA 11Department of Physics, Lafayette College, Easton, PA 18042, USA 12National Observatory, 1003 Lopezville Rd., Socorro, NM 87801, USA 13Department of Physics, Hillsdale College, 33 E. College Street, Hillsdale, Michigan 49242, USA 14Department of Astrophysical and Planetary Sciences, University of Colorado Boulder, Boulder, CO 80309, USA 15Infinia ML, 202 Rigsbee Avenue, Durham NC, 27701 16NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA 17Department of Physics, McGill University, 3600 University St., Montreal, QC H3A 2T8, Canada 18Department of Physics and Astronomy, University of British Columbia, 6224 Agricultural Road, Vancouver, BC V6T 1Z1, Canada 19Physical Sciences Division, University of Washington Bothell, 18115 Campus Way NE, Bothell, WA 98011, USA 20 arXiv:2101.02716v1 [astro-ph.GA] 7 Jan 2021 Center for Gravitation, Cosmology and Astrophysics, Department of Physics, University of Wisconsin-Milwaukee, P.O. Box 413, Milwaukee, WI 53201, USA 21Center for Interdisciplinary Exploration and Research in Astrophysics (CIERA), Northwestern University, Evanston, IL 60208 22School of Physics and Astronomy, Rochester Institute of Technology, Rochester, NY 14623, USA 23Laboratory for Multiwavelength Astrophysics, Rochester Institute of Technology, Rochester, NY 14623, USA 24Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA 25Department of Astronomy & Astrophysics, University of Toronto, 50 Saint George Street, Toronto, ON M5S 3H4, Canada 26Green Bank Observatory, P.O. Box 2, Green Bank, WV 24944, USA 27Department of Astronomy, University of California Berkeley, Campbell Hall, Berkeley, CA 94720

Corresponding author: Maria Charisi [email protected] 2 The NANOGrav Collaboration

28Center for Computational Astrophysics, Flatiron Institute, 162 5th Avenue, New York, New York, 10010, USA 29Department of Physics, University of Connecticut, 196 Auditorium Road, U-3046, Storrs, CT 06269-3046, USA 30Dunlap Institute for Astronomy and Astrophysics, University of Toronto, 50 St. George St., Toronto, ON M5S 3H4, Canada 31National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA 32Institute of Physics, E¨otv¨osLor´andUniversity, P´azm´anyP. s. 1/A, 1117 Budapest, Hungary 33Space Science Division, Naval Research Laboratory, Washington, DC 20375-5352, USA 34Department of Physics, Oregon State University, Corvallis, OR 97331, USA 35Centre for Astrophysics and Supercomputing, Swinburne University of Technology, P.O. Box 218, Hawthorn, Victoria 3122, Australia 36Department of Physics and Astronomy, Oberlin College, Oberlin, OH 44074, USA

(Received January 11, 2021; Revised; Accepted) Submitted to ApJ

ABSTRACT Supermassive black hole binaries (SMBHBs) should form frequently in galactic nuclei as a result of mergers. At sub- separations, binaries become strong sources of low-frequency gravita- tional waves (GWs), targeted by Pulsar Timing Arrays (PTAs). We used recent upper limits on con- tinuous GWs from the North American Nanohertz Observatory for Gravitational Waves (NANOGrav) 11 yr dataset to place constraints on putative SMBHBs in nearby massive galaxies. We compiled a comprehensive catalog of ∼44,000 galaxies in the local universe (up to redshift ∼0.05) and pop- ulated them with hypothetical binaries, assuming that the total mass of the binary is equal to the SMBH mass derived from global scaling relations. Assuming circular equal-mass binaries emitting at NANOGrav’s most sensitive frequency of 8 nHz, we found that 216 galaxies are within NANOGrav’s sensitivity volume. We ranked the potential SMBHBs based on GW detectability by calculating the total signal-to-noise ratio (S/N) such binaries would induce within the NANOGrav array. We placed constraints on the chirp mass and mass ratio of the 216 hypothetical binaries. For 19 galaxies, only very unequal-mass binaries are allowed, with the mass of the secondary less than 10 percent that of the primary, roughly comparable to constraints on a SMBHB in the . Additionally, we were able to exclude binaries delivered by major mergers (mass ratio of at least 1/4) for several of these galaxies. We also derived the first limit on the density of binaries delivered by major mergers purely based on GW data.

Keywords: editorials, notices — miscellaneous — catalogs — surveys

1. INTRODUCTION GW regime is of paramount importance for understand- Supermassive black hole binaries are a natural con- ing galaxy evolution (Taylor et al. 2019). 8 10 sequence of galaxy mergers, since massive galaxies are The most massive binaries (10 − 10 M ) are strong known to host supermassive black holes (SMBHs) (Kor- sources of GWs in the nanohertz band (∼ 1 − 100 nHz) mendy & Ho 2013; Haehnelt & Kauffmann 2002). In the and are promising targets for Pulsar Timing Arrays post-merger galaxy, the SMBHs evolve through dynam- (PTAs). PTAs measure precisely the times of arrival ical friction and stellar interactions, and eventually form (TOAs) of pulses for a set of very stable millisecond pul- a gravitationally bound binary. At small (milli-parsec) sars. GWs passing through the Milky Way distort the separations, the binary evolution is dominated by the -pulsar distance, and induce coherent deviations in emission of gravitational waves (GWs). Even though the the TOAs. If such deviations are measured over multi- main steps of binary evolution were proposed decades ple pulsars in the array with a characteristic quadrupole ago (Begelman et al. 1980), there are significant theo- correlation signature, then GWs can be detected (De- retical uncertainties in the path of the SMBHs to their tweiler 1979; Hellings & Downs 1983). final coalescence (e.g., see Colpi 2014; De Rosa et al. Currently, three collaborations are involved in the 2019 for reviews). Therefore, detecting binaries in the effort to detect nanohertz GWs: the North Amer- ican Nanohertz Observatory for Gravitational waves (NANOGrav, Ransom et al. 2019), the European Pul- ∗ NANOGrav Physics Frontiers Center Postdoctoral Fellow sar Timing Array (EPTA, Desvignes et al. 2016) and the limits on smbhbs within 500 mpc 3

Parkes Pulsar Timing Array (PPTA, Kerr et al. 2020). EPTA’s sensitive frequencies, with the mass of the sec- Together along with other emerging PTAs, they form ondary SMBH a few percent that of the primary. a wider consortium, known as the International Pulsar In this paper, we expand on this work considering the Timing Array (IPTA, Perera et al. 2019). most recent (and most sensitive) CW upper limits from PTAs are expected to detect two main types of GW NANOGrav, derived from the 11 yr dataset (Aggarwal signals: (1) the stochastic GW background (GWB), i.e., et al. 2019; Arzoumanian et al. 2018a). This dataset the incoherent superposition of many unresolved sig- spans from 2004 to 2015 and contains TOAs of 45 pul- nals from a population of binaries, and (2) continuous sars. Of those, only 34 were observed for at least 3 yr and waves (CWs), i.e., monochromatic GWs from individual were included in the present CW analysis. At the most SMBHBs. Additional (albeit less likely) signals, such as sensitive frequency (fGW = 8 nHz), the sky-averaged −15 1 bursts with memory from SMBHBs, primordial GWs, upper limit on the GW strain is h0 < 7.3 × 10 . For or cosmic strings may also be detected (Burke-Spolaor comparison, the sky-averaged upper limits used in SM16 −14 −14 et al. 2019). As the sensitivity of PTAs is continuously were h0 < 1.7 × 10 and h0 < 1.1 × 10 (at 10 nHz) improving, the first detection is expected in the near fu- from PPTA and EPTA, respectively. ture (Rosado et al. 2015; Taylor et al. 2016; Mingarelli With the NANOGrav 11 yr upper limits, we have al- et al. 2017; Kelley et al. 2018). ready excluded the presence of a circular equal-mass 9 Even before the first detection, PTA upper limits on binary, with chirp mass M > 1.6 × 10 M , emitting the GWB already provide interesting astrophysical in- in the frequency range between 2.8 nHz to 317.8 nHz, sights. For instance, Sesana et al.(2018) demonstrated in the Cluster (Aggarwal et al. 2019). Here, we that the GWB inferred from the population of binary extend the astrophysical inference based on the 11 yr candidates identified as with periodic variability dataset beyond the . For this, we con- (Graham et al. 2015; Charisi et al. 2016) is in tension struct a comprehensive galaxy catalog, which contains with the current upper limits. This suggests that the all the massive nearby galaxies, using the Two Micron samples of photometrically identified candidates likely All-Sky Survey (2MASS). We carefully determine the contain false detections. Holgado et al.(2018) reached galaxy distances and the masses of their SMBHs. We a similar conclusion for the population of with select a subset of 216 galaxies which are in the sensi- periodic variability (Sandrinelli et al. 2018). tivity volume of NANOGrav, i.e., they would emit de- Similarly, the upper limits on CWs provide constraints tectable GWs if they hosted equal mass circular binaries on binary candidates. For instance, a SMBHB was sug- at 8 nHz. Note that the galaxies examined in SM16 are gested in the galaxy 3C 66B, because the radio core a subset of catalog that we use here. of the galaxy shows an elliptical motion (Sudou et al. We rank the above galaxies based on the GW de- 2003) with a period of 1.05 yr. Since this is a rela- tectability, calculating the total S/N the tentative SMB- tively nearby system (at ∼85 Mpc), it could produce HBs induce to the NANOGrav pulsars. Subsequently, GWs detectable by PTAs. Our recent study using the we run the upper limits analysis in the direction of the NANOGrav 11 yr dataset constrained the chirp mass of galaxies and convert the GW strain limits to constraints 9 the potential SMBHB to less than 1.65 × 10 M (Ar- on the mass ratio (using the distance and the mean zoumanian et al. 2020), improving by a factor of four SMBH mass from the catalog we constructed). For the on previous limits based on a single pulsar (Jenet et al. five top ranked galaxies, we derive upper limits sampling 2004). directly on the mass ratio, incorporating the uncertainty Furthermore, Schutz & Ma(2016) [hereafter SM16] in the total mass. We also place constraints on binaries examined two samples of local galaxies for putative bi- delivered by major mergers, examining frequencies for naries. They focused on galaxies with SMBHs, the mass which binaries with a mass ratio of q > 1/4 can be ex- of which has been measured dynamically (McConnell cluded and derive a limit on the number density of such & Ma 2013), and galaxies from the MASSIVE survey, binaries at 8 nHz. which consists of the most massive early-type galaxies The paper is organized as follows. In § 2, we de- within 100 Mpc (Ma et al. 2014). Based on upper lim- scribe the properties of the galaxy catalog and the se- its from the PPTA’s first data release (Zhu et al. 2014), lection of the sub-sample of galaxies that lie within the and the most recent data release from EPTA (Babak et al. 2016), they placed contraints on the mass ratio 1 From the same dataset, the limit on the amplitude of the GWB, of hypothetical binaries in the above sample. They con- at a frequency of f = 1yr−1 for a fiducial power-law GWB spec- −15 cluded that for a handful of nearby massive galaxies only trum with a slope of -2/3 is AGWB < 1.45×10 (Arzoumanian very unequal-mass binaries are plausible at PPTA’s and et al. 2018b). 4 The NANOGrav Collaboration

NANOGrav sensitivity volume. In § 3, we summarize (e.g., due to infall to clusters). Therefore, it is necessary the statistical analysis of the NANOGrav data. In § 4, to correct for the distortions in the local velocity field we present the constraints on putative SMBHBs in these or use direct distance measurements which are available galaxies. We compare our results with previous related for several local galaxies. For the latter, we used the fol- studies and discuss future directions in § 5 followed by lowing catalogs: (A) Cosmicflows-3 (Tully et al. 2016) a brief summary in § 6. extracted from the the Extragalactic Distance Database (Tully et al. 2009), and (B) the galaxy groups catalog 2. GALAXY CATALOG (Crook et al. 2007). We compiled a comprehensive catalog of nearby galax- Cosmicflows-3 is a compilation of distances of nearby ies. Our starting point was the 2MASS Redshift Survey galaxies, which were obtained with high quality meth- (2MRS, Huchra et al. 2012), which consists of 43,533 ods. These include methods that rely on standard can- galaxies, the redshifts of which have been measured dles, such as: (1) the Cepheid period-luminosity rela- spectroscopically. Our choice of 2MASS for constructing tion, in which the intrinsic luminosity is determined a census of galaxies in the local Universe was motivated from the pulsation period of the star; (2) the Tip of by the fact that, as a near-IR survey (K-band; 2.2µm), the Red Giant Branch, in which the brightest red giants it is minimally affected by interstellar extinction. In in the Hertzsprung-Russel diagram are used as standard fact, 2MRS is 97.6% complete to a distance of 300 Mpc, candles; and (3) Type Ia supernovae. Cosmicflows-3 also and magnitudes mK ≤ 11.75 (Huchra et al. 2012). incorporates distances from additional methods, includ- In order to determine whether NANOGrav can probe ing: (4) the Tully-Fisher relation, an empirical correla- a tentative binary in the above galaxies, we needed to tion between the rotation velocity of spiral galaxies and calculate the GW strain such a binary would produce. their intrinsic luminosity; (5) the fundamental plane of The amplitude of the GW strain hGW is given by elliptical galaxies, a bivariate correlation between the lu- 5/3 2/3 minosity, the effective radius and the dispersion velocity 2(GM) (πfGW) hGW = , (1) of the galaxy; and (6) the surface brightness fluctuations c4D method, which relies on the variance of the observed D is the luminosity distance, f the observed GW GW light distribution (e.g., distant galaxies have lower vari- frequency, and M the chirp mass, ance across pixels and appear smoother). Among the  q 3/5 galaxies in the 2MRS sample, 8,625 were covered in M = M , (2) (1 + q)2 tot Cosmicflows-3. For galaxies not included in Cosmicflows-3, we used where Mtot is the total mass and q the mass ratio (e.g., the distance estimates from the catalog of galaxy groups. see Finn & Thorne 2000). The observed frequency In this catalog, a “friends-of-friends” algorithm was em- and chirp mass are related to the rest-frame values as ployed to determine galaxy groups by identifying neigh- fr = fGW × (1 + z) and Mr = M/(1 + z). However, boring galaxies with similar distances. The galaxy dis- since our catalog includes only relatively nearby galax- tances were determined by applying a flow-field model, ies (z < 0.05), we set 1 + z ' 1. We note that as the following Mould et al.(2000). More specifically, the sensitivities of PTAs improve, and in turn the distances heliocentric velocities were first converted to the Local to which they can probe increase, redshifts should be Group frame. Subsequently, the galaxies in the vicin- incorporated in future analyses. ity of major local clusters (i.e., the Virgo Cluster, the Therefore, for each galaxy an estimate of the distance Shapley , and the Great Attractor) were as- and the mass of the SMBH (which we attributed to the signed the velocity of the respective cluster (see Crook total mass of the binary) was required. Since at milli- et al. 2007 and Mould et al. 2000 for details). Finally, an parsec separations, it is almost impossible to directly additional correction was added to account for the grav- resolve two SMBHs (D’Orazio & Loeb 2018), assigning itational pull towards the aforementioned major galaxy the mass of a central SMBH (either measured or esti- clusters. We used the High-Density-Contrast (HDC) mated from scaling relations) to be the total mass of catalog from Crook et al.(2007), in which a galaxy was the binary is the best possible alternative. Below we considered a group member when the density contrast describe how we estimated these quantities. was 80 or more, and for groups with more than two 2.1. Galaxy Distances members, we used the mean distance of the members. The group catalog provided the distances for another In the local volume, converting redshift to distance is 4,533 galaxies. not trivial because the spectroscopically measured veloc- ities are significantly affected by the motion of galaxies limits on smbhbs within 500 mpc 5

For the remaining galaxies (i.e., galaxies not included catalog or they were classified as irregular. For this sub- in any of the above two catalogs), we followed the pre- set, we did not use the MBH-σ relation. scription from Mould et al.(2000), described above, to If a measurement of the dispersion velocity was not correct for velocity distortions in the local volume. We available (or we excluded it for reasons mentioned −1 −1 adopted a nominal value of H0 = 70 km s Mpc , and above), we used the MBH-Mbulge relation to estimate accordingly adjusted the distances from Cosmicflows-3 the SMBH mass. We first used the K-band magnitude and galaxy groups, in which the adopted values were to estimate the stellar mass, which in turn was used to 75 km s−1 Mpc−1 and 73 km s−1 Mpc−1, respectively. In calculate the bulge mass. In particular, we determined our calculations, we did not include the distance uncer- the absolute magnitude MK from the observed mK as tainties, primarily because they are negligible compared M = m − 5 × log (D) − 25 − 0.11 × A , (4) to the uncertainties of the mass estimates (see below). K K 10 v 2.2. SMBH Masses where D is the distance in Mpc and Av is the extragalac- tic extinction, which we calculated from the reddening For the SMBH masses, we used direct measurements relation R = A /E = 3.1 (Fitzpatrick 1999), with when available and global scaling relationships other- V V B−V the color term E provided by the 2MRS catalog. wise. Dynamical mass measurements are observation- B−V Next, we obtained the total stellar mass of the galaxy ally expensive, and are limited only to a small num- M using the correlation (Cappellari 2013) ber of galaxies. We used the catalog compiled by Mc- ∗

Connell & Ma(2013), which includes 72 galaxies with log10(M∗) = 10.58 − 0.44 × (MK + 23) . (5) high-quality dynamical mass measurements, from ob- servations of stellar, gaseous, or maser kinematics. We At this step, we excluded 21 quasars, since the K-band added to this catalog recent measurements for the galax- luminosity does not correspond to the stellar luminosity, ies NGC 4526, M60-UCD1, NGC 1271, NGC 1277, but is dominated by dust emission (Barvainis 1987). and NGC 1600 (Davis et al. 2013; Seth et al. 2014; Next, we estimated the bulge mass as Mbulge = Walsh et al. 2015, 2016; Thomas et al. 2016). We also fbulgeM∗, with fbulge the fraction of the stars that re- included 29 galaxies from the Active Galactic Nuclei side in the bulge. For early-type galaxies, we assigned (AGN) Black Hole Mass Database,2 the masses of which their total stellar masses to their bulges (fbulge = 1), were determined with spectroscopic reverberation map- whereas for late-type galaxies, we used the bulge-to- ping (Bentz & Katz 2015). However, the above dynam- total mass ratio, determined from image decomposition ical tracers probe spatial scales that cannot distinguish (Weinzirl et al. 2009), which depends on the galaxy’s if the mass corresponds to one or two SMBHs, and thus sub-class. The bulge fraction also depends on the exis- reflect the total enclosed mass. tence of a galactic bar (e.g., Weinzirl et al. 2009, Fig- For the remaining galaxies, we estimated the SMBH ure 14), but, since the 2MRS catalog typically does not masses from correlations with observable quantities, like specify whether a galaxy has a bar, we used the average value of barred and unbarred galaxies for each sub-class. the MBH-σ or the MBH-Mbulge correlations. First, we extracted measurements of the velocity dispersion σ In Table1, we summarize the adopted values of bulge- to-total mass ratio (or equivalently the bulge fraction for 3,300 galaxies from the Lyon-Meudon Extragalac- 3 tic Database (LEDA), using the Extragalactic Distance fbulge), as a function of the galaxy type. Additionally, Database to cross-correlate our sample with LEDA. We the galaxy types are unknown for ∼ 40% of the galax- ies in the catalog. For those, we estimated their bulge used the MBH-σ relation from McConnell & Ma(2013) fractions assuming two different scenarios for the galaxy −1 log10(MBH) = a1 + b1 log10(σ/200 km s ) , (3) type (see § 2.3). We estimated the SMBH mass using the correlation with (a1, b1) = (8.39, 5.20) for early-type (elliptical and from McConnell & Ma(2013), S0) galaxies and (a1, b1) = (8.07, 5.06) for late-type (spi- ral) galaxies. Of the 3,300 galaxies, we excluded 810 11 log10(MBH) = a2 + b2 log10(Mbulge/10 M ) , (6) galaxies because their velocities were beyond the range −1 from which the correlation was established (80 km s – with (a2, b2) = (8.46, 1.05). The MBH-σ and the MBH- −1 400 km s ) or their dispersion measurements had sig- Mbulge correlations have intrinsic scatters of  = 0.38 nificant uncertainty (>10%). Additionally, the morpho- logical types of 284 galaxies were not included in the 3 We note that none of the late-type galaxies was included in NANOGrav’s sensitivity volume and thus the details of the as- signed bulge mass do not have a significant impact on our final 2 http://www.astro.gsu.edu/AGNmass/ results. 6 The NANOGrav Collaboration

Table 1. Summary of methods for estimating SMBH nside=8) from Aggarwal et al.(2019). Therefore, a masses. In the second part of the table, we focus on the sub- binary could be detectable by NANOGrav, if set of galaxies the SMBH mass of which was determined from the MBH-Mbulge correlation. We show the adopted mean hGW (q = 1, fGW = 8 nHz) ≥ h95 (Ra, Dec) (7) bulge-to-total mass ratio (fbulge) as a function of galaxy type from Weinzirl et al.(2009). In Table2, we present the entire 2MRS galaxy catalog, with the estimated SMBH masses and galaxy distances Method Number of galaxies (including the method with which each quantity was Dynamical methods 77 calculated) and the relevant GW upper limits at fGW = Reverberation mapping 29 8 nHz. Excluded quasars 21 Among the galaxies with known types, there are 194 9 10 MBH-σ 2,206 galaxies with SMBH mass of 10 − 10 M in the vol- 4 MBH-Mbulge 41,200 ume NANOGrav can probe (up to 500 Mpc). The vast

Galaxy Type Number of galaxies fbulge majority of those are early-type galaxies with only four Unknown type∗ 18,575 1 and 0.31 classified as S0/a. Next, we examined the unclassified Known type 22,625 galaxies, and estimated their masses examining two pos- sible scenarios for the galaxy type: (1) We assumed Elliptical 8,791 1.00 that they are late-type Sa galaxies with f = 0.31, S0 1,363 0.25 bulge which corresponds to the most massive SMBHs for the Sa 2,237 0.31 case of spiral galaxies. Under this assumption, none of Sab 1,626 0.29 them are within NANOGrav’s reach. (2) We assumed Sb 2,709 0.17 that they are early-type galaxies (fbulge = 1), in which Sbc 1,908 0.13 case, another 84 galaxies enter NANOGrav’s sensitivity Sc 2,838 0.17 volume. We visually inspected those 84 galaxies, us- Scd 693 0.09 ing data from the online database Simbad.5 We found Sd 221 0.14 that 22 of those are indeed early-type galaxies, classified Sdm 136 0.15 as brightest cluster galaxies (BCGs). Six are classified Sm 103 0.15 as quasars/AGN, which we excluded because we cannot ∗ If the morphological type of a galaxy was not specified, reliably estimate the stellar mass from the K-band mag- we examined two possible scenarios (see § 2.3). nitude (§ 2.2). The remaining 56 galaxies are impossible to classify based on the available images. We kept these and  = 0.44, respectively. We used these scatters as a galaxies as a separate sub-sample, the analysis of which proxy for the uncertainties in the mass estimates, but we is presented in the Appendix. did not include additional sources of uncertainty (e.g., Figure1 (bottom panel) shows the 194 galaxies, along due to galaxy classification, bulge fraction, etc). with the 22 BCGs among the subset with initially un- known type, that are within NANOGrav’s sensitivity 2.3. Galaxies in the NANOGrav Volume volume, color-coded according to their distance with the marker size denoting the estimated SMBH mass. As ex- We calculated the GW strain amplitude from eq. (1), pected, there are significantly more galaxies in the most using the estimates of the SMBH masses (regarded as sensitive half of the sky, where most of the systematically the total masses of the binaries) and the distances ob- monitored pulsars are located. In the least sensitive half tained above. We considered a galaxy to be within the of the sky, binaries can be detectable only if they reside volume that NANOGrav can probe, if the GW strain in very massive or in relatively nearby galaxies. of the tentative binary in the most optimistic scenario Additionally, we calculated the number of galaxies in terms of mass ratio and binary orbit, i.e., equal-mass that NANOGrav can probe considering the uncertainty binary in a circular orbit emitting GWs at NANOGrav’s in the mass estimates. Assuming the total mass of the most sensitive frequency, h (q = 1, f = 8 nHz), GW GW binary to be equal to M − σ and M + σ, we found was equal to or higher than the upper limit from the BH BH 20 and 2,313 galaxies with known morphological clas- 11 yr dataset. However, as shown in Fig.1 (top panel), sification within the sensitivity volume. When we also the sensitivity of NANOGrav depends on the spatial dis- tribution of pulsars on the sky and is highly anisotropic. 4 9 10 We thus compared the GW strain to the upper limit NANOGrav can detect binaries with total mass of 10 −10 M up to an average distance of ∼ 15 − 725 Mpc, respectively. in the direction of the source, h95 (Ra, Dec), using a 5 skymap with resolution of 768 pixels (healpix with http://simbad.u-strasbg.fr/simbad/ limits on smbhbs within 500 mpc 7

Table 2. 2MRS galaxy catalog with estimates of the SMBH mass and distance, along with the GW strain upper limit towards the direction of each galaxy.

2MASS Name∗ Dist Mass GW Method∗∗ Limit −15 [Mpc] [M ][×10 ] J00424433+4116074 0.82 8.15 4.22 1 1 J00473313−2517196 3.81 7.74 3.76 1 4 J09553318+6903549 3.87 7.90 5.51 1 1 J13252775−4301073 3.92 7.77 3.31 1 1 J13052727−4928044 3.99 7.22 3.28 1 4 ......

∗JHHMMSSss±DDMMSSs ∗∗Method used to calculate the distance and the SMBH mass. See online catalog for a detailed description.

large discrepancy between the above numbers. We fur- ther explore this issue in § 5.3. 3. GW SEARCH METHODOLOGY As shown in Fig.1, the GW signal induced by a SMBHB on a PTA depends on the position of the source with respect to the position of the pulsars in the array. For this reason, we derived the GW upper limits towards the direction of each galaxy. In particular, we modeled the timing residuals δt as 100 200 300 400 500 Distance [Mpc] δt = M + nwhite + nred + s , (8) where M represents the timing model (with M the design matrix for the linearized timing model, and  Figure 1. Top: The 95% upper limit on the GW strain am- a vector of timing model parameters), nwhite and nred plitude for circular SMBHBs at fGW = 8 nHz as a function describes the white and red noise of the pulsars, re- of sky position in equatorial coordinates from the analysis of spectively, and s the GW signal, which depends on the the NANOGrav 11 yr dataset (Aggarwal et al. 2019). The dimensionless GW strain amplitude from eq. (1), the stars indicate the pulsars used in that analysis. Bottom: GW polarization, the source location (i.e., the so-called Spatial distribution of 216 galaxies (194 among the galaxies of known type and 22 BCGs identified among the initially antenna-pattern), etc. Details on the analysis can be unclassified galaxies), within the NANOGrav sensitivity vol- found in Aggarwal et al.(2019) and Arzoumanian et al. ume. Colors and marker sizes reflect galaxy distances and (2020). estimated masses of the central SMBHs, respectively. Galax- We computed the upper limits performing a Bayesian th ies in the least sensitive half of the sky (hGW below the 50 analysis with fixed GW frequency, and repeated the percentile from the map of the top panel) are outlined with analysis for 11 distinct frequencies linearly spaced in the red circles. interval [2.75, 10] nHz. Each Bayesian analysis requires ∼ 103 CPU Hours due to the large number of parameters included the galaxies of unknown type, there are still that must be included in the model of eq. (8). Therefore, 20 galaxies in the volume for the lower mass bound, in order to reduce the computational cost, instead of an- whereas for the high mass bound the number becomes alyzing each galaxy individually, we divided the sky into 2,338 and 4,545 for the two scenarios described above 192 pixels of equal area (healpix with nside=4) and (i.e. late vs early type galaxies). However, in this case, conducted the analysis only for the 110 pixels containing it is impractical to visually examine over 2,000 galax- galaxies in our sample. Nearby pixels have very similar ies to verify whether they are indeed elliptical galaxies. upper limits, and thus performing an upper limit anal- The impact of the mass uncertainty is obvious from the ysis for each galaxy individually would be significantly 8 The NANOGrav Collaboration Period [yr] In most cases, we modeled the timing residuals from 20 16 12 8 eq. (8) sampling the GW signal directly in GW strain -14 h . The uncertainty in the mass estimates was not th GW 95 percentile included in this analysis. However, for a small subset of -16 50th percentile th galaxies (which host the binaries with the highest S/N), 5 percentile we expressed the GW signal in eq. (8) in terms of mass ratio and sampled directly in q, using the distribution 8.5 of the total mass (with its uncertainty) as a prior. The reason for limiting this analysis to only five galaxies is 8 again computational efficiency. 0.3 Major Merger Excluded (q<0.25) 4. RESULTS

0.1 4.1. Galaxy Ranking based on signal-to-noise ratio

3 4 5 6 7 8 9 10 Identical binaries (in terms of total mass, mass ratio, Frequency [nHz] and distance) may have different probabilities of detec- tion depending on their coordinates (Fig.1). We quan- Figure 2. Illustration of the 95% upper limits on the GW strain (top panel), the chirp mass (middle panel) and the tified the probabilities of detection by calculating the mass ratio (bottom panel) as a function of GW frequency S/N for putative binaries in the 216 galaxies that fall and orbital period for a circular SMBHB for the galaxy M49. within the sensitivity volume of NANOGrav. (The Ap- pendix presents the S/N of the 56 galaxies of unknown more expensive computationally and would provide only type that are detectable by NANOGrav, only under the incremental improvements. assumption that they are early type galaxies.) Using the upper limits of the relevant pixel, we then We calculated the total S/N of a binary, adding the calculated the constraints on hypothetical binaries as a S/Nj in each individual pulsar, following Rosado et al. function of the GW frequency, or equivalently the or- (2015). For this purpose, we assumed equal mass bi- bital period of a circular binary. In Fig.2, we illus- naries in circular orbits, and a fixed GW frequency of trate the steps of the analysis for the galaxy M49 (or 8 nHz. Since the S/N depends on the inclination i, the 2MASS J12294679+0800014) in our sample, an early initial phase Φ0, and the GW polarization angle ψ, we type galaxy at a distance of 17.2 Mpc with a SMBH randomly draw these parameters from uniform distri- 9 butions, in the range [−1, 1] for cos i, [0, π] for ψ and mass of 2.4 × 10 M . The top panel shows the distri- th th [0, 2π] for Φ0. We estimated the average total S/N from bution (5 to 95 percentile) of the GW strain, hGW. Using eq. (1), and the galaxy distance we calculated in 1000 realizations for each putative binary. In Table3, § 2.1, we converted the distribution (and in turn the up- we ranked the binaries from the highest to lowest based on their average S/N. per limits) of hGW to the distribution of the chirp mass, which we present in the middle panel of Fig.2. We then 4.2. Constraints on individual galaxies solved eq. (2) for the mass ratio using the mean SMBH mass from § 2.2 as the total mass of the binary We used the NANOGrav 11 yr dataset to derive con- straints on hypothetical SMBHBs in the 216 galaxies identified in § 2. Even though this dataset allows us 1 − p1 − 4(M/M )5/3 q = tot (9) to probe frequencies up to ∼317.8 nHz, we limited our p 5/3 1 + 1 − 4(M/Mtot) analysis to 10 nHz for computational efficiency. We fur- ther explain this choice towards the end of the section. This provided the distribution of the mass ratio as a As mentioned above, in order to avoid unnecessary com- function of frequency, which is shown in the bottom putations, we calculated the GW strain upper limits in panel of the figure. Next, we examined the existence 110 pixels (out of the 192 in which we divided the sky). binaries delivered by major galaxy mergers with mass In Fig.3, we show the 95% upper limits on the mass ratio at least 1/4. For the range of NANOGrav frequen- ratio (which corresponds to the solid line in the bottom cies where the derived mass ratio upper limit was less panel of Fig.2) as a function of GW frequency for eight than 1/4, such binaries can be excluded (see § 4.4). This galaxies in the sample. The four galaxies on the top range is indicated with an arrow at the bottom panel of and bottom panels are distributed close to the most and Fig.2. The distinct frequencies we analyzed are indi- least sensitive sky location, respectively. We randomly cated by vertical gray lines in the figure. selected these galaxies, in order to illustrate a wide range limits on smbhbs within 500 mpc 9

0.8 600 1

0.6 J1812+0730 [9.7, 201] 0.9 95 400 0.8 0.4 200 0.7

0.2 0.6 Distance [Mpc] Distance J0209+3547 [9.8, 158] 0.5 J0329+3947 [9.9, 103] 0.6 0.4 400 0.3

0.4 Mass Ratio Upper Limit, q Limit, Upper Ratio Mass 200 0.2 0.2

Distance [Mpc] Distance 0.1 0 0 9 10 4 5 6 7 8 9 10 10 10 Frequency [nHz]

Figure 3. Constraints on the mass ratio versus GW fre- Figure 4. Galaxy total mass vs galaxy distance color-coded quency (or orbital period) for galaxies, randomly selected by 95% upper limit on the mass ratio at fGW = 8 nHz. The from the most sensitive part of the sky (top panel) and the top panel shows the most sensitive half of the sky, whereas least sensitive part of the sky (bottom panel). The abbrevi- the bottom panel illustrates galaxies in the least sensitive ated galaxy names, along with their properties (logarithmic half, which are outlined with red circles as in Fig.1. SMBH mass in solar masses and distance in Mpc) are given in the legend. Since it is not practical to show the mass ratio con- straints as a function of frequency for the entire sample, of total masses and distances of the hypothetical bina- in Fig.4 we present the 95% upper limits for a fixed ries; these quantities are shown in the brackets next to frequency of 8 nHz for the 216 galaxies in our sample, the abbreviated galaxy names in the legend. For a sig- with the top/bottom panel illustrating galaxies in the nificant fraction of the galaxies we can place stringent most/least sensitive half of the sky. There is an obvious constraints on the mass ratio, for several GW frequen- trend from yellow to dark blue points, which indicates cies. Therefore, if there is a circular binary in the center that increasingly unequal mass binaries are required, of these galaxies, the mass of the secondary SMBH must when the distances decrease for fixed mass (vertically be only a few percent that of the primary. We tabulate down) and when the total masses increases for fixed dis- the constraints on mass ratio in Table3. tance (horizontally to the right), as expected. The same It is obvious from Figs.2, and3 that the mass ratio general trend is repeated in the least sensitive half of the upper limit is a declining function of the frequency at sky, but the overall curve is shifted down, which means low frequencies, whereas at high frequencies (beyond a that we are limited to more massive and more nearby critical frequency), the function becomes flat and the galaxies. We emphasize that we selected this particular constraints on the mass ratio are independent of fre- frequency (fGW=8 nHz) for two reasons: (1) this is the quency. The flattening of the mass ratio upper limit most sensitive frequency of NANOGrav — for this rea- at high frequencies can be derived analytically, since son, the galaxy sample was initially selected assuming for frequencies above 8 nHz, the GW strain upper limit tentative binaries emitting GWs at 8 nHz, as shown in 2/3 has a power-law dependence on frequency h95(f) ∝ f Fig.1, and (2) this frequency corresponds to the flat (see Aggarwal et al. 2019). From eq. (1) it then follows part of the mass ratio upper limit curve and thus our that the chirp mass upper limit for f>8 nHz is indepen- conclusions can be extended to higher frequencies. dent of the frequency, M95(f) = Const, and thus from eq. (9) the mass ratio upper limit is also constant for 4.3. Sampling in mass ratio high frequencies, i.e., q (f) = Const. This is numer- 95 For the five top ranked galaxies based on their S/N, ically demonstrated in Fig.5, where we extended the we run the upper limits analysis sampling the likelihood analysis to higher frequencies. As a result, our con- of eq. (8) directly in the mass ratio. This allows us to clusions can be extended to higher frequencies without incorporate the uncertainty in the total binary mass as additional computations, which explains our decision to a prior. In Fig.5, we show the 95% mass ratio upper restrict the analysis to 10 nHz. limits as a function of frequency. For these five galaxies, we run the analysis for the full range of NANOGrav fre- 10 The NANOGrav Collaboration

Table 3. The 95% mass ratio upper limits as a function of frequency for the galaxies in the NANOGrav volume ranked by the mean S/N of putative equal-mass circular binaries at 8 nHz. The full table is available online.

∗ 2MASS Name Mass Dist S/N q95 q95 q95 q95 q95 q95 q95 q95 q95 q95 q95 ∗∗ [M ] [Mpc] [2.75] [3.47] [4.20] [4.92] [5.65] [6.37] [7.10] [7.82] [8.55] [9.27] [10.0] J13000809+2758372 10.32 112.2 240.1 0.15 0.06 0.05 0.04 0.02 0.01 0.02 0.01 0.01 0.01 0.01 J12304942+1223279 9.79 17.7 188.8 0.20 0.16 0.08 0.04 0.03 0.03 0.02 0.02 0.02 0.02 0.02 J04313985−0505099 10.23 63.8 135.9 0.42 0.16 0.10 0.06 0.05 0.03 0.03 0.03 0.03 0.03 0.03 J12434000+1133093 9.67 18.6 115.4 1.00 0.21 0.13 0.07 0.05 0.04 0.03 0.02 0.02 0.02 0.02 J13182362−3527311 9.89 53.4 99.9 0.55 0.25 0.09 0.04 0.04 0.05 0.04 0.02 0.03 0.02 0.02 ......

∗JHHMMSSss±DDMMSSs. ∗∗ GW frequencies in nHz.

of NANOGrav is limited (e.g., at ∼30 nHz, which corre- 1 sponds to a period of 1 year and is caused by fitting the 0.5 pulsars’ positions). [NGC 5062] In Fig.5, we also show with red dashed lines the limits derived from the GW strain using the mean SMBH as the total mass of the binary in § 4.2 for comparison. As J1243+1133 95 0.5 expected, including the uncertainty leads to weaker up- [NGC 4649] per limits, which is more pronounced for galaxies with highly uncertain mass measurements, like NGC 4889 and NGC 5062.7 The other three galaxies have dynam- 0.5 [NGC 1600] ically measured SMBH masses and thus relatively small uncertainties. The upper limits here are slightly higher, but comparable to the ones obtained above, especially J1230+1223 at higher frequencies (flat part of the upper limit curve). 0.5 Mass Ratio Upper Limit, q Limit, Upper Ratio Mass [M87] Additionally, the constraints we derive for putative bi- naries in these three galaxies are significant with only very unequal-mass binaries allowed, i.e., the secondary J1300+2758 0.5 SMBH should be at least 10 times less massive than the [NGC 4889] primary for the majority of frequencies we examined. 0 We further explore the impact of the uncertainty in the 10 100 total mass in § 5.3. Frequency [nHz] 4.4. Constraints on major galaxy mergers Figure 5. The 95% upper limits on mass ratio for the top five galaxies based on S/N. The blue curves show the upper We explored whether the galaxies in our sample could limits derived by sampling the likelihood directly in mass host a binary delivered by a major merger. Major merg- ratio, whereas the red dashed curves show the upper limits ers are considered the mergers between galaxies with derived from the GW strain/chirp mass upper limits, which similar mass (i.e., minimum mass ratio of 1/4) and play ignore the uncertainty in the total mass. an important role in galaxy formation and evolution (Volonteri et al. 2003; Kelley et al. 2017). Since the quencies up to the cutoff frequency, which depends on mass of the central SMBH is tightly correlated with the the binary total mass.6 This figure also confirms that galaxy mass, we assume that a major merger would re- the constraints are roughly constant at high frequencies, sult in a binary with a mass ratio q > 1/4. This means except for specific frequencies, for which the sensitivity that major galaxy mergers are also significant for PTAs, since they produce promising binaries for GW detection.

6 For the most massive binaries, the highest frequencies NANOGrav can probe are not relevant; if we set the cut- 7 We note that the mass uncertainty of NGC 4889 is large, even off frequency at the respective innermost stable circular orbit though it was directly measured with dynamical methods. On the 2 (ISCO), RISCO = 6GMtot/c , for a binary with total mass other hand, the large uncertainty for NGC 5062 is unsurprising, log (M /M ) = 10.32 (like our top ranked galaxy NGC 4889), 10 tot since its mass was estimated from the MBH-σ relationship, which the cut-off frequency is at ∼220 nHz. has significant intrinsic scatter. limits on smbhbs within 500 mpc 11

can probe binaries from major mergers; for binaries of Distance [Mpc] fixed chirp mass, i.e., fixed mass ratio (q=1/4) and to- J1521+3040 tal mass (see eq.1), the GW upper limit can be con- J0329+3947 verted to a maximum distance. Due to heterogeneous J1442+4150 N5739 350 sensitivity, the maximum distance is different for each

J1604+2355 N6051 pixel, and thus the volume NANOGrav can probe is a J1036+5837 N3286 shell of irregular shape. For each pixel, we calculated 300 J1511+0620 the maximum distance and the respective spherical vol- J2003+5120 ume that NANOGrav would cover if the sensitivity was J1334+3446 N5228 J0224+4237 250 homogeneous. Since the pixels have equal area, we ap- J2215+3717 N7242 N6876 proximated the NANOGrav volume with a sphere of an N6861 effective volume equal to the average spherical volumes N5094 200 J1548+0850 that correspond to each pixel. Then, we calculated the number of galaxies within this volume. Dividing the J2328+1718 N7681 J1812+0730 number of galaxies with the effective volume of the shell N5516 150 N1407 we get the NANOGrav limit on the density of galaxies J1229+0800 M49 J1455+1151 that could host binaries delivered by major mergers at J1856+5222 N6732 100 N5419 8 nHz. J1813+1849 N6587 Next, we examined several discrete values for the to- N5328 9 10 J1144+1956 N3842 tal binary mass in the range of 10 − 10 M , where N5062 50 all the galaxies identified in § 2.3 lie. As the binary to- J1243+1133 M60 N1600 tal mass increases, the distance NANOGrav can probe J1300+2758 N4889 J1230+1223 M87 0 becomes larger, which allows more galaxies to enter the 2 4 6 8 10 sensitivity volume. However, a counter effect is that Frequency [nHz] massive galaxies, with higher SMBH masses (e.g., with

Figure 6. Range of frequencies for which a binary with mass log10(MBH /M ) > 9.5) are extremely rare. In Fig.7, ratio q > 1/4 delivered by a major galaxy mergers can be we show with a black solid line the NANOGrav limit on excluded. Each bar corresponds to a galaxy, the abbreviated the density of binaries produced by major mergers. The coordinates of which are shown on the left. If the galaxy is errorbars denote the range of densities, if we consider included in one of the major galaxy catalogs, like the Messier the 1σ uncertainty in the SMBH mass. For comparison, or the New General Catalog, the respective names are also we also show the overall galaxy density above a certain included. The color coding and the size of the circles on the right reflects the galaxy distance and the SMBH mass, fixed mass for spherical volumes with radii of 100, 300 respectively, as in Fig.1. and 500 Mpc. We see that the density of galaxies is decreasing, when we examine a larger volume (from a radius of 100 Mpc to a radius of 500 Mpc). This may Based on the constraints on the mass ratio derived in mean that the universe has higher density locally, po- § 4.2, we excluded binaries delivered by major mergers tentially because we consider a small number of galaxies for galaxies where q95 < 1/4. In Fig.6, we show the concentrated in local clusters (Virgo Cluster at 16 Mpc, range of frequencies for which we were able to exclude at 120 Mpc). An alternative explanation the existence of binaries from major mergers for the 50 is that the 2MRS catalog may not be complete out to top-ranked galaxies based on S/N. Each bar corresponds 500 Mpc, even though this scenario is unlikely, especially to a distinct galaxy, the abbreviated name of which is for the most massive galaxies that we consider here. shown on the left. The color coding reflects the distance We emphasize that this is the first constraint on major of the binary and the size of the marker on the right the galaxy mergers derived directly by GW data. However, total mass of the binary (similar to Fig.1). It is obvious it is worth pointing out that there are several caveats from the figure that the most informative constraints are directly connecting a major merger to a binary with derived for relatively nearby galaxies (within 100 Mpc). q > 1/4. The main caveats are summarized below: (1) Motivated by the above findings, we estimated the The scaling relations between the SMBH and the host number density of nearby galaxies, for which we can ex- galaxy have significant scatter, and thus the binary mass clude binaries delivered by major mergers (q95 < 1/4) at ratio does not necessarily reflect the mass ratio of the NANOGrav’s most sensitive frequency (fGW=8 nHz), as initial galaxies. (2) The mass of each SMBH may evolve a function of the binary total mass. For this, we first es- significantly in the post-merger galaxy, meaning that the timated the maximum volume out to which NANOGrav 12 The NANOGrav Collaboration

10-5 6 10.5 NANOGrav Limit 10 100Mpc 5

300Mpc 9.5 ] 3 500Mpc 9 4 8.5

3 8 7.5 Present Catalog 2 7 Mingarelli et al., 2017

Density [Galaxies/Mpc Density 6.5 Schutz & Ma, 2016 1 NANOGrav volume 6 0 100 200 300 400 500 0 Distance [Mpc] 9.2 9.3 9.4 9.5 9.6 9.7 Figure 8. SMBH mass versus distance for the ∼45,000 galaxies in the 2MRS catalog (grey triangles), compared with Figure 7. Density of binaries delivered by major galaxy the galaxy samples used in SM16 (blue diamonds), and in mergers (q > 1/4) as a function of total binary mass. M17 (black squares). We also indicate the 216 galaxies that The shaded region shows the excluded densities based on lie within the volume NANOGrav can probe with magenta NANOGrav constraints at a frequency of 8 nHz. The lines stars. show the galaxy density within spherical volumes of radii 100 Mpc (solid line), 300 Mpc (dashed line), and 500 Mpc respectively. In 2.3, we found that 24 of these galaxies (dotted line). § are within the NANOGrav volume at 8 nHz. Since each array has distinct and heterogeneous sensitivity, which binary mass ratio at the final stages may not be directly depends on the distribution of pulsars each collabora- linked to the initial mass ratio of the two SMBHs. For tion systematically monitors, it is not surprising that instance, if the secondary SMBH accretes at a higher each array can probe a different number of binaries. For rate, as seen in binaries embedded in circumbinary disks, instance, PPTA is more sensitive in the southern hemi- the mass ratio tends to equalize (Farris et al. 2014; Si- sphere, where most of the PPTA pulsars lie (e.g., Fig. 2 wek et al. 2020). (3) In the mass ratio upper limits, we in SM16). did not include the uncertainty of the SMBH mass es- In Fig.9, we show the mass ratio constraints for four timate. Therefore, the q < 1/4 cutoff is used merely as of the top galaxies in SM16 and compare with the 95% proxy for a binary produced by a major . upper limits we obtained in § 4.2. We see that the NANOGrav upper limits are comparable or slightly bet- 5. DISCUSSION ter (e.g., NCG 4889, NGC 4649) in the flat part of 5.1. Comparison with SM16 the upper limit curve. On the other hand, at the low- SM16 used PPTA and EPTA upper limits and per- est frequencies, our limits are less constraining. The formed a similar analysis on a sub-set of galaxies. In par- differences between the results of the two studies are ticular, they focused on galaxies with dynamical mass mainly due to a combination of the following reasons: measurements (77 galaxies) and galaxies from the MAS- (1) There are slight differences in the assumed proper- SIVE survey (116 galaxies). The total sample consisted ties of the putative binaries in the two studies. The of 185 galaxies, since eight of the MASSIVE galaxies total masses of the binaries in these four galaxies are had dynamical mass measurements. In Fig.8, we show identical, but the distances in our study are slightly with blue diamonds the distribution of these galaxies larger. All else being equal, this would naturally lead as a function of distance and SMBH mass. We com- to slightly weaker upper limits for the binaries consid- pare them with the sample of galaxies in this paper il- ered here. (2) At the most sensitive frequency of 8 nHz, lustrated with gray triangles, and the galaxies in Min- the sky averaged 95% upper limit of NANOGrav is ap- −15 garelli et al.(2017) shown with black squares (see be- proximately h0 < 7.3 × 10 , which is 1.4 times lower −14 low). We also summarize the galaxy selection and prop- than the EPTA limit (h0 < 10 ), and 2 times lower −14 erties, along with key differences between the samples than the PPTA limit (h0 < 1.7 × 10 ) used in SM16. in Table4. Of the 185 galaxies, 33 were in the sen- The improvement in the GW limits should lead to more sitivity volume of PPTA and 19 in the EPTA volume, stringent constraints on the mass ratio and partly ex- limits on smbhbs within 500 mpc 13

Table 4. Comparison of the galaxy catalog with related previous studies. SM16 M17 Current Paper Sample 185 5,110 43,533 Ma et al.(2014) 2MRS Catalog (early-type, 2MRS Catalog

McConnell & Ma(2013) D < 225 Mpc, MK < −25) (entire catalog) Distance estimates (1) Groups Catalog (1) Groups Catalog (1) Cosmicflows-3 (2) Flow-field model (2) Groups Catalog (3) Flow-field model SMBH mass estimates (1) Dynamical methods (1) Dynamical methods (1) Dynamical methods

(2) MBH-Mbulge (2) Reverberation mapping

(3) MBH-σ

(4) MBH-Mbulge In NANOGrav volume 24 106 216

Mingarelli et al.(2017) [hereafter M17] modeled the J1300+2758 J1230+1223 [M87] local GW sky, using galaxy merger rates from Illustris, [NGC 4889]

0.1 NANOGrav and the local massive galaxies as an observational an- 95 EPTA chor. For this, they extracted a sample of 5,110 mas- PPTA 0.05 sive early-type galaxies from 2MRS,8 and calculated the probability of each galaxy hosting a GW-emitting SMBHB system. These galaxies are illustrated with J1243+1133 black squares in Fig.8. [NGC 1600] [NGC 4646] 0.1 Even though the starting point of both studies is the 2MRS catalog, the goals of the two studies are comple- Mass Ratio Upper Limit, q Limit, Upper Ratio Mass 0.05 mentary, which explains why different selection criteria and cuts were employed. For instance, M17 assessed the likelihood of PTAs detecting GWs from individual 4 6 8 4 6 8 10 sources, through multiple realizations of the local uni- Frequency [nHz] verse and thus the completeness of the galaxy sample is Figure 9. Mass ratio upper limits from the NANOGrav important. On the other hand, we present a purely ob- 11 yr dataset (solid lines), compared to upper limits from servational approach, in which we used the current GW PPTA (dashed lines) and EPTA (dotted lines), for four upper limits to place constraints on circular binaries. galaxies in SM16. For our purposes, compiling the most comprehensive catalog that includes all the galaxies that NANOGrav plains why NANOGrav performs better at higher fre- can probe is crucial. quencies. (3) PPTA and EPTA have increased sensitiv- Examining the galaxies in the M17 catalog, we found ity at low frequencies, because some of the pulsars have that only 106 lie in the NANOGrav sensitivity volume, a long baselines of timing observations (longer than 17 yr). factor of two lower compared to the 216 from the catalog Therefore, the limits derived with the NANOGrav 11 yr that we use here. This discrepancy can be understood dataset are less constraining at these low frequencies. in light of the selection criteria and calculation of the (4) The NANOGrav upper limits vary significantly from galaxy properties (SMBH masses, and distances). Below the most sensitive to the least sensitive sky location we summarize the key differences between the catalogs −15 −14 (from h0 < 2.0 × 10 to h0 < 1.34 × 10 ). This (but see also Table4): means that the relative position of the galaxy to the pulsars in the array is important and thus discrepan- 1. We examined the entire 2MRS catalog, whereas cies between the derived limits from the three arrays M17 imposed cuts on the absolute K-band magni- are expected. An example of this is NGC 1600, which is tude (M < −25) and the galaxy distance (D < located at NANOGrav’s least sensitive part of the sky, K 225 Mpc). The magnitude cut has no effect on where EPTA has at least four pulsars in its vicinity.

8 https://github.com/ChiaraMingarelli/nanohertz GWs/blob/ 5.2. Comparison with Mingarelli et al.(2017) master/galaxy data/2mass galaxies.lst 14 The NANOGrav Collaboration

the sample of interest (i.e., galaxies with binaries assumed SMBH mass and in turn the GW strain detectable by NANOGrav), as it mostly excludes are overestimated. low-mass galaxies. However, the distance cut is significant, because it excludes massive galaxies at 5.3. Uncertainty in total mass relatively large distances, but within NANOGrav’s This work presented a study on how the NANOGrav capabilities (e.g., see Fig.8). data can be used to place constraints on SMBHBs in 2. We used a multi-layered approach for the esti- massive nearby galaxies. For most galaxies, we derived mates of SMBH masses. The difference with M17 the chirp mass upper limits from the GW strain, and lies in the inclusion of mass estimates from re- converted them to mass ratio upper limits using the verberation mapping for AGN, and the estimates mean SMBH mass (measured or estimated from scaling relations) as the total mass of the binary. As explained from the MBH-σ relationship. Both methods above, the main reason for this approach is computa- provide better estimates compared to the MBH- tional efficiency. However, one of the caveats is that it Mbulge relationship, because the former provides dynamical mass measurements, whereas the latter ignores the uncertainty in the total binary mass. As we relies on fewer assumptions (see also § 5.3).9 demonstrated in § 4.3, incorporating the SMBH mass uncertainty deteriorates the mass ratio upper limits, 3. For the distance estimates, we started from with larger uncertainties having a stronger impact on the Cosmicflows-3 catalog, since it compiles the these constraints. highest-quality distance measurements available Here we further explore the effect of the mass uncer- in the literature. For galaxies not included in tainty on the mass ratio upper limits. For this, we de- Cosmicflows-3, we used the catalog of galaxy rived the upper limits from the GW strain (as in § 4.2), groups (Crook et al. 2007) or corrected for the lo- but instead of using the mean SMBH mass in eq. (9), cal velocity field (Mould et al. 2000), as in M17. we sampled the total mass from the entire mass distribu- For the latter subset, the distances in the two cata- tion, which includes the uncertainty. Next, we compared logs are in excellent agreement, with the exception with the upper limits we obtained sampling directly in of galaxies in the vicinity of the Great Attractor, the mass ratio in § 4.3, where the mass uncertainty was to which we assigned the velocity of the cluster. included as a prior in the analysis. We find that when For the subset of galaxies in Cosmicflows-3, the the uncertainty is small (e.g., NGC 4649, NGC 1600, distances in M17 are overestimated. M87), the upper limits from the different methods are in excellent agreement. However, if the mass uncertainty 4. We examined the entire 2MRS catalog, whereas is large (e.g., NGC 5062, NGC 4889), the mass ratio M17 selected early-type (elliptical and S0) galax- upper limits tend to be higher when we sample directly ies. Additionally, the classification in M17 was in mass ratio, which means that the two methods are based on the extragalactic database Hyperleda, not always equivalent. We carefully examined the cause whereas we used the classification from the 2MRS for this discrepancy for the galaxy NGC 4889 at a fixed catalog itself. We note that a comparison of frequency of 8 nHz, which we illustrate in Fig. 10. We galaxy types in the two catalogs indicates that see that the posterior distribution of the total mass is not all 5,110 galaxies in M17 are classified as skewed towards smaller values compared to the prior, early-type in the 2MRS classification; a fraction of which is shown with a dotted histogram. This results them (∼16%) are classified as late-type galaxies, in a tail in the mass ratio distribution, which leads to whereas ∼15% are unclassified. With visual in- significantly higher 95% mass ratio upper limits. spection of images in Simbad, we confirmed that Currently, this is a significant limitation, since sam- at least some of the galaxies in the M17 sample pling the mass ratio for each galaxy is computationally are indeed late-type (spiral) galaxies. Although expensive. This is why we chose to perform this type of it is impossible to determine whether this affects analysis only for five tentative binaries. We defer a sim- their overall conclusions, if spiral galaxies are mis- ilar analysis for the entire sample to a future study. Ad- classified as elliptical, the general trend is that the ditionally, for the vast majority of galaxies, the SMBH mass (attributed to the total mass of the binary) was estimated from global scaling relationships, the intrinsic 9 There are several steps involved in estimating the SMBH mass from the MBH-Mbulge relationship (estimation of stellar mass scatter of which inevitably results in large uncertain- from the K-band magnitude, estimation of the bulge fraction, ties. The uncertainty is small only for galaxies with etc), which may add significant uncertainty. precise dynamical mass measurements, which are obser- limits on smbhbs within 500 mpc 15

at the high mass end) compared to other correlations (Shankar et al. 2016). Adopting a shallower correlation would result in lower-mass binaries and weaker mass- ratio limits. Even though all of the above affect the de- termination of the mean SMBH mass (and in turn the mass ratio upper limits), they are likely negligible com- pared to the effect of the uncertainty due to the instrinsic scatter in the MBH-σ and MBH-Mbulge correlations, as shown above. Another potential source of uncertainty in estimat- ing the mass ratio limits is related to the distance de- termination. In this work, we did not include the un- certainty in the distance estimates, mainly because this uncertainty is typically smaller than that of the SMBH mass (especially when obtained from scaling relations). Additionally, the dependence of the GW strain on the distance is weaker compared to the dependence on the chirp mass. Overall, even if the impact of the distance uncertainty is likely negligible, it should be incorporated in future work. Future analysis should also explore the Figure 10. Posterior distribution of total mass and mass uncertainty in the value of the Hubble constant, which ratio for galaxy NGC 4889 obtained in 4.3. The prior § is necessary for the distance calculations. distribution of the total mass is also shown with the dotted histogram. Last but not least, in our analysis we assumed circular binaries, and thus our constraints cannot be generalized to non-circular binaries. In reality, binary orbits may de- vationally demanding and limited to a small number of viate from circular due to high eccentricity at formation, galaxies. Such observations are unlikely to become read- or because the eccentricity is excited as the binary in- ily available for a large number of galaxies in the near- teracts with the surrounding gaseous environment (Ar- future. However, in this study, we also included a small mitage & Natarajan 2005). (We note, however, that subset of AGN with SMBH masses measured from rever- GW emission tends to circularize the binary orbits.) In beration mapping. This is a promising direction, since eccentric binaries the GW strain is split among multiple planned multi-epoch spectroscopic surveys, like SDSS-V harmonics, resulting in weaker GW signal. In practice, (Kollmeier et al. 2017), will soon provide SMBH mass this means that if the galaxies we analyzed hosted ec- estimates for a large number of AGN through reverber- centric binaries, the limits we would derive would be less ation mapping. This is a particularly exciting develop- constraining that the ones we presented here for circu- ment for future studies, as AGN are promising candi- lar binaries. In a future study we will derive the upper dates for hosting SMBHBs. limits examining eccentric binaries, especially since non- 5.4. Additional assumptions and limitations circular binaries may be relatively common. The mass ratio upper limits depend on the total mass of the binary (eq.9), which was assumed to be equal 6. SUMMARY to the mass of the SMBH and for the vast majority of We analyzed the NANOGrav 11 yr dataset to derive galaxies was calculated from scaling relationships. In limits on tentative binaries in nearby galaxies. For this, § 2.2 and 2.3, we described the assumptions required we assembled a comprehensive sample of ∼44,000 galax- to determine the SMBH mass, such as the bulge frac- ies, for which we estimated the distance and the mass tion, the galaxy classification, the existence of a bar, etc. of the central SMBH, starting from the 2MRS catalog. We also demonstrated the challenges in determining the We found that 216 massive and relatively nearby galax- SMBH mass, when the galaxy type is unknown. An- ies are within the volume NANOGrav can probe, i.e., other effect is related to the choice of the SMBH mass if they hosted equal-mass circular binaries emitting at correlations; throughout the analysis, we have adopted 8 nHz (the most sensitive frequency of NANOGrav), the the scalings (MBH-σ and MBH-Mbulge) from McConnell GWs would be detectable by NANOGrav. We ranked & Ma(2013). These correlations may provide higher the tentative binaries in these galaxies based on the to- SMBH mass estimates (due to measurements of SMBHs tal S/N. Subsequently, we obtained mass ratio upper 16 The NANOGrav Collaboration limits as a function of frequency for each of those galax- ACKNOWLEDGMENTS ies. The upper limits for several galaxies are stringent and only binaries with very unequal masses are allowed. Author Contributions: We list specific contributions This is comparable to constraints placed on a potential to this paper below. MC led the work on this paper, SMBHB in the center of Milky Way. However, we also compiled the galaxy catalog, derived the mass ratio up- demonstrated that the uncertainty in the total mass of per limits and wrote the manuscript. SJV run the GW the binary can significantly weaken the upper limits and analysis. CPM, TJWL and JS contributed with signif- needs to be taken into account in future studies. Addi- icant discussions in shaping the project. NANOGrav tionally, we placed limits on binaries delivered by major data is the result of the work of dozens of people over galaxy mergers and constrained the density of such bi- the course of more than thirteen years. ZA, PBD, MED, naries at NANOGrav’s most sensitive frequency, which TD, JAE, ECF, EF, PAG, MLJ, MTL, RSL, MAM, CN, is the first such constraint based on GW data alone. DJN, TTP, SMR, PSR, RS, IHS, KS, and JKS devel- As the PTA upper limits improve (with more pul- oped the 11-year data set. All authors are key contribut- sars being systematically monitored for longer periods ing members to the NANOGrav collaboration. of time, and employing improved statistical techniques), The NANOGrav project receives support from Na- we will be able to place more stringent limits on GW tional Science Foundation (NSF) Physics Frontiers Cen- strain. This, in turn, will improve the limits on ten- ter award number 1430284. MC and SRT acknowledge tative nearby binaries. In particular, the combination support from NSF, award number 2007993. Part of this of data from individual PTAs in an IPTA dataset will research was carried out at the Jet Propulsion Labo- provide not only improved upper limits, but also a more ratory, California Institute of Technology, under a con- homogeneous sensitivity on the sky. The enhanced sen- tract with the National Aeronautics and Space Admin- sitivity will allow us to probe a larger volume extending istration. We are grateful for computational resources beyond the local universe. As a result, a larger number provided by the Leonard E. Parker Center for Gravi- of galaxies will become accessible for this type of study, tation, Cosmology & Astrophysics at the University of while the limits for the current galaxies will be improved Wisconsin-Milwaukee, which is supported by NSF Grant further. In addition, NANOGrav is planning targeted PHY-1626190. Data for this project were collected us- pulsar searches towards the direction of promising galax- ing the facilities of the Green Bank Observatory and ies (e.g., promising massive elliptical galaxies, or binary the Arecibo Observatory. The Green Bank Observatory candidates identified in electromagnetic searches). This is a facility of the National Science Foundation oper- is significant, since proximity to a pulsar with high qual- ated under cooperative agreement by Associated Uni- ity data has the potential to improve the mass ratio up- versities, Inc. The Arecibo Observatory is a facility of per limits. All of above will lead to an overall better the National Science Foundation operated under coop- understanding on the existence of SMBHBs in the local erative agreement by the University of Central Florida universe and beyond. in alliance with Yang Enterprises, Inc. and Universidad Metropolitana. limits on smbhbs within 500 mpc 17

Software: enterprise (Ellis et al. 2017), PTMCMCSampler (Ellis & van Haasteren 2017)

APPENDIX

lated the mass of the central SMBH and assigned it to the total mass of the binary. Since we primarily used the global scaling relationship MBH-Mbulge, it is neces- sary to identify the morphological type of the galaxy, because the fraction of the stars in the bulge depends on this. However, the galaxy classification is unknown for about 40% of the galaxies in 2MRS. For those we ex- amined two scenarios: (1) the galaxies are early type, so fbulge = 1 and (2) they are late-type with fbulge = 0.31. We found that 56 galaxies of unknown type would be detectable by NANOGrav under the assumption they are early-type galaxies. Here we present the results for this sub-set of galaxies. In Fig. 11, we show the distribution of these 56 galax- Figure 11. Spatial distribution of 56 galaxies of unknown ies on the sky. We see that these galaxies are dis- type that could be within the NANOGrav sensitivity volume tributed relatively close to the , where if they are elliptical galaxies. Colors and marker sizes similar NANOGrav’s sensitivity is maximum and typically are to Fig.1. at large distances. This could explain the scarcity of high quality images that would allow their morphologi- In § 2.3, we examined whether a galaxy lies in the cal classification. sensitivity volume of NANOGrav. For this, we calcu-

REFERENCES

Aggarwal, K., Arzoumanian, Z., Baker, P. T., et al. 2019, Charisi, M., Bartos, I., Haiman, Z., et al. 2016, MNRAS, ApJ, 880, 116, doi: 10.3847/1538-4357/ab2236 463, 2145, doi: 10.1093/mnras/stw1838 Armitage, P. J., & Natarajan, P. 2005, ApJ, 634, 921, Colpi, M. 2014, SSRv, 183, 189, doi: 10.1086/497108 doi: 10.1007/s11214-014-0067-1 Arzoumanian, Z., Brazier, A., Burke-Spolaor, S., et al. Crook, A. C., Huchra, J. P., Martimbeau, N., et al. 2007, 2018a, The Astrophysical Journal Supplement Series, ApJ, 655, 790, doi: 10.1086/510201 235, 37, doi: 10.3847/1538-4365/aab5b0 Davis, T. A., Bureau, M., Cappellari, M., Sarzi, M., & Arzoumanian, Z., Baker, P. T., Brazier, A., et al. 2018b, Blitz, L. 2013, Nature, 494, 328, ApJ, 859, 47, doi: 10.3847/1538-4357/aabd3b doi: 10.1038/nature11819 —. 2020, arXiv e-prints, arXiv:2005.07123. De Rosa, A., Vignali, C., Bogdanovi´c,T., et al. 2019, https://arxiv.org/abs/2005.07123 NewAR, 86, 101525, doi: 10.1016/j.newar.2020.101525 Babak, S., Petiteau, A., Sesana, A., et al. 2016, MNRAS, Desvignes, G., Caballero, R. N., Lentati, L., et al. 2016, 455, 1665, doi: 10.1093/mnras/stv2092 MNRAS, 458, 3341, doi: 10.1093/mnras/stw483 Barvainis, R. 1987, ApJ, 320, 537, doi: 10.1086/165571 Detweiler, S. 1979, ApJ, 234, 1100, doi: 10.1086/157593 Begelman, M. C., Blandford, R. D., & Rees, M. J. 1980, D’Orazio, D. J., & Loeb, A. 2018, ApJ, 863, 185, Nature, 287, 307, doi: 10.1038/287307a0 doi: 10.3847/1538-4357/aad413 Bentz, M. C., & Katz, S. 2015, PASP, 127, 67, Ellis, J., & van Haasteren, R. 2017, doi: 10.1086/679601 jellis18/PTMCMCSampler: Official Release, Burke-Spolaor, S., Taylor, S. R., Charisi, M., et al. 2019, doi: 10.5281/zenodo.1037579 A&A Rv, 27, 5, doi: 10.1007/s00159-019-0115-7 Ellis, J. A., Vallisneri, M., Taylor, S. R., & Baker, P. T. Cappellari, M. 2013, ApJL, 778, L2, 2017, https://github.com/nanograv/enterprise: doi: 10.1088/2041-8205/778/1/L2 enterprise. https://github.com/nanograv/enterprise 18 The NANOGrav Collaboration

Farris, B. D., Duffell, P., MacFadyen, A. I., & Haiman, Z. Ransom, S., Brazier, A., Chatterjee, S., et al. 2019, in 2014, ApJ, 783, 134, doi: 10.1088/0004-637X/783/2/134 Bulletin of the American Astronomical Society, Vol. 51, Finn, L. S., & Thorne, K. S. 2000, PhRvD, 62, 124021, 195. https://arxiv.org/abs/1908.05356 doi: 10.1103/PhysRevD.62.124021 Rosado, P. A., Sesana, A., & Gair, J. 2015, MNRAS, 451, Fitzpatrick, E. L. 1999, PASP, 111, 63, doi: 10.1086/316293 2417, doi: 10.1093/mnras/stv1098 Graham, M. J., Djorgovski, S. G., Stern, D., et al. 2015, Sandrinelli, A., Covino, S., Treves, A., et al. 2018, A&A, MNRAS, 453, 1562, doi: 10.1093/mnras/stv1726 615, A118, doi: 10.1051/0004-6361/201732550 Haehnelt, M. G., & Kauffmann, G. 2002, MNRAS, 336, Schutz, K., & Ma, C.-P. 2016, MNRAS, 459, 1737, L61, doi: 10.1046/j.1365-8711.2002.06056.x doi: 10.1093/mnras/stw768 Hellings, R. W., & Downs, G. S. 1983, ApJL, 265, L39, Sesana, A., Haiman, Z., Kocsis, B., & Kelley, L. Z. 2018, doi: 10.1086/183954 ApJ, 856, 42, doi: 10.3847/1538-4357/aaad0f Holgado, A. M., Sesana, A., Sandrinelli, A., et al. 2018, Seth, A. C., van den Bosch, R., Mieske, S., et al. 2014, MNRAS, 481, L74, doi: 10.1093/mnrasl/sly158 Nature, 513, 398, doi: 10.1038/nature13762 Huchra, J. P., Macri, L. M., Masters, K. L., et al. 2012, Shankar, F., Bernardi, M., Sheth, R. K., et al. 2016, ApJS, 199, 26, doi: 10.1088/0067-0049/199/2/26 MNRAS, 460, 3119, doi: 10.1093/mnras/stw678 Jenet, F. A., Lommen, A., Larson, S. L., & Wen, L. 2004, Siwek, M. S., Kelley, L. Z., & Hernquist, L. 2020, arXiv ApJ, 606, 799, doi: 10.1086/383020 e-prints, arXiv:2005.09010. Kelley, L. Z., Blecha, L., & Hernquist, L. 2017, MNRAS, https://arxiv.org/abs/2005.09010 464, 3131, doi: 10.1093/mnras/stw2452 Sudou, H., Iguchi, S., Murata, Y., & Taniguchi, Y. 2003, Kelley, L. Z., Blecha, L., Hernquist, L., Sesana, A., & Science, 300, 1263, doi: 10.1126/science.1082817 Taylor, S. R. 2018, MNRAS, 477, 964, Taylor, S., Burke-Spolaor, S., Baker, P. T., et al. 2019, doi: 10.1093/mnras/sty689 BAAS, 51, 336. https://arxiv.org/abs/1903.08183 Kerr, M., Reardon, D. J., Hobbs, G., et al. 2020, arXiv Taylor, S. R., Vallisneri, M., Ellis, J. A., et al. 2016, ApJL, e-prints, arXiv:2003.09780. 819, L6, doi: 10.3847/2041-8205/819/1/L6 https://arxiv.org/abs/2003.09780 Thomas, J., Ma, C.-P., McConnell, N. J., et al. 2016, Kollmeier, J. A., Zasowski, G., Rix, H.-W., et al. 2017, Nature, 532, 340, doi: 10.1038/nature17197 arXiv e-prints, arXiv:1711.03234. Tully, R. B., Courtois, H. M., & Sorce, J. G. 2016, AJ, 152, https://arxiv.org/abs/1711.03234 50, doi: 10.3847/0004-6256/152/2/50 Kormendy, J., & Ho, L. C. 2013, ARA&A, 51, 511, Tully, R. B., Rizzi, L., Shaya, E. J., et al. 2009, AJ, 138, doi: 10.1146/annurev-astro-082708-101811 323, doi: 10.1088/0004-6256/138/2/323 Ma, C.-P., Greene, J. E., McConnell, N., et al. 2014, ApJ, Volonteri, M., Haardt, F., & Madau, P. 2003, ApJ, 582, 795, 158, doi: 10.1088/0004-637X/795/2/158 559, doi: 10.1086/344675 McConnell, N. J., & Ma, C.-P. 2013, ApJ, 764, 184, Walsh, J. L., van den Bosch, R. C. E., Gebhardt, K., et al. doi: 10.1088/0004-637X/764/2/184 2015, ApJ, 808, 183, doi: 10.1088/0004-637X/808/2/183 Mingarelli, C. M. F., Lazio, T. J. W., Sesana, A., et al. —. 2016, ApJ, 817, 2, doi: 10.3847/0004-637X/817/1/2 2017, Nature Astronomy, 1, 886, Weinzirl, T., Jogee, S., Khochfar, S., Burkert, A., & doi: 10.1038/s41550-017-0299-6 Kormendy, J. 2009, ApJ, 696, 411, Mould, J. R., Huchra, J. P., Freedman, W. L., et al. 2000, doi: 10.1088/0004-637X/696/1/411 ApJ, 529, 786, doi: 10.1086/308304 Zhu, X.-J., Hobbs, G., Wen, L., et al. 2014, MNRAS, 444, Perera, B. B. P., DeCesar, M. E., Demorest, P. B., et al. 3709, doi: 10.1093/mnras/stu1717 2019, MNRAS, 490, 4666, doi: 10.1093/mnras/stz2857