A pair of UV nuclei or a compact star forming region near the active nucleus in Mrk~766?
JJ. Astrophys. Astr. (0000) :
A pair of UV nuclei or a compact star forming region near the activenucleus in Mrk 766?
P. P. Deka , G. C. Dewangan , K. P. Singh and J. Postma Inter-University Centre for Astronomy and Astrophysics (IUCAA), SPPU Campus, Pune, India, Indian Institute of Science Education and Research Mohali, Knowledge City, Sector 81, Manauli P.O., SAS Nagar,140306, Punjab, India, Department of Physics & Astronomy, University of Calgary, 2500 University Dr. NW, Calgary, AB T2N 1N4,Canada. * Corresponding author. E-mail: [email protected] received 31 October 2020; accepted 16 Dec 2020
Abstract.
We report the discovery of a bright, compact ultraviolet source at a projected separation of 1.1 kpcfrom the known active galactic nucleus (AGN) in Mrk 766 based on
AstroSat / UVIT observations. We performradial profile analysis and derive the UV flux almost free from the nearby contaminating sources. The new sourceis about 2.5 and 5.6 times fainter than the AGN in the far and near UV bands. The two sources appear as a pairof nuclei in Mrk 766. We investigate the nature of the new source based on the UV flux ratio, X-ray and opticalemission. The new source is highly unlikely to be another accreting supermassive black hole in Mrk 766 as it lacksX-ray emission. We find that the UV / Optical flux of the new source measured at four di ff erent bands closely followthe shape of the template spectrum of starburst galaxies. This strongly suggests that the new source is a compactstar forming region. Keywords. galaxies—active galactic nuclei—star formation.
1. Introduction
Presence of dual or multiple compact and luminoussources in the central regions of galaxies are rare.Mergers of galaxies can naturally lead to the formationof dual and / or multiple compact sources such as thoseobserved in Ultraluminous infrared galaxies. Simula-tions of mergers of gas-rich disk galaxies show thatvery massive, compact and highly luminous star clus-ters can form from the strongly disturbed gas disks.Consist of young stars, these clusters appear as severalbright cores in the central kilo-parsec region of galaxies(Matsui et al ., 2012).Another popular interpretation of dual compactsources is the presence of double accreting supermas-sive black holes in the central regions. It is now widelyaccepted that almost every galaxy has a supermassiveblack hole (SMBH) at its center whose mass can beestimated through various techniques like central stel-lar velocity dispersion observations, reverberation map-ping observations, etc. (Schneider, 2006). During themerging of galaxies, their respective SMBHs are alsoexpected to come closer due to gravitational attractionand finally coalesce. The whole process can be classi-fied into three phases (Merritt & Milosavljevic, 2005): 1) After the two galaxies merge, the two SMBHs movetowards the center of the newly formed galaxy andform a binary pair by losing their angular momentumthrough dynamical friction. 2) The orbit is further hard-ened by slingshot ejection of stars whose orbits crossesthe binary by the process of three body interaction andthereby moving the two SMBHs closer to each other. 3)When the separation between the two SMBHs is smallenough such that emission of gravitational waves candominate the other forms of energy loss, the two blackholes merge. There are many key questions to answer,e.g., till when the two SMBHs retain their individualaccretion disks and when they start sharing a commonaccretion disk, at what rate the accretion takes place andthe rate of growth of the individual SMBHs etc. Also,after merging, the anisotropically emitted gravitationalwave gives a kick velocity to the final merged blackhole due to which it gets ejected from the center (Ko-mossa, 2012). For su ffi cient kick velocities, the mergedblack hole can even get completely ejected from thehost galaxy, though its probability is very small (Ko-mossa, 2012).Observing multiple luminous, compact sources inthe nuclear regions of galaxies and finding their natureis crucial to understand galaxy evolution and mergers © Indian Academy of Sciences 1 a r X i v : . [ a s t r o - ph . GA ] J a n J. Astrophys. Astr. (0000) : of supermassive black holes. The exceptional high spa-tial resolution of the
Chandra
X-ray telescope and theHubble Space Telescope (
HST ) have led to the first dis-covery of dual nuclei in galaxies (Komossa et al ., 2002;Junkkarinen et al ., 2001; Ballo et al ., 2004). A numberof other techniques have also been used (see Komossa& Zensus (2014) for a review). Here we used high res-olution of
AstroSat / UVIT and discovered a compact,bright UV source near the well known active nucleusin Mrk 766. In Section 2., we present
AstroSat obser-vations and the reduction of UVIT data. In Section 3.,we perform spatial analysis of UVIT images and inves-tigate the nature of the new UV source in Section 4, andsummarize our findings in Section 5. AstroSat / UVIT observations and the data reduc-tion
We observed Mrk 766 with
AstroSat as a part of theSXT Guaranteed programme during 4-6 February 2017with the SXT as the primary instrument for an exposuretime of 50 ks. Here we present the UVIT data only.We used the broadband filters FUV / BaF2 (F154W)and NUV / Silica (N242W) and acquired photon count-ing data. We obtained the Level1 data from the
As-troSat data archive and we processed them using theUVIT pipeline CCDLAB (Postma & Leahy, 2017). Wegenerated cleaned images for each orbit, aligned themand created merged image for each filter. This re-sulted in net exposure time of 33.4 ks (NUV / Silica) and27 ks (FUV / BaF2). We derived the astrometric solutiontransforming the image coordinates to the world coor-dinates using the astrometry.net package (Lang et al .,2010). We show the NUV and FUV images of Mrk 766in Figure 1. In order to show the relative intensities ofthe two UV sources, we created 2D histogram views ofthe NUV image in two di ff erent scales, namely linearand square root. We show the histograms in Figure 2.To identify the position of known active galactic nu-cleus in Mrk 766 and to look for X-ray counter part ofthe new UV source, we generated a composite three-color image using FUV / BaF2, NUV / Silica and
Chan-dra
X-ray images. We obtained the processed andcleaned
Chandra / ACIS event data (obsID:1597) fromthe HEASARC archive . The composite three-colorimage of Mrk 766 is shown in Figure 1. Clearly the cen-tral bright UV source with strong X-ray emission is theknown AGN. We refer the AGN as the primary source https://astrobrowse.issdc.gov.in/astro_archive/archive/Home.jsp https://heasarc.gsfc.nasa.gov/cgi-bin/W3Browse/w3browse.pl and the nearby UV source as the secondary source. Wedid not find significant X-ray emission at the positionof the secondary UV source.We measured the positions of the primary source(AGN) and the secondary source in both the FUV andNUV images using the centroiding algorithm availablewith the SAOImage / DS9 tool. We list the positions inTable 1.
3. Spatial analysis and measurement of UV flux
The separation between the AGN and secondary sourceis 4 .
211 arcsec. Though the peak positions of the twosources are well separated, the wings of the two PSFsoverlap. Moreover, the two sources are located in thecentral regions where the di ff use emission from thegalaxy is strong. In addition, there are other galaxy fea-tures such as the bar and star forming clumps possiblyassociated with spiral arms in the central regions. Thesmall separation and a number of features complicatemeasurement of flux from the two sources. Here we usea simplistic approach to separate the emission from thetwo sources. We extract radial profiles and fit with thePSFs of point sources and profile of the di ff use emis-sion and background level. Given the complexity ofthe spatial structure, our method will be approximate.Unlike optical images, UV images of galaxies are notsmooth due to the presence of star forming clumps andpossible non-uniformity of internal reddening and eventhe 2D profile fitting using tools such as the GALFITmay not yield accurate results. We postpone such a de-tailed analysis to a future paper.Our analysis consists of the following steps in se-quence. First we extract radial profiles centered on oneof the two sources in each image. Thus we generatefour radial profiles for the two sources in the NUV andFUV images. We then fit the radial profiles to obtainthe count rates of the AGN and the secondary source inthe NUV and FUV bands. If we take the ratio of theNUV and FUV count rates for each source, this willrepresent the slope of the spectrum of the source. Inthe final step, we compare these ratios with the slopesof spectrum of known sources such as quasars and star-burst galaxies. Below, we describe our analysis in thesesteps.3.1 Radial profile fitting and count-rate ratio analysis
The radial profiles in the FUV and NUV bands cen-tered on each of the two sources were derived using theimage display and astronomical data visualization toolSAOImage / DS9. In each case, the source at the centerof the radial profile was fitted with a 1D Mo ff at function(PSF model for UVIT) and the o ff -centered source was . Astrophys. Astr. (0000) : Figure 1 . Composite three color NUV / Silica (red), FUV / BaF2 (green) and Chandra / HETG X-ray (blue) image ofMrk 766.
Table 1.
Positions of the primary and secondary UV sources
Source FUV NUV α (J2000) δ (J2000) α (J2000) δ (J2000)AGN 12h18m26.5s + + + + J. Astrophys. Astr. (0000) :
Figure 2 . Surface plot of NUV emission from Mrk 766 on linear (left panel) and square root (right panel) scales showingcomplex spatial structure. fitted with a 1D Gaussian. The contribution from thegalaxy in the form of di ff use emission was fitted withan 1D exponential function and finally the overall con-stant background was fitted with a constant 1D func-tion. The fitting was performed using Sherpa whichis inside Chandra’s data reduction and fitting package
CIAO . Before going into the results of the fitting pro-cess, we would like to state here that due to variouscomponents present and resolved in the UVIT imagesof the galaxy (e.g., the bar in the central region, the ex-tended spiral arm etc.), our simple models for fittingthe radial profiles didn’t prove to be su ffi cient. Conse-quently, to get acceptable values of the fit statistics, wehad to add systematic errors to our data, which resultedin increased error bars in the best-fit parameters. Belowwe give the form of the di ff erent profile functions usedto fit the components of the radial profile.Mo ff at: m ( x ) = A (cid:104) + (cid:16) x − x γ (cid:17) (cid:105) − β Gaussian: g ( x ) = A exp (cid:104) − x − x ) FWHM (cid:105) Exponential: e ( x ) = A exp(coe ff ( x − x ))Constant: c ( x ) = c (1)In what follows, m1 would indicate a Mo ff at function,g1 or g2 would indicate Gaussian profiles, e1 would indicate an exponential profile and c1 would indicate aconstant background profile.3.1.1 FUV radial profile analysis for the primarysource
For fitting the primary source, we fixed the β and γ parameters of the Mo ff at function (m1) by fit-ting a radial profile extracted from a field star (PSFmodelling). The remaining features were fitted withcomponents as described above. Table 2 lists the fittedparameters along with their 89.041% confidence inter-vals. Also, we had to add 3.5% systematic error to geta good χ / dof = /
20. Figure 3 shows the modelfitted data along with its residual. We integrated thefitted Mo ff at function from the position of the peak ofthe Mo ff at function to a radius of 25 pixels ( ∼
10 arc-sec) which encompasses greater than 95% of the energy(Tandon et al ., 2020) and divide by the exposure timeto get the number of counts per second (CPS) from theprimary (see Table 2).3.1.2
FUV radial profile analysis for the secondarysource
Again for fitting the secondary source, wefixed the β and γ parameters of the Mo ff at function(m1) from the stellar profile. Table 3 shows the val-ues of the fitted parameters along with their 89.041%confidence intervals. We had to add 7% systematic er-ror to get a reasonable value of reduced χ = . . Astrophys. Astr. (0000) : Table 2 . Best-fit parameters from the FUV radial profileanalysis for the primary sourceParam Type Best Valuem1.A thawed 187.5 ± x frozen 0.6m1. γ frozen 2.6m1. β frozen 1.9c1. c thawed 2.7 ± x thawed 10.1 ± ± ± x thawed 0.6 ± ± ff thawed − . ± . ± + x linked (m1. x ) 0.6CPS 0 . ± . Figure 3 . FUV radial profile of the primary source, thebest-fitting model and residuals. the residuals. We integrated the fitted Mo ff at functionfrom the position of the peak of the Mo ff at function to aradius of 25 pixels and divide by the exposure time andobtained the count rate of 0 . ± .
02 counts s − for thesecondary.3.1.3 NUV radial profile analysis for the primarysource
Similar procedures as described above forthe FUV band were followed for fitting the NUVradial profiles for the primary as well as the sec-ondary. Table 4 lists the best-fit parameters alongwith their 89.041% confidence intervals. We had toadd 2% systematic error to get a reasonable value of χ / dof = /
24. In this case, we didn’t freeze the Mof-fat parameters as the bright AGN was good enough toestimate the γ and β parameters. Figure 5 shows the Table 3 . Best-fit parameters from the FUV radial profileanalysis for the secondary sourceParam Type Best Valuem1.A thawed 72.6 ± x frozen 0.6m1. γ frozen 2.6m1. β frozen 1.9c1. c thawed 0 ± x thawed 10.0 ± ± ± x linked (g1. x ) 10.0e1.A thawed 20.9 ± ff thawed − . ± . ± . Figure 4 . FUV radial profile of the secondary source, thebest-fitting model and the residuals. model fitted data along with its residual. We derivedan NUV count rate of 3 . ± . NUV radial profile analysis for the secondarysource
In this case, we had to fit an additional Gaus-sian (g2) to account for the AGN. Table 5 below showsthe values of the fitted parameters along with their89.041% confidence intervals. Again, we had to add2% systematic error to get a reasonable value of re-duced χ = .
09. In this case, Mo ff at parameters werefixed based on the stellar radial profile analysis. Asbefore, we derived an NUV count rate of 0 . ± . − for the secondary source. Figure 6 showsthe model fitted data along with its residual.Table 6 summarizes the results obtained from radialfitting. J. Astrophys. Astr. (0000) :
Table 4 . Best-fit parameters from the NUV radial profileanalysis for the primary sourceParam Type Best Valuem1.A thawed 5353.3 ± x thawed 0.5 ± γ thawed 2.6 ± β thawed 2.2 ± c thawed 44.3 ± ± ± x thawed 10.2 ± x linked (m1. x ) 0.5e1.A thawed 1384.7 ± ff thawed − . ± . ± . Figure 5 . NUV radial profile of the primary source, thebest-fitting model and the residuals.
Count-rate ratio analysis
The nature of any source can be inferred from its spec-trum. In the absence of spectrum, the colors of an ob-ject become useful. We calculate the ratio of count ratesin the NUV / Silica and FUV / BaF2 bands for the AGNand the secondary source. These ratios are equivalentto colours. Generally two colors are used to classify orinfer the type of an object. In addition to the UV col-ors, we use X-ray / optical flux. In Table 7, we list theUV colours of the two sources. As described earlier,we will compare the ratios listed in Table 7 with thecorresponding ratios obtained from standard spectra ofquasars and starburst galaxies. Table 5 . Best-fit parameters from the NUV radial profileanalysis for the secondary sourceParam Type Best Valuem1.A thawed 1145.3 ± x frozen 0.6m1. γ frozen 1.7m1. β frozen 1.6c1. c thawed 21.7 ± ± ± x thawed 10.0 ± x thawed 11.2 ± ± ff thawed − . ± ± ± x thawed 11.4 ± ± Figure 6 . NUV radial profile of the secondary source, thebest-fitting model and the residuals.
4. Nature of the secondary source
Comparison with composite quasar spectrum
We used the composite quasar spectrum derived byVanden Berk et al . (2001) using SDSS spectra of over2200 quasars in the redshift range from 0.044 to 4.789.Since this spectrum has practically zero extinction, butour observations were reddened by both the Galacticextinction and internal reddening in Mrk 766, we needto redden the composite quasar spectrum before calcu-lating the count rates in NUV and FUV.The ’non-standard’ extinction caused by the AGNenvironment of Mrk 766 was applied with the help ofempirical formula given in Czerny et al . (2004). Weused a color excess of E(B-V) = α to H β . Astrophys. Astr. (0000) : Table 6.
Results from Radial fitting Source Systematic error χ / dof CPSPrimary FUV 3 .
5% 1.0589 0.23 ± .
0% 1.11381 0.09 ± .
0% 1.07013 3.9 ± .
0% 1.0872 0.7 ± Table 7.
NUV-to-FUV count ratios
Source Observed count-rate ratio Internal E(B-V) Predicted ratioPrimary 17 ± ± z = . et al ., 1989). We assume this reddened and redshiftedquasar composite spectrum as our model spectrum foractive nucleus in Mrk 766. We calculated the predictedUVIT count rates for our model spectrum using the ef-fective area of a filter using the following equation: CPS = (cid:90) f λ ( hc /λ ) A e f f ( λ ) d λ (2)Using the e ff ective areas for the FUV / BaF2 andNUV / Silica filters, we calculated the predicted count-rate ratio for the composite quasar spectrum to be 18 . ff erent than the ob-served ratio for the secondary source. The internal ex-tinction with E(B-V) = .
36 derived from the Balmerdecrement is appropriate for the AGN in Mrk 766. Itis possible that the secondary source su ff ers with a dif-ferent internal extinction. If we do not redden the com-posite quasar spectrum with the internal extinction, wepredict a count-rate ratio of 7 .
8. We list the predictedcount-rate ratios in Table 7.From Table 7, we find that the count-rate ratio ofthe primary source is within the predicted ratio for thecomposite quasar spectrum. Thus, our analysis impliesthat the primary source is indeed an AGN which in turnverifies the correctness of our methodology. The ob-served count-rate ratio of the secondary source devi-ates significantly from the expected value for an AGNwith similar internal reddening as the primary source.But interestingly, the observed ratio for the secondarysource matches well with the AGN ratio if there is nointernal reddening. Thus, our analysis clearly rules out the secondary source to be a background AGN or anaccreting SMBH embedded in the galaxy Mrk 766.4.2
Estimation of X-ray flux and detectability withChandra
From our analysis in previous sections, it is clear thatthe count-rate ratio of the secondary source is consis-tent with the expected ratio for an unabsorbed AGN.This possibility can be tested by estimating the ex-pected count rate in the X-ray band and comparing itwith the upper limit from the
Chandra data. In orderto predict the expected X-ray flux, we use the optical toX-ray flux ratio α ox which is the ratio of flux densitiesat 2500Å and 2 keV. We first calculate the 2500Å fluxdensity using the observed count rates in the FUV andNUV bands.We converted the FUV and NUV count rates to fluxdensities at the mean wavelengths of the filter bandpassusing the relation (Tandon et al ., 2017): f λ ( ergs cm − s − Å − ) = CPS × UC (3)The unit conversion factor (UC) was derived from thezero point magnitude (ZP) given in Tandon et al . (2020)and using the relation (Tandon et al ., 2017):ZP = − . ( UC × λ m ) − .
407 (4)Where, λ m is the mean wavelength of the filter in Å.With λ m = ZP = . UC = . × − ( ergs s − cm − Å − ) / ( counts s − )(5)The observed count rate of 0 . − for the sec-ondary source is converted to f λ (2418Å) = . × − ergs cm − s − Å − .Since we have the observed flux at 2418 Å, we de-reddened it first from the extinction A λ at 2418 Å due J. Astrophys. Astr. (0000) : to Milky Way using CCM89. Then we simply trans-ferred the wavelength and the corresponding flux to therest frame of Mrk 766 by multiplying the flux by (1 + z)(z = + z)which gave us λ = f λ (2500 Å) = . × − ergs / cm / s / Å.The optical to X-ray index is given by (Sobolewska et al ., 2009), α ox = − log [( f keV / f )]2 .
605 (6)where L keV and L are in ergs cm − s − Hz − .We converted f λ (2500Å) to f ν (2500Å) and using α ox = .
37 (Lusso, E. et al ., 2010) to find f ν (2 keV) = . × − ergs cm − s − Hz − or f E = . × − photons / cm / s / keV. Using a power law model with X-ray photon index Γ = . N H = . × cm − along the line of sightto Mrk 766, we calculated 0 . −
10 keV band X-rayflux, f X = . × − ergs cm − s − . We converted thisflux to Chandra / HETG count rate of 0 . − using the WebPIMMS tool. This is the expected countrate if the secondary source were an unabsorbed AGN.We used the Chandra observation (ObsID:1597)with an exposure time of 89 ks. There is no X-raysource at the location of the secondary source. Wecalculated the 4 σ upper limit of 128 counts or 0.0014counts s − , which is less than the predicted count rate.Thus, due to the lack of X-ray emission, the secondaryis highly unlikely to be another accreting SMBH inMrk 766.Another possibility is that the secondary sourcecould be a compact region of enhanced star formation.We suspected this possibility based on the ratio imagewhere the secondary source does not stand out. We cre-ated a ratio image, shown in Figure 7, by dividing eachpixel value in the NUV image by the correspondingpixel value in the FUV image after scaling the NUV im-age to have the same exposure time as the FUV image.We find similar ratios at the position of the secondarysource as in other parts of the galaxy except at the lo-cation of the primary. This suggests that the emissionprocess responsible for the secondary source is likelysimilar to that of the di ff use emission from the otherparts of the galaxy which is likely from a population ofyoung, massive stars resulting from star formation. Figure 7 . Ratio image obtained by dividing the NUV imageby the FUV image after the NUV image was scaled to havethe same exposure time as that of the FUV image.
Table 8 . Ratio obtained from UV templates of starburstgalaxiesTemplate E(B-V) RatioStarburst 1 E ( B − V ) < . Starburst 2 0 . < E ( B − V ) < . Starburst 3 0 . < E ( B − V ) < . Starburst 4 0 . < E ( B − V ) < . Comparison with the spectra of starburst galaxies
To investigate further if the secondary source is actuallyan enhanced star-forming region, we estimated typicalNUV-to-FUV count-rate ratios for starburst galaxies.We obtained the template spectra of starburst galaxiesfrom Kinney et al . (1996) derived for di ff erent valuesof the internal extinction, E(B-V). These spectra are al-ready corrected for Galactic extinction. We calculatedthe count-rate ratios for the template starburst spectrausing Eqn. 2 by using the flux densities at the meanwavelengths of our FUV, NUV filters. We did not ap-ply any additional internal reddening. The predictedratios are listed in Table 8. We see that the observedcount-rate ratio for the secondary source is very similarto that derived for the starburst templates with internalextinction E ( B − V ) < .
21. If the secondary source isindeed a compact star forming region with optical / UVspectrum similar to the starburst templates, we expectsignificant optical emission.4.4
HST images and calculation of flux
We searched for the counterpart of the secondarysource in
HST images. We detected multiple com-pact sources near the position of the secondary sourcein the
HST images of Mrk 766 acquired with di ff er-ent instrument and filter combinations. Because ofthe excellent PSF of the HST , the secondary source . Astrophys. Astr. (0000) :
Figure 8 . Spectra of starburst galaxies scaled to match theflux density measured with the FUV / F154W filter at themean wavelength 1541Å. Filled circles show flux densitiesmeasured with the NUV / N242W filter ( λ mean = / ACS F330W filter ( λ central = . / WFC3F547M filter ( λ central = that appeared as a point source in UVIT images, nowappeared as three distinguishable point sources. Fig-ure 9 shows the secondary source marked with cir-cles in the HST image in the F330W filter. We in-cluded the three sources and performed aperture pho-tometry and obtained the count rates for the secondarysource in the
HST / ACS F330W filter (central wave-length = . HST / WFC3 F547M filter (cen-tral wavelength = . HST images.We found f λ (3363Å) = . × − ergs cm − s − forF330W filter and f λ (5475Å) = . × − ergs cm − s − for the F547M filter.In order to compare these flux densities measuredwith the HST and that expected from the secondarysource assuming it to be a compact star forming region,we re-scaled the starburst template spectra to the mea-sured flux density at 2418 Å (shown in section 4.2) withthe UVIT / NUV. We then compared the UV and opti-
Figure 9 . Selected regions for aperture photometry on thesecondary source. cal flux densities with the re-scaled starburst templatespectra in Figure 8. We find that the measured flux atfour di ff erent wavelengths follow the shape of the star-burst template spectra. This clearly suggests that thesecondary source is indeed a compact star forming re-gion.
5. Conclusion
We identified a bright far and near UV source at a pro-jected distance of ∼ . Chandra
X-ray observation, and
HST imagesin the near UV and optical bands. The lack of X-rayemission in the
Chandra image at the location of thesecondary source makes it highly unlikely to be an ac-creting SMBH. Further, the UV / optical flux measuredat four di ff erent bands closely follow the shape of thestarburst template spectra. Therefore we conclude thatthe secondary is most likely a compact star forming re-gion. Acknowledgements
This publication uses the UVIT data from the As-troSat mission of the Indian Space Research Organ-isation (ISRO), archived at the Indian Space ScienceData Centre (ISSDC). The UVIT data are processedby the payload operations centers at IIA, Bangalore.This research has made use of UVIT pipeline (CCD-LAB) developed at University of Calgary for UVIT de-velopment and science support. The scientific resultsreported in this article are based in part on observa-tions made by the Chandra X-ray Observatory, data ob-tained from the Chandra Data Archive. This research isbased on observations made with the NASA / ESA Hub-
J. Astrophys. Astr. (0000) : ble Space Telescope obtained from the Space TelescopeScience Institute, which is operated by the Associationof Universities for Research in Astronomy, Inc., underNASA contract NAS 5–26555. This research has madeuse of the python and julia packages. This research hasmade use of the SIMBAD database, operated at CDS,Strasbourg, France.
References
Ballo, L., Braito, V., Della Ceca, R., et al . 2004, TheAstrophysical Journal, 600, 634Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989,The Astrophysical Journal, 345, 245Czerny, B., Li, J., Loska, Z., & Szczerba, R. 2004,Monthly Notices of the Royal Astronomical Society,348, L54Gonzalez Delgado, R. M., & Perez, E. 1996, MonthlyNotices of the Royal Astronomical Society, 278, 737Junkkarinen, V., Shields, G. A., Beaver, E. A., et al .2001, The Astrophysical Journal, 549, L155Kinney, A. L., Calzetti, D., Bohlin, R. C., et al . 1996,The Astrophysical Journal, 467, 38Komossa, S. 2012, Advances in Astronomy, 2012,doi:10.1155 / / et al . 2002,The Astrophysical Journal, 582, L15Komossa, S., & Zensus, J. A. 2014, Proceedings of theInternational Astronomical Union, 10, 13–25Lang, D., Hogg, D. W., Mierle, K., Blanton, M., &Roweis, S. 2010, The Astronomical Journal, 139,1782Lusso, E., Comastri, A., Vignali, C., et al . 2010, A&A,512, A34Matsui, H., Saitoh, T. R., Makino, J., et al . 2012, TheAstrophysical Journal, 746, 26Merritt, D., & Milosavljevic, M. 2005, Living Reviewsin Relativity, 8, doi:10.12942 / lrr-2005-8Postma, J. E., & Leahy, D. 2017, Publications of theAstronomical Society of the Pacific, 129, 115002Schneider, P. 2006, Extragalactic Astronomy and Cos-mology, An Introduction, doi:10.1007 / et al . 2017,The Astronomical Journal, 154, 128Tandon, S. N., Postma, J., Joseph, P., et al . 2020, TheAstronomical Journal, 159, 158Vanden Berk, D. E., Richards, G. T., Bauer, A., et alet al