A Catalog of Post-starburst Quasars from Sloan Digital Sky Survey Data Release 7
aa r X i v : . [ a s t r o - ph . GA ] M a r D RAFT VERSION F EB
28, 2018.
Preprint typeset using L A TEX style AASTeX6 v. 1.0
A CATALOG OF POST-STARBURST QUASARS FROM SLOAN DIGITAL SKY SURVEY DATA RELEASE 7 P ENG W EI ( 魏 鹏 ) , Y ANG G U ( 顾 洋 ) , M ICHAEL
S. B
ROTHERTON , Y ONG S HI AND Y ANMEI C HEN School of Astronomy and Space Science, Nanjing University, Nanjing 210023, China Nanjing Foreign Language School, Xianlin Campus, Nanjing 210023, China Department of Physics and Astronomy, University of Wyoming, Laramie, WY 82071, USA Key Laboratory of Modern Astronomy and Astrophysics (Nanjing University), Ministry of Education, Nanjing 210023, China
ABSTRACTWe present a catalog of nearby (z ≤ δ absorption ( EW ≥ )indicating a significant contribution of an intermediate-aged stellar population formed in a burst of star forma-tion within the past 1 Gyr, which makes them potentially useful for studying the co-evolution of supermassiveblack holes and their host galaxies. Keywords: galaxies: active – galaxies: interactions – galaxies: starburst – quasars: general INTRODUCTIONSupermassive black holes (SMBHs) exist at the centersof essentially all massive galaxies (Kormendy & Richstone1995). Quasars, the most luminous Active Galactic Nuclei(AGNs), are powered by the accretion onto SMBHs. Theyare not only an important phase of SMBH growth, but alsorepresent a key stage in the life cycle of massive galaxies(Heckman & Best 2014). AGN feedback can terminate starformation in the host galaxy and mass accretion onto theSMBH (Fabian 2012, and references therein).Theoretical studies have suggested two mechanisms re-sponsible for triggering starbursts and AGNs. First, inthe early universe, major-mergers ignite the the most lu-minous quasars and starbursts(e.g., Bouwens et al. 2009;Treister et al. 2012). Second, in recent epochs, the main fuel-ing mechanisms may be entirely driven by secular processes(e.g., Hopkins & Hernquist 2009).Recently, several studies have indicated that AGN and star-burst activity may not be coeval. For example, Davies (2007)studied a sample of local AGN with IFU data, and found thatstrong AGNs are not present when the central stellar popula-tions are of the order of a few 10 Myr old, implying a delayabout 50-100 Myr between the onset of central stellar massgrowth and subsequent rapid black hole growth. Other re-cent studies (e.g., Schawinski et al. 2009; Wild et al. 2009,2010; Yesuf et al. 2014) investigated AGN activity relative tothe integrated stellar populations of local galaxies and sug- [email protected] gested that AGNs become stronger after the stellar popula-tions reach a few 100 Myr old.Quasars with massive poststarburst hosts may also rep-resent a vital intermediate phase before total star forma-tion quenching, and may provide insight on the connec-tion between AGNs and starbursts. Various observationslend support for the evolutionary scenario from ULIRGsto quasars to dead ellipticals (e.g., Sanders et al. 1988;Magnelli et al. 2011; Murphy et al. 2011). A significant frac-tion of advanced merger (U)LIRGs (Guo et al. 2016 ) areobscured transiting post-starburst galaxies (see Yesuf et al.2014)„ which are the starting point of the fast evolutionarytrack.The host galaxies of luminous AGNs are foundto have intermediate-age stellar populations (e.g.Canalizo & Stockton 2013; Matsuoka et al. 2015), whichmay correspond to the stage after gas-rich mergers in themerger driven co-evolutionary scenario. Zhang et al. (2016)found that the specific star formation rates (sSFRs) decreasefrom ULIRGs to obscured 2MASS quasars (Shi et al. 2014)to unobscured PG quasars. On the other hand, infraredbright QSOs and narrow-line Seyfert 1 (NLS1) galaxies arepossibly in the early phase of the evolution from ULIRGs todead ellipticals (Hao et al. 2005).Post-starburst quasars (PSQs) simultaneously show thespectral signatures of quasars and post-starburst stellar pop-ulations (Brotherton et al. 1999). The observed spectra ofPSQs show the power-law continuum and the broad emissionlines, as well as strong Balmer absorption,characteristic ofintermediate age stellar populations (Brotherton et al. 1999,2002). W
EI ET AL .A small sample of luminous PSQs, spectroscopically se-lected from the SDSS at z ∼ Spitzer mid-infrared spectroscopy(Wei et al. 2013).. The selection criteria of this PSQs sam-ple used the Balmer absorption lines and the Balmer breakof the quasar spectra in order to identify the post-starburstsignatures.These aforementioned studies found that PSQs have a het-erogeneous host population with different stellar properties(starburst mass and age) and AGN properties (SMBH massand Eddington fractions). They concluded that the PSQswith early-type hosts likely evolved via major mergers, whilethose with spiral hosts were triggered by secular processes.In this work, we compile a large sample of PSQs based oncareful AGN-host decomposition. Large samples of post--starburst quasars, covering large areas, have not yet beenidentified before . In the rest of this section we briefly reviewworks relevant to post-starbursts quasars.Kauffmann et al. (2003b) found that a significant fractionof high-luminosity type 2 AGN have experienced a starburstin the recent past. Likewise, Goto (2006) identified 840 type2 AGNs with post-starburst hosts from the SDSS, and foundthat the fraction of post-starburst type 2 AGNs is at least 4.2%of all galaxies in their sample.Similarly, Matsuoka et al. (2015) identified 191 type 1quasars at z < field of the SDSS ReverberationMapping (RM) project. They derived host stellar propertiessuch as age and mass, and found that half of the quasars inthis sample show post-starburst signatures.Melnick et al. (2015) compiled a sample of 72 nearbyPSQs from the SDSS DR7Q catalog. Unlike Matsuoka et al.(2015), who used a spectral decomposition and mea-sured EW(H δ ), Melnick et al. (2015) directly measured theEW(H δ ) from the observed quasar spectra without estimat-ing the dilution of this absorption line by the correspondingquasar emission line. In addition, their lenient (EW(H δ ) > )cut cannot reject large number of quasars which are hosted bynormal star-forming galaxies.This work is different from preceding works for the fol-lowing reasons: 1) Our work focuses on the post-starburstquasars, which are brighter than M i < − mag. In con-trast, several large sample studies only focused on post-starburst hosts of the low-luminosity type 2 AGNs (cf.Kauffmann et al. 2003b; Goto 2006; Yesuf et al. 2014). 2)We use the spectral decomposition analysis of our spectra toaccount for the dilution of post-starburst features by the AGNcomponent. Several previous investigations did not includesuch analysis (cf. Cales et al. 2011, 2013; Melnick et al.2015). 3) We applied a stricter criterion of (EW(H δ ) > ) toselect the pure post-starburst hosts with larger post-starburststellar populations fractions and effectively reject the nor- mal star-forming hosts (cf. Goto 2006; Melnick et al. 2015).4) The parent sample of this work has larger areal cover-age than that of SDSSRM quasar sample (cf. Matsuoka et al.2015). Our spectral decomposition method is similar to thatof Matsuoka et al. (2015).Our sample of quasars and the selection criteria for PSQsare described in subsection 2.1. Details of the spectral fit-ting method are explained in subsection 2.2. We test ourmethod in subsection A.1. Results and discussion are givenin section 3 & 4. At last, we summarize our results insection 5. We will discuss a more detailed study of the prop-erties of PSQs in a subsequent paper. We adopt cosmologicalparameters H =
70 km s − Mpc − , Ω m = . , and Ω Λ = . throughout this paper. SAMPLE SELECTION AND FITTING METHODFor our sample of PSQs, we begin with the SDSS DR7Quasar Catalog (DR7Q, Schneider et al. 2010), which con-sists of the 105,783 SDSS quasars with M i < − mag andat least one broad optical emission line with FWHM ≥ − . The absolute magnitude limit in DR7Q catalogwas calculated by correcting the BEST i-band PSF magni-tude measurement for Galactic extinction (using the mapsof Schlegel et al. 1998) and assuming that the quasar spec-tral energy distribution in the ultraviolet-optical can be rep-resented by a power-law with α = − . (see details inSchneider et al. 2010).Our PSQ sample inherits this absolute magnitude cut fromthe parent DR7Q sample. This absolute magnitude cutis close to the classical quasar threshold M B = − ma g (Schmidt & Green 1983), which corresponds to M i ∼− ma g (Matsuoka et al. 2015). However, both AGN andhost can contribute to the i-band PSF luminosity. For an ob-served PSQ spectrum, the fraction of light contributed by thehost galaxy must be large enough to show their spectral sig-natures. Due to this compound nature of PSQs, this abso-lute magnitude cut ( M i < − ma g ) means our sample is notcomplete down to a specific AGN luminosity or a stellar pop-ulation luminosity. 2.1. Sample selection
Strong Balmer absorption arises in galaxies that experi-enced a recent burst of star formation triggered . − Gyrpreviously with subsequent rapidly quenching. After a ma-jor starburst epoch has ceased, the aging population of thestarburst will produce a characteristic post-starburst spectralsignature with strong Balmer lines in absorption. The EW ofthe H δ absorption line, one of the star-formation history in-dicators, allows us to constrain mean stellar ages of galaxiesand the fractional stellar mass formed in bursts over the pastfew 100 Myr.However, unlike the selection of post-starburst galaxies orpost-starburst type 2 AGNs, a selection criterion based onthe Balmer absorption line of observed quasar spectral de- C ATALOG OF
PSQ
FROM
SDSS DR7 3 (a) (b)
Figure 1 . Examples of the spectral decomposition. The black line indicates the observed spectra, while the dark green line shows the total modelspectrum. The other lines represent the individual model components, namely, the quasar power-law continuum (purple), the Fe II emission(red), the gas emission from the BLR (blue), and the stellar emission (light green). pends both on the age of the stellar populations as well as theamount of quasar dilution. Hence, for the most reliable mea-surements of the host components (H δ host ), it is important toapply spectral decomposition methods.The initial selection criteria for the spectra for which weapply spectral decomposition are: • S / N ∼ ≥ • z < • EW(H δ ) ≥ − S / N ∼ is the continuum signal-to-noise ratio be-tween the rest-wavelengths of 4150 and 4250 Å. This S/Ncriterion requires that spectra have enough S/N to decomposerobustly, and approximately corresponds to m i ≤ . ma g .We limit the redshift range to z < m i = . ma g (i.e. S / N ∼ ) has an absolute magni-tude M i = − . ma g . If the host is a massive galaxy with M i = − ma g , there is only 25% light contributed by thehost which is only 2.5 times the noise.The EW(H δ ) used for this initial step is the direct measure-ment of the total observed quasar spectrum. EW(H δ ) ≥ − ensures that the broad Balmer emission line from the quasarBLR is not too strong to hide the Balmer absorption line fromthe host.There are 1682 quasar spectra that satisfy the above crite-ria in the SDSS DR7Q catalog. We performed fitting and de-composition (see next section) of the spectra of those quasars.We exclude 438 objects with the fraction of light contributedby AGN ( f AGN ) larger than 75%, because the light con-tributed by the host may be comparable to the noise, whichmakes the spectral fitting results unreliable. After a carefulcheck by eye, 80% (987/1244) spectra have successful spec-tral fitting and decomposition. W
EI ET AL .For classification as a post-starburst quasar, one additionalcriterion from the decomposition result below has to be ful-filled: EW(H δ host ) ≥ . According to the tests in appendixA, EW(H δ host ) is the most robust parameter of the host com-ponent that we can recover with our spectral decomposition.It is pointed out that, although our decomposition providesquantitative measurements of mean stellar age, in the rest ofthis paper we use EW(H δ host ) as the primary diagnostic pa-rameter.We found that about 21% of the objects (208 out of 987)have rest-frame equivalent withs of H δ host greater than 6Åin absorption, which is the final PSQs sample as listed inTable 1. Figure 2 displays the distribution of their redshiftsand absolute magnitudes in SDSS i band ( M i ). Our PSQfraction is significantly higher than the 4.2% post-starburstAGNs studied by Goto (2006) in a volume-limited sampleof SDSS galaxies, where they derive the fraction of all AGNwith strong H δ in absorption, while our sample is restrictedto objects classified as quasars (f AGN ≤ and6 PSQs are in the 38 PSQs sample of Cales et al. (2013) .Kauffmann et al. (2003b) find that 95% of galaxies withEW(H δ ) > have experienced a burst with a mass fractiongreater than 5% during the last 2 Gyr. If using a more re-laxed H δ absorption criterion (i.e. EW(H δ ) > Å), it is dif-ficult to distinguish post-starbursts from normal star-forminggalaxies with constant SFR. Therefore, the selection post-starburst galaxies also uses a limit on nebular emission torestrict the amount of on-going star formation. However,Falkenberg et al. (2009) found this approach to be too nar-row to cover the full range in post-starburst populations. Re-cent investigations (e.g., Wild et al. 2009; Yesuf et al. 2014)have identified the precursors of post-starburst galaxies haveshown that AGNs are more common in these objects but thatthere is a significant time lag between the starburst and theAGN phase. In fact, while requiring nebular lines to beweak only identifies objects in the quenched post-starburst(Yesuf et al. 2014, QPSB) phase, but overlooks the quench-ing post-starburst galaxies in transit between the starburststage and the fully QPSB stage Furthermore, limits on nebu-lar emissions also exclude any post-starburst galaxies hosting AGNs since AGNs can power nebular emission lines. Giventhese issues, the non-negligible number of AGNs found inpost-starburst galaxies should cause us to reconsider the rela-tive importance of the presence of an AGN in post-starbursts.We should not be too surprised that only 1% of all galaxiesare quenched post-starbursts galaxies (Wong et al. 2012), be-cause of the underestimation of the transiting post-starburstswith ongoing star formation or AGN activity .We therefore did not place limits on [OII] λ or H α emission. The only constraint we used is that the equiv-alent width of H δ in absorption is EW(H δ host ) ≥ α lines. From the figure 6 ofKauffmann et al. (2003a), we can see there still exist star-burst components at EW(H δ ) > . In future work , we willuse [OII]/[OIII] to gauge the ionization level, and use mid-infrared data and the SED of PSQ to estimate the SFRs ofPSQs, and to determine the fraction of starbursts in our sam-ple.
Figure 2 . Redshifts and i-band absolute magnitudes of the DR7Qquasars (small dots) and 208 post-starburst quasars (large dots). Thedotted line is our redshift cut (z < Table 1 . The properties of Post-starburst Quasars.
SDSS DR7 NAME RA (2000) DEC (2000) z i M i f AGN H δ H δ host (deg) (deg) (mag) (mag) (Å) (Å)(1) (2) (3) (4) (5) (6) (7) (8) (9)000104.92+160101.2 0.27 16.02 0.45 18.97 -23.36 0.60 0.71 6.21002815.78+004247.5 7.07 0.71 0.31 18.27 -23.30 0.39 5.93 9.45002959.03+150817.2 7.50 15.14 0.21 18.04 -22.40 0.57 0.00 7.52021652.47-002335.3 34.22 -0.39 0.30 18.29 -23.31 0.53 -1.24 7.66031715.10-073822.3 49.31 -7.64 0.27 18.32 -22.93 0.51 0.74 8.32 Table 1 continued C ATALOG OF
PSQ
FROM
SDSS DR7 5
Table 1 (continued)
SDSS DR7 NAME RA (2000) DEC (2000) z i M i f AGN H δ H δ host (deg) (deg) (mag) (mag) (Å) (Å)(1) (2) (3) (4) (5) (6) (7) (8) (9)032628.53-002741.4 51.62 -0.46 0.45 18.75 -23.72 0.65 0.90 8.50072837.76+443646.0 112.16 44.61 0.28 18.35 -23.07 0.27 2.66 6.92074310.24+461014.5 115.79 46.17 0.27 17.99 -23.32 0.21 2.65 6.38074621.06+335040.7 116.59 33.84 0.28 18.08 -23.26 0.23 3.54 6.66074844.87+441422.0 117.19 44.24 0.43 18.72 -23.51 0.41 4.40 9.17...N OTE — This table is available in its entirety in a machine-readable form in the online journal. A portion is shownhere for guidance regarding its form and content.
Spectral-fitting method
In order to disentangle the AGN contribution from that ofthe host galaxy, we performed a spectral fitting and decompo-sition. We utilized the IDL MPFIT to model the host stellarpopulations and AGN contributions to the DR7 SDSS spec-tra. The χ minimization technique of MPFIT simultane-ously models multiple components: S ( λ ) = aA ( λ, σ Fe ) + bB ( λ, σ B ) + cC ( λ, α λ ) + Õ i = d i SSP i ( λ, σ ∗ ) (1) S ( λ ) is the observed spectrum. A ( λ, σ Fe ) denotes the UVand optical iron emission line blends, which are modeled us-ing the UV and optical Fe II templates derived from I Zw1 (Vestergaard & Wilkes 2001; Boroson & Green 1992). Weconvolved a Gaussian of width σ Fe with the Fe II templatesto simulate different velocity widths. B ( λ, σ B ) representsthe templates of Balmer emission lines and Balmer contin-uum broadened by convolving with a Lorentzian of width σ B with a specific Balmer decrement. The AGN power-low continuum is assumed to be C ( λ ) = λ α λ . The slope α λ is assumed to extend from − Í i = d i SSP i ( λ, σ ∗ ) represents the starlight componentmodeled by the 6 Simple Stellar Populations templates withthe solar metallicity and different age (30 Myr, 100 Myr,300 Myr, 1 Gyr, 3 Gyr and 10 Gyr), which had been builtup from the spectral template library of (Bruzual & Charlot2003, hereafter BC03). These were broadened by convolvingthe spectra with a Gaussian of width σ ∗ to match the stellarvelocity dispersion of the host galaxy.The fitting was performed by minimizing χ with σ Fe , σ B , − α λ , σ ∗ , a , b , c , and d i being non-negative free parameters.We give plots of the run corresponding to the best fit of firsteight objects as examples in Figure 1. We note that we did not employ younger SSP templates(e.g., <
30 Myr) because they may be degenerate with thequasar power-law continuum. We emphasize that we onlyuse our fitting to approximately reconstruct the host spectra.This fitting does not do detailed modeling of the stellar pop-ulations of the hosts. A more sophisticated model may usemore SSP templates of varying ages and metallicities thanwe did.The narrow emission lines from the interstellar medium(ISM) and the AGN narrow-line region (NLR) are maskedout during the fitting. Our default models do not in-corporate the effects of dust extinction. The amountof intrinsic extinction observed in unobscured quasars isusually small (Richards et al. 2003; Hopkins et al. 2004;Salvato et al. 2009; Matute et al. 2012) and its effect is lim-ited. For half of our sample, the rest-frame spectral rangedoesn’t cover blueward of the Mg II line, because of the red-shift distribution of our sample. Hence we did not fit the MgII line. RESULTS
We constructed a catalog of 208 PSQs. Most of themare identified as such for the first time here.
We used spec-tral decomposition of AGN and stellar population of the hostgalaxies. We defined post-starbursts as host galaxies withEW(H δ host ) ≥ . Since we are interested in PSQs as a classas well as their typical characteristics, we derived compositeoptical spectra and composite SEDs.3.1. Composite spectra of PSQs
We computed the composite optical spectra of the totalsample of 208 PSQs by rebinning the spectra to the restframe using accurate redshifts, normalizing the spectra atrest-frame 5500Å, and applying a median stacking. A pre-vious study (Cales & Brotherton 2015, hereafter CB15) alsocreated composite spectra of PSQs, but the AGN luminosityand post-starburst criteria are different between our sampleand theirs. In Figure 3, we compare our composite spectrum W
EI ET AL . F λ ( A r b i t r a r y U n i t s ) SDSS QSOs (V01)Division: SDSS QSOs / Power−law FitPSQsCB15Residual: Division − PSQs F λ ( A r b i t r a r y U n i t s ) QSOsPSQsCB15
QSOsPSQsCB15 (a)(b) (c)
Figure 3 . (a) Composite optical spectra of PSQs in our sample, that of Cales & Brotherton (2015) and that of SDSS QSOs (Vanden Berk et al.2001). They are normalized at 5500 Å. The red line indicates the SDSS QSOs spectra which was removed the power law continuous, while thegreen line shows the total model spectrum. (b) The detailed view of H β region. The the red line shows the composite spectrum of PSQs (blue)and the blue one shows SDSS QSOs (red). They are 5100Å continuum subtracted and are normalized at the peak flux of H β . (c) The detailedview of H α region, and the colors are assigned similarly to (b). The 6800Å continuum flux is subtracted and the spectrum are normalized at thepeak flux of H α . of PSQs with that of SDSS QSOs (Vanden Berk et al. 2001,hereafter V01) and that of CB15.Our PSQs spectrum has a relatively flat continuum com-pared to V01. In order to better compare our spectrum to theV01 spectrum, we normalized the V01 spectrum by divid-ing out its power-law continuum. Our spectrum is similar tothe normalized V01 and CB15 spectra at the red end. At theblue end, we can see the obvious the Balmer high-order ab-sorption lines in our PSQ spectrum similar to the spectrum ofCB15, which shows the post-starburst characteristics, unlikeV01 which only has Ca K absorption line. We can also seethe Fe emission features in our spectrum. The H α and H β profiles are very similar to those of V01, which indicates thatthey have similar widths and asymmetries of broad Balmer emission, similar ratio of narrow Balmer line to broad Balmerline, and the ratio of SII to H α .3.2. Composite SED of PSQs
We also derived the composite SED of the PSQ sample,where the optical photometric data are from SDSS DR7,the near-ultraviolet (
NUV ) magnitudes are from the
GALEX satellite (Martin et al. 2005), near-infrared magnitudes arefrom 2MASS, and the mid-infrared magnitudes are from
WISE .We created the composite SED of the PSQs by normaliz-ing their spectra at rest-frame 4.6 µ m and applying a mediancombine at rest-frame wavelengths. We repeated the proce-dure for the quasar sample. In Figure 4, we show compositeSED of PSQs and the SED templates of Seyferts 2, QSOs C ATALOG OF
PSQ
FROM
SDSS DR7 7
Figure 4 . The composite SEDs in ν F ν vs λ of our full quasar sampleand PSQ subsample, compared with the SED template of Seyferts2, QSOs (SWIRE Template Library; Polletta07), and that of post-starburst galaxies (Goto 2007), normalized at the WISE µ m W2band. (SWIRE Template Library; Polletta et al. (2007)), and post-starburst galaxies (Goto 2007).It is clear that the PSQs lie between Seyfert 2 and QSOSEDs in the mid-infrared, especially between 3.4 µ m and12 µ m where they are closer to QSOs. The bumps at opticaland near-infrared wavelengths are due to a significant stellar fraction in our PSQs. The flat mid-infrared SED of post-starburst galaxies is very similar to our quasar and QSO tem-plates, which can be fitted with a hot dust component plus anAGN model (Sajina et al. 2012). The 22 µ m excess of PSQsmight correspond to the dust heated by young stars createdduring ongoing star formation. DISCUSSION4.1.
Modeling of composite SEDs
We performed the detailed modeling of the compositeSEDs in Figure 4 with an AGN SED fitting code to deter-mine the age of stellar component and the properties of thestar-forming. The code is AGNfitter (Calistro Rivera et al.2016), an open-source code allows the user to disentanglecomponents responsible for their emission robustly througha fully Bayesian Markov Chain Monte Carlo algorithm to fitthe SEDs of AGNs. A large library of theoretical, empirical,and semi-empirical models are used to characterize both theAGN and its host radiation simultaneously. Four physical ra-diation components constitute the model: stellar populations,cold dust in star-forming regions, an emitting accretion diskand a torus of hot dust surrounding the AGN. AGNfitter al-lows the user to derive the integrated luminosities, dust atten-uation parameters, stellar masses, and SFRs, by calculatingthe posterior distributions of numerous parameters governingAGNs’ physics through a fully Bayesian treatment of errorsand parameter degeneracies. We use the mean value of theprobability density function (PDF) as the final value of eachparameter (see Table 2), while the 16th and 84th percentilesof the distribution give the associated uncertainty.
Table 2 . Comparisons of fundamental physical properties
Type PDF percentiles τ Age M ⋆ L IR ( − ) SFR IR (Gyr) (log yr) (log M ⊙ ) ( erg s − ) (M ⊙ yr − )(1) (2) (3) (4) (5) (6) (7)16% 1.83 7.96 10.32 40.74 0.00PSQs 50% 5.69 8.74 10.58 44.19 6.0284% 12.02 9.21 10.82 44.59 15.1216% 3.47 8.80 10.74 40.22 0.00QSOs 50% 7.87 9.32 10.92 42.38 0.0984% 12.11 9.64 11.05 44.28 7.46 Figure 5 shows the composite SEDs and the AGNfitter fits.We scaled the composite SEDs to the median 4.6 µ m lumi-nosity of the corresponding sample. The starburst (green),host galaxy (orange), the hot-dust emission (purple) and the“big-blue bump” (blue) components combine for the com-posite model (red), which is shown in the panels in Figure 5.We randomly pick eight different realizations from the poste-rior PDFs and over-plot the corresponding component SEDs in order to visualize the dynamic range of the parameter val-ues included in the PDF. In the Table 2 and Figure 5, wecan see our composite PSQ SED shows younger stellar age,shorter exponential star formation history (SFH) timescale τ ,and higher specific star formation rates (sSFR; SFR/M ⋆ ) thanthose of the QSOs sample. The fitting result of age ∼ EI ET AL . (a) (b) Figure 5 . SED fitting of the composite SED of PSQs (left) and QSOs (right): The black circular markers with error bars indicate the observedphotometric data. The other lines represent the individual model components, namely, starburst component (green), host galaxy component(orange), the hot-dust emission (purple) and the BBB template (blue), while the red line shows the linear combination of these, i.e., the ’totalSED’. Eight different realizations which are picked randomly from the parameter posterior PDFs are plotted to show the uncertainties of theparameters on the SEDs.5.
SUMMARYWe constructed a catalog of 208 post-starburst quasars(PSQs) from the SDSS DR7 quasar catalog. The PSQs haveH δ absorption equivalent width EW(H δ host A. APPENDIX MATERIALA.1.
Mock spectra test
In order to evaluate the reliability of our method, we generated 5000 mock spectra by combining spectral models of quasarpower-law continuum, Fe emission template, broad Balmer emissions and 6 SSPs host stellar population as described insubsection 2.2, and 10% random noise. We built a probability distribution function (PDF) of each input parameter by a lin-ear interpolation of the measured distribution of that parameter. The fitting results of the free parameters are plotted in Figure A1.We found that the output best-fitting f
AGN is reliable, and the scatter of EW(H δ host ) is not greater than 0.6Å until the AGN comesto dominate the spectral flux (f AGN > ). We therefore excluded the objects with f AGN > which their host results are C ATALOG OF
PSQ
FROM
SDSS DR7 9considered unreliable. In Figure A1 (c), there are some patterns in the the region between -2Å and 2Å. The reason is that whenthe Balmer absorption features are not obvious, e.g. the age of host is relatively old, our fitting result may reach the boundaryvalue of -2Å. But this problem of an old age host does not affect the results of this investigation.We also applied our spectral-fitting method for the sample of C13 to further test our method. Although our sample and theirsonly partially overlap because the AGN luminosity and post-starburst criteria are different, we still obtained a tight correlationbetween our EW(H δ host ) measurements and the starburst ages of PSQs from C13 (see Figure A2), showing that our methods areat least consistent with this previous careful work. REFERENCES Bergvall, N., Marquart, T., Way, M. J., et al. 2016, A&A, 587, A72Boroson, T. A., & Green, R. F. 1992, ApJS, 80, 109Bouwens, R. J., Illingworth, G. D., Franx, M., et al. 2009, ApJ, 705, 936Brotherton, M. S., Grabelsky, M., Canalizo, G., et al. 2002, PASP, 114, 593Brotherton, M. S., van Breugel, W., Stanford, S. A., et al. 1999, ApJL, 520,L87Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Cales, S. L., & Brotherton, M. S. 2015, MNRAS, 449, 2374Cales, S. L., Brotherton, M. S., Shang, Z., et al. 2011, ApJ, 741, 106—. 2013, ApJ, 762, 90Calistro Rivera, G., Lusso, E., Hennawi, J. F., & Hogg, D. W. 2016, ApJ,833, 98Canalizo, G., & Stockton, A. 2013, ApJ, 772, 132Davies, R. I. 2007, MNRAS, 375, 1099Fabian, A. C. 2012, ARA&A, 50, 455Falkenberg, M. A., Kotulla, R., & Fritze, U. 2009, MNRAS, 397, 1940Goto, T. 2006, MNRAS, 369, 1765—. 2007, MNRAS, 381, 187Hao, C. N., Xia, X. Y., Mao, S., Wu, H., & Deng, Z. G. 2005, ApJ, 625, 78Hopkins, P. F., Strauss, M. A., Hall, P. B., et al. 2004, AJ, 128, 1112Kauffmann, G., Heckman, T. M., White, S. D. M., et al. 2003a, MNRAS,341, 33Kauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003b, MNRAS, 346,1055Kormendy, J., & Richstone, D. 1995, ARA&A, 33, 581Magnelli, B., Elbaz, D., Chary, R. R., et al. 2011, A&A, 528, A35Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005, ApJL, 619, L1 Matsuoka, Y., Strauss, M. A., Shen, Y., et al. 2015, ApJ, 811, 91Matute, I., Márquez, I., Masegosa, J., et al. 2012, A&A, 542, A20Melnick, J., Telles, E., De Propris, R., & Chu, Z.-H. 2015, A&A, 582, A37Murphy, E. J., Chary, R.-R., Dickinson, M., et al. 2011, ApJ, 732, 126Polletta, M., Tajer, M., Maraschi, L., et al. 2007, ApJ, 663, 81Richards, G. T., Hall, P. B., Vanden Berk, D. E., et al. 2003, AJ, 126, 1131Sajina, A., Yan, L., Fadda, D., Dasyra, K., & Huynh, M. 2012, ApJ, 757, 13Salvato, M., Hasinger, G., Ilbert, O., et al. 2009, ApJ, 690, 1250Sanders, D. B., Soifer, B. T., Elias, J. H., Neugebauer, G., & Matthews, K.1988, ApJL, 328, L35Schawinski, K., Virani, S., Simmons, B., et al. 2009, ApJL, 692, L19Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010, AJ, 139, 2360Shen, Y., Richards, G. T., Strauss, M. A., et al. 2011, ApJS, 194, 45Shi, Y., Rieke, G. H., Ogle, P. M., Su, K. Y. L., & Balog, Z. 2014, ApJS,214, 23Treister, E., Schawinski, K., Urry, C. M., & Simmons, B. D. 2012, ApJL,758, L39Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001, AJ, 122, 549Vestergaard, M., & Wilkes, B. J. 2001, ApJS, 134, 1Wei, P., Shang, Z., Brotherton, M. S., et al. 2013, ApJ, 772, 28Wild, V., Heckman, T., & Charlot, S. 2010, MNRAS, 405, 933Wild, V., Walcher, C. J., Johansson, P. H., et al. 2009, MNRAS, 395, 144Wong, O. I., Schawinski, K., Kaviraj, S., et al. 2012, MNRAS, 420, 1684Yesuf, H. M., Faber, S. M., Trump, J. R., et al. 2014, ApJ, 792, 84Zhang, Z., Shi, Y., Rieke, G. H., et al. 2016, ApJL, 819, L27
EI ET AL . AGN0 −0.3−0.2−0.1−0.00.10.20.3 f A G N − f A G N rms = 0.039rms = 0.041 −2 0 2 4 6 8 10 12H δ [Å]−6−4−20246 H δ − H δ [ Å ] rms = 0.490rms = 0.947rms = 0.600 AGN0 −6−4−20246 H δ − H δ [ Å ] (a)(b)(c) Figure A1 . Comparison between the input parameter with the difference of the output from input parameter (a) input AGN fraction and thedifferent between output and input AGN fraction, (b) input AGN fraction and the different between output and input H δ host , (c) input H δ host andthe different between output and input H δ host . The black dots represent the spectra with reliable fits of the stellar component (output AGNfraction f AGN < AGN < δ host > AGN > C ATALOG OF
PSQ
FROM
SDSS DR7 11
Figure A2 . The EW(H δ host ) measurements plotted against the starburst ages of PSQs from C13. There is a linear correlation with a Pearson rvalue of − −−