Docstoc
EXCLUSIVE OFFER FOR DOCSTOC USERS
Try the all-new QuickBooks Online for FREE.  No credit card required.

A redshift-dependent fraction of broad absorption line quasars

Document Sample
A redshift-dependent fraction of broad absorption line quasars Powered By Docstoc
					  A redshift-dependent fraction of
   broad absorption line quasars
                                James T. Allen
                            Institute of Astronomy, Cambridge


                                     Paul Hewett
                                   Natasha Maddox
                                   Gordon Richards
                                   Vasily Belokurov



James T. Allen, IoA Cambridge                               Oxford, 2nd December 2009
                                Outline
 ●   Broad absorption line quasars (BALQSOs)
 ●   SDSS DR6 BALQSO catalogue
 ●   Selection effects
      –   BAL trough detection probability
      –   SDSS targeting completeness
 ●   Intrinsic BALQSO fraction
 ●   Implications for models


James T. Allen, IoA Cambridge                Oxford, 2nd December 2009
        Broad Absorption Line Quasars
                (BALQSOs)




 ●   Intrinsic absorption with velocity widths and
     blueshifts of 1000s of km s-1
 ●   Most commonly seen in high-ionisation lines,
     e.g. CIV 1549, SiIV 1397
 ●   Seen in ~10% of quasars
James T. Allen, IoA Cambridge            Oxford, 2nd December 2009
        Broad Absorption Line Quasars
                (BALQSOs)
 ●   Blueshifts and velocity
     widths imply outflows
     relatively close to the
     supermassive black hole
 ●   Origins unclear – orientation
     effect, evolutionary stage or
     separate class of quasar?
 ●   Need to measure fBAL to
                                     Urry & Padovani (1995)
     constrain possibilities


James T. Allen, IoA Cambridge              Oxford, 2nd December 2009
                          Balnicity Index (BI)

                                                            
                                                         f v
                                         −1
                                −3000 km s
                      BI=∫−25000 km s                 1−       C dv
                                                          0.9
                                             −1




James T. Allen, IoA Cambridge                                    Oxford, 2nd December 2009
                          Balnicity Index (BI)
                                Where should the
                                 continuum be?




James T. Allen, IoA Cambridge                      Oxford, 2nd December 2009
     Non-Negative Matrix Factorisation
                 (NMF)
 ●   Blind source separation technique: rewrites a
     data array as a linear combination of
     components
                                V=WH
             data = weights X components
 ●   The number of components produced is less
     than the number of input spectra



James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
                Reconstructing BALQSOs




   ~500 “clean” input quasars




James T. Allen, IoA Cambridge      Oxford, 2nd December 2009
                Reconstructing BALQSOs




   ~500 “clean” input quasars
                                NMF
                                        ~10 components




James T. Allen, IoA Cambridge         Oxford, 2nd December 2009
                Reconstructing BALQSOs




                                     ~10 components




James T. Allen, IoA Cambridge      Oxford, 2nd December 2009
                Reconstructing BALQSOs




                                     ~10 components

   quasar spectra




James T. Allen, IoA Cambridge      Oxford, 2nd December 2009
                Reconstructing BALQSOs




                                                                          ~10 components
                         NMF applied to weights only (not components)
   quasar spectra




James T. Allen, IoA Cambridge                                           Oxford, 2nd December 2009
                    Catalogue construction
 ●   Make components and reconstructions in
     redshift bins of width z = 0.1
 ●   Automated masks
 ●   Empirical corrections for heavily dust-reddened
     objects
 ●   ~0.5% spectra require manual intervention
 ●   NMF reconstructions generated for each quasar
     in SDSS DR6 (~100 000)


James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
                                Results

                                           z = 1.737
                                          BI = 13120 km s-1




                                           z = 1.752
                                          BI = 489 km s-1




James T. Allen, IoA Cambridge                     Oxford, 2nd December 2009
                                Results

                                           z = 2.509
                                          BI = 6876 km s-1




                                           z = 2.537
                                          BI = 74 km s-1




James T. Allen, IoA Cambridge                     Oxford, 2nd December 2009
                            BALQSO fraction
 ●   Count BALQSOs and non-BAL quasars
                                                  N BAL  BI , z , M i ,...
                      f BAL  BI , z , M i ,...=
                                                    N total  z , M i , ...




James T. Allen, IoA Cambridge                                            Oxford, 2nd December 2009
                            BALQSO fraction
 ●   Count BALQSOs and non-BAL quasars
                                                  N BAL  BI , z , M i ,...
                      f BAL  BI , z , M i ,...=
                                                    N total  z , M i , ...


                   Observed fraction
                                              ≠
                      Intrinsic fraction
James T. Allen, IoA Cambridge                                            Oxford, 2nd December 2009
                                Selection effects
 ●   Two main sources:
      –   Misidentification of objects within SDSS
            ●   Never previously calculated
      –   Quasar target selection for SDSS spectroscopic
          survey
            ●   Estimated by Reichard et al. (2003) using average
                observed BALQSO properties




James T. Allen, IoA Cambridge                           Oxford, 2nd December 2009
        Noise dependence of BALQSO
                identification
 ●   In general, decreasing the SNR decreases the
     quality of the NMF reconstruction
 ●   Even for a perfect reconstruction, the measured
     BI depends on the SNR




James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
        Noise dependence of BALQSO
                identification
 ●   In general, decreasing the SNR decreases the
     quality of the NMF reconstruction
 ●   Even for a perfect reconstruction, the measured
     BI depends on the SNR




James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
        Noise dependence of BALQSO
                identification
 ●   In general, decreasing the SNR decreases the
     quality of the NMF reconstruction
 ●   Even for a perfect reconstruction, the measured
     BI depends on the SNR




James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
        Noise dependence of BALQSO
                identification
 ●   In general, decreasing the SNR decreases the
     quality of the NMF reconstruction
 ●   Even for a perfect reconstruction, the measured
     BI depends on the SNR




James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
      Measuring detection probabilities
 ●   Insert synthetic broad absorption into real
     quasars
 ●   Vary SNR using SDSS sky spectra
 ●   Calculate NMF reconstruction
 ●   Is the BAL feature detected?




James T. Allen, IoA Cambridge             Oxford, 2nd December 2009
      Measuring detection probabilities
 ●   Detection probability depends most strongly on
     redshift, absorption depth, and SNR
     (~luminosity)




James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
           SDSS quasar target selection
 ●   Quasars are selected for SDSS spectroscopy
     based on their photometric properties
 ●   Absorbing material changes the photometric
     properties of a quasar
 ●   So the completeness for BALQSOs is not the
     same as for non-BAL quasars
 ●   Quantify effect using model quasar spectra



James T. Allen, IoA Cambridge           Oxford, 2nd December 2009
                      Model quasar spectra


                   Base model (Maddox et al. 2008)




James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
             BAL troughs

                   Base model (Maddox et al. 2008)




James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
                                      Dust reddening
             BAL troughs

                   Base model (Maddox et al. 2008)




James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
                                      Dust reddening
             BAL troughs

                   Base model (Maddox et al. 2008)

                                          UV continuum
                                           adjustments




James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
                                      Dust reddening
             BAL troughs

                   Base model (Maddox et al. 2008)

      Lyman limit
                                          UV continuum
       absorbers
                                           adjustments




James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
                                      Dust reddening
             BAL troughs

                   Base model (Maddox et al. 2008)

      Lyman limit
                                          UV continuum
       absorbers
                                           adjustments


       Magnitudes at all redshifts and luminosities

James T. Allen, IoA Cambridge                 Oxford, 2nd December 2009
                      Model quasar spectra
         Magnitudes at all redshifts and luminosities




James T. Allen, IoA Cambridge              Oxford, 2nd December 2009
                      Model quasar spectra
         Magnitudes at all redshifts and luminosities

          Scatter according to typical SDSS errors




James T. Allen, IoA Cambridge              Oxford, 2nd December 2009
                      Model quasar spectra
         Magnitudes at all redshifts and luminosities

          Scatter according to typical SDSS errors

            Process through SDSS target selection




James T. Allen, IoA Cambridge              Oxford, 2nd December 2009
                      Model quasar spectra
         Magnitudes at all redshifts and luminosities

          Scatter according to typical SDSS errors

            Process through SDSS target selection

  Weight each
   variable

                   SDSS Completeness (z, L, BAL)

James T. Allen, IoA Cambridge                Oxford, 2nd December 2009
                                Intrinsic fraction

                                 Observed numbers




James T. Allen, IoA Cambridge                        Oxford, 2nd December 2009
                                Intrinsic fraction

                                 Observed numbers
    Detection probabilities

                                Numbers within SDSS




James T. Allen, IoA Cambridge                         Oxford, 2nd December 2009
                                Intrinsic fraction

                                 Observed numbers
    Detection probabilities

                                Numbers within SDSS
                                          SDSS completeness

                      Intrinsic numbers (or fraction)



James T. Allen, IoA Cambridge                         Oxford, 2nd December 2009
                                Intrinsic fraction




James T. Allen, IoA Cambridge                        Oxford, 2nd December 2009
                                Intrinsic fraction




James T. Allen, IoA Cambridge                        Oxford, 2nd December 2009
                    Implications for models
 ●   Strong redshift dependence is not compatible
     with purely orientation-based models
 ●   Covering fraction may vary as quasar evolves
 ●   Covering fraction may depend on other redshift-
     dependent factors
 ●   Better statistics may help (e.g. BOSS)
 ●   Need better understanding of physics of BAL
     outflows: multiple transitions, other wavelengths


James T. Allen, IoA Cambridge            Oxford, 2nd December 2009
                                Conclusions
 ●   We have generated a catalogue of BALQSOs
     based on NMF reconstructions (to be made
     publicly available)
 ●   Selection effects are very important:
      –   Misidentification of observed quasars
      –   SDSS completeness
 ●   After corrections, intrinsic BALQSO fraction
     shows strong redshift dependence – highest at
     high z, minimum at z~2.6

James T. Allen, IoA Cambridge                     Oxford, 2nd December 2009