Why would you want accurate quasar redshifts? Paul Hewett (IoA, Cambridge) James Allen (U. Sydney)
Jan 18, 2018
Why would you want accurate quasar redshifts?
Paul Hewett (IoA, Cambridge)
James Allen (U. Sydney)
Outline• Motivation: quasar clustering, host-galaxy and local
quasar environment, relation to Lyα-absorbers and the inter-galactic medium, outflow properties of quasars themselves
• Current redshift accuracy for high-redshift, z>2.0, quasars and what do we need?
• Decomposing spectra into components to do better - redshifts using Independent Component Analysis applied to the ~1900A CIII]+SiIII]+AlIII complex
• Comparison to BOSS DR10 pipeline and PCA redshifts• What is now possible
Galaxies2dF Galaxy Survey redshift wedge
Redshift due to `Hubble flow’component plus`peculiar velocity’ due to influence of mass inhomogeneities over age of Universe
For optical galaxy spectra can reach σ~30km/s
For Hubble Constant ~70km/s/Mpcapprox 0.5Mpc λObserved = λRestframe×(1+z) with z=redshift
Want same for quasars but also have ability to study relation of quasars to gas along line-of-sight: probes host-galaxy, local environment and inter-galactic medium
(Michael Murphy)
Last decade produced revolution in quasar surveys – Sloan Digital Sky Survey (SDSS)- DR7 100000 quasars (2007); DR12 (BOSS) 450000 quasars (2014)- 3D quasar clustering and 3D quasar-absorber studies possible- Quasar host-galaxy and environment studies also viable
Wild+2008
Blackhole, fed by accretion disk with associated Broad Line Region (BLR) clouds and more distant Narrow Line Region (NLR) cloudsCloud distances ~1 parsec (BLR), 1000 parsec (NLR) with associated velocities ~5000 and 1000km/s respectively
NASA
• Rest-frame ultraviolet and optical quasar spectrum • Continuum (from accretion disk) broad and narrow emission lines• Encouraging in that emission from common elements evident• At low-redshift, see host-galaxy spectrum and quasar• Can work up to higher redshift using emission lines with rest-frame
wavelengths >2500Ǻ - find redshift accuracy no worse than σ~170km/s• BUT at shorter wavelengths, equivalent to redshift z>2.0 for ground-based
spectra, quasar spectral energy distribution (SED) shows significant variations
• Observationally, strong asymmetries (blueshifts) evident for high-ionization emission lines (Gaskell 1982, Carswell, Tytler,…), i.e the broad emission lines have different shapes
• One explanation invokes presence of `disk winds’ with material at high outflow velocities contributing to the emission-line profiles
• Systematic dependence on viewing orientation, L/LEddington , luminosity relative to Eddington luminosity,…
• Not a subtle effect – 3000km/s shifts mean redshifts awry by up to 100× galaxy redshift errors [~15Mpc Hubble flow]
• True for optical spectra of z>2 quasars – key epoch and essential for quasar-IGM studies
• Relationship between quasar and environment also severely compromised
Observed frequency distribution of redshift differences, , for 23 800 C iv absorbers using both SDSS (blue) and HW (red) redshifts for quasars with redshifts 1.55< z <3.5.
Paul C. Hewett, and Vivienne Wild MNRAS 2010;405:2302-2316
© 2010 The Authors. Journal compilation © 2010 RAS
• Low ionization lines, including MgII 2800 provide stable reference – good for redshifts z<2.0 but z>2.0 key
• Hewett+Wild (2010) scheme major improvement for SDSS DR7 but essentially based on a single-”template”
• All single-template schemes have no information on spectra differences – hence SED-dependent systematic errors
• Amplitude of systematics >1000km/s, whereas want ~200km/s for host-galaxy, environment, clustering,…investigations
• Widely recognised that a natural solution involves– Parametrize quasar spectra into a number of “components”– Each quasar represented by sum of components with different weights – Reconstruct spectrum of each quasar determining component weights
and redshift similtaneously– Allows for SED-variation, e.g. component(s) might include `outflow’
signature and be present in different amounts from quasar to quasar
Quasar Redshifts: Status
• Spectra as a linear combination of components S = W C
spectra = weights × components
Good if C<<S• Differing rules/constraints on “component” derivation
– Principal Component Analysis (PCA)– Independent Component Analysis (ICA)
• Mean Field ICA (Allen, Hewett+ 2013 MN 430 3510)– Very different from most ICA implementations– Priors, constraints on components possible– Extremely compact (i.e. #components small)– Example using SDSS low-z Post starburst galaxies (“answer
known”)
Decomposing Spectra
• For the SDSS BOSS quasar survey [450000 quasars nearly all with z>2]
• Two main redshift estimates – Z_VI (pipeline/visual inspection) – Z_PCA (Paris+ 2011,2012) hope to be SED-independent
• Z_PCA unbiased, σ~750km/s relative to MgII 2800 – not great• Statistical analysis using Lyα-forest
– Cross-correlation with Lyα-forest gives Z_VI=231km/s low, Z_PCA=154km/s low (Font-Ribera 2013) [both +/-30km/s]
• Both BOSS schemes use quasar spectra down to 1400Ǻ• MFICA components [6] applied to just the low ionization CIII]
+SiIII]+AlIII complex. ~2400 BOSS quasars, @z~2.5 where MgII visible [use well-behaved MgIIλ2800 as reference]
SDSS DR12 Quasar Redshifts
Current Status
• Confirmation of BOSS-projects own redshift uncertainty determinations
• MFICA CIII]-complex redshifts reduce errors to σ~200km/s (cf. current σ~750km/s)– Possible for SDSS DR7 and BOSS to z<4.0– Possible for any quasar spectrum where CIII]-
complex covered• Accurate individual errors from an MCMC scheme
using component weight and redshift errors – on the way
• Quasar environments (host-galaxy, group/clusters,…) and absorber outflow properties as a function of black-hole mass, L/LEddington, radio-properties,…
• Example from Wild (2009) for MgII 2800 absorbers – host(?), outflow and intervening absorber components defined
• Full SDSS DR7 and DR12 analysis will produce vast improvement in statistics
Summary
• Many astrophysical investigations involving quasars at z>2 limited by errors in redshift determinations
• MFICA CIII]-complex redshifts reduce errors to σ~200km/s (cf. current σ~750km/s)– Possible for SDSS DR7 and BOSS to z<4.5– Possible for any quasar spectrum where CIII]-
complex covered