SDSS

The SDSS is Two Surveys

The Fuzzy Blob Survey

The Squiggly Line Survey

The Site

The telescope•2.5 m mirror

1.3 MegaPixels

$150

4.3 Megapixels

$850

100 GigaPixels

$10,000,000

Digital Cameras

CCDs

CCDs: Drift Scan Mode

NGC 450

NGC 1055

NGC 4437

NGC 5792

NGC 1032

NGC 4753

NGC 60

NGC 5492

NGC 936

NGC 5750

NGC 3521

NGC 2967

NGC 5719

UGC 01962

NGC 1087

NGC 5334

UGC 05205

UGC 07332

UGCA 285

Arp 240

UCG 08584

NGC 799NGC 800

NGC 428

UGC 10770

Measuring Quantities From the Images:Measuring Quantities From the Images:The Photo pipelineThe Photo pipeline

Most People use Magnitudesm = –2.5 Log (flux) + C

How do you measure brightness?

We use Luptitudes

m = –2.5 ln (10) [asinh( ) + ln(b)] f/f0

2b

OK, but how do you measure flux?

Isophotal magnitudes:What we don’t do

OK, but how do we measure flux?

Petrosian Radius:Surface brightnessRatio =0.2

Petrosion flux:Flux within 2 Petrosian Radii

Some Other Measures

PSF magnitudes

Fiber magnitudes

Galaxy Models

de Vaucouleurs magnitudes:assume profile associated with ellipiticals

Exponential magnitudes:Assume profile associated with spirals

I=I0 exp {-7.67[(r/re)1/4]}

I=I0 exp {-1.68(r/re)}

Model magnitudes pick best

Which Magnitudes to Use?

Photometry of Distant QSOs

PSF magnitudes

Colors of Stars PSF magnitudes

Photometry of Nearby Galaxies

Petrosian magnitudes

Photometry of Distant Galaxies

Petrosian magnitudes

Other Image Parameters

• Size

• Type

psfMag – expMag > 0.145

• Many hundreds of others

SPECTRA

OBAFGKMLT

h

e

ine

irl/Guy

iss

eong

ime

Galaxy Spectra

Galaxies =Star+gas

QSO spectra

Z=0.1

QSO spectra

Z=1

QSO spectra

Z=2

QSO spectra

Z=3

QSO spectra

Z=4

QSO spectra

Z=5

Types of MapsTypes of Maps

• Main Galaxy Sample• LRG sample• Photo-z sample• QSO sample• QSO absorption systems• Galactic Halo• Ly-α systems• Asteroids• Space Junk

EDR PhotoZ

Tamás BudaváriThe Johns Hopkins

University

István Csabai – Eötvös University, Budapest

Alex Szalay – The Johns Hopkins University

Andy Connolly – University of Pittsburgh

Template fittingTemplate fittingComparing known

spectra to photometry

++ no need for calibrators, physics in templates

++ more physical outcome, spectral type, luminosity

–– template spectra are not perfect, e.g. CWW

Empirical methodEmpirical methodRedshifts from calibrators

with similar colors

++ quick processing time

–– new calibrator set and fit required for new data

–– cannot extrapolate, yields dubious results

Pros and Cons

Empirical Methods

• Nearest neighbor– Assign redshift of closest calibrator

• Polynomial fitting function– Quadratic fit, systematic errors

• Kd-tree– Quadratic fit in cells

z = 0.033 z = 0.027 z = 0.023

Template Fitting

• Physical inversion– More than just redshift– Yield consistent spectral type,

luminosity & redshift– Estimated covariances

• SED Reconstruction– Spectral templates that match

the photometry better– ASQ algorithm dynamically

creates and trains SEDs

Ltype

z

u’g’r’i’z’

Trained LRG Template

• Great calibrator set up to z = 0.5 – 0.6 !

• Reconstructed SED redder than CWW Ell

Trained LRG Template

Photometric Redshifts• 4 discrete templates

– Red sample z = 0.028• z > 0.2 z = 0.026

– Blue sample z = 0.05

• Continuous type– Red sample z = 0.029

• z > 0.2 z = 0.035

– Blue sample z = 0.04

• Outliers– Excluded 2% of galaxies

• Sacrifice?– Ell type galaxies have better

estimates with only 1 SED– Maybe a decision tree?

z = 0.028

z = 0.029 z = 0.04

z = 0.05

PhotoZ Plates

• The Goal– Deeper spectroscopic sample

of blue SDSS galaxies• Blind test• New calibrator set

• Selection– Based on photoz results– Color cuts to get

• High-z objects• Not red galaxies

Plate 672

• The first results– Galaxies are indeed

blue

– … and higher redshift!

• Scatter is big but…– … that’s why needed the

photoz plates

LRGs

z = 0.085

Plate 672

• Redshift distributions compare OK# of g = 519– Photometric redshifts (Run 752 & 756)

– Spectroscopic redshifts (Histogram scaled)

Measures of the Clustering

• The two point correlation function ξ(r)

• The power Spectrum

• N-point Statistics

• Counts in Cells

• Topological measures

• Maximum Likelihood parameter estimation

Constraining CosmologicalParameters from Apparent Redshift-space Clusterings

Taka MatsubaraAlex Szalay

Redshift Survey Data → or →

Constraining Cosmological Parameters

(Traditional) Quadratic Methods

)(kP )(r,... , , , , , , 8BM nbh

• Effective for spatially homogeneous, isotropic samples.• However, evaluation of in real (comoving) space is not straightforward. (z-evolution, redshift-space distortion)

)(kP

)()2,1( r

2z

1z

12 r

),,()2,1( 1221 zzRedshift-space:

:space-real ,1z

Example:

Anisotropy of the clustering

Velocity distortions

real space redshift space

Finger-of-God (non-linear scales)

Squashing by infall (linear scales)

pec0 vrHcz

b/6.0

Geometric distortions (non-small z) real space redshift space

)(zH

)(zd A

)1()1()1()( M2

M3

0 zzHzH

z

A zH

zdH

Hzd

0M0

M0 )(1sinh

1

1)(

Likelihood analysis of cosmological parameters without direct determination of or)(kP )(r

LLL ii || (Bayesian)

,...,,,,,, , 8M nbhbii

x

Linear regime → : Gaussian, fully determined by a correlation matrix

modeljiijC

Huge matrix ← a novel, fast algorithm to calculate Cij for arbirtrary z : under development

|iL

Results

single determination

Normal ±3% ±19% ±16% ±4% ±2% ±0.5% ±0.5%

Red ±2% ±4% ±9% ±2% ±1% ±0.3% ±0.4%

QSO ±14% ±15% ±76% ±20% ±14% ±5% ±6%

M MB b8nh

simultaneous determination (marginalized)

Normal ±14% ±57% ±51% ±2%

Red ±9% ±10% ±33% ±0.9%

QSO ±170% ±75% ±360% ±69%

M MB b

Direct determinations of cosmological parameters

A novel, fast algorithm to calculate correlation matrix in redshift space

Normal galaxies : dense, low-z, small sample volume

QSOs : sparse, high-z, large sample volume

Red galaxies : intermediate → best constraints on cosmological parameters

Summary

0.00.0 1.0

1.0

0.8

0.6

0.4

0.2

0.2 0.4 0.6 0.8

ΩM

ΩΛ

Visualization

• CAVE VR system at Argonne National Laboratory

• SDSS VS v. 1.0 Windows based visualization system

• Tool directly tied to the skyserver for general visualization of multi-dimensional data

Accessing the Data

• Two databases

• Skyserver (MS SQL)– Skyserver.fnal.gov

• SDSSQT– Download from www.sdss.org

• Lab astro.uchicago.edu/~subbarao/chautauqua.html