Constructing Mock Galaxy Catalogs with ADDGALS

1
Constructing Mock Galaxy Catalogs with ADDGALS

• Michael Busha, KIPAC, Stanford University
• rencontres de Moriond, 18 March 2008
• Risa Wechsler (Stanford/KIPAC), Ben Koester (U.
  of Chicago)

2
Outline

• Outline
• Motivation
• The requirements of a mock galaxy catalog for
  large photometric surveys and why we need a new
  method to do this.
• ADDGALS The algorithm
• Adding galaxies to a dark matter lightcone
  simulation based on local dark matter density.
• ADDGALS Validation
• How well the algorithm reproduces luminosity
  functions, color distributions, and correlation
  functions
• Ongoing projects with the catalogs
• Calibrating Cluster Finders
• DES analysis
• 3-point correlation function
• Future Prospects
• Refining the overdensity criteria
• Pushing to higher redshifts

3
(No Transcript)
4
Motivation -- Optical Cluster Surveys

• Upcoming photometric surveys such as DES and LSST
  have the potential to constrain cosmological
  parameters
• Large cluster samples with photometric redshifts
  give us a handle on the halo mass function
• Multiple mass-observable relations (richness,
  lensing, luminosity, ...)
• Wide, deep surveys will also let us constrain
  galaxy evolution by measuring the HOD
• In order to do this we need to
• Understand how galaxies are distributed with
  respect to the dark side of the universe, i.e.
  P(NgalMhalo)
• Understand how observed galaxy properties relate
  to the underlying distribution, i.e.
  P(NobsNgals) or P(NobsMhalo)
• Create mock galaxy sets so that analysis tests
  can be compared to a truth to get a handle on
  mass-observable relations and more generally
  characterize the selection function.

5
Ways to make Mock Catalogs

• A mock galaxy catalog for a large-scale
  photometric survey should have
• A magnitude or volume limited sample of galaxies
  5,000-40,000 sq.deg to z 1.5
• Accurate BCG and cluster galaxy properties
• Accurate colors and photometric redshift
  estimates
• Current methods for constructing mocks
• Semi-analytic models (SAMs, Croton et al 2005,
  Sommervil et al 20xx)
• Galaxy properties are set by halo merger
  histories.
• HOD/CLF (Berlind Weinberg 2002, van den Bosch
  et al 2003, Tinker et al 2005 )
• Constrain a parameterization for P(NgalsMhalo).
  Need to resolve subhalos with vmax 150 km/s
• Match Galaxies to subhalos (Kravtsov et al 2004,
  Conro et all 2006)
• Create a relation between subhalo properties,
  such as vmax, and galaxy luminosities. Need well
  resolved subhalos
• Use the local dark matter overdensity ADDGALS

SAM Subhalos HOD ADDGALS
Mrgt-20 2x1011 6x1011 4x1010 109
Mrgt-19 1012 1012 6x1011 1010
Number of particles needed to make a magnitude
limited 5000 sq deg mock out to z 1.4 (3Gpc/h)
6
ADDGALS Adding Density Determined GAlaxies to
Lightcone Simulations

• Specify a Luminosity function to generate a list
  of galaxies with Mrs.
• Assign the galaxies to a dark matter particle in
  a simulated lightcone with based on local dark
  matter density, P(dmMr).
• dm is the dark matter density smoothed on an
  appropriate scale.
• The Probability function is measured from
  clustering of subhalos and galaxies and
  parameterized to give a specified luminosity
  dependent 2-point function, ?(r,Mr).
• Note The galaxies know nothing about the halo
  distribution, or even the dark matter
  distribution on scales less than the smoothing!
• Galaxy colors are added by mapping SEDs from a
  training set of real galaxies (SDSS) to match the
  measured local galaxy density.

Galaxies from Millennium Simulation
All Particles
Probability
Mr -19
Mr -21
Mr -22
Rd - the distance enclosing dm
7
ADDGALS and BCGs

• One thing that is not accurately reproduced are
  the BCGs -- theyre not like other galaxies!
• We enforce that every resolved halo has a galaxy
  according to the relation (Vale Ostriker 2006,
  Zheng, Coil, Zehavi 2007)
• Scatter in the luminosity is fixed at 0.15
  (Hansen et al 2008)
• BCGs are taken from the Luminosity Function
  before any other galaxies are assigned, and added
  at the bright end as needed.
• With this modification, ADDGALS can create a
  galaxy distribution down to an arbitrary
  magnitude limit so long as
• The smoothing mass, dm, is resolved with at least
  10 particles
• The masses of halos hosting BCGs of interest are
  resolved with 20 particles

Hansen et al 2008
Vale Ostriker relation
ADDGALS Input
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Applications
Photo-z estimation (Farrens et al, in prep)

• HOD comparisons -- how well can different codes
  recreate an input truth.
• Optical Cluster Finding -- characterizing the
  behavior of 8 cluster finders (including BCG,
  red sequence, matched filter, Voronoi-Delaunay
  tessellation, ...)
• DES
• Generating a transfer function to use for
  groupfinding.
• Need Errors in magnitudes (photo-zs) that we
  can use create a training set for cluster
  finders.
• Ultimately we want a full transfer function that
  will include foreground contamination and
  telescope response so that we can push through
  the entire DES analysis pipeline and do a
  (blind?) parameter estimation.
• 3-point correlation function from SDSS

3-point function (Marin et al, in prep)
12
(No Transcript)
13
Conclusions

• ADDGALS works!
• The algorithm is able to reproduce a specified
  luminosity function and magnitude-dependent
  clustering in a low resolution dark matter
  simulation based on local dark matter
  overdensities.
• Colors can be accurately mapped from a training
  set based on local galaxy densities.
• BCGs must be inserted in a way that depends on
  host halo mass.
• The algorithm has been well verified by
  comparisons with maxBCG galaxy catalogs
• Much work is going into calibrating cluster
  finders and other tools for upcoming large-scale
  photometric galaxy surveys.