Galaxy Formation, Theory and Modelling

1
Galaxy Formation, Theory and Modelling

• Shaun Cole (ICC, Durham)

Collaborators Geraint Harker John Helly Adrian
Jenkins Hannah Parkinson
ICC Photo Malcolm Crowthers
25th October 2007
2
Outline

• An Introduction to the Ingredients of Galaxy
  Formation Models
• Recent improvements/developments
• Dark matter merger trees (Parkinson, Cole
  Helly 2007)
• Modelling Galaxy Clustering
• Constraints on s8
  (Harker, Cole Jenkins 2007)
• Conclude

3
Galaxy Formation Physics
Dark Matter

• The hierarchical evolution of the dark matter
  distribution
• The structure of dark matter halos
• Gas heating and cooling processes within dark
  matter halos
• Galaxy mergers
• Star formation and feedback processes
• AGN formation and feedback processes
• Stellar population synthesis and dust modelling

Gas
4
The hierarchical evolution of the dark matter
distribution

• Lacey Cole trees (extended Press-Schechter)
• Simulation from the Virgo Aquarius project
• Parkinson, Cole and Helly trees

5
The hierarchical evolution of the dark matter
distribution

• Millennium Simulation (movie and merger trees)
• Lacey Cole trees
• Parkinson, Cole and Helly trees

6
The hierarchical evolution of the dark matter
distribution

• Lacey Cole trees (extended Press-Schechter)
• Simulation from the Virgo Aquarius project
• Parkinson, Cole and Helly trees

7
EPS Merger Trees (Lacey Cole 1993, Cole et al
2000)
8
Parkinson, Cole and Helly 2007

• Parkinson, Cole and Helly 2007

Insert an empirically motivated factor into this
merger rate equation
9
Sheth-Tormen or Jenkins universal mass function
is a good fit to N-body results at all
redshifts. Thus we require

• Very nearly consistent with the universal
  Sheth-Tormen/Jenkins Mass Function

10
The structure of dark matter halos
NFW profiles, but with what concentration
Neto et al 2007
11
Gas heating and cooling processes within dark
matter halos

• Standard Assumptions
• Gas initially at virial temperature with NFW or
  b-model profile
• All gas within cooling radius cools
• Improved models being developed (McCarthy et al)
• Initial power law entropy distribution
• Cooling modifies entropy and hydrostatic
  equillibrium determines modified profile.
• Explicit recipe for shock heating

Helly et al. (2002)?
12
Galaxy mergers

• Galaxy orbits decay due to dynamical friction
• Lacey Cole (1993)
• Analytic
• Point mass galaxies
• Orbit averaged quantities
• Jiang et al 2007 (see also Boylan-Kolchin et al
  2007)

13
Star formation and feedback processes

• Rees-Ostriker/ Binney cooling argument cannot
  produce M break
• Feedback needed at faint end

Benson Bower 2003
14
AGN formation and feedback processes

• SN feedback not enough as we must affect the
  bright end
• AGN always a sufficient energy source but how is
  the energy coupled
• Demise of cooling flows
• Benefits LF modelling as heats without producing
  stars

Bower et al 2006
15
Stellar population synthesis and dust modelling
Star Formation Rate and Metallicity as a Function
of Time IMF assumption
Library of Stellar Spectra
Convolution Machine
Dust Modelling
Galaxy SED
16
Stellar population synthesis and dust modelling

• Many Stellar Population Synthesis codes (eg
  Bruzual Charlot, Pegase, Starburst99) are quite
  mature. But they arent necessarily complete.
• Maraston (2005) showed that TP-AGB stars can make
  a dominant contribution in the NIR.

Maraston 2005
17
Dust

• Dust makes galaxies appear fainter, and
  typically redder
• Also re-emits absorbed energy at longer
  wavelengths (dominating the SED at these
  wavelengths)?
• Dust has been treated with various degrees of
  sophistication

No Dust
Diffuse Screen
Diffuse in Disk
Diffuse Clouds
18
Semi-analytic Modelling
Semi-Analytic Model
19
Semi-analytic N-body Techniques

• Harker, Cole Jenkins 2007
• Use a set of N-body simulations with varying
  cosmoligical parameters.
• Populate each with galaxies using Monte-Carlo DM
  trees and the GALFORM code.
• Compare the resulting clustering with SDSS
  observations and constrain cosmological
  parameters.
• Particles in 300 Mpc/h box

Benson
20
Harker, Cole Jenkins 2007

• Two grids of models with
• and varying
• Achieved by rescaling particle masses and
  velocities (Zheng et al 2002)

-- Grid 1
-- Grid 2
21
Harker, Cole Jenkins 2007

• For each (scaled) N-body output we have two
  variants of each of three distinct GALFORM
  models.
• Low baryon fraction (Cole et al 2000)
• Superwinds (Baugh et al 2005 aka M)
• AGN-like feedback (C2000hib)
• Each model is adjusted to match the
• observed r-band LF.

22
Select a magnitude limited sample with the same
space density as the best measured SDSS
sample. Compare clustering and determine best
fit.
23

• Comparison of models all having the same .
• Clustering strength primarily dependent on
• I.E. Galaxy bias predicted by the GALFORM model
  is largely independent of model details.

24
The constraint on
25
How Robust is this constraint?

• For this dataset the error on (including
  statistical and estimated systematic
  contributions) is small and comparable to that
  from WMAP estimates.
• The values do not agree, with WMAP3 preferring
  (Spergel et al 2007)
• If the method is robust we should get consistent
  results for datasets with different luminosity
  and colour selections.

26
The constraint on
from b-band 2dFGRS data
Norberg 2002
27

• None of the models produce observed dependence of
  clustering strength on luminosity over the full
  range of the data.

More modelling work required.
28
Conclusions

• Significant improvements in our understanding and
  ability to model many of the physical processes
  involved in galaxy formation have been made in
  recent years.
• They are not yet all incorporated in
  Semi-Analytic models
• Big challenges remain in modelling stellar and
  AGN feedback
• Clustering predictions from galaxy formation
  models can be more predictive and provide more
  information than purely statistical HOD/CLF
  descriptions.
• Comparisons with extensive survey data can place
  interesting constraints on galaxy formation
  models and/or cosmological parameters