Semi-analytics and mock catalogues

1
Semi-analytics and mock catalogues as tools to
observe ideas

1. Semi-analytic modelling of galaxy formation The
   long way from first principles to the
   distribution of galaxy properties

II. Mocking the Universe
Construction, limitations and examples of mock
catalogues
2
Semi-analytic modelling of galaxy formation
Y a des progrès à faire du côté de la
gastrophysique F. R. Bouchet
Jérémy Blaizot (MPA)
3
Large-scale surveys
To what extent are galaxies tracers of
DM Physical sampling (bias) observational
selection
Colless et al., 2001
4
From low to high redshifts
SAMs and mocks provide a means to connect
populations of galaxies selected in different
ways at different redshifts (e.g. LBGs/BXs/etc.
from Steidels group)
Driver et al. 1998
5
Observations at different wavelengths
SAMs and mocks help establish the connection
between populations of galaxies selected at
different wavelengths
HST
ISO
Sources 15mm
Sources 6.7mm
The ISO-HDF Project (Mann et al.)
6
Last but not least

• On top of these motivations, there is the
  increasing need to produce realistic catalogues
  that can be used
• to prepare forthcoming observations
• to validate analysis techniques used on real
  obs.
• to check/understand biases uncertainties (e.g.
  cosmic variance)

7
Structure formation
Dark matter hierarchical structure formation
Given initial conditions and a cosmological
model, we know how to describe the formation of
dark matter structures with N-body simulations.
8
Structure formation N-body simulations
9
It all happens in haloes
Semi-analytics neglect the impact of baryons on
the formation of large scale structures, and can
thus be described a posteriori within the
hierarchy of haloes and their evolution. The
hybrid approach exploits our best way to describe
structure formation N-body simulations.
10
Galaxy formation relevant processes
Star formation (threshold, efficiency, IMF, )
Cooling (metallicity, structure, )
AGNs (BH growth, feedback, )
Dust (formation, distribution, heating cooling,
)
Galaxy formation evolution
Galaxy interactions (morphological
transformations, starbursts, intracluster stars,
Winds (IGM heating, enrichment, SN feedback, etc)
Stellar evolution (spectro-photometric evolution,
yields, SN I/II rates,)
11
Layout
I. Implementation of the hybrid approach
II. Limitations of SAMs
III. Example Brightest cluster galaxies
12
Layout
I. Implementation of the hybrid approach
II. Limitations of SAMs
III. Example Brightest cluster galaxies
13
From particles to haloes
From particles to  haloes 
Halo identification (FOF) and characterisation
(Mass, Spin, Energies, etc.)
14
(Sub-)Halo finders
Identification of sub-structures from the density
field (only)
SUBFIND (Springel et al. 2001) ADAPTAHOP
(Aubert et al. 2004)
15
From particles to halo merger trees
From particles to  haloes 
Halo identification (FOF) and characterisation
(Mass, Spin, Energies, etc.)
From density evolution to merger trees
Construction of a full merger tree (mergers,
accretion, fragmentation, evaporation)
16
Example of a Clusters tree
17
Semi-analytics
18
Cooling (source term)
Assume hydrostatic equilibrium ( isothermal)
temperature and density profile.
Cooling time (function of radius)
White Rees (1978)
Binney (1977), Silk (1977)
Mass of gas that actually cools
Free-fall radius
Note cooling rates are sensitive to the heavy
elements content of the gas (Z).
19
Cooling (source term)
Transition at 1012Msun (with some redshift
dependency)
Kravtsov et al.
20
Star formation feedbacks
Star formation rate (highl redshifts ?)
SSFR
Supernovae feedback (highly uncertain)
or not
Kennicutt (1998)
Metal enrichment (hyper-highly uncertain)
Sgas
Fixed yield ? Instantaneous recycling ?
Instantaneous mixing ?
21
Galaxy mergers - galaxy morphologies
Galaxies spiral down haloes potential wells due
to dynamical friction. When they reach the center
they merge with the central galaxy.
Bulge formation
Disrupted disk (m1 m2)
100
Major mergers
Fraction of progenitor disk mass tranfered to
descendents bulge.
50
Minor mergers
No bulge (m1 gtgt m2)
0
m2 / m1
1
0
22
Spectral energy distributions
Final SED is the sum of SEDs of stars formed all
along the hierarchical history

• stellar evolutionary tracks (Padova tracks,
  Genova, a-enhancement ? )
• stellar spectra library
• IMF (Chabrier, Kennicutt, Salpeter )
• - Extinction/emission by dust.

23
THE result
spirals
ellipticals
Stellar mass
Gasstars
SFR
24
THE result
25
Frequently asked questions

• Do you resolve galaxies ?
• NO ! Galaxies in a SAM are vectors Mstar,
  etc,
• How many parameters do you fit ?
• I wish I knew Lucky we dont fit
• What do you get that you didnt put in by hand ?
• A quantitative estimate of the coupled evolution
  of a set of processes (each put by hand) within
  a complex system of boundary conditions (merger
  trees).

26
SAM Cinema
Semi-analytic galaxies
D.M. density
John Helly (Durham http//www.virgo.dur.ac.uk/)
27
Layout
I. Implementation of the hybrid approach
II. Limitations of SAMs
III. Example Brightest cluster galaxies
28
Chosing a simulation
Trade-off between - Mass resolution (ability
to describe history faint objects) - Volume
(ability to describe rare objects)
29
Effects of mass resolution (1/3)

• completeness limit galaxies in small mass haloes
  are missing.

Halo mass resolution
Galics 1 1.6 1011Msun
Galics 3 2.8 109Msun
30
Effects of mass resolution (2/3)

• completeness limit galaxies in small mass haloes
  are missing.

1010 MO
1011 MO

• redshift limit beyond zlim, there are no
  resolved haloes.

1012 MO
1013 MO
31
Effects of mass resolution (3/3)

• completeness limit galaxies in small mass haloes
  are missing.

• redshift limit beyond zlim, there are no
  resolved haloes.

• history resolution properties of new galaxies
  are not realistic

32
Other limitations

• Each step of the post-processing involve
  approximations that do not disapear even if the
  results fit the observations !
• halo finder N-body describes exactly the
  (non-linear) evolution of a density field
  haloes are not so exact
• halo merger trees following sub-structures is
  a delicate business
• galaxy mergers largely unknown (both when
  how)
• metals production, transport
• SEDs if you dont believe in BC03 or
  Chabriers IMF

33
Layout
I. Implementation of the hybrid approach
II. Limitations of SAMs
III. Example Brightest cluster galaxies
34
Brightest Cluster Galaxies (BCGs)
Brightest (and central) galaxies of the most
massive haloes of the Universe (typically Mhalo
1015 Msun)
Selection of clusters (e.g. with LX), so far
possible up to z 1
BCGs are the galaxies with the richest merger
trees
35
Brightest Cluster Galaxies (BCGs)
De Lucia Blaizot (2006)
36
Brightest Cluster Galaxies (BCGs)
De Lucia Blaizot (2006)
37
Brightest Cluster Galaxies (BCGs)
2 x 2 Mpc (comoving)
38
Brightest Cluster Galaxies (BCGs)
Mass growth 3 since z1 (along the main
branch)
Infered mass growth 3 since z1 (total)
High-z BCGs are do not end up in local BCGs
39
Brightest Cluster Galaxies (BCGs)
The monolithic approximation (isolated evolution
or one-branch tree) is wrong in general and
should not be used to try to assess evolutionary
links between galaxy populations observed at
different redshifts.
The proper way to go is to reproduce
observational selections on the model galaxies,
using mock catalogues, and then go back to the
model to understand the (hierarchical) links
between galaxies selected in different ways.
40
SAMs mock catalogues for interpreting
observations
Jérémy Blaizot (MPA)
41
Selections
Colless et al., 2001
42
Selections, selections
SAMs Mocks help establish the connection
between populations of galaxies selected at
different wavelengths
HST
ISO
Sources 15mm
Sources 6.7mm
The ISO-HDF Project (Mann et al.)
43
Selections, selections, hierarchical evolution
SAMs and mocks provide a means to connect
(statistically) populations of galaxies selected
in different ways at different redshifts (e.g.
LBGs/BXs/etc. from Steidels group)
44
General framework
Observations
Theoretical Framework
Surveys Galaxy samples _at_ diff. z l
Physical model (ingredients Recipes)
Hybrid implementation Some comparison to obs.
Mock Catalogues
45
Layout
I. Construction of mock catalogues
II. Limitations of mock catalogues
III. Example 2 Lyman Break Galaxies
IV. Just do it
46
Layout
I. Construction of mock catalogues
II. Limitations of mock catalogues
III. Example 2 Lyman Break Galaxies
IV. Just do it
47
Inputs for mock catalogues
Series of napshots at zsnap zi (i 1, , N)

• Observer-frame (zsnap) absolute magnitudes and
  their derivative
• positions / velocities
• size(s), inclination
• IDs

48
Tiling boxes basics
49
Tiling boxes replications
50
Tiling boxes random tiling
Random tiling
dec.
Supresses replication effects and some of the
signal (see later)
r.a.
51
Example 1 mock SDSS stripe
21 lt r lt 22
20 lt r lt 21
19 lt r lt 20
18 lt r lt 19
52
Pre-observation maps
discs
Scale to galaxy size
Scale to galaxy size
Rotate grid for orientation
Projection on the final grid
Spheroids
53
Example 2 mock V-band deep field
HDF
Johnson V filter
6 arcmin
3 arcmin
54
SkyMaker (E. Bertin)
55
Layout
I. Construction of mock catalogues
II. Limitations of mock catalogues
III. Example 2 Lyman Break Galaxies
IV. Just do it
56
Correlation functions
Excess probability of finding a pair of galaxies
at a given separation, relative to a random
distribution.
Data-Data
Random-Random
Field-to-field variance (in counts) average of
x over field-size
57
Random tiling bias
Random pairs
58
Random tiling bias
100 Mpc/h
R.T. bias present around r0, but well
understood. Finite volume effects (integral
constraint) comes in at larger scales
12 Mpc/h
Analytic estimate
59
Finite-volume effects correlation function
100 Mpc/h
A simulation does not contain fluctuations
(clustering) on scales larger than Lbox
20 Mpc/h
60
Finite-volume effect cosmic variance
Simulation volume should be gtgt light-cone volume
61
Layout
I. Construction of mock catalogues
II. Limitations of mock catalogues
III. Example 2 Lyman Break Galaxies
IV. Just do it
62
LBG selection (at z3)
(e.g.) Adelberger et al. (1998)
Pure photometric selection good test for the
model and mock-catalogue methodology
Blaizot et al. (2004)
63
LBG counts and cosmic variance
Clustering of LBGs dominates cosmic variance up
to (at least) 1 deg.
64
LBGs physical properties
Cest  ca va  !
Steidels team
30 of LBGs intense SF is triggered by mergers
65
Link to local galaxies (1/2)
The Epoch of Galaxy Formation, Baugh et al. 1998
LBGs
z
z
66
Link to local galaxies (2/2)
z 3
z 0
77 of z3 LBGs end up in E or S0 at z 0 35 of
local E or S0 have a LBG progenitor at z 3
E S0 with LBG prog. at z3
LBGs at z3
Other E S0
Sp with LBG prog. at z3
67
Layout
I. Construction of mock catalogues
II. Limitations of mock catalogues
III. Example 2 Lyman Break Galaxies
IV. Just do it
68
Online stuff
69
Online stuff
http//www.g-vo.org/Millennium/
(Gerard Lemson)