Title: Formation and structure of dark matter halos in N-body and SPH simulations
1Formation and structure of dark matter halos in
N-body and SPH simulations
Sino-France Workshop Dark Universe Sep 2005_at_
CPPM, France
- Wei-Peng Lin
- The Partner Group of MPI for Astrophysics,
Shanghai Astronomical Observatory, CAS, P.R.China
2Introduction of our group
- The Partner Group of Max-Planck-Institute for
Astrophysics, Shanghai Astronomical Observatory
was founded in year 2000 through the exchange
program between CAS and Max-Planck Society (MPG).
The goal of establishing this group is to create
an active research group which will play an
important role in promoting cosmological research
in China, in enhancing the existing exchanges
between Chinese and German astronomers, in
training outstanding young cosmologists. - The group is carrying out research on numerical
simulations of galaxy formation and on
statistical analysis of large scale structures. - Group Head Dr Yi-Peng Jing. We now have 6
faculty members, 8 graduate students and several
visitors.
3Our Interests
- Dark Energy
- Large Scale Structure, statistic, 3PCF, PVD
- Galaxy Formation, semi-analytical model, HOD
- Weak lensing, cosmic shear, spin-spin correl.
- Strong lensing, giant arc
- Sunyav-Zedovich effect, x-ray
- Simulations, N-body, SPH
- Halo formation, structure, angular momentum
- Quasar Absorption Line Systems
4Contents
- Part I The formation-time distribution of halos
in N-body simulations - Part II The structure of halos in N-body
simulations - Part III The structure of halos in N-body/SPH
simulations
5Why and what to do?
Part I The formation-time distribution of halos
(Lin, Jing Lin 2003, MNRAS)
- The blue-color problems of dwarf galaxies
- Small halo forms earlier than large one, thus
stars form earlier and they are old,
metal-rich, red! - Theories of galaxy formation can hardly solve
this problem!
6F.C.van den Bosch 2002 (MNRAS 332, 456)
red
?
blue
The impact of cooling and feed back on disc
galaxies
7Questions
- How many fraction of dwarf galaxies form at
low redshift? - From tidal debris or just newly form out of
over-dense regions? -
- Can theories predicted consistent results with
N-body simulations?
8Press-Schechter formalism
What theory?
- Extended PS theory,e.g.,Lacey Cole (1993)
- simple linear growth of over-density field.
- simple threshold for over-dense regions to
collapse and form virial objects. - predict the formation of haloes, mass
function, conditional mass function, halo
formation redshift, halo survival time, halo
merger rate, etc.
9PS formalism
- Has been used to construct galaxy Merger Trees
in semi-analytical models of galaxy formation
cooling, star formation, feedback, yield, outflow
(super-wind), etc. - Kind of Successful!
- Shortcoming no environmental effect, no
interaction between different scales, non-linear
evolution of structures
10The formation-redshift distribution of dark
matter haloes
- Example EPS approach by Lacey Cole (1993)
- A parent halo with mass of M2, if define its
formation time as the epoch when its largest
progenitor have half of the mass, the conditional
probability is -
Conditional mass function
11Let t2t0, M2M0,and
12So that we can derive the halo formation time
tf ?zf
13(No Transcript)
14zf
15M1.66x1013 M?/h
1650
17The improvement of the excursion set approaches
- Ellipsoidal collapseSheth Tormen 1999 Sheth,
Mo Tormen 2001, Sheth Tormen 2002 - Non-spherical collapse boundary (Chiueh Lee
2001, Lin, Chiueh Lee 2002) 6-D random walks
18EC or NCB models
- For EC/NCB models, the threshold is higher for
smaller haloes. Not a constant! - The moving barrier for EC model
- The unconditional mass function and conditional
mass function are modified
19Black EPS Red EC Blue NCB
20- previous comparison with simulationsunconditiona
l/conditional mass function,formation time(mainly
for high mass haloes,because of low resolution)
21F.C.van den Bosch 2002
22The N-body simulations
- ?CDM?m0.3, ?? 0.7
- Box A 25 h-1Mpc (small haloes), B 100
h-1Mpc(sub-M haloes), C 300 h-1Mpc(Large haloes) - CDM power spectrum ?0.2,?80.9/1.0/1.0
- Total Number of Particles A/B 2563, C 5123
- Mass of particles A 7.7x107h-1M?, B
4.9x109h-1M?, C 1.67x1010 h-1M? - P3Msoftening 2.5 h-1kpc
- Time-steps/outputs A 5000/165 B 600/30
C1200/36
23Definitions of halo and formation redshift
- FOF group method to select haloes The points
with min-potential as halo center spherical
virial halo assumption - The formation redshiftwhen the largest
progenitor for the first time has half of the
parent-halos mass, the redshift at this epoch is
defined as the formation redshift of the parent
halo.
24Methods
- Particle tracing methodsselect a parent halo,
find its member particles, trace these particle
back at the last output step and check if they
inhabit in some progenitor haloes, calculate the
fraction of member particles inside each
progenitor halo, and so on - Calculate the redshift distribution possibility
of the formation redshifts and compare with
theory predictions
2525 Mpc/h
10-3 to10-2 M
Green Simulations Black EPS Red EC Cyan
NCB
2 realizations
26Results for small mass haloes
- In contrast to the anticipations, the formation
redshifts of small haloes are averagely larger
than the theoretical predictions by EPS - At low redshifts, the prediction by ellipsoidal
collapse (EC) are consistent with simulated
results at high redshifts, the EPS prediction is
better, while EC/Non-spherical collapse boundary
model (NCB) predict too large fraction of haloes
formed. - The simulated profile of formation redshift
distribution is narrow but higher than
prediction, and shift to higher redshift.
27More results
- We found 1015 small haloes once sink
- into some big halo within its half virial
radius and then come out. - These strong interaction may trigger star
bursts and form lots of young stars (thus make
the color blue), however, the physics for gas
procedures is complex.
28Discussion
- If simulation results are believable, the
blue-color problem of dwarf galaxies can not be
solved directly (formation shifts to higher
redshift). - other ways to solve the problems
- 1. Even if the fraction of haloes formed at
low redshifts is small, however they posses
enough number of blue dwarf galaxies in
observations. - 2. When small haloes formed at high redshifts,
they are pre-heated, gas temperature is too high
to be cooled down to form stars, i.e. the star
formation was delayed.
29Discussion..
- 3. Gas in small haloes was stripped off at high
redshifts, thus can not form large amount of
stars They accrete gas again at some lower
redshift to form stars (so that the stars are
young, mental poor and blue). - 4. Other possibilities, for example
environmental effects, star formation by galaxy
interaction, other unknown physics, etc.
30100 Mpc/h
0.03 to 0.3 M
Sub-M haloes
3 realizations
31300 h-1Mpc 5123 particles
0.17 to 8.74 M
1 realizations
32More to be done
- The improvement of conditional mass function to
lower mass end (in progress by using simulation
with 10243 particles). - The survival probability of haloes.
- The dynamical evolution of haloes.
33Part II The structure of haloes in N-body
simulations
NFW density profile
34Example of NFW fitting
35Redshift evolution of Cvir From top to bottom
z0, 0.5, 1.0, 2.0
36Black1012M?, slop -0.99 0.08 Red1013M? ,
slop -0.94 0.08 Green1014M?, slop -0.90 0.08
37Black1012M?, slop -0.91 0.07 Red1013M? ,
slop -0.88 0.07 Green1014M?, slop -0.82
0.07 Blue Curve progenitor
38Zhao et al.(2003)found there is close correlation
between rs and Ms for main progenitor haloes The
same relation was found for all haloes solid
(z0) 1.96 dot (z1.0) 1.93 dash
(z2.0) 1.72 Here rs is in physical scale
LCDM
39Zhao et al. 2003
This relation has been used to predict halo
concentration accurately
40The relation of halo structure and formation epoch
- As a halo formed earlier, its environmental mass
density is higher, therefore its core is denser
and more compact, hence with bigger concentration
factor cvir - cvir ?(1zf)0.6,the dependence is much more
stronger than that on halo mass(?M-0.1) its
scatter span reflects the span of halo
formation-time distribution. - Other reasons of scatter of cvirdeviation from
NFW, fitting errors, sub-structure,
non-equilibrium halo, halo ellipsoidal halo, etc.
41Cvir ?(1zf)0.603Mvir-0.065
Dependence on formation redshift Formed earlier,
when mass density is higher, halo core is more
compact Dependence on halo mass Larger halo has
averagely smaller formation-redshift
42Part III The halo structure in N-body/SPH
simulations
- The dynamical interaction between baryonic matter
and DM - Would the relatively small fraction of gas has
impact on the distribution of dark matter in
halo? (adiabatic/with cooling/with star
formation) - Who will win, dynamical friction of big galaxy
clumps sinking in to halo center or adiabatic
compress effect? - Two body heating, as artificial fact in
simulations?
43The problems of the central distribution of
matter of dark haloes are hot topic
- The central density profiles have cusps in
(CDM) N-body simulations, while observations of
galaxies and clusters show at least some objects
have shallow density profiles and even have
core-like structures. - SIMP, WIMP, WDM?
-
44- Why to study the density profiles of clusters of
galaxies? No strong effects from complex star
processes, relatively clean and simple in some
sense. - So far, people have just begun to study the
structures of haloes by simulations with gas and
to investigate dynamical interaction of dark
matter with gases components. - Only 16 percents of mass in baryons (WMAP
results) Weak interaction between DM and baryon
particles
45- Observations of galaxy clusters(Sand et al.
2002,2003)concluded - in the central part of clusters of galaxies, the
density profiles are more flat than NFW profile,
i.e., ? ? r-0.5. - However, Bartelmann Meneghetti 2004, Dalal
Keeton 2004 weaken this constrain by taking into
account the non-spherical structures of haloes.
46- Assuming a NFW halo and simulating the infall of
galaxies, El-Zant et al.(2002, 2004) found the
dynamical friction on the galaxies can transfer
orbital energy to and heat up DM in the central
part of the halo, thus make shallow density
profile. - Counter effect adiabatic compression from
baryonic matter. (Blumenthal et al. 1986, Mao, Mo
White 1998, Rasia et al. 2004) the adiabatic
contraction of baryon can transfer energy from DM
to gas therefore make the density profile
steeper. - So, we use hydro-dynamical N-body simulations to
find whether the dark matter profile can be
affected by gaseous components.
47(No Transcript)
48(No Transcript)
49(No Transcript)
50Our simulations
- A set of simulations one is adiabatic, one with
weak cooling and another with strong cooling.
Each have 1283 DM and 1283 gas particles. - A pure DM simulation provides control sample.
- All realizations have the same initial condition.
- We selected the first 12 biggest haloes
(cluster-size). - An additional high-resolution simulation with
2563 DM and 2563 gas particles using Gadget
(Springel, Yoshida White 2001) to study the
adiabatic case.
511283 P3M
Mgas2.4E9M? Mdm2.2E10M?
522563 Gadget Tree-code
Mgas3.0E8M? Mdm2.8E9M?
53(No Transcript)
54(No Transcript)
55Fitting from 2 virial radii
1283 simul.
Over-cooling?
2563 simul.
56- results
- We find that adiabatic compression can make the
DM density profile steeper even if the dynamical
friction effect has been taken into account in
the simulations. - In simulations with cooling, DM density profiles
become even steeper than in adiabatic case. - The additional simulation using Gadget and with
2563 DM and 2563 gas particles confirm our result
with low-resolution .
57Implications
- If our results are correct, the overall density
profiles of haloes remains NFW form but with
larger concentration factors and the DM-only
profiles become even steeper. This may have
effects in the observations of gravitational
lensing.
58Discussions
- Why in El-Zant et al (2002,2004) simulations,
they got flat density profile? - The possible reasons are
- a very strong working assumption is that
there has been already a NFW halo where galaxy
clumps spiral in. In fact, the hierarchical
growth of halo by merger and accretion were
omitted - in their simulation, adiabatic compression
effect and tidal stripping were not taken into
account.
59Adiabatic Compression
Dynamical Friction
Winner in our simulations
60Discussions
- We need simulations with higher resolution to
confirm our results. There could be some
resolution effects, for example, gas particle are
much lighter then DM particles in the
simulations, softening length is too large, etc. - Over-cooling problems thermal feedback,thermal
conduction, AGN, particle annihilation, etc. - Dynamical friction and tidal stripping on
substructures and/or luminous systems.
61Discussions
- Two-body heating (Steinmetz White 1997)
- gas particle are much lighter than DM
particles in the simulations
62Yoshikawa, Jing Suto 2000
63Works in progress
- Simulations with 2563 gas particles and 5123 DM
particles. Particle mass of gas and DM will be
almost comparable. Simulation done! - Using simulations with star formation and
feedback. With 5123 DM and the same number of
gas particles. Simulation done! - Re-simulations of some regions with much
higher-mass and force resolutions. Outside, DM
only. In preparation
64(No Transcript)
65Parallel simulations in Shanghai Supercomputer
Center
- SSC 2048 Processors (512 nodes, Myrinet), once
ranked among Top 10 (we were permitted to use 512
CPUs) - Simulations done so far
- 10243 DM, 5123 DM 2563 GAS (adiabatic) , 5123
DM 5123 GAS (adiabatic/star formation), all with
the same IC, 100 h-1Mpc - simulations with DE
- Simulations in preparation
- re-simulations, 10243 DM 10243 GAS
(adiabatic/star formation, 300 h-1Mpc)
66- Thanks for patience!
- Welcome to visit the Partner Group of MPA in
Shanghai Observatory - and welcome for collaborations!