박창범 ( 고등과학원 )

& 김주한 ( 경희대학교 ), J. R. Gott (Princeton, USA), J. Dubinski (CITA, Canada)

한국계산과학공학회 창립학술대회 2009. 9. 12

Cosmological N-Body Simulation of Cosmic Structure Formation

Simulation of Cosmic Structure Formation

1. Purpose of cosmological simulations

2. Structure

3. Requirements

4. Recent achievement

History of the Universe

우주 조망도

관측 가능한 시공간

대폭발 특이면빛 분리면

암흑시대

재이온화시기

현재

high-z시대

Comparison between simulated universes with the real universe

Cosmological model

Initial linear density fluctuation

Non-linear gravitational evolutionGalaxy biasing

Redshift space distortionPast light cone effectsSurvey characteristics

Simulated sample of 'galaxies'

The real universe

Observed sample of galaxies

Statistical test of adopted models

Calibration of systematic effects

Prediction of new phenomena

Simulation of Cosmic Structure Formation

1. Purpose

2. Structure of cosmological simulations

3. Requirements

4. Recent achievements

Execution of Cosmological Simulations

0-1. Simulation code

N-body gravity code: PM, P3M, Tree, PM-Tree codes

Hydro code: 1. particle - SPH

2. mesh (AMR) - ENZO, FLASH

0-2. Dynamic ranges

Mass range: simulation particle, collapsed object, total # of particles

Spatial range: simulation box size, force resolution

Structure of Cosmological Simulations

1. Initial conditions

Cosmological model: power spectrum, high-order correlations

Generation of initial conditions on a simulation mesh

2. Evolution & Intermediate analyses

Snapshot, past-light cone, collapsed object data at predetermined epochs

3. Post analyses

Merger tree construction

Assignment of physical values to collapsed objects

Comparison with observations

Growth of Structures from initial Density Fluctuations

13.7b

11.8b

7.7b tb=0

Simulation of Cosmic Structure Formation

1. Purpose

2. Structure

3. Requirements of cosmological simulations

4. Recent achievements

Dream Reality 1 Realty 2 Reality 3

cosmic structures to simulate

stars ~ horizon stars ~ galaxy* subgalactic objects ~ large-scale structure

galaxy ~ cosmological scale

Spatial scales to resolve

0.1pc~104Mpc

1017~1028 cm

0.1pc~100kpc

1017~1023cm

few kpc ~ few 100Mpc

1022~1027cm

few 10kpc~3000Mpc

1023~1028cm

Spatial dynamic range : initial conditions / force resolution**

1011 106 / 105 105 / 104 105 / 104

Dynamic range in mass (# of simulation particles)

1033 1015 1012 1012

(memory~40TB)

* Galaxy, a city of stars, is the building block of the universe

** Force resolution 10 times higher than the mean particle separation assumed

Requirements for cosmological simulations

Simulation of Cosmic Structure Formation

1. Purpose

2. Structure

3. Requirements

4. Recent achievements

Date particles scales machine

2004. 8 20483 0.05~1024 Mpc

0.275~5632 Mpc

IBM SP3 at KISTI, 128 CPUs, 0.9 TB memory

2008. 8* 41203 0.16~6592 Mpc Sun Blade at KISTI, 1648 CPUs, 2.4TB memory

planned** 100003 0.13~14000 Mpc 40TB memory

Major simulations of the KIAS astrophysical simulation group

: LCDM model, PMTree N-body code, galaxy ~ cosmological scale

* Horizon Run

** Horizon Run-II, a trillion particle simulation

( 김주한 & 박창범 2004)

Simulation of bright galaxies and large-scale structure --> SDSS Main galaxy

survey

T H E H O R I Z O N R U N

Kim, Park, Gott & Dubinski (2009)

http://astro.kias.re.kr/Horizon_Run

Here

Now

Universe seen along the past light cone

Deco

uplin

g

Ep

och

Dark

A

ges

The First

Obje

cts

HI +

+

He

p +

e- + +

He

Reionization Epoch

Structure Formation & Evolution

Acceleration (Dark Energy

Dominated)

Deceleration (Matter

Dominated)

Inflatio

n

SDSS Main

( 김주한 , 박창범 , Gott & Dubinski 2009)

Simulation of very bright galaxies and large-scale structure; the 1st simulation out to the horizon --> SDSS-III Luminous Red Galaxy survey

Evolution of the

number of simulatio

n particles

Horizon Run

Millennium Run

Horizon Run-II

Millennium Run

Horizon Run

Horizon Run-II

Evolution of the spatial

dynamic range

box size

resolution scale

Comparison between simulated universes with the real universe

Cosmological model

Initial linear density fluctuation

Non-linear gravitational evolutionGalaxy biasing

Redshift space distortionPast light cone effectsSurvey characteristics

Simulated sample of 'galaxies'

The real universe

Observed sample of galaxies

Statistical test of adopted models

Calibration of systematic effects

Prediction of new phenomena

KSG-VAGC DR7 sample

A tour of the real universe : SDSS galaxies

Final SDSS DR7 Main Galaxy Sample (2008)

[Choi et al. 2009]

The Sloan Great Wall (Gott et al. 2005)

The CfA Great Wall (Geller and Huchra 1989)

The Cosmic Runner (Park et al. 2005)

SDSS2006

CfA1986

Voids (blue - 7% low), filaments/clusters (red - 7% high) => Sponge !! (Gott et al. 2008)

SDSS2006

A mock survey of massive halos out to z=0.6 simulating the SDSS-III Luminosity Red Galaxy Survey

that will finish in 2014.

[the Horizon Run (Kim et al. 2009)]

Summary

1. The cosmological N-body gravity simulation is limited by memory, and hydrodynamics simulation is limited by both memory and speed.

2. In the near future a trillion particle N-body gravity simulation will be made for the first time.

3. The low-resolution cosmological simulation is now reaching the horizon scale.

4. The high-R cosmological simulations are increasing the box size, and the low-R simulations are increasing the force resolution toward smaller scales.

5. Cosmological simulations are indispensable for understanding the observed universe, guiding new surveys, and predicting new scientific findings.