fine tuning (or not) in structure formation simulations · 2017-07-19 · 4 large scale...

Fine tuning (or not) in structure formation simulations

Debora SijackiDebora Sijacki IoA & KICCIoA & KICC CambridgeCambridge

The Physics of Fine Tuning ConferenceThe Physics of Fine Tuning ConferenceJune 20 2017June 20 2017

2 From the CMB to the large scale structure

The initial conditions are directly observable

cosmic time

Millennium simulationSpringel et al. 2005

Planck 2015 results

3 Cosmological simulations of structure formation

Millennium XXL

40 yrs!

4

Large scale distribution of matter (and galaxies)

Boylan-Kolchin et al. (2009)

Hierarchical growth of dark matter halos

Springel et al. 2006

Angulo et al. 2012

Abundance of the rarestobjects in the Universe

Cosmological simulations of structure formation

5

Frenk et al. 1999

Cosmological simulation of a galaxy cluster (DM only)


6

Frenk et al. 1999



7

Sembolini et al. 2016 (Nifty)



8

Sembolini et al. 2016 (Nifty)



9 But what about all the relevant physics?

radiative cooling and heating processesstar formationsupernovae feedback and stellar windsblack holes and AGN heatingnon-ideal plasma effectsnon-thermal pressure supportmagnetic fields,...

Perseus cluster, Fabian et al.

For ideal inviscid gasEuler equations: conservation laws for mass, momentum and energy

State vector:

Flux vector:

Hydrodynamical simulations

Equation of state:

Uncertainties in hydro solvers of different codes used to simulate galaxy formation

Much more careful code comparisons are needed!Improvements in basic code solvers

Eulerian mesh-based codes

(+ AMR)

Lagrangian particle-based Codes (SPH)

FLASH

ENZO

RAMSES

ART

ATHENA

GADGET

GASOLINEHYDRA

GODUNOVSPH


Non-radiative Cosmological simulation of a galaxy cluster (DM + GAS)

Frenk et al. 1999


Non-radiative Cosmological simulation of a galaxy cluster (DM + GAS)

Frenk et al. 1999

SPH simulations:power-law entropy profiles

GRID-based simulations:cored entropy profiles


Discrepancy between SPH and grid entropy profiles

Springel et al. 2010

Mitchell et al. 2009

FUNDAMENTAL IMPLICATIONS FOR UNDERSTANDING ASTROPHYSICS

Standard SPH simulations flawed (Sijacki et al. 2012)

nIFTy clusters comparison project; Sembolini et al. 2016

15

The Aquila comparison project Scannapieco et al. 2012

9 different codes, 13 runs with the same ICs but different physics“Despite the common halo assembly history, we find large code-to-code variations in the stellar mass, size, morphology and gas content of the galaxy at z=0, due mainly to the different implementations of star formation and feedback.

Physical (& numerical) modeling uncertainties

16



17



NEED FOR LARGE-SCALE COSMOLOGICAL NEED FOR LARGE-SCALE COSMOLOGICAL SIMULATIONS WITH A LARGE, STATISTICAL SIMULATIONS WITH A LARGE, STATISTICAL SAMPLE OF GALAXIESSAMPLE OF GALAXIES

18Current state-of-the-art in cosmological simulationsThe Eagle Project (Schaye et al. 2015) The Horizon AGN project (Dubois et al. 14)

Massive Black II (Khandai et al. 2015)Magneticum (Dolag et al. 2014)

19The Illustris project

Box size = 106.5MpcMin cell size = 48pc3 x 1820^3 dark matter particlesgas cellspassive tracers -> 18 billion8192 cores, 19 MCPUh

Physics:

primordial & metal line cooling+ self-shieldingstellar evolutionstellar feedbackgas recyclingchemical enrichmentblack hole growth & feedback

DM DENSITY with overlaid GAS VELOCITY

20

COSMIC STAR FORMATION RATE DENSITY

Genel, Vogelsberger, Springel, Sijacki, et al., MNRAS, 2014

The Illustris project

21

GALAXY MORPHOLOGIESThe Illustris project

Genel et al., MNRAS, 2014Vogelsberger et al., MNRAS, 2014 see also e.g. EAGLE, HORIZON AGN, MASSIVE BLACK and

MAGNETICUM projects

22

BHs in IllustrisBH MASS – BULGE MASS RELATION

Kormendy & Ho, 2013: best fitcircles: ellipticals; stars: spirals with bulges; squares: pseudo bulges

Sijacki et al, 2015

23

GAS DENSITY MAPS + VORONOI MESHNO REFINEMENT MODERATE AGRESSIVE

Curtis & Sijacki, MNRAS, 2015

How (dramatic) change in resolution affects the physics?

24

VORONOI MESHNO REFINEMENT MODERATE AGRESSIVE


Resolving flows onto BHs

25

SAME BH FEEDBACK AT DIFFERENT RESOLUTIONS LEADS TOVERY DIFFERENT OUTFLOW

hot outflow

cold disk inflow


Resolving flows onto BHs

26Powerful QSO outflow in a massive disk galaxy at z ~ 5

Curtis & Sijacki, MNRAS Letter 2015

Carniani et al. 2013 ALMA data of a QSO/SMG at z = 4.7

SAME BH FEEDBACK AT DIFFERENT RESOLUTIONS LEADS TO VERY DIFFERENT GALAXY MORPHOLOGY

(have we understood morphological evolution of galaxies and quenching?)

27

Smith, Sijacki & Shen, 2017, in prep.

How (dramatic) change in resolution affects the physics?

2000M⊙ 200M⊙

20M⊙

SAME SUPERNOVA FEEDBACK AT DIFFERENTRESOLUTIONS LEADS TO COMPLETELYDIFFERENT GALAXY MORPHOLOGIES

28

ConclusionsLessons learned:

1. Accuracy of hydro solvers still needs to be improved No “fine tuning” of baryonic physics to “cover up” for errors in hydro →

solvers

2. Sub-grid physics uncertainties still very large! Free parameters of sub-grid models “fine tuned” for specific observables→ Other results are in principle predictions, but….→

a) Different set of baryonic physics can lead to similar z = 0 results(redshift evolution is different) DEGENERACIES→

b) Same baryonic physics at different resolutions may lead to different results WHAT DO WE LEARN ABOUT PHYSICS?→

3. Next generation sub-grid models for SF and BH physics needed in large cosmological simulations

spatial resolution requirements daunting→ more cross-talk with “small-scale” community →

fine tuning (or not) in structure formation simulations · 2017-07-19 · 4 large scale...

Documents