How feedback shapes the galaxy stellar mass function

Transcription

1 How feedback shapes the galaxy stellar mass function Eagle, owls and other Gimics Institute for Computational Cosmology Ogden Centre for Fundamental Physics Durham University, UK and University of Antwerp Belgium Dirac 1 1

2 Outline: Motivation Introduction cosmology 101: forming structures cosmology 102: forming galaxies. The need for subgrid physics EAGLE subgrid physics implementation in Gadget star formation, cooling, and feedback (SNe and AGN) Lessons learned from the precursors: Owls and Gimic (How) Do supernova regulate starformation? Parameter selection (tuning) methodology 2 2

3 Galaxy formation Aims: 2 pc How do galaxies form? How do they evolve? Which physical processes operate? x Basic paradigm Dark haloes form Cool(ed) gas forms discs Discs fragment to form stars 20 kpc x Multi-scale/complex/rich problem 200 Mpc 3 3

4 Motivation Simulations follow evolution Physical understanding 4 Which modelling needs improving? 4

5 Multi-scale/complex/rich problem 2 Mpc 40 Mpc 20 kpc 5 5

6 Observed distribution of galaxies Anglo-Australian 2-degree field redshift survey 6 CfA survey 6

7 Millennium simulation + semi-analytical model 7 7

8 Gravitational build-up of dark matter structures is solved problem Nature, Nature,

9 The abundance of haloes as function of mass (and redshift) is known log number / dex / Mpc**3 log halo mass 9 Springel+05,Heitmann+13, 9

10 How do galaxies form inside their halo? White & Rees

11 Number / dex / Mpc**3 log stellar mass 11 Baldry+12 11

12 Abundance matching Too hot to cool? Reed Baldry+12 12

13 Abundance matching Ratio M*/Mhalo log Stellar mass 13 log Halo mass Qi+10,Behroozi+, Tom Leauthaud+ Theuns 13

14 Halo mass function and galaxy luminosity functions have very different shapes Dark halos (const M/L) galaxies Feedback or gastrophysics is key Benson+ 14

15 Galaxy formation Aims: 2 pc How do galaxies form? How do they evolve? Which physical processes operate? x Basic paradigm Dark haloes form Cool(ed) gas forms discs Discs fragment to form stars 20 kpc x Multi-scale/complex/rich problem 200 Mpc 15 15

16 Outline: Introduction cosmology 101: forming structures cosmology 102: forming galaxies. The need for subgrid physics EAGLE subgrid physics implementation in Gadget star formation, cooling, and feedback (SNe and AGN) Lessons learned from the precursors: Owls and Gimic (How) Do supernova regulate starformation? Parameter selection (tuning) methodology 16 16

17 EAGLE project Leiden: Liverpool Chicago HITS: ICC-Durham MPE HITS MPA Dirac 2 17 ICC 17

18 subgrid physics added to Gadget-3 Star formation Galactic winds AGN feedback Z+J(nu) dependent cooling Stellar evolution 18 18

19 Commercial break: your talk will continue in 20 seconds 19

20 1. Element-by-element cooling (and heating) in the presence of UV/X-ray background With ionizing background from gals & AGN Without ionizing background Wiersma et al 08 20

21 2. Star formation implementation (and the origin of the Kennicutt-Schmidt law) Σ SFR Σ n gas (n = 1.4 ± 0.15) Local: same galaxy Global: different galaxies Calzetti et al Kennicutt 98 21

22 Star formation guarantees the simulated galaxies follow the imposed Kennicutt-Schmidt law Schaye 04 22

23 3. Implementation of winds: Springel, Kay+, Scanapiecco+,Oppenheimer+,Kawata+,Tornatore+,Teyssier+ Schaye & Dalla Vecchia 08, 12 23

24 Evidence for galactic winds: At low z: M82 At high z: Pettini et al In absorption 24

25 Supernova feedback leads to expulsion of gas out of galaxy Dark halos (const M/L) galaxies GIMIC simulation 25 25

26 4. Stellar evolution Stellar initial initial mass function (Chabrier) Stellar lifetimes Luminosities (BC models) Stellar yields Type I SNe Type II SNe AGB stars Few+12, Tornatore+07,Oppenheimer +06,Kawata+13,Scannapieco

27 5: AGN implementation Dark matter haloes determine the masses of supermassive black holes C. M. Booth 1 and Joop Schaye 1 1 MN, 2010 BH grows such that it produces a constant amount of feedback Ė = ɛ f ɛ r ṁ accr c 2 = ɛ fɛ r 1 ɛ r ṁ BH c 2, BH accretion rate energy from accreting BH injected into surrounding gas Mass of BH is not set by accretion rate, but by its feedback efficiency 27 Di Tom Matteo+08 Theuns 27

28 28 28

29 29 29

30 Halo mass function and galaxy luminosity functions have very different shapes Dark halos (const M/L) galaxies Feedback or gastrophysics is key 30

31 Subgrid parameters Heating/cooling Epoch of reionisation, UV/X-ray background, selfshielding Star formation KS-parameters, threshold, H2 - Z dependence? Stellar evolution Stellar initial mass function, yields, life-times Supernova feedback Coupling SNe to gas, heating/wind parameters AGN feedback Seed mass, accretion rate, feedback efficiency 31

32 The challenges of theory/numerical simulations: Scales: Box Size = 50 Mpc, bulge size = 1kpc need ( )^3 resolution elements Mean density = 10-7 cm -3, star formation starts at 100 cm density contrast Age of Universe 13.7 Gyr, sound-crossing time bulge:1 Myr require 10 4 steps Physics: Gas cooling follow synthesis of elements, effects of radiation star formation magnetic fields, dust, shielding feedback from stars supernovae, cosmic rays Black-hole formation feedback from black holes Observables! 32 32

33 Outline: Introduction cosmology 101: forming structures cosmology 102: forming galaxies. The need for subgrid physics EAGLE subgrid physics implementation in Gadget star formation, cooling, and feedback (SNe and AGN) Lessons learned from the precursors: Owls and Gimic (How) Do supernova regulate starformation? Parameter selection (tuning) methodology 33 33

34 .. as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns -- the ones we don't know we don't know." 34 34

35 Schaye+10 Galaxies-Intergalactic Medium Interaction Calculation I. Galaxy formation as a function of large-scale environment. Robert A. Crain 1,2, 1,3, Claudio Dalla Vecchia 4, Vincent R. Eke 1, Carlos S. Frenk 1, Adrian Jenkins 1, Scott T. Kay 5, John A. Peacock 6 Frazer R. Pearce 7, Joop Schaye 4, Volker Springel 8, Peter A. Thomas 9, Simon D. M. White 8 & Robert P. C. Wiersma 4 (The Virgo Consortium) Institute for Computational Cosmology, Department of Physics, University of Durham, South Road, Durham, DH1 3LE, UK Crain

36 Suite of simulations: GIMIC/OWLS Galaxy-Intergalactic Medium Interaction Calculation Zoomed simulations of 5 spheres picked from the Millennium Simulation Combine LSS with high numerical resolution 36

37 The physics driving the cosmic star formation history Joop Schaye, 1 Claudio Dalla Vecchia, 1 C. M. Booth, 1 Robert P. C. Wiersma, 1, 2,3 Marcel R. Haas, 1 Serena Bertone, 4 Alan R. Duffy, 1,5 I. G. McCarthy, 6 and Freeke van de Voort 1 OverWhelmingly Large Simulations: periodic boxes (25,100Mpc) with range of physics (50+models) 37 37

38 Name Box Size (Mpc/h) Comment DBLIMFCONTSFV /25 Top-heavy IMF above n H > 30 cm 3, v w = 1618 km s 1 DBLIMFV /25 Top-heavy IMF above n H > 30 cm 3, v w = 1618 km s 1, Σ (0) = M yr 1 kpc 2 DBLIMFCONTSFML14 100/25 Top-heavy IMF above n H > 30 cm 3, η = DBLIMFML14 100/25 Top-heavy IMF above n H > 30 cm 3, η = , Σ (0) = M yr 1 kpc 2 REFERENCE 100/25 EOS1p0 100/25 Isothermal equation of state, particles with n H > 30 cm 3 are instantaneously converted into stars if they are on the equation of state EOS1p67 25 Equation of state p ρ γ, γ = 5/3 IMFSALP 100/25 Salpeter IMF, SF law rescaled MILL 100/25 Millenium cosmology (WMAP1): (Ω m, Ω Λ, Ω b h 2, h, σ 8, n, X He ) = (0.25, 0.75, 0.024, 0.73, 0.9, 1.0, 0.249) NOAGB NOSNIa 100 AGB & SNIa mass & energy transfer off NOHeHEAT 25 No He reheating NOSN 100/25 No SNII winds, no SNIa energy transfer NOSN NOZCOOL 100/25 No SNII winds, no SNIa energy transfer, cooling uses initial (i.e., primordial) abundances NOZCOOL 100/25 Cooling uses initial (i.e., primordial) abundances REIONZ06 25 Redshift reionization = 6 REIONZ12 25 Redshift reionization = 12 SFAMPLx3 25 Σ (0) = M yr 1 kpc 2 SFAMPLx6 25 Σ (0) = M yr 1 kpc 2 SFSLOPE1p75 25 γ KS = 1.75 SFTHRESZ 25 Metallicity-dependent SF threshold SNIaGAUSS 100 Gaussian SNIa delay distribution (efficiency: 2.56 %) WDENS 100/25 Wind mass loading and velocity determined by the local density WML1V /25 η = 1, v w = 848km s 1 WML4 100/25 η = 4 WML8V η = 8, v w = 300km s 1 WPOT 100/25 Momentum driven wind model (scaled with the potential) WPOTNOKICK 100/25 Momentum driven wind model (scaled with the potential) without extra velocity kick = 2 x local velocity dispersion WVCIRC 100/25 Momentum driven wind model (scaled with the resident halo mass) 38 38

39 The basics. What do we want? What can we do? Does it work? What did we learn? Where do we go from here? 39 39

40 Observed NGC

41 Simulated Observed Gadget simulation NGC

42 SKIRT + EAGLE M Baes (Gent) 41 41

43 Hubble Deep Field 42 42

44 Hubble Deep Field Gadget deep field 42 42

45 We have a Hubble sequence! but why? Fraction Disc/total 43 Bulgedominated Discdominated Simulations have > 400 galaxies of MW mass and more, with 10 5 or more particles in them each. 43

46 We have a Hubble sequence! but how do you classify simulated galaxies? bulge-disc decomposition on image? bulge-disc decomposition? colours? 44 44

47 Why do we have a Hubble sequence? Sales+12 Martig+12 key: alignment of angular momentum 45 in forming galaxy 45

48 Tully-Fisher relation log Vmax logv80 McCarthy log M* 46

49 Shapes of rotation curves circular velocity [km s -1 ] low mass observed high mass radius [kpc] 47 McCarthy+12 47

50 Why do simulated galaxies follow a TF-relation? M*/M200 key: correct efficiency, efficient feedback log M* 48 McCarthy+12 48

51 Diskyness of Aquila galaxies Different codes codes 49 Scannapieco+12 49

52 X-ray haloes of MW-like galaxies Gas in haloes of spirals has a reasonable X-ray luminosity and metallicity (long a stumbling block in semi-analytical models) Zx Lx Zgas Crain+10 LK Lx Figure 3. The mass-weighted relation of GIMIC galaxies. Galax- 50 Crain+12 50

53 Why does gas in haloes of spirals have a reasonable X-ray luminosity and metallicity? gas density actual gas profiles isothermal NFW R/R

54 The Universe in (HI) absorption 52 52

55 HI in absorption (log) Number of absorbers Simulation! Lya-forest Lyman-limit Damped systems Altay + (log) Column 53 density of absorber 53

56 Forest LLS DLAs 54 Van de Voort+12 54

57 Star formation is self-regulating by feedback Star formation rate different star formation choices yield same SFR Total mass 55 Haas+12 55

58 different star formation choices yield same SFR ISM fraction.. because galaxies modify their reservoir of star forming gas Total mass 56 56

59 ratio between models... which then affects statistics of high HI DLAs DLA column density 57 Altay+13 57

60 Simulations give correct abundance of HI emitters Abundance of emitters DM simulation observed (Alfalfa) Hydro-simulation HI line width (km/s) 58 Sawala+12 58

61 Why do simulations give correct abundance of HI emitters? low-mass haloes lose baryons and dark matter mass due to 59 feedback Sawala+12 59

62 Mass function Figure 3. LX-LK!""#$%&'()*$&+),-#%.)$/01#2%34$50 Stellar halo DLA 5,-#%.)06#7%&1)/#48'35100)9%" Tully-Fisher,-#%.)0"1&3#$0&$"4&-B4+1)319"1#%36#1)9%" AGN in groups C!5%B.01)3'1)>;)9$03)9%00421)8#$6"0)45)1%&')D;;)' -D *"&) EFGH)0496B%34$5)/$I)J* K;; L)M)I)D; DM * 065 N 60 HI distribution Figure 3. Shows the HI CDDF over ten orders of magnitude in column density. On the low column density end, we have supplemented our grid calculation with Voight Profile fitting of random lines of sight through the box. This allows us to overcome the collapsing of multiple low column density systems into a single higher column density system. One can estimate the point at which the collapsed grid deviates from the true abundance by noting it matches the VP fit at Log NHI One can also see the importance of shielding above the optically thick limit. In addition, our results agree with the survey of Prochaska et al (POW09 on plot) for Lyman Limit Systems. 60

63 Simulation issues: stellar mass function Log Number density of galaxies stars too old specific sfr too low at z=0 too many faint gals Log Stellar mass power-law mass function 61 61

64 Tune feedback to shape galaxy stellar mass function Log Number density of galaxies too many faint gals feedback more efficient at low mass? Log Stellar mass stars too old specific sfr too low at z=0 feedback more efficient at high z? power-law mass function AGN feedback? Di Matteo+05, Crotton+06, Bower

65 in all cases: efficient feedback is key low mass galaxies: SNe in higher mass galaxies: AGN feedback more efficient at high z Does feedback behave like this? Why? 63 63

66 White & Rees 78, Dekel & Silk Martin Tom Stringer Theuns 64

67 65 Theuns MartinTom Stringer 65

68 light-curve fitting HYPERNOVAE AND OTHER BLACK-HOLE- FORMING SUPERNOVAE Ken ichi Nomoto, 1,2 Keiichi Maeda, 1 Paolo A. Mazzali, 2,3 Hideyuki Umeda, 1 Jinsong Deng, 1,2 Koichi Iwamoto, 4 Hypernovae and Other Black-Hole-Forming Supernovae Martin Tom Stringer Theuns 66

69 Inject energy as hot gas? As a wind? Some combination? cooling rate depends strongly on density and temperature 67 67

70 Naive implementation depends directly on numerical scheme cooling rate strongly dependent on density (and hence resolution) how much is the heating temperature? Give all energy to 1 particle-> T= K SPH: give energy to 48 particles -> T= K cooling rates differ by factor 10! 68 68

71 Physics of SN blast wave (cold) ejecta reverse shock: thermally driven shell interior and shell cool: snowplough 69 69

72 Arbitrary re-heating Heat just stellar ejecta Sedov: similarity solution depends only on injected energy But: amount of cooling determines transition to snowplough and hence effectiveness of feedback Numerical overcooling makes problem worse (Creasey+11) 70 70

73 Sod shock similarity solution with cooling Gas temperature Correct answer Numerical result Position along shock tube 71 Creasey+, 11 71

74 30 Jupiter mass resolution Computational volume of 10^9 cells has same mass as single Eagle particle Slice perpendicular to disc plane Creasey+13 FLASH 72 72

75 Wind is a series of overlapping rare faction waves Wind accelerates away from plane due to thermal driving Creasey

76 Feedback efficient log Mass loading Feedback inefficient log Disc surface density 74 Creasey+13 74

77 dense disc: SNe feedback inefficient sparse disc: SNe feedback efficient exponent depends on disc model SN feedback (much) less efficient in big discs Creasey+,

78 Summary: Introduction cosmology 101: forming structures cosmology 102: forming galaxies. The need for subgrid physics EAGLE subgrid physics implementation in Gadget star formation, cooling, and feedback (SNe and AGN) Lessons learned from the precursors: Owls and Gimic (How) Do supernova regulate starformation? Parameter selection (tuning) methodology 76 76

79 Subgrid parameters Heating/cooling Epoch of reionisation, UV/X-ray background, self-shielding Star formation KS-parameters, threshold Stellar evolution Stellar initial mass function, yields, life-times Supernova feedback Coupling SNe to gas, heating/wind parameters AGN feedback Seed mass, accretion rate, feedback efficiency 77

80 log stellar mass Parameter turning: re-heating T, and efficiency-mhalo relation 3. Efficient BH feedback 1: Efficient SN feedback 2. Inefficient SN feedback log halo mass 78 78

81 log stellar mass Parameter turning: re-heating T, and efficiency-mhalo relation 3. Efficient BH feedback 1: Efficient SN feedback 2. Inefficient SN feedback log halo mass

82 Eagle: Stellar mass function 79 79

83 Eagle: Stellar mass function 79 79

84 Eagle: Specific star formation rate 80 80

85 Eagle: Specific star formation rate 80 80

86 Eagle: star formation history 81 81

87 Eagle: star formation history 81 81

88 Eagle: Tully-Fisher relation 82 82

89 Eagle: Tully-Fisher relation 82 82

90 Eagle: M* versus BH-mass 83 83

91 Eagle: M* versus BH-mass 83 83

92 Summary: Introduction cosmology 101: forming structures cosmology 102: forming galaxies. The need for subgrid physics EAGLE subgrid physics implementation in Gadget star formation, cooling, and feedback (SNe and AGN) Lessons learned from the precursors: Owls and Gimic (How) Do supernova regulate starformation? Parameter selection (tuning) methodology 84 Institute for Computational Cosmology Ogden Centre for Fundamental Physics Durham University, UK and University of Antwerp Belgium 84