Measuring galaxy environments with group finders: Methods & Consequences

Transcription

1 Measuring galaxy environments with group finders: Methods & Consequences Diego STALDER Doc student INPE, São Jose dos Campos, Brazil & IAP Marina TREVISAN Postdoc, IAP Manuel DUARTE former IAP Doc student Euge DÍAZ-GIMÉNEZ IATE, Cordoba, Argentina

2 utline Motivations of Group Finders Review of Group Finders FoF, matched-filter, Voronoi, Yang, MAGGIE Do Group Finders give results? surface density & LS velocity dispersion profiles environmental trends Are Group Finders so bad that they blur or bias our knowledge of environmental effects? Do group properties strongly depend on Ωm? 2

3 Why are group finders useful? Study individual groups Statistics of environmental effects on galaxies Galaxy morphology, structure, kinematics, gas & dust content, luminosity & stellar mass functions, fertility, chemistry, = f (global environment, local envt, large-scale envt, redshift) Cosmological tools evolution of group/cluster mass function velocity fields around groups 3

4 Why use ptical group finders? X-rays suffer least from projection effects X-rays are expensive! LX T 3 M 2 Difficult to blindly detect low-mass groups SZ low sensitivity Y M T M 5/3 Lensing least affected by systematics Lensing is ~ cheap! Difficult to blindly detect low-mass groups ptical group finders = cheapest way to blindly detect groups! 4

5 What should group finders provide? Positions (centers) Mean redshifts Group luminosities & stellar masses Group total masses (Global Environment) Galaxy positions and line-of-sight velocities in group (Local Environment) Galaxy membership (Probabilistic?) 5

6 Review of Group Finders this talk: ~ limited to spectroscopic surveys! 6

7 How to extract real-space groups from redshift-space data? real space GM, Biviano & Murante 10 Moore, Frenk & White 93 redshift distortions vlos / v200 +κ σv κ σv cz = H0D + vpec redshift space Dlos / r200 = overdensity / critical density

8 Group finders incomplete list! for spectroscopic galaxy samples Frequentist Friends-of-Friends Voronoi Tessellation Dendrograms Huchra & Geller 82 Marinoni+02 Tully 87 Prior-based Matched Filter Yang MAGGIE Kepner+99 Yang, Mo, van den Bosch +05, 07 Duarte & Mamon 15 8

9 Friends of Friends (FoF) survey edge survey edge 9

10 Friends of Friends (FoF) transverse linking length line-of-sight linking length Dimensionless linking lengths in terms of mean nearest neighbor separation: b = LL/ n(z) 1/3 10

11 ptimal FoF linking lengths Duarte & Mamon 14 mean transverse link for =200 & Ωm=0.25 = 0.07 max (95% c.l.) transverse link mean line-of-sight link b k b? = vmax v v v vir r b = ' 11 for vmax/σv=1.65 (95%) 11

12 Voronoi tessellation Background regions: expect a=area/ Area follows f(a) a 3 e 4a Kiang+66 Area threshold f(a) a 2.89 e 3.89a Soares-Santos+11 groups/clusters = connected regions w Area below threshold FoF on low area cells! 1st application to z-space data: Marinoni+02 Ramella+01 12

13 Dendrograms Materne 78 link pairs by values of max(li,lj)/rij 3 Tully 87 13

14 Matched filter Postman+96 (2D) Kepner+99 (2D,2+½D,3D) Convolve data with filter using position (prior on surface density profile) redshift (Gaussian prior on distribution of vls) or magnitudes (LF prior) or or photo-zs (Gaussian prior) 14

15 Yang et al. s Halo-based Group Finder g(r, v z )= NFW (R) exp v 2 z 2 2 LS > 10 c Univ H 0 Yang, Mo & van den Bosch 04; Yang+07 Domínguez Romero, García Lambas & Muriel 12 group masses (hence virial radii) from: FoF group luminosities (1st pass: M=300L) Halo Abundance Matching (next passes) Accurate group masses (global environment), BCG at center (local environment) weaknesses LS velocity dispersion profile should be convex in log-log (not cst) LS velocity distributions not Maxwellian (outer radial vel. anisotropy) ad hoc threshold for membership (10) imprecise correction for lum. incompleteness (for SDSS flux-limited sample) hard group assignment is unstable 15

16 MAGGIE: Duarte & Mamon 15 Models & Algorithms for Galaxy Groups, Interlopers & Environment probabilistic P (R, v z )= g halo (R, v z ) g halo (R, v z )+g ilop (R, v z ) interlopers halo interlopers more realistic ghalo from ΛCDM 3D model with anisotropic velocities NFW z z z z z σr(r) from solving Jeans equation β(r) from cosmo simulations 16

17 P (R, v z )= g halo (R, v z ) g halo (R, v z )+g ilop (R, v z ) MAGGIE: interlopers Duarte & Mamon 15 z z DM particles in HD simulation: Borgani+04 SAM: Guo+11 Mamon, Biviano & Murante 10 Duarte & Mamon 15 17

18 MAGGIE: Mamon & Duarte 15 Models & Algorithms for Galaxy Groups, Interlopers & Environment group masses by Halo Abundance Matching on central galaxy luminosity or stellar mass (1st pass) on total group luminosity or stellar mass (next passes) groups extracted from D- & L-complete subsamples group properties = sums weighted by probabilities 18

19 Testing Group Finders 19

20 How can group finders go wrong? group fragmentation secondary fragments bring down group purity reduced galaxy completeness group merging reduced group completeness reduced purity of galaxy membership 20

21 Friends-of-Friends optimization Duarte & Mamon 14 Nest 3 & Ntrue 3 & unflagged SAM: Guo+11 Galaxy H: Huchra & Geller 82 R: Ramella_89 t: Trasarti-Battistoni 98 E: Eke+04 B: Berlind+06 T: Tago+10 R: Robotham+11 T: Tempel+14 : optimal (theoretically)

22 bias (dex) inefficiency (dex) FoF optimization: 0.9 mass accuracy Duarte & Mamon H: Huchra & Geller 82 R: Ramella_89 t: Trasarti-Battistoni 98 E: Eke+04 B: Berlind+06 T: Tago+10 R: Robotham+11 T: Tempel+14 : optimal (theoretically) 0.4 Best compromise is: b = 0.07 & b// = 1.1 theoretical for mean separation Robotham

23 Tests: Group Fragmentation mocks SDSS galaxy catalog with errors on luminosities (0.08 dex) & stellar masses (0.2 dex) matching extracted & true groups by most luminous (L) or massive in stars (M) member only unflagged groups Ntrue 3 & Nest 3 fraction of extracted groups = secondary fragments of true groups Duarte & Mamon 15 FoF-M (solid) FoF-L (dashed) Yang-M Yang-L MAGGIE-M MAGGIE-L FoF clusters: high probability of being secondary fragment! M-based: more physical than L-based, but higher (systematic) errors

24 Group total mass accuracy only unflagged primary-fragment groups Ntrue 3 & Nest 3 Duarte & Mamon 15 FoF-M (solid) FoF-L (dashed) Yang-M Yang-L MAGGIE-M MAGGIE-L FoF masses biased low by 0.15 to 0.5 dex, 0.3 dex at hi mass mass accuracy M = 13: 0.35 (FoF), 0.32 (Yang), 0.28 M = 14: (FoF), 0.23 (Yang), 0.20 (MAGGIE) 24

25 Euclid Cluster Finders Euclid: deep mainly based on photo-zs Euclid Cluster Finder Challenge (4 versions) Maurogordato & Biviano 8 algorithms on mock galaxy catalogs (SAM & HD) with photo-z errs few galaxies will have spec-zs 25

26 Cluster Finder Challenge 3 on Durham SAM mock Maurogordato, Biviano < z < 1 FoF FoF matched-filter purity purity log Mmin/M : completeness < z < 2 wavelet wavelet Best algorithms: FoF, matched filter, then wavelets completeness 26

27 Group properties vs. group finder 27

28 Surface density profiles of SDSS groups N 5 log M/M > 13.1 Mr < 19 2 Σ [N 200 r 200 ] 1e 03 1e 02 1e 01 1e+00 Yang (c = 1.75) FoF Tempel (c = 2.66) Σ NFW (R) Group center: BCG Marina Trevisan R / r 200 FoF & Yang consistent with NFW for 0.04 (F) or 0.08 (Y) < R/r200 < 1 28

29 Surface density profiles of SDSS groups N 5 log M/M > 13.1 Mr < 19 2 Σ [N 200 r 200 ] 1e 03 1e 02 1e 01 1e+00 R / r 200 Yang (c = 1.75) FoF Tempel (c = 2.66) MAGGIE (c = 1.93) Σ NFW (R) ΣNFW sph (R) Marina Trevisan Group center: BCG FoF & Yang consistent w NFW for 0.04 (F) or 0.08 (Y) < R/r200 < 1(F) or 1.2 (Y) MAGGIE consistent with NFW-in-sphere for 0.05 < R/r200 < 1 29

30 Line-of-sight velocity dispersion profiles returned σv/v200: Yang: too high FoF: too low MAGGIE: just right At high richness: Yang & FoF get worse! 30

31 Line-of-sight velocity dispersion profiles 0.2 dex bias in r200 & v dex bias in Yang! correct for 0.3 dex mass bias = 0.1 dex bias in r200 & v200 FoF-M (solid) FoF-L (dashed) Yang-M Yang-L MAGGIE-M MAGGIE-L At high richness: Yang & FoF get worse! Yang r200 biased high by 0.6 dex? 31

32 log Mgroup/M Environmental trends log L/L Stalder+ in prep. fblue SDSS f blue (R) =f 1 R R + a blue R/R200

33 Group Finders on mock-sam Stalder+ in prep. FoF Yang MAGGIE Perfect log ablue = PRELIMINARY! check scaling w L & M + log L10 + log M13 f blue (R) =f 1 R R + a blue f = + log L10 + log M13 differences of ~ 0.2 dex in quenching projected radii perfect mocks have lower quenching radii: i.e. less efficient quenching (!?) 33

34 Group Finders on SDSS Stalder+ in prep. PRELIMINARY! check scaling w L & M FoF Yang MAGGIE log ablue = + log L10 quenching radii 10x lower than in SAM + log M13 R f blue (R) =f 1 R + a blue ablue log = a 1 + a 2 log r 200 L f 1 = b 1 + b 2 log L L L + b 3 log f = + a 3 log Mgroup M Mgroup M + log L10 + log M13 34

35 Do group properties depend on Ωm? 35

36 What fraction of mock CGs are physically dense? (DM) simulation Díaz-Giménez & Mamon 10; Díaz-Giménez, GM+12 SAM physically dense Reference 2 3 virialized SAM (!) Mamon 87 40% Mamon 86, MS Bower+06 77% DíazG & GM MS Croton+06 73% DíazG & GM MS De Lucia & Blaizot 07 58% DíazG & GM MS-II Guo+11 69% DíazG MS Guo+11 56% DG+ in prep MS Henriques+12 53% DG+ in prep MS Planck Henriques+15 31% DG+ in prep Gary Mamon (IAP), ptimal grouping algorithms & recent advances on Compact Groups, Bologna, 19 Sep 2014, Evolving Galaxies in Evolving Environments 36

37 What fraction of mock CGs are physically dense? (DM) simulation Díaz-Giménez & Mamon 10; Díaz-Giménez, GM+12 SAM physically dense Reference 2 3 virialized SAM (!) Mamon 87 40% Mamon 86, MS Bower+06 77% DíazG & GM MS Croton+06 73% DíazG & GM MS De Lucia & Blaizot 07 58% DíazG & GM MS-II Guo+11 69% DíazG MS Guo+11 56% DG+ in prep MS Henriques+12 53% DG+ in prep 1/2 2/3 MS Planck CGs physically Henriques+15 dense (90% within 31% virialized DG+ groups) in prep 1/3 1/2 chance alignments (80% within virialized groups) Gary Mamon (IAP), ptimal grouping algorithms & recent advances on Compact Groups, Bologna, 19 Sep 2014, Evolving Galaxies in Evolving Environments 37

38 What fraction of mock CGs are physically dense? (DM) simulation Díaz-Giménez & Mamon 10; Díaz-Giménez, GM+12 SAM physically dense Reference 2 3 virialized SAM (!) Mamon 87 40% Mamon 86, MS Bower+06 77% DíazG & GM MS Croton+06 73% DíazG & GM MS De Lucia & Blaizot 07 58% DíazG & GM MS-II Guo+11 69% DíazG MS Guo+11 56% DG+ in prep MS Henriques+12 53% DG+ in prep MS Planck Henriques+15 31% DG+ in prep higher Ωm more CGs by chance alignments (now 70%!) expect more chance alignments within filaments Hernquist, Katz & Weinberg 95 Gary Mamon (IAP), ptimal grouping algorithms & recent advances on Compact Groups, Bologna, 19 Sep 2014, Evolving Galaxies in Evolving Environments 38

39 Conclusions z-distortions no group finder can be perfect: fragmentation, etc. Prior-based group finders are much better for nearby spec-z surveys group finders lead to results LS velocity dispersion profile quenching radii use Group Finders e.g. w GGA (FoF, Yang, MAGGIE) public release early 17 Environmental effects NT washed out by imperfect group finders(?) SDSS quenching radii 10x smaller than expected from Guo+11 SAM Compact groups: mocks with higher Ωm: 2x less frequent 1.5x more contaminated by chance alignments (now > 50%!) 39