1. The 2dF galaxy redshift survey John Peacock & the 2dFGRS team Harvard, October 1999

2. Outline  Brief history  2dF Survey motivation & design  Survey data  Spectral classification  Results: –Luminosity function –Correlation function  Galaxy bias & future issues

3. History of galaxy clustering  1930s: Hubble lognormal cell counts  1940s/50s: Eyeball surveys (Shane & Wirtanen, Zwicky, Abell…)  1970s: Correlation functions r mass = r 0 (1+d) x A (r) = (1) Autocorrelation function (2) Two-point correlations r prob(pair) = 2 dV 1 dV 2 [1 + x 2 (r) ] dV 1 dV 2 x A (r) = x 2 (r) ? - only if Poisson Clustering Hypothesis is true

4. Meaning of clustering Neyman Scott & Shane (1953): random clump model r ~ r -a (r < R) ” x ~ r -(2a-3) obs: x ~ r -1.8 ” a = 2.4? Modern view: gravitational instability of (C)DM - sheets, pancakes, filaments, voids...

6. Las Campanas Redshift Survey ~25000 z’s CfA/SSRS z-survey ~15000 z’s Redshift Surveys

7. The 2dF Galaxy Redshift Survey  Aim for LCRS X 10 = 250,000 z’s  Increase sky coverage to get fully 3D sample –Measure >100-Mpc power –Test Gaussian nature of linear fluctuations –Measure redshift-space distortions  Increase sampling density –Spatial distribution for different galaxy types –Tests of theories for biased galaxy formation

8. The 2dF Galaxy Redshift Survey  Goal: complete 1:1 galaxy sample out to 300 Mpc –3D sampling (cf. LCRS = 2D)  2dF strategy: –2000 deg 2 –250,000 galaxies, most SGP –400 per 1 hr shot (fibre optics) –97% completeness (tiling) –Photographic selected –b j <19.45 (extinction corrected) –z~0.15 –4m shared telescope (AAT) –50,000 QSO candidates SDSS strategy: –10,000 deg 2 –900,000 most NGP –640 (fibres) –96% (tiling) –CCD selected –r <18.2 –z~0.1 –2m dedicated –100,000 QSOs

9. 2dFGRS Survey Team  Australian team members: Matthew Colless, Joss Bland-Hawthorn, Russell Cannon, Warrick Couch, Kathryn Deeley, Roberto De Propris, Karl Glazebrook, Carole Jackson, Ian Lewis, Bruce Peterson, Ian Price, Keith Taylor.  British team members: Steve Maddox, John Peacock, Shaun Cole, Chris Collins, Nicholas Cross, Gavin Dalton, Simon Driver, George Efstathiou, Richard Ellis, Carlos Frenk, Ofer Lahav, Stuart Lumsden, Stephen Moody, Peder Norberg, Shai Ronen, Mark Seabourne, Robert Smith, Will Sutherland, Helen Tadros.

10. 2dFGRS parameters  Galaxies: b J  19.45 from revised APM  Total area on sky ~ 2000 º  250,000 galaxies in total, 93% sampling rate  Mean redshift ~ 0.1, almost all with z < 0.3

11. 2dFGRS geometry NGP SGP NGP 75  x7.5  SGP 75  x15  Random 100x2  Ø ~70,000 ~140,000 ~40,000 ~2000 sq.deg. 250,000 galaxies Strips+random fields ~ 1x10 8 h -3 Mpc 3 Volume in strips ~ 3x10 7 h -3 Mpc 3

12. Tiling strategy ‘2dF’ = ‘two-degree field’ = 400 spectra Efficient sky coverage, but variable completeness High completeness through adaptive tiling: multiple coverage of high-density regions

13. Sampling: 2dF vs LCRS 2dFGRS (~93%) LCRS (~25%)

14. Calibrating photometry

15. Photometric calibrations  Extensive photometric calibrations now available from various sources.  Preliminary results suggest small scale error in the APM magnitudes.  Re-calibration makes faint objects fainter relative to bright objects.  The z-survey limit is shifted ~0.2mag fainter.  Further photometry being obtained/reduced to confirm this result. B J (APM) B(CCD) B J (APM)-B(CCD)

16. Recalibrated number counts Old APM counts Recalibrated APM counts

17. Deep CCD photometry  Driver et al. are using the Wide Field Camera on the INT to get deep CCD images.  Imaging covers a strip of 0.68°x75° (i.e. 50deg 2 ) on equator from about 10 h to 15 h.  Photometry to g=25,  g =26 mag arcsec -2.  Check photometric accuracy, completeness.  Will generate surface brightness profiles and morphologies for ~10,000 2dFGRS gals. MGC 2dF

18. The 2dF site Prime Focus

19. The 2dF facility

20. 2dF on the AAT

21. Configuring fibres >12 arcsec spacing; 15 degree bend <10 seconds to position each fibre

22. Data pipeline: real-time X-corr z’s

23. Example spectra

24. Survey status - August 1999  Observed: –227/1093 fields –58764 targets –4037 repeats  Redshifts/IDs: –53192 (91% complete) –50180 galaxies –2993 stars, 19 QSOs

25. Redshift yield  The median redshift yield is 93%.  10% of fields have a yield less than 80%.  30% of fields have a yield less than 90%.  After ADC s/w fix, good conditions routinely give yields >95%.  Reliability: of 1404 z’s in overlap with LCRS, only 8 disagree (99.4% agree).

26. Completeness  Redshift completeness is >90% for b J <19 but drops to 80-85% at b J =19.45.  Completeness is similar in NGP and SGP strips.  Completeness as a function of magnitude varies with the overall completeness of the field.  Selection function depends on (at least) overall completeness and magnitude.

27. 24.3 mag arcsec -2 Surface brightness selection  Use mean SB above detection isophote.  SB limit varies from field to field.  Approximate mean SB limit 24.3 mag arcsec -2.  Objects without z’s have faint mags, but are not esp. LSB.  SB limit varies with absolute magnitude: e.g. at b J =19.45, more luminous galaxies have higher z and larger (1+z) 4 SB and K(z) corrections.

28. Bivariate Brightness Distribution  Correct observed SB- magnitude distribution for incompleteness and volume sampled (as functions of M and  )  intrinsic  (M,  ).  Main features: –Giants have bounded ~Gaussian 2D distn. –Dwarfs are steeply increasing at faint L. –Giant LSB galaxies rare. –Sharp HSB cutoff at all luminosities. –BBD falls off before low- SB selection limit. –Isophotal corrections? Observed Intrinsic

29. Survey mask NGP SGP Cutouts are bright stars and satellite trails.

30. Selection mask NGP SGP ‘Bitten-cookie’ effect from missing overlap tiles. 0% 100% 0% 100%

31. Stellar contamination Contamination by objects with z~0. SGP NGP Typical level of stellar contamination is <5%. 0% 20% 0% 20%

32. Survey rate  At least some data were obtained on 61/99 of nights so far allocated to survey.  Over all nights with any data, the mean number of fields/night is 4.5.  Averaged over year, expect to get 7 fields for each completely clear night = 3000 z’s per night  Full survey requires about 100 clear dark nights, or all dark time for 1 year. In practice 2dFGRS uses about 1/3 AAT dark and will take 3 years

33. Cone diagram: all declinations

34. Cone diagram: 4-degree wedge

35. The big picture 2dFGRS The 2dF galaxy + QSO redshift surveys 50180 galaxies 6824 QSOs

36. Redshift distribution Mean redshift =0.11; almost all z<0.3. N(z) still shows significant clustering.

37. (mean K-corrections)  2 fit to 1/Vmax LF Overall luminosity function STY Schechter fit gives  -1.2 (due to clustering? non- Schechter form?) Small numbers at M B >-14.

38. Mean spectrum PC1 PC2 PC3 Early Late Early Late Spectral classification by PCA  Apply Principal Component analysis to spectra.  PC1: emission lines correlate with blue continuum.  PC2: strength of emission lines without continuum.  PC3: strength of Balmer lines w.r.t. other emission.  Classify spectral types in PC1-PC2 plane using sample of Kennicutt to set bounds.  Further work: –effect of spectro-photometric errors; –self-classification algorithms; –calibration against spectral models.

39. STY fit 1/V max LF Early Late  M* All LFs by spectral type  For 12,000 galaxies with PCA types, fit LFs by type.  1/V max LFs have less-steep faint ends than STY fits of Schechter functions.  From early to late types… –M* gets fainter: -19.6  -18.9 –  gets steeper: -0.7  -1.7  Overall M* brighter than M* of any type; Schechter function not adequate fit.  Evidence for upturn at faint end of LF. Overall STY Schechter fit Sum of Schechter fits to each type

40. Early types (1,2)Late types (3,4,5) Galaxy distribution by type

41. 2D correlations x(s, p) r s p

42. Model comparison - I Flattening depends on b=W 0.6 /b (d gal = b d mass ) infall Fingers of God

43. Model comparison - II Analyze x(s,p) into Legendre polynomials: get b from quadrupole-to-monopole ratio -P 2 /P 0 = (4b/3 + 4b 2 /7) / (1 + 2b/3 + b 2 /5) Linear Damping by fingers of God

44. Projected correlations X(r) APM w(q) deprojection works well to r = 20 h -1 Mpc (cosmic variance matters on larger scales)

45. The CDM clustering problem Non-monotonic scale-dependent bias LCDMtCDM Jenkins et al. 1998 ApJ 499, 20 b 2 = x g / x m

46. The CDM Mach-number problem Pairwise dispersion too high LCRS

47. Numerical galaxy formation Durham Munich Santa Cruz Edinburgh...

48. Antibias in LCDM Benson et al. astro-ph/9903343

49. Dark-matter haloes and bias Moore et al: r = [ y 3/2 (1+y 3/2 ) ] -1 ; y = r/r c

50. Correlations from smooth haloes PS++ mass function and NFW++ halo profile gives correct clustering LCDM tCDM Lin NL APM

51. Halo occupations depend on mass PS++ mass function wrong shape for cluster/group LF Correct weighting of low-mass haloes predicts antibias LCDM

52. Summary  2dF survey status –Over 50,000 redshifts (20% of survey) –Expect 100,000 March 2000 –250,000 March 2001  Preliminary results –Luminosity functions by spectral type –Correlations and redshift-space distortions  Future issues –Clustering on 100-Mpc scales –Gaussian nature of density field –Clustering by spectral type and luminosity –Detailed tests of halo-based bias models