1. Impact of intrinsic alignments on cosmic shear
Sarah Bridle, UCL (London)
Concepts: II, GI
Observations
Theoretical models and simulations
Impact on dark energy

2. Bluff your way in intrinsic alignments: Use these phrases with confidence!
‘GI partially cancels the cosmic shear signal’
‘II can be removed by removing close pairs’
‘GI comes from dark matter aligning a nearby galaxy and shearing a distant galaxy’
‘GI is easier to measure than II’
‘Removing elliptical galaxies will reduce GI’
‘Removing IA requires 4 times better photozs’

3. Intrinsic Alignment EquationsII GI SE GG
Crittenden, Natarajan, Pen, Theuns 2000 but see Hirata & Seljak 2004

4. Gravitational Lensing
Light from a distant galaxy is bent around a mass

5. Face-on view Galaxy image is Magnified
Stretched tangentially around the mass (shear)

6. Cosmic lensing Lensing by large-scale structure of the Universe
Cosmic shear
Cosmic magnification

7. Lensing by dark matter causes nearby galaxies to appear aligned
Cosmic shear
Face-on view
Gravitationally
Sheared (G)
Gravitationally
Sheared (G)
Lensing by dark matter causes nearby galaxies to appear aligned

8. Face-on view (zoomed out)
Cosmic magnification
Face-on view (zoomed out)
Increased number counts behind a mass
(If number counts increase as go fainter)
Broadhurst, Taylor & Peacock 1995; Bartelmann 1995; Dolag & Bartelmann 1997; Sabz et al 1997; Moessner & Jain 1998; Jain 2002; Barber & Taylor 2003; Takada & Hamana 2003; Zhang & Pen 2005
Benitez & Martinez-Gonzlez 1997; Benitez et al 2001; Gaztanaga 2003; Scranton et al 2005;

9. Intrinsic alignments (II)
Croft & Metzler 2000, Heavens et al 2000, Crittenden et al 2001, Catelan et al 2001, Mackey et al, Brown et al 2002, Jing 2002, Hui & Zhang 2002

10. Intrinsic alignments (II) Face-on view
Intrinsically
Aligned (I)
Intrinsically
Aligned (I)
Dark matter distribution causes galaxies to align
Adds to cosmic shear signal

11. Intrinsic-shear correlation (GI)
Hirata & Seljak 2004;
See also Heymans et al 2006, Mandelbaum et al 2006, Hirata et al 2007

12. Intrinsic-shear correlation (GI) Face-on view
Gravitationally
sheared (G)
Intrinsically
aligned (I)
Galaxies point in opposite directions
Partially cancels cosmic shear signal

13. GI from galaxies themselves
Bridle & Abdalla 2007

14. GI from galaxies themselves Face-on view
Significant on scales ~< 8 arcmin

15. Intrinsic Alignment EquationsII GI
=0? GG
Crittenden, Natarajan, Pen, Theuns 2000 but see Hirata & Seljak 2004

16. Observing intrinsic alignments (II)
Do galaxies point at each other?
Need lots of physically close galaxies
Don’t want contamination from cosmic shear!
Use a low-z survey
e.g. SuperCOSMOS, SDSS
Calculate shear correlation function rp

17. Mandelbaum et al 2006

18. Observing intrinsic alignments (II)
Mandelbaum et al 2006

19. Observing intrinsic alignments (GI)
GI - do galaxies point at high mass clumps?
Need mass map – or substitute galaxy map
= are galaxies pointing at other galaxies?
Calculate shear-galaxy correlation function
Find all pairs at given separation. Find mean radial ellipticity of one galaxy. rp

20. Observing intrinsic alignments (GI)
Low luminosity
Weak signal
High luminosity
Significant detection
Brown et al 2002; Heymans et al 2004; Trujillo, Carretero, Patiri 2006; Mandelbaum et al 2006; Hirata et al 2007

21. Observing intrinsic alignments (GI)
Very strong signal
for luminous red galaxies
Brown et al 2002; Heymans et al 2004; Trujillo, Carretero, Patiri 2006; Mandelbaum et al 2006; Hirata et al 2007

22. Theoretical models Tidal torque theory Linear alignment model
N-body simulations
Fitting formulae

23. Tidal torque theory Quadratic alignment model
Spiral galaxies
Apparent ellipticity determined by angular momentum
Angular momentum is from external tidal fields
Contribution to GI vanishes in linear theory
Crittenden, Natarajan, Pen, Theuns 2000; Catelan, Kamionkowski, Blandford 2001; Hui & Zhang 2002
Hoyle 1949; Peebles 1969; Doroshkevich 1970; White 1984; Peacock & Heavens 1985; Barnes & Efstathiou 1987; Heavens & Peacock 1988; Porciani et al 2002

24. Linear alignment model
Elliptical galaxies
Galaxy ellipticity is stretched along potential curvature
Note similarity with lens equations
Catelan, Kamionkowski, Blandford 2001

25. Intrinsic-shear (GI)
Bridle & King 2007
Hirata et al 2007

26. N-body simulations Best model - Cosmic shear Heymans et al
Heavens, Refregier & Heymans 2000; Croft & Metzler 2000; Jing 2002; Heymans et al 2006

27. Fitting formula
HRH* (Heymans et al 2004 based on Heavens, Refregier & Heymans 2000)

28. Intrinsic-intrinsic (II)

29. Cosmic shear two point tomography

30. Cosmic shear tomography

31. Tomographic Lensing Power Spectra

32. Intrinsic Alignment EquationsII GI
=0? GG
Crittenden, Natarajan, Pen, Theuns 2000 but see Hirata & Seljak 2004

33. Ellipticity power spectra

34. Ellipticity power spectra
Lensing sensitivity function for z bin i
Source redshift distribution in z bin j

35. Linear alignment model

37. Effect on cosmic shear of changing w by 1% Cosmic Shear Intrinsic
Alignments (IA)
Don’t explain details of whether GI, II
Normalised to Super-COSMOS
Heymans et al 2004

38. Effect on cosmic shear of changing w by 1% Intrinsic Alignments (IA)
If consider only w
then IA bias on w
is ~10%
If marginalise 6
cosmological
parameters
then IA bias on w
is ~100% (+/- 1 !)
Intrinsic
Alignments (IA)
Move along degeneracies -> 100%
We have to consider intrinsic alignments, and remove them carefully
See Bridle & King 2007 for details

39. Removal of intrinsic alignments
Intrinsic – intrinsic (II)
Weight down close pairs (King & Schneider 2002, Heymans & Heavens 2003, Takada & White 2004)
Fit parameterized models (King & Schneider 2003, Bridle & King 2007)
Shear – intrinsic (GI)
Fit parameterized models (King 2005, Bernstein DETF. Bridle & King 2007)
Redshift weighting (Schneider)
The effect I just described can be classified as a type of shear-intrinsic (GI) intrinsic alignment. Various methods have been suggested for removing the intrinsic alignments.
The intrinsic-intrinsic alignment is the alignment of close pairs of galaxies.
So we can remove it by simply removing all close pairs!
Or simultaneously fit a model for the intrinsic alignments at the same time as fitting the cosmological model.
Its much harder to remove the shear-intrinsic correlation.
For the intrinsic-intrinsic correlation we need to remove all close pairs of galaxies.
Whereas, to first order, for the shear-intrinsic term we need to remove all the pairs that aren’t close, which would leave us with no galaxies!
Lindsay King and Gary Bernstein have suggested fitting parameterised models, and on Monday Peter briefly described a new redshift weighting method.
Whichever method you use, the quality of the redshift information is absolutely crucial.
So Lindsay King and I recently investigated the effect of photometric redshift errors on the model fitting techniques.
Redshift quality is crucial!

43. Dark energy Figure of Merit
Perfect redshifts
Least flexible model considered
FoM is improved!
Redshift
dependence
of IA (# bins) 2 3 5
No Intrinsic Alignments
Dark energy Figure of Merit
Reasonable model? (14 IA pars)
Similar FoM to no IA case
Very flexible (100 IA pars)
FoM is roughly halved
Scale dependence of IA (# bins)

44. Dark energy Figure of Merit
Perfect redshifts
Redshift
dependence
of IA (# bins) 2 3 5
Dark energy Figure of Merit
Scale dependence of IA (# bins)

45. Realistic photozs σz=0.05(1+z)
Redshift
dependence
of IA (# bins) 2 3 5
Dark energy Figure of Merit
Scale dependence of IA (# bins)

46. No Intrinsic Alignments
FoM / FoM(specz)
Relatively flat
Bridle & King arXiv:
(e.g. Hu 1999, Ma, Hu, Huterer 2006, Jain et al 2007,
Amara & Refregier )
Photoz error σz / (1+z)

47. FoM / FoM(specz) Photoz error σz / (1+z)
Reasonable model? (14 IA pars)
Very flexible (100 IA pars)
FoM / FoM(specz)
Bridle & King arXiv:
Photoz error σz / (1+z)

48. Redshift accuracy σz / (1+z)
With intrinsic alignments, need 4x better redshifts to get 80% of the DE information
0.8
FoM / FoM(specz)
Bridle & King arXiv:
0.02
0.08
Redshift accuracy σz / (1+z)

49. Bluff your way in intrinsic alignments: Use these phrases with confidence!
‘GI partially cancels the cosmic shear signal’
‘II can be removed by removing close pairs’
‘GI comes from dark matter aligning a nearby galaxy and shearing a distant galaxy’
‘GI is easier to measure than II’
‘Removing elliptical galaxies will reduce GI’
‘Removing IA requires 4 times better photozs’

50. More research needed! Theoretical
Can IA be predicted accurately enough to help?
How complicated are IA as fn (scale, z)?
Necessary to model IA as fn(magnitude, type)?
Can intrinsic cosmic magnification be removed?
Observational
How do intrinsic alignments (IA) evolve with z?
How big are IA for blue galaxies?
Correlated with other properties e.g. photozs?
Can cosmic magnification overcome e.g. extinction?

51. END

52. Shearing by elliptical galaxy halos
Plan:
Calculate shear from elliptical halo
Calculate contribution to shear correlation fn
Average over a population of lenses
Compare with cosmic shear signal
Consider effect of halo profile
Investigate redshift dependence
Bridle & Abdalla 2007

53. Shear correlation function
z1=0.3 z2=0.8
Cosmic shear signal
NFW
Shear correlation function
Average over population
visible to R=24 ^

54. Shear correlation function
z1=0.3 z2=0.8
Cosmic shear signal
Singular isothermal
ellipsoid
NFW
Shear correlation function
Average over population
visible to R=24 ^

55. Shear correlation function
zlens=0.3 zsource=0.8
Bridle & Abdalla
M200=1x1012 h-1 Mo
Shear correlation function

56. How good to photozs need to be to remove intrinsic alignments?
Plan:
Remove GI, II by marginalising over some flexible model
Look at the effect of GI, II on dark energy errors
Dependence on flexibility of model?
Dependence on photoz errors?
Bridle & King 2007

57. σz / (1+z)

60. Contribution to ellipticity correlation function:
Bridle & Abdalla
Contribution to ellipticity correlation function:
Average shear around circular annulus
Does not average to zero →net contamination

61. Shear correlation function
z1=0.3 z2=0.8
Cosmic shear signal
Bridle & Abdalla
Shear correlation function
Average over population
visible to R=24

62. Shear correlation function
z1=0.3 z2=0.8
Cosmic shear signal
Bridle & Abdalla
Shear correlation function
Average over population
visible to R=24
Change in
cosmic shear signal
for  w = 0.05
We do need to consider this effect at scales smaller than about 10 arcmins

63. Intrinsic-shear (GI)
Hirata et al 2007
Bridle & King 2007