Shear measurements in simulated SNAP images with realistic PSFs Hkon Dahle, Stephanie Jouvel, JeanPa

1
Shear measurements in simulated SNAP images with
realistic PSFs Håkon Dahle, Stephanie Jouvel,
Jean-Paul Kneib, Eric Prieto, Sebastien Vives,
Bruno MilliardLaboratoire dAstrophysique de
Marseille (LAM), France
2
Motivation

• STEP-like simulations for SNAP, with PSF (800nm)
  from optical design
• Include effects of
• jitter, charge diffusion, sky background,
  detector noise (no CTE or diffraction spikes)
• Dithering and re-sampling
• KSB analysis Shear recovery as function of
  focus, distance from optical axis

3
Skymaker simulations

• 34x34 grid of galaxies in 4 simulated
  exposures ditherered, re-sampled
  and combined
  0.1 pixels
  --gt 0.06 pixels 34132
• PSF from 49 stars
• All galaxies have the same intrinsic ellipticity
  (e 0.25), but random orientations
• 5 magnitude bins (25.0,26.0,27.0,28.0,28.5)
• 3 size bins (galaxies of scale 0.1, 0.2 and
  0.4)

4
PSF variation with focus/position
5
Note ellipticity dependence on smoothing scale
(PSF wings are more elliptical)
0.34o 0.57o 0.8o
0.34o off-axis 0.57o off-axis 0.8o off-axis
6
0.34o 0.57o 0.8o
Note that e1e20 PSFs are not necessarily
circularly symmetric
0.34o off-axis 0.57o off-axis 0.8o off-axis
7
0.34o 0.57o 0.8o
0.34o off-axis 0.57o off-axis 0.8o off-axis
8
0.34o 0.57o 0.8o
0.34o off-axis 0.57o off-axis 0.8o off-axis
9
Analysis
KSB analysis (measuring stellar ellipticities
and polarizabilities at same scale as these
quantities are measured for each galaxy)
Select faint galaxies (6 lt S/N lt 250), which
tend to carry most of the shear signal in real WL
analyses Note The magnitude values have an
arbitrary zeropoint and should not be taken
literally. S/N values are more meaningful
10
Bias of shear measurements
Size 0.1(blue), 0.2(red), 0.4(green)
11
ShearKaiser Squires Broadhurst Ap.J. 449
460-475 1995 (KSB)

• Polarization is characterized by a vector (e1,
  e2)
• e1 (QxxQyy)/(QxxQyy)
• e2 2Qxy/(QxxQyy)
• The Qij are Gaussian weighted second moments of
  the intensity distribution
• NOTE In these simulations, all galaxies have the
  same intrinsic e sqrt(e12 e22)
• The simulated galaxies of different sizes have
  their ellipticities diluted by the PSF to a
  varying extent --gt concentric circles in e1-e2
  space
• Noisier (fainter) galaxies make fuzzier circles
• Anisotropic PSFs make them slightly
  non-concentric non-circular

12
Shear recovery

• Apply standard KSB methods (later modifications
  by Luppino Kaiser 1997 and Hoekstra et al.
  1998) to recover the intrinsic values of e1 , e2
  (see fig)
• From these, calculate the mean value of the
  modulus ltegt ltsqrt(e12 e22)gt
• Define bias as the ratio of the output value to
  the input value of the ellipticity
• Define uncertainty as the rms scatter around
  ltegt

13
Bias of the shear measurements (ellipticity
modulus after PSF correction, relative to input
ellipticity) Green 0.34o off-axis Blue 0.57o
off-axis Red 0.8o off-axis
14
Uncertainty of the shear measurements Green
0.34o off-axis Blue 0.57o off-axis Red 0.8o
off-axis
15
Mean value of each shear component (should in
principle be zero circle indicates 1 sigma
uncertainty).
16
This is work in progress

• Currently probing features/limitations of KSB
  more than SNAP ?
• Would be useful to compare to another method for
  shape measurement
• Simulations with finer sampling of different
  focus values (checking how smoothly uncertainty
  bias vary as function of focus).

17
(No Transcript)