General information

Information specific to latest implementation for GREAT08 Challenge LowNoise_Blind

Information specific to latest implementation for GREAT08 Challenge RealNoise_Blind

Information specific to earlier implementations for GREAT08 Challenge:

N/A

Information specific to GREAT08 Challenge:

Prior

The prior functional form derived from the data (see makeprior and updateprior in the code and the README) is different from the functional 

form in Kitching et al. (2008) in two ways:

• The D parameter is ignored, since as shown in Kitching et al. this a poorly fitted parameter and has a small effect on the shape of the prior. For STEP 1 and 2 D >> 1 so it effectively had no effect on the prior functional form.
• The centroid of the prior is now a free parameter
  • for real data the centroid of the intrinsic ellipticity distribution is expected to be at e1=0, e2=0
  • in a simulation this may not be the case and the intrinsic distribution may be non-zero centred
  • Note that when the prior is updated using updateprior the non-central parameters are ignored since in estimating the shear the intrinsic distribution is assumed to be zero-centered and the sensitivity calculation (Miller et al., Kitching et al.) takes into account the bias introduced due to this assumption.

So the functional form fitted using makeprior is

where the free parameters are B, C, x1, x2 and A is calculated from a normalisation.

Weights

In lensfit the shear is calculated from the following estimator

where s_i is the sensitivity and <e_i> is the expectation value of the ellipticity (calculated by integrating the probability over e). The w_i are weights that can in general be applies to the estimator. In the papers Miller et al. and Kitching et al. we use w_i=1. For the data version of lensfit we use a flux dependent measure of the weight. Because the GREAT08 flux distribution is not the same as real data this breaks down for these simulations (submitted Q value goes down from 116 to approx 100). The optimal weight should be something that is orientation independent, we are investigating this now (Feb-Mar 2009).