Approaching the Cramér–rao Bound in Weak Lensing With PDF Symmetrization

ABSTRACT Weak lensing statistics is typically measured as the weighted sum of shear estimators or their products (shear–shear correlation). The weighting schemes are designed with a view to minimizing the statistical error without introducing systematic errors. It would be ideal to approach the Cramér–Rao bound (the lower bound of the statistical uncertainty) in shear statistics, though it is generally difficult to do so in practice. The reasons may include difficulties in galaxy shape measurement, inaccurate knowledge of the probability distribution function (PDF) of the shear estimator, misidentification of point sources as galaxies, etc. Using the shear estimators defined by Zhang et al., we show that one can overcome these problems, and allow shear measurement accuracy to approach the Cramér–Rao bound. This can be achieved by symmetrizing the PDF of the shear estimator, or the joint PDF of shear estimator pairs (for shear–shear correlation), without any prior knowledge of the PDF. Using simulated galaxy images, we demonstrate that under general observing conditions, this idea works as expected: it minimizes the statistical uncertainty without introducing systematic error.