Shears from shapelets

Accurate measurement of gravitational shear from images of distant galaxies is one of the most direct ways of studying the distribution of mass in the universe. We describe an implementation of a technique that is based on the shapelets formalism. The shapelets technique describes PSF and observed images in terms of Gauss-Hermite expansions (Gaussians times polynomials). It allows the various operations that a galaxy image undergoes before being registered in a camera (gravitational shear, PSF convolution, pixelation) to be modeled in a single formalism, so that intrinsic ellipticities can be derived in a single modeling step. The resulting algorithm, and tests of it on idealized data as well as more realistic simulated images from the STEP project, are described. Results are very promising, with attained calibration accuracy better than four percent (1 percent for round PSFs) and PSF ellipticity correction better than a factor of 20. Residual calibration problems are discussed.