Our main paper runs showed significantly larger alpha bias than our pre-run, particularly for the additive shear bias. We have demonstrated that a significant portion of this bias came from a bug in meds.MEDSMaker which does not save the PSF image Jacobian correctly by replacing the cutouts w/ the true PSF image in #92. We now need to implement an equivalent fix for PSF estimates such as PSFEx and PIFF.
Our main paper runs showed significantly larger alpha bias than our pre-run, particularly for the additive shear bias. We have demonstrated that a significant portion of this bias came from a bug in
meds.MEDSMakerwhich does not save the PSF image Jacobian correctly by replacing the cutouts w/ the true PSF image in #92. We now need to implement an equivalent fix for PSF estimates such as PSFEx and PIFF.