Closed bd-j closed 5 years ago
@lgarrison has made great progress on this (#22 and #23) Remaining issue is probably the currently slow calculation of spline interpolations of gaussian galaxy amplitudes as a fn of sersic and rhalf
In addition to the splines, a number of the source gaussian covariance matrices have the form a*eye(2)
so matmuls with those could be converted to scalar multiply
closing this and opening an issue specifically for the Sersic splines.
Speed up conversions from on-sky gaussians to on-image gaussians. This is currently a bottleneck due to inefficient python loops and linear algebra calls that could probably be made faster for the 2 x 2 matrices used here.