Fix Version/s: None
Sprint:Alert Production S17 - 3
In order to implement the "mean zero point" functionality of PhotoCalib, I need to compute the integral over the bounding box. BoundedField does not include facilities for integration/differentiation. Fortunately, my primary usage is ChebyshevBoundedField, which has a relatively straight-forward recurrence relation for integration.
I suggest an interface like the following on BoundedField, with a virtual "not implemented" default:
// Compute the integral of this function over its bounding-box.
double integrate() const;
DM-9192 Implement new spatially-variable PhotoCalib model
Jim Bosch Thanks a lot for your help getting this together. Please let me know if you'd rather we have a third pair of eyes.
Jonathan Sick I did an "interesting" thing with a test docstring and I'm curious to get your opinion on it. Specifically, see the docstring for testIntegrateBox, although there are changes to other docstrings that might be worth your look.
Jenkins run in progress: https://ci.lsst.codes/job/stack-os-matrix/22668/
Comment on the Mathematica code block at https://github.com/lsst/afw/pull/197/files#r108538321 tl;dr it won't work exactly, but it's not a big deal in practice.
Thanks Jonathan Sick. I redid it as you suggested.
Note: Jenkins run passed.
Thanks for the quick review! I believe I fixed all your comments.
Merged and done.
Have to do this before I can finish