[PIPE2D-230] Remove at least one of the spline implementations in drp_stella Created: 24/Jul/17 Updated: 03/Nov/18 Resolved: 03/Nov/18 |
|
| Status: | Done |
| Project: | DRP 2-D Pipeline |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Story | Priority: | Normal |
| Reporter: | rhl | Assignee: | Unassigned |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Description |
|
I discovered that there are two separate spline interpretations in drp_stella, and we should remove at least one of them. There are unnecessary copies of arrays in FiberTraces.cc that should be removed when we clean up the interpolation. There is functional spline interpolation in {lsst::afw::math::Interpolate}} , but it only supports std::vector<double>. If it is appropriate to use doubles wherever we need to interpolate in drp_stella we should definitely switch to use the afw implementation. |
| Comments |
| Comment by rhl [ 26/Jul/17 ] |
|
I also note that both of these splines seem to be "natural" splines (with vanishing second derivative at the first/last knots). This is the simplest to code, but almost always a bad choice in practice. |
| Comment by swinbank [ 29/Jul/17 ] |
|
Per our meeting of 2017-07-28, we agreed that RHL would replace existing the spline implementations with a "not-a-knot" implementation which he has lying around. We will then file an LSST-side ticket to get the afw spline implementation cleaned up, in the hope that we can ultimately switch to that. |
| Comment by rhl [ 01/Aug/17 ] |
|
I have replaced the "spline.cc" implementation with a clean implementation which supports not-a-knot; this was done on |
| Comment by hassan [ 03/Nov/18 ] |
|
Done already according to RHL's recent comments on this ticket. |
| Comment by rhl [ 03/Nov/18 ] |
|
I don't think this ticket is done (but didn't check). I cleaned things up, and added a not-a-knot, but I am not sure that the other versions are fine, and we certainly didn't switch to afw |
| Comment by price [ 03/Nov/18 ] |
|
I believe there's only one spline implementation in drp_stella these days, which is RHL's. |
| Comment by price [ 03/Nov/18 ] |
|
Agreed with RHL that this can be closed. |