stscijgbot-jp
commented
3 months ago Comment by Timothy Brandt on JIRA:
6/16/24 @ 15:30: I updated the accompanying notebook with a few more example sets of limits, deleted the duplicate issue on GitHub, and made a slight update to the issue. It looks like if any set of limits is entirely contained within another set, the results may be incorrect. I do not know if results can be incorrect more generally. It is a bit difficult for me to follow the logic of the original routine, but it seems like it may assume that both the lower and the upper limits of the sequence of intervals are both in ascending order after sorting by the lower limits only, which is equivalent to assuming that no interval is entirely contained within any other interval. If this assumption ever breaks in the JWST pipeline then incorrect results, and incorrect extracted spectra, may follow.
Issue JP-3660 was created on JIRA by Timothy Brandt:
extract1d._coalesce_bounds does not appear to behave as intended when the first segment contains the other ones, for example, if the input is [[1, 100], [5, 50], [3, 60]] the output will be [[1, 60]] where I am nearly certain that [[1, 100]] is intended, and for input of [[1, 100], [5, 27], [6, 28], [40, 50]] the output will be [[1, 28], [40, 50]] where again I am nearly certain that [[1, 100]] is the intended output. This issue seems to be present whenever any interval is contained within another. The routine is also opaque (to my eyes at least) and inefficient in using lists for everything; we could consider rewriting it more simply with arrays. See the attached Jupyter notebook for more examples of unexpected behavior and a proposed rewrite.