Details
-
Enabler
-
Must have
-
None
-
Data Processing
-
-
-
Inter Program
-
3
-
3
-
8
-
2.667
-
Team_NZAPP
-
Sprint 5
-
-
-
-
18.2
-
Stories Completed, Integrated, Outcomes Reviewed, Satisfies Acceptance Criteria, Accepted by FO
-
-
SPO-1592
Description
Who?
- SDP pipeline developers
What?
This feature was originally split from SP-2332, as we were trying to address the boundary between processing functions ("C" functions communicating using in-memory data and python bindings) and - essentially - higher-level workflow steps (possibly internally distributed processes communication via the buffer).
- So for this feature, we would like to prototype a deconvolution processing function according to HIPPO's C interface, drawing on work done for
SP-2332andSP-1860 - Consider how/if deconvolution should be decomposed into a set of functions.
- As usual, we would be specifically interested in distributed deconvolution - i.e. that we can do this on facets separately
- To this end, consider how to define masks in case we want to choose facet borders strategically (as in WSClean) - probably similar metadata discussion as with
SP-2332, but for in-memory data models (data structures).
Why?
- Deconvolution is of tremendous importance for the imaging pipeline - especially as it is comparatively hard to parallelise.
- Therefore having processing functions available is going to be of great value