Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-10260

Add "value+error" object to somewhere

    Details

    • Type: Improvement
    • Status: To Do
    • Resolution: Unresolved
    • Fix Version/s: None
    • Component/s: base

      Description

      PhotoCalib has a Quantity Measurement, which is a simple Struct containing a value and its error, as that's clearer than dealing with a std::pair or tuple. This seems like a useful thing to have more generically (including vectorized versions and maybe a templated version that can manage Point s etc.), but it's not clear where it should live. We probably need something like lsst.primitives for such things (it's not really an afw.geom or an afw.math, it's lower than both).

      I've filed this on base, mostly because I'm not sure where else to put it. I don't think this needs an RFC, but I wanted to file a ticket so it doesn't get forgotten (I wasn't going to do the work to lift it somewhere as part of DM-9192).

        Attachments

          Issue Links

            Activity

            Hide
            jbosch Jim Bosch added a comment -

            This is definitely at least a convention that we should standardize, and it might be a generally useful class or set of classes, but we should think about higher-dimensional quantities; the standard uncertainty on a coordinate pair, for instance, is a 2x2 symmetric matrix. You can find an effort to do something along these lines for fluxes, centroids, and shapes in meas_base's *Utilities.h files. I think the right level of specialization for these classes is probably somewhere between a one-size-fits all templated Quantity struct and the measurement-specific things in meas_base.

            Show
            jbosch Jim Bosch added a comment - This is definitely at least a convention that we should standardize, and it might be a generally useful class or set of classes, but we should think about higher-dimensional quantities; the standard uncertainty on a coordinate pair, for instance, is a 2x2 symmetric matrix. You can find an effort to do something along these lines for fluxes, centroids, and shapes in meas_base's *Utilities.h files. I think the right level of specialization for these classes is probably somewhere between a one-size-fits all templated Quantity struct and the measurement-specific things in meas_base.
            Hide
            krughoff Simon Krughoff added a comment -

            I'm going to weigh in now so I make sure to get it done. I am pretty uncomfortable with the term Quantity as that already has a well defined meaning in the astropy world (and I like their meaning). I would advocate for UncertainValue or Measurement as both get across the concept of value+error better than the word quantity in my opinion.

            Show
            krughoff Simon Krughoff added a comment - I'm going to weigh in now so I make sure to get it done. I am pretty uncomfortable with the term Quantity as that already has a well defined meaning in the astropy world (and I like their meaning). I would advocate for UncertainValue or Measurement as both get across the concept of value+error better than the word quantity in my opinion.

              People

              • Assignee:
                Unassigned
                Reporter:
                Parejkoj John Parejko
                Watchers:
                Jim Bosch, John Parejko, John Swinbank, Krzysztof Findeisen, Paul Price, Russell Owen, Simon Krughoff
              • Votes:
                0 Vote for this issue
                Watchers:
                7 Start watching this issue

                Dates

                • Created:
                  Updated:

                  Summary Panel