Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-429

Make NoiseReplacer outputs reproduceable

    Details

      Description

      We need a way to get back the noise-replaced Exposure as it was when a particular source was measurement, after the measurement has been run, without having to run noise-replacement on all the previous objects again.

      There is already code in afw::math::Random to output its state as a string; I think we should probably just save this string in the output catalog. This will require some API changes to allow the NoiseReplacer to modify the schema and set a field in the output records.

        Attachments

          Issue Links

            Activity

            Hide
            jbosch Jim Bosch added a comment -

            1. Yes.

            2. Yes - but I haven't thought much at all whether multiplying by a constant is better or worse than adding a constant.

            3. I think in this case, maybe we just use the config multiplicative/additive term by itself.

            4. By default, though having a (non-default) option to go back to the current behavior would be good.

            5. Yes.

            Show
            jbosch Jim Bosch added a comment - 1. Yes. 2. Yes - but I haven't thought much at all whether multiplying by a constant is better or worse than adding a constant. 3. I think in this case, maybe we just use the config multiplicative/additive term by itself. 4. By default, though having a (non-default) option to go back to the current behavior would be good. 5. Yes.
            Hide
            pgee Perry Gee added a comment -

            I have redone the changes, according to this last comment. The changes are in u/pgee/DM-429 in meas_base and pipe_tasks.

            You did not reply to my last email about the persistence to the metadata being unnecessary, but please read that and comment on it.

            Show
            pgee Perry Gee added a comment - I have redone the changes, according to this last comment. The changes are in u/pgee/ DM-429 in meas_base and pipe_tasks. You did not reply to my last email about the persistence to the metadata being unnecessary, but please read that and comment on it.
            Hide
            jbosch Jim Bosch added a comment -

            You did not reply to my last email about the persistence to the metadata being unnecessary, but please read that and comment on it.

            Sorry - I read and essetially agreed with it. The persistence to the metadata is incomplete and unnecessary, but harmless and possibly informative.

            Show
            jbosch Jim Bosch added a comment - You did not reply to my last email about the persistence to the metadata being unnecessary, but please read that and comment on it. Sorry - I read and essetially agreed with it. The persistence to the metadata is incomplete and unnecessary, but harmless and possibly informative.
            Hide
            jbosch Jim Bosch added a comment -

            Looks great, just two comments:

            • We need to make the same sort of changes for forced measurement.
            • Please add a description for exposureId to the Doxygen comments for SingleFrameMeasurementTask.run().

            No need to send back to me for review - I think we've converged on what's needed (sorry it took so long).

            Show
            jbosch Jim Bosch added a comment - Looks great, just two comments: We need to make the same sort of changes for forced measurement. Please add a description for exposureId to the Doxygen comments for SingleFrameMeasurementTask.run() . No need to send back to me for review - I think we've converged on what's needed (sorry it took so long).
            Hide
            pgee Perry Gee added a comment -

            Oops, sorry. I did do forcedMeasurement and tested it, but didn't transfer it to my checkin directory.

            Show
            pgee Perry Gee added a comment - Oops, sorry. I did do forcedMeasurement and tested it, but didn't transfer it to my checkin directory.

              People

              • Assignee:
                pgee Perry Gee
                Reporter:
                jbosch Jim Bosch
                Reviewers:
                Jim Bosch
                Watchers:
                Jim Bosch, Kian-Tat Lim, Perry Gee
              • Votes:
                0 Vote for this issue
                Watchers:
                3 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:

                  Summary Panel