Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-11229 Investigate GLMM convergence
  3. DM-11230

Investigate divergence in spectral normalization

    Details

    • Type: Technical task
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: meas_deblender
    • Labels:
      None
    • Templates:
    • Story Points:
      5
    • Sprint:
      DRP F17-2
    • Team:
      Data Release Production

      Description

      In hyperspectral mixing, a similar but easier to solve problem then a deblender, we normalize the spectra in exactly the same way that we do in the deblender, namely requiring that the values of each component's spectrum are positive and sum to one.

      What we notice is that if we perform the normalization as a linear operator, the algorithm quickly diverges. Understanding why this happens is likely to improve our understanding of deblender failures, so this ticket aims to understand the reason for this divergence.

        Attachments

          Activity

            People

            • Assignee:
              fred3m Fred Moolekamp
              Reporter:
              fred3m Fred Moolekamp
              Reviewers:
              Peter Melchior
              Watchers:
              Fred Moolekamp, Peter Melchior
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Summary Panel