Uploaded image for project: 'Request For Comments'
  1. Request For Comments
  2. RFC-491

Make MatchPessimisticB the default astrometry matcher in place of MatchOptimisticB


    • Type: RFC
    • Status: Implemented
    • Resolution: Done
    • Component/s: DM
    • Labels:


      Several failure modes have been found in the MatchOptimisticB blind astrometry matcher, specifically in dense fields with thousands of references objects to match to and similar numbers of detected sources. At these high densities, false positive matches for the Optimistic pattern matcher (OPMb) are no longer rare.
      To address these problems we have created a more generalized version of the OPMb algorithm called MatchPesimisticB and which is described in DMTN-031. This DMTN also compares the performance of the two algorithms and shows that across a large dynamic range of reference and source densities MatchPesimisticB performs just as well as MatcherOptimisticB in low density environments and has much higher success rate for high densities. In terms of algorithm timing, MatchPesimisticB is roughly ~20% slower than MatchOptimisticB despite being implemented in Python vs MatchOptimisticB which is written in C++.
      With these performance metrics in hand, this RFC would like to change the stack default algorithm to that of MatchPesimisticB instead of MatchOptimisticB for blind astrometry matching. This respresents a change to the default config of AstrometryTask in meas_astrom.


          Issue Links



              • Assignee:
                cmorrison Chris Morrison
                cmorrison Chris Morrison
                Chris Morrison, Colin Slater, Jim Bosch, John Swinbank, Michael Wood-Vasey, Paul Price
              • Votes:
                0 Vote for this issue
                6 Start watching this issue


                • Created:
                  Planned End:

                  Summary Panel