fgcmLoadReferenceCatalog doesn't work for single band data

[AstroJacobLi]

Hello! I'm trying to reduce some of the HSC data using the LSST Science Pipelines, using $DRP_PIPE_DIR/pipelines/HSC/DRP-RC2.yaml. Everything went smoothly, except for the FGCM part in Step 2cde:

WARNING 2024-12-25T00:03:47.340-05:00 py.warnings (fgcmBuildStarsTable:{instrument: 'HSC'})(warnings.py:109) - /projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/meas_algorithms/g592df8196b+7381939c19/python/lsst/meas/algorithms/sourceSelector.py:349: RuntimeWarning: divide by zero encountered in true_divide
  signalToNoise = flux/err

INFO 2024-12-25T00:03:47.341-05:00 lsst.fgcmBuildStarsTable.fgcmLoadReferenceCatalog.referenceSelector (fgcmBuildStarsTable:{instrument: 'HSC'})(sourceSelector.py:666) - Selected 18430/42177 references
WARNING 2024-12-25T00:03:47.345-05:00 py.warnings (fgcmBuildStarsTable:{instrument: 'HSC'})(warnings.py:109) - /projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/fgcmcal/g2ffcdf413f+0ec844bbf9/python/lsst/fgcmcal/fgcmLoadReferenceCatalog.py:213: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  fgcmRefCat = np.zeros(np.sum(selected), dtype=[('ra', 'f8'),

ERROR 2024-12-25T00:03:47.346-05:00 lsst.ctrl.mpexec.singleQuantumExecutor (fgcmBuildStarsTable:{instrument: 'HSC'})(singleQuantumExecutor.py:279) - Execution of task 'fgcmBuildStarsTable' on quantum {instrument: 'HSC'} failed. Exception IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed
ERROR 2024-12-25T00:03:47.373-05:00 lsst.ctrl.mpexec.mpGraphExecutor ()(mpGraphExecutor.py:492) - Task <TaskDef(lsst.fgcmcal.fgcmBuildStarsTable.FgcmBuildStarsTableTask, label=fgcmBuildStarsTable) dataId={instrument: 'HSC'}> failed; processing will continue for remaining tasks.
Traceback (most recent call last):
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/ctrl_mpexec/g1d38f3a6c4+57975668de/python/lsst/ctrl/mpexec/mpGraphExecutor.py", line 462, in _executeQuantaInProcess
    self.quantumExecutor.execute(qnode.taskDef, qnode.quantum)
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/ctrl_mpexec/g1d38f3a6c4+57975668de/python/lsst/ctrl/mpexec/singleQuantumExecutor.py", line 152, in execute
    result = self._execute(taskDef, quantum)
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/ctrl_mpexec/g1d38f3a6c4+57975668de/python/lsst/ctrl/mpexec/singleQuantumExecutor.py", line 277, in _execute
    self.runQuantum(runTask, quantum, taskDef, limited_butler)
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/ctrl_mpexec/g1d38f3a6c4+57975668de/python/lsst/ctrl/mpexec/singleQuantumExecutor.py", line 559, in runQuantum
    task.runQuantum(butlerQC, inputRefs, outputRefs)
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/fgcmcal/g2ffcdf413f+0ec844bbf9/python/lsst/fgcmcal/fgcmBuildStarsTable.py", line 301, in runQuantum
    fgcmStarIdCat, fgcmStarIndicesCat, fgcmRefCat = self.fgcmMatchStars(visitCat,
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/fgcmcal/g2ffcdf413f+0ec844bbf9/python/lsst/fgcmcal/fgcmBuildStarsBase.py", line 495, in fgcmMatchStars
    fgcmMakeStars.makeReferenceMatches(self.fgcmLoadReferenceCatalog)
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/fgcm/g322c47f462/lib/python/fgcm/fgcmMakeStars.py", line 845, in makeReferenceMatches
    refCat = refLoader.getFgcmReferenceStarsSkyCircle(meanRA, meanDec, rad,
  File "/projects/HSC/LSST/stack_20220527/conda/envs/lsst-scipipe-5.1.0/share/eups/Linux64/fgcmcal/g2ffcdf413f+0ec844bbf9/python/lsst/fgcmcal/fgcmLoadReferenceCatalog.py", line 232, in getFgcmReferenceStarsSkyCircle
    fgcmRefCat['refMag'][:, :] = 99.0
IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed

I think the reason is very simple -- I only have one band (HSC-R2), and the following code doesn't work for 1-band situation, because len(filterList)=1:

fgcmcal/python/lsst/fgcmcal/fgcmLoadReferenceCatalog.py

Line 215 in cab2d05

('refMag', 'f4', len(filterList)),

.

Changing it to the following would fix the issue:

fgcmRefCat = np.zeros(np.sum(selected), dtype=[('ra', 'f8'),
                                                ('dec', 'f8'),
                                                ('refMag', 'f4', [len(filterList)]),
                                                ('refMagErr', 'f4', [len(filterList)])])

Maybe I shouldn't run FGCM for single-band data. But please let me know if you have any suggestions on bypassing this error!

Thanks!