This page provides a place for the AP team to keep track of past and future configuration investigations.


packageOutput locationconfiguration parameterticketsdiscussioninvestigation complete?
ip_diffim
doScaleTemplateVariance = True


No.  Need to determine if should be set to True by default

Nov. 2020: old backlog ticket to investigate scaling down  the variance in some cases needs attention.   Will place

meas_algorithms/project/sullivan/saha/

Use psfex as the PSF determiner (now the default)

import lsst.meas.extensions.psfex.psfexPsfDeterminer

config.charImage.measurePsf.psfDeterminer["psfex"].spatialOrder=2
config.charImage.measurePsf.psfDeterminer["psfex"].psfexBasis="PIXEL_AUTO"
config.charImage.measurePsf.psfDeterminer["psfex"].samplingSize=0.5
config.charImage.measurePsf.psfDeterminer='psfex'

Ian Sullivan increasing reduced chisq for PSF candidates to allow crowded field processing (from 3 to 20–enough to have about 100 candidates)

Yes.  Superseded by change to PSFEx.
pipe_tasks
calibrate.photoCal.match.referenceSelection.magLimit.fluxField="i_flux"
calibrate.photoCal.match.referenceSelection.magLimit.maximum=22.0

Meredith Rawls uses this for obs_decam reruns since mid-2019 because sensible defaults are set only in obs_subaru. Also see this Community post.

Yes, but defaults still unset for obs_decam.  Meredith to ticket.


calibrate.astrometry.doMagnitudeOutlierRejection

Lauren MacArthur added this during DC2 processing, but value depends on whether refcats have a reasonable range, whether there are refcat vs. science color differences

Current consensus (July 12, 21) is that this is too dataset-dependent to set a single default.

ip_diffim
 self.makeKernel.kernel.active.spatialBgOrder = 2
Shiu Liu (DESC DIA)'s analysis (10/4/2022) suggests ~50% reduction in artifacts going to a spatial order of 4