Details
-
Type:
Improvement
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: jointcal
-
Labels:
-
Story Points:6
-
Epic Link:
-
Sprint:AP S18-6, AP F18-1, AP F18-2, AP F18-3, AP F18-4, AP F18-5
-
Team:Alert Production
Description
The astrometric fitter includes a pedestal on the measurement errors (AstrometryFit.cc::tweakAstromMeasurementErrors(), because it appeared the position error for bright sources was too small. We can compare (plot/statistics) the residuals vs. errors for the photometry, to see whether we will need a similar error pedestal there. This would be a good thing to put in the photometry notebook that uses the chi2 contributions files.
Attachments
Issue Links
Activity
Per my conversation just now with Jim Bosch, the first attempt at a photometric error pedestal should be proportional to source flux, and equivalent to roughly 0.05 magnitudes (~5% in flux) to start, and then reducing to a tighter level (estimated systematic flux uncertainty is 0.01 mags (~1% in flux)).
Dan Taranu: can you please do this medium (~200 lines) review?
The jupyter notebook was me planing around with looking at residuals (with the new TRACE output in I added), but I didn't end up making the plots that were specifically listed in the ticket description. The code I added was useful enough that we want to incorporate it, but it defaults to an error pedestal of 0, and thus has no effect, but we'll probably want to turn it on in the future (with different values for different surveys) after deeper analysis of jointcal's output.
Sure.
Looks good to me, assuming that the OSX run completes without issues.
Thanks for the review: it was good to better document those regexes.
Merged and done.
Adding this to the sprint: the PDR1 phtoometry continues to have nonlinear-like problems that might be fixed by a pedestal on the errors, so I need to look into this. Plus, Michael Wood-Vasey says that the validate_drp plots strongly imply a systematic error floor that is not being accounted for in the singleFrame results.