Created a copy of CALIB-20220615 (built for PIPE2D-1045) to new calib folder CALIB-20220615 and added m-band detectorMap and fiberProfiles with doSlitOffset=True. Compared residuals by running reduceArc for visit 63351 using this newly generated calib folder: , and the next most recent calib folder containing m-band calibs, CALIB-SuNSS-2021-06-16: . Vertical striping is clearly diminished in the new calib set.

m-band detector map has been generated by bootstrapping from simulator detectormap using 21-fiber Ne arc and then running reduceArc on SuNSS visit 63350. Key stats below:

reduceArc.fitDetectorMap INFO: Final fit: chi2=56831.010268 dof=23890 xRMS=0.049894 yRMS=0.069913 (0.003298 nm) xSoften=0.028745 ySoften=0.042386 from 12017/13225 lines
reduceArc.fitDetectorMap INFO: Final result: chi2=30246.331401 dof=23388 xRMS=0.021701 yRMS=0.060891 xSoften=0.000000 ySoften=0.033281 from 12017 lines

New calib folder CALIB-20220622 available on the Princeton tiger cluster under /projects/HSC/PFS/Subaru, at the summit (/data/drp/sm1-5.2/) and Hilo (/work).

As Arnaud and Yabe-san have discovered, the CALIB-20220622 calib is problematic and has been removed. Running reduceExposure against this calib resulted in strange pfsArm spectra as shown in v78699-20220622.png and v78705-20220622.png. Compare this with the result of processing with the previous official calib, CALIB-PFI-20211220 (v78699-20211220.png, v78705-20211220.png).

This is due to an incorrect profile radius set when constructing the fiber profiles. Using profiles.profileRadius=3 repair.doCosmicRay=False when running constructFiberProfiles leads to the expected spectra shown in v78699-20220630.png and v78705-20220630.png.

The updated calib, CALIB-20220630 is already available on the Princeton cluster. This will also be copied over to to the summit and Hilo.

Calib CALIB-20220630 also copied to the summit and Hilo.