IPP Progress Report for the week 2011.02.14 - 2011.02.18

(Up to IPP Progress Reports)

Eugene Magnier

I spent much of the week working on the missing fields from the psphotStack output that are needed by PSPS. I added the ability to write out the full set of covariances for all model parameters, and a few other fields (petrosian fill factors, fix the error values in the petrosian analysis, added standard deviations to the radial aperture fluxes). I also created a better test suite of this stuff which helped fix an error in the radial aperture flux calculations. I also added the radial aperture fluxes to the single-image version of psphot (previously it was only available in the psphotStack version). I generated a test sample for MD04, using 8 sets of raw exposures and running to the stack creation. I will be running psphotStack on this set to produce output for PSPS to test.

I also re-ran the astrometry and photometry analysis on the full-sky DVO reference database. The photometry analysis had been running very slowly, and I traced that to an image-map optimisation which I had added to relastro but forgot to port to relphot. When that was done, the full-sky relative photometry took about 30 hours to run. I had also finished the relative astrometry analysis, but discovered that an important config table defining the relative weights of PS1 and 2MASS detections was wrong (credit to Roger Lin for pushing that). I re-ran relastro as well, and that took about 24 hours as well. I have also run the relative photometry analysis on the MD field database. At this point, I need to assess the results of the calibration analysis, and merge the MD and 3pi zero point analysis (using the program 'uniphot').

I spent some time looking into diffraction spikes and glints. Inspired by results from MOPS, I looking at the quality of our diffraction spike masking. I realised that the masks were sometimes a small amount off in rotation, and for very bright stars, this could lead to significant misses. I asked Chris to look into improving the model. At John Tonry's urging, I examined the positions of some stars causing a particularly nasty glint. This appears to come from part of the L3 support structure, and John believes he can shadow it with a better top mask.

Serge Chastel

  • MOPS + IPP-MOPS czar
  • Fixed truncate problem of MOPS data.
  • MOPS SVM for asteroids: accuracy ~ 85%, missed ~ 10%, false alarms ~ 7%... But data were partly flaky (because of the truncature problem)
  • Spent most of my time on the unfinished https://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/nebulous-performance-analysis Still need to 1) tidy up/document/test code before committing it to tools; 2) finish adding comments in that link;

Heather Flewelling

  • magictests are done
    • stacks are fixed, so M31 and MD01 magic numbers are now okay.
    • bathtub rings trigger magic now
    • sent Chris the ppImage config file for this magictest/staticmask/darkmask combination
    • ran into problem with ~4 diffs: Data error code: 36b6
    • ran into odd problems with dark master for < 2009-12-09 data -- we were using DARK_PREMASK and now we are using DARK, and DARKTIME is unreliable for random cells for this period (DARK_PREMASK doesn't use DARKTIME, DARK did)
    • slowed down because ipp005 was offline, and ipp005 had the copy of the static masks (need to replicate still).
    • various issues related to nebulous orphans/widows
    • summary http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/magictest_3pi_results
  • attempted to answer questions on new dvodb
  • additional processing
    • queued MD04, MD05, MD05 for jtrp. MD04, MD05 are warped, stacked, and cleaned. MD06 still needs to warp and stack.
    • queued mopstestsuite.20110214 for Ken, and published to IPP-MOPS-TEST2
  • installed ipp-20101215 tag on ifaps1, set as default, cleaned up ~/ippconfig
  • detrends

Roy Henderson

out sick

Bill Sweeney

  • Made some changes to cleanup to make it more reliable.
  • Debugged and fixed several problems with the new dependency checking.
  • Brought the magicDSRun.state 'goto_restored' back to life. This can be used to swap uncensored images back in place. This is useful if data corruption errors happen during destreaking.
  • fixed some bugs in raw data distribution. Started posting censored raw images of STS observations for MPG.
  • fixed bug in get image postage stamp requests that I introduced when I changed distribution to store files in nebulous.
  • added a feature to the distribution client that allows user to set a flag to tell the system to not extract files from the bundles.
  • Investigated reports that detections are missing from the ThreePi? DVO catalog. It looks like something went wrong and the merges happened without the mini-dvo dbs being sorted. This affects observations since around December 15.

Chris Waters

  • diskspace work: Identified files that have not been properly cleaned, despite having been through the cleanup process. Continued bin-by-8 of laser data. Ran neb-shuffle with updated ranking method to better balance cluster disks.
  • Burntool: Reworked burntool process to run more reliably. Wrote scripts to help monitor registration speed and allow easier fixes of faults.
  • detection efficiency: did final CDFs for data processed with old deteff code (without covariance fix). Made preliminary CDFs with corrected code.
  • diffraction spikes: Measured position angle offset from diffraction spike masks and actual diffraction spikes. Implemented variable width masks to cover wider and longer spikes.
  • Started work on new GPC1 format files to match changes in camera configuration, to allow for time based pattern correction.