IPP Progress Report for the week 2010.02.01 - 2010.02.05

(Up to IPP Progress Reports)

Eugene Magnier

Heather Flewelling

  • stacked/diffed/magicked y+w chips
  • investigated some statistics on the linear dark and the polynomial dark.
  • found the streakie chips which cause problems for magic, and created a wiki http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/RowPatternCorrectionOTAs
  • investigated VerifyStreaks on more magicverify labels (checked to see causes of streaks, both before and after various iterations of IPP build)
    • band/observation dependence- different observations (md, vs 3pi) and different bands had different #s of streaks, and for different reasons
  • investigated chip 24 and chip 15 with ppImage.
  • created small test set for magic:
    • images for renormalization bug, burntool, pattern subtraction, and a couple of SS images.
  • created larger test set for magic, consisting of the worst 200 and the best 100 (roughly).
  • processed subset of SVS for MOPS
  • installed ippMonitor on ifaps1 machine
  • lots of reverting/stacktooling/etc for processing and pantasking.

Bill Giebink

Roy Henderson

  • Numerous meetings with Conrad and Jim regarding data integrity, speed, error handling etc
  • Now that I'm able to churn out data in bulk, speed is seen as a real issue (up to 10 minutes to create FITS detections from a 'busy' smf source) so...
  • Re-wrote FITS I/O stuff to use cfitsio directly, rather than psLib. Now more than twice as fast: remaining bottleneck is access to DVO db
  • Worked with Conrad on his 'dxlayer' including:
    • potential designs for error-handling
    • transferring some tasks to ippToPsps code so that dxlayer can remain ignorant of db schema (will save on maintenance when there are schema changes)
    • wrote fast code to convert FITS extensions to csv files using the low-level 'bulk copy' cfitsio functions (under 10secs for 70Mb FITS, which take minutes to write)

Paul Price

Bill Sweeney

Chris Waters

  • Identified that fringe needs to be applied after masking to ensure the correct scaling is found.
  • Expanded nightly stacking code to handle automation of science for an entire night. This has not been fully checked and debugged.
  • Detrend cleanup needed to be rewritten, and detrend processed and residual image cleanup is progressing.
  • Burntool. Upgraded to new version of the code and confirmed that it resolved the issues found in the old version.
  • Checked that difference image signal to noise distributions were consistent with Gaussian distributions.
  • Began study of diffraction spike lengths, in an attempt to expand the masking to cover them completely.