IPP Progress Report for the week 2010.03.29 - 2010.04.02

(Up to IPP Progress Reports)

Eugene Magnier

We had a somewhat abbreviated week with the holiday on Friday and a power outage in Manoa on Wednesday. I spent much of the week working on issues related to astrometry and proper motion measurements. I defined a new DVO format to support additional fields in the average position table needed to check on the quality of the proper motion measurement. I adjusted 'relastro' so that all objects which are fitted for proper motion are also fitted for a static position, and the chisq from both fits are saved in the database. I also added flags to track which fits had been tested and which fit was kept. I also realized that the position errors from the proper motion fit are coupled to the velocity errors, so that the reported errors are amplified if the reference epoch is far from the fitted time range. Previously, I had been fitting with a reference epoch of J2000 in all cases, but this gives odd results for PS1-only proper motion fits. For example, if we fit proper motion to the years 2009 and 2010 referenced to 2000, the error on the mean position include the nominal mean position (say, 15 mas) plus the nominal velocity errors (say 15mas/year) times the extrapolation (10 years). Although one could back out these correlations, I decided it would makes like easier to perform the fit relative to the mean epoch. That in turn requires the database to track the mean epoch.

I added a new capability to DVO to allow extraction of average parameters relative to a set of supplied coordinates. This function allows for inter-DVO database comparisons, eg extracting the average magnitudes from a 2MASS or USNO DVO database for objects extracted from a PS1 database. I have been using this feature to compare the proper motion fits using only PS1 vs PS1+2MASS. There are some features to this comparison which still need to be explored -- there are a subset of correlated proper motion values which have the wrong slope (< 1).

I have also made progress on the stack photometry analysis framework. It looks like I will have an initial working version of that code by the end of the next week.

Heather Flewelling

  • vacation Mon-Wed.
  • moved into new office on Thus
  • working on ISP issues

Bill Giebink

  • Dealt with DNSM hydraulic fluid leak
    • Fixed leak by tightening cover plate
    • Refilled fluid to proper level
    • Cleaned up spilled oil
  • Purchased spare oil filters for DNSM hydraulics
  • Restarted DNSM system after power failure
  • Posted photos of PS2 dome/cooling system
  • Traced cabling from meteorological stations on Airglow and posted photos
  • MHPCC work
    • Replaced motherboard/CPU fans in IPP 018 and 037
    • Installed new RAID card in IPP037
    • Updated BIOS setting for serial console ops on IPP 018 and 037
    • Put IPP 008, 018, and 037 in cabinets
    • Diagnosed array fan error in IPP014

Roy Henderson

  • PSPS
    • Loaded yet more simulation data to Maui cluster after changes by Sue at JHU
    • Committed schema changes: some merge issues which took a little time
    • Loaded data on JHU sandbox with new schema: dealing with inevitable issues
    • PSPS meeting regarding PSI interface
  • IPP
    • Learned about role of processing czar
    • Finished changes to ippToPsps after PSPS schema change
  • Other stuff
    • Root canal surgery on Tuesday
    • Power cut on Wednesday

Paul Price

  • SMF file corruption:
    • Sifting through SMF files to identify bad files.
    • Discovered new destreaking bug: some exposures have more than 50 chips masked in SMF, not due to NFS glitches. Occurs when skycell to chip mapping isn't identical between diff inputs. Fixed magicdstool.
  • Added code for automated deletion of diff convolved images
  • Shepherding M31 and STS data through the pipeline
  • Added extra fields to MOPS product
  • Chasing down pipeline bugs
    • Discovered pole bug in getstar or astrometric reference catalogue --> Gene
    • Changed psastro to flag exposure as bad quality when astrometry fails (since this will generally be due to there being insufficient stars), to prevent these from being reverted and repeated forever.
    • Marked SweetSpot.nightlyscience chip/cam/warp on TJD5269 as dropped because the camera warmed up (30 degrees!) and some exposures were making psphot go crazy. We need protection in the pipeline against this sort of thing!
    • Little fixes to prevent aborts and assertion failures all over
    • Newton-Raphson method for radius determination in pmModel_PS1_V1.c is failing and producing aborts; fixed by Gene
  • Bundled Microtest magic clusters from 2010-03-26 and sent to Camera team
  • Queued 4045 gri 3pi exposures from Demo Month for reprocessing; mainly for MOPS
  • Discovered that at least some 3pi warp-warp diffs are being done against themselves. These are identifiable by fault=1, due to out-of-bounds variance renormalisation (factors ranging from 0.14 to 0.0003).

Bill Sweeney

  • Most of the week was spent being processing shepherd.
    • Recovered STS template warps that were destreaked by mistake.
    • searched for data that can be cleaned up to make room for new processing
    • debugged causes of processing failures and had Paul fix them
    • managed labels and survey task inputs to match the ever changing priorities
    • various changes to the ippTools to help manage data flow.
  • Reorganized the way the distribution bundles are distributed over the cluster to avoid uneven use of space.
  • Installed test postage stamp server and data store to avoid debugging on the live system.
  • Modified the stub detectability query implementation to be compatible with current workflow.
  • Half a holiday on Friday.

Chris Waters

  • vacation Wed+Thu
  • sent out notification of detrend distribution
  • defined MD07 reference stacks