IPP Progress Report for the week 2012.04.02 - 2012.04.06

(Up to IPP Progress Reports)

Eugene Magnier

  • spent time helping Chris with false positives & writing up the current state for the PS1SC.
  • Looked a bit into the Kron mags using Peter D. test set and applying a very simple moments / flux analysis with user-controlled apertures. I can confirm that I recover the model mags (with expected kron aperture bias) if I use the provided Kron radius. I do not recover the Kron radius if I start with the Kron radius; I need to use a larger aperture. Michael Wood-Vasey has volunteered to look at psphot issues with me, so I pointed him at some of these measurements.
  • discussions with Edouard Bernard & Eddie S. about the faint-end bias & psphot vs DAOphot comparisons. Edouard's DAOphot measurements show the faint end bias. Looking at the DAOphot-reported sky levels, however, gives pretty strong evidence that we cannot blame the psphot-measured sky on the faint end bias. First, the sky values found by DATAphot (after psphot has removed the sky model) are, as expected, generally quite close to zero, with a possible small (0.2-0.5 DN) positive offset. Second, the faint-end-bias has more or less the same amplitude for a given magnitude value (in this dataset), and do not seem to correlate with the small changes in the reported DAOphot sky levels. Finally, and most conclusive, the DAOphot sky levels are, if anything, generally slighlyt positive, while the faint end bias has the sense that the psphot values are too small -- this is the wrong direction. I think need to look elsewhere. I've asked Edouard to send me his measurements for a comparison dataset.
  • looked into some of the remaining outlier images in the big 3pi dvo database. It looks like I marked detections as 'poor' if they had the 'OFF_CHIP' flag raised. For a subset of images, this flag was raised for all measurements, so those images could never be tied to the rest of the database through relative photometry. I'm also looking into some of the other images that failed to be pulled into the overall relative photometry system, but this is going a bit slowly.

Serge Chastel

  • Set up Pittsburgh postage-stamp server at IfA.
  • Still working on ippb02 but nothing exciting to report

Heather Flewelling

  • answered various liason questions - still have a few outstanding:
    • nial isp
    • bertrand/shannon want lap dvodb
  • ssp
    • copied off data from atrc, ingesting into ssp database
    • we need a proper datastore
  • dvodb
    • built new SAS.20120403 off of czw.SAStest label
    • monitoring LAP progress
    • tweaked input files to automatically make CNP.V2 - these cycle every 15 days as there are few CNP exposures per night
    • cleaned up my messed and commited my code changes
    • blocked on copying minidvodbs to ipp022 - lack of space
  • jt
    • u bands stacks are ugly - lots of crosstalk, lots of discussions with jt/gene about this
      • I used the same warps as did Gene - there are minimal differences between the ipp codes I used and Gene used - the problem is the crosstalk present in the raw images
    • md nightlystacks for Mar2012 (v2 tess) are warped, need to be stacked

Roy Henderson

  • lots of work to populate object colors from DVO tables using row numbers as index
    • worked on stored procedure with cursor to get around DVO mags stored by row number issue, but ended up...
    • adding row-count columns to cps and object tables, indexing then joining. Faster
    • method to query Photcodes table and get index of each available filter, which defines order in cps file
    • method to get mags from cps file given list of filters (above)
    • method to calculate colors from mags
    • calculating flux in Janskys
    • various other fields populated
  • adapted ODM and cleanup programs to incorporate new object batches
  • writing metadata for object batches to ipptopsps database
  • tested new version of java jre, which fixed an oustanding bug in certain ippToPsps programs. Good news.
  • now, all programs can choose to use 'all' configs, meaning they will cycle through all active configs. handy for metrics and polling programs
  • began setting up ippToPsps to run as IPP user
  • incorporated nice user prompted config maker into all programs
  • created and loaded IN, P2, ST and OB batches in order to test OB loading on sandbox
  • lots of documentation, including work on UML diagram for ippToPsps code

Mark Huber

  • ppStack: finished tracing problematic inputs to produce a sample for defining additional cuts to apply. Started a larger sample of MD04 skycells near edge.
  • ppSub: test runs with new refcat.

Bill Sweeney

  • Modified psphot to read in the "exposure number" images produced during stacking and use the value at the peak pixel for each source to set the cmf column N_FRAMES. This was a bit more complicated than it sounds given the very powerful but somewhat intricate system that the IPP uses to manage images.
  • Added new table "skycell" to the gpc1 database. This will contain an entry for each skycell containing the celestial and galactic coordinates for each skycell. Wrote a program to read a tessellation and write out the data (dumpskycells) and another program to manage the table ("sctool" - short for skycell tool)
  • Investigated circumstances surrounding the crash of our web proxy. Didn't find anything suspicious
  • Created build and installed new operational tag ipp-20120404
  • Implemented minor change to psphotStack to include "stack_id" in output file names.

Chris Waters

  • Dark studies: http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/Background_Dark_Model
  • Noisemap/false positive studies:
    • Changing the binning areas in the noisemap changes the results, but not in a dramatic way that would account for the observed remaining false positive results.
    • Confirmed that as the noise increases in a noisemap cell, it also becomes increasingly non-Gaussian. However, it does not seem to be so non-Gaussian as to explain the false positive rate.
    • Looked at the trend of false positives with changes in detector GAIN. This works in the way we want, but although the cells with presumed bad gain values have a higher false positive rate, the small number of these cells prevents them from having the majority of the false positives.
    • Confirmed that the false positive rate decreases dramatically if we enforce a 5-sigma detection limit on measurements.
  • Change to psastro to record reference catalog name in the PSREFCAT header keyword.
  • Confirmed date range for various dark/noisemap changes, and queued new detrends to match those changes.
  • Confirmed that the A/B mode dark variations do not correspond to the status of the glycol.