Core Minutes 3/31/2015

• ScienceTools: (Jim)

  • Made new tag including effective area systematics as requested by FSSC.
  • Eric Charles has made new externals for healpix and wcslib and modified several packages which used them as described in a Confluence page. ST builds with the new arrangement and test programs are working.
  • Added background estimator to ST; now in LATEST

  See Science Tools Development Notes for current activity.

• FSSC: (Joe) Not much to report; currently in testing cycle for new release.

• New IGRF model (Michael) has been volunteered to look into the following (thanks to him for providing this description):

  Nicola Mazziotta found that with the new astro [updated to use IGRF-12 coefficients, tagged astro-03-15-07] the value for the geo B field starts to deviate from the values obtained with the old astro and from common sense from 2010 on. So a guess would be that there is something wrong with the implementation. I checked:

  1. B-field web form on the ngdc web site: the B-fields there are reasonable, so it's our code.
  2. compared the new "definitive" geo model with the old extrapolated one, there are minor differences.
  3. removed superfluous data files; no change.
  4. changed formatting of data files, older files have no leading spaces in the line, new files instead had pretty formatting, no change

  So, I wait for inspiration during the night and continue tomorrow.

• Reprocessing: (Tom) There is not yet a definite release date for Pass8 so he's holding off on a penultimate backfill for another week.

• Hardware: (Tom) The Fermi user disk is under stress, and not because of abuse by users. There is an nfs problem causing large numbers of connections not to be released in a timely fashion. Since the problem is not reproducible in a simple, clear-cut way Redhat won't help debug it. The only way currently to clear this up is to reboot the entire batch farm. This can take quite a while (days) because of long-running jobs.

• C&A (Leon)

  • The first public release of Pass8 will come with instructions not to use the data below 100 MeV. Given that and the 80-degree limit on “zmax” (maximum angle from the azimuth) the limb is not in the field of view, so those templates are not needed. This is just for the public release; collaborators will still have access to all the usual stuff.
  • The latest comparisons, between Vela “on/off” data and the pass8 model appear to agree within 3% in the validity range, for front vs back. For the data samples cut on PSF the agreement is more like 10%. This validation, and others based on stacked AGN, will be used to set official systematic errors, TBD.
  • Bill has started looking at how well we reconstruct low-energy events, with a view to doing a study of the polarization of the gammas. For this purpose, you need to correctly reconstruct 2 tracks, and measure the plane of the pair conversion.

    Both pass7 and pass8 patrec have low efficiency (~20%) for reconstructing a two-track vertex, with pass7 doing a bit better. New strategies, such as combining both patrecs, are under investigation.

• Core talk:

  • CHS migrating to RHEL 6 (Joanne) Steve Tether is working on this. He's come by with some questions about our use of SCons. He is making good progress and may be close to done. (Richard) How does it happen that they're using SCons for CHS builds? (Joanne) CHS relies on several of our packages:

previous

minutes index

next

J. Bogart, Last Modified: