Science Tools Working Group
We met last week, 12 attendees (F de Palma, T Stephens, J Peachey, N Giglietto, F Gargano, J Chiang, S Digel, E Winter, D Band,F Giordano, A Caliandro, M Hirayama).
Dave Davis and Eric Winter provided recap of the feedback to the GSSC from the GLAST Users Group from their 'beta test' of the Science Tools. The Beta Test was not limited to the GLAST Users Group members, in that the GSSC also recruited some other likely users.
The GSSC has been able to verify installations working at about 10 institutions representing 17 Beta Test participants, on a remarkable variety of Linux versions, thanks to a lot of work. In making distributions in the HEASARC hmake system, Eric found that he had to make some small changes in some packages to get the 64-bit builds to work. He is planning to enter these changes as JIRA issues for the package owners in our repository. Eric reports that the tools cannot be ported to Mac OSX 10.5 before a ROOT build for that version is made. [Update from Eric W.: "we have a 10.4 binary running on 10.5 now"]
Dave Davis said that only 5 feedback reports had been about the tools themselves, rather than installation issues. Apparently none of the comments were about bugs; some were comments on the documentation. So right now it looks like no major issues will arise from the Beta Test. This is ok.
The current release version of the Science Tools remains v9r5p4.
Data products: No news. We had a long discussion about whether TDMIN/TDMAX keywords should be included in the FT1 files. These are optional in the HEASARC specification; they are defined to contain the minimum and maximum values of a given column. Some tools, e.g., ds9, use them if they are present to set scaling parameters without having to read through the data. The GSSC data server has been configured to add them to the FT1 files. This is semi-problematic right now as they are not part of the agreed-on format and in particular gtselect does not interpret or update them. This does not break any of the Science Tools but we'll move to a resolution with the GSSC.
Discussion on this issue led to a discussion on delivery of the Science Tools to the GSSC, and what constitutes a part of the distribution intended for release by the GSSC to the public. Eric and Jim have converged on a current list of tools; this will get posted in...Confluence. More importantly, Jim has defined a checkout package, ST_dist, for the tools for the GSSC. Eric also has a proposed procedure for joint agreement on additions/deletions from the set of distributed tools. I think that it will fly; details later.
Databases and related utilities
No news
Likelihood analysis
No news
GRB tools
No news
Pulsar tools
Masa reports that he and James "are still working on the time handling classes (bug fixes and clean-ups)"
Observation simulation
From Jim: "I fixed a bug in FitsTransient that Johann encountered for very fast transients (i.e., GRBs) that was related to roundoff errors for small time intervals being added to our 9 digit MET values (celestialSources/genericSources v1r11p2)."
Max's recently (Monday) checked in a new version of PulsarSpectrum (v2r5) that should fix a bug in handling environment variables that point to the location of files of pulsar parameters: "now the input method is more robust for preventing loops and wrong input files". He has also posted a tutorial.
User interface and infrastructure (& utilities)
From Jim: "I updated evtClassDefs to use the pass-5 style event (non-) classifications so that FT1 files coming out of L1Proc correspond to the P6_v1 IRFs."
Source Catalog
Did not meet last week. Will meet this week, primarily for discussion of issues related to source association and identification.
Science Tools Working Group
We will have a meeting this week (tomorrow). The topics will include a recap of the feedback to the GSSC from the GLAST Users Group from their 'beta test' of the Science Tools.
Since May 15 the current release version of the Science Tools is v9r5p4. Here are the changes since v9r5p3. It has several behind-the-scenes changes that you should not notice plus a couple of small changes that you might. See below
Data products: No news
Databases and related utilities
No news
Likelihood analysis
No news
GRB tools
No news
Pulsar tools
From Masa: "We are still working on time handling classes." This is toward the end of handling multiple timing models for binary pulsars
Observation simulation
Jim diagnosed an odd problem with the 'obssim3' simulations of the Big Run sky model as result from the microQuasar source reseting the random number seed. The symptom was that changing the seed given to gtobssim had no effect on the sequence of events that was output. Richard has fixed microQuasar to no longer do this, but the change is not yet in a release of the Science Tools.
For long gtobssim runs (of order 1 year) with a "complete" sky model (e.g., obssim3), we still encounter duplicate event times, probably owing to the intrinsic single precision of the random number generators we are using. See the discussion from obssim 1. As per Toby's suggestion, we have implemented a 26 micro second deadtime: photons with arrival times within this window relative to the previous photon will be discarded. This should result in a loss of a few hundred photons over the course of a year. observationSim v8r1p2, for now still only in ST LATEST.
Max's recent additions of environment variables to control the writing of log files by PulsarSpectrum is in v9r5p4 of the Science Tools; I don't know any more details at this point. Presumably Max also is working on updating the documentation in the User Workbook.
User interface and infrastructure (& utilities)
Toby's update to the astro package - mentioned in last week's report, is in the v9r5p4 release. It makes the code more flexible regarding what orders of magnitude to expect for the spacecraft position information.
Source Catalog
Met last week. The technical topics included a presentation by Markus on initial work on a tool to fit extended sources. Toby gave a presentation about unbinned vs. binned spectral fits.
Science Tools Working Group
We won't have a meeting this week this week; there's not a lot of new news and the Catalog group, Service Challenge Steering Committee, and the Cosmic-Ray Electrons groups will be putting the 8-10 am Pacific time slot to good use. The GSSC has received a lot of feedback from GUG members who participated in the Beta Test; this may be a topic at the next Science Tools meeting. Surprise users are reporting success with the Science Tools.
The current version of the Science Tools is v9r5p3. Here are the changes since v9r5p2. The noteworthy one is that Jim has included CALDB files with v1 of the Pass 6 IRFs, the update being Riccardo's improvement to the fitting of the energy redistribution profiles.
Data products: No news
Databases and related utilities
No news
Likelihood analysis
No news
GRB tools
No news
Pulsar tools
From Masa: "James and I are still working on improving the time handling classes. It has been
slower than usual because I have been busy with other things." This is toward the end of handling multiple timing models for binary pulsars
Observation simulation
No news
User interface and infrastructure (& utilities)
Toby has updated astro package - in this version "the EarthCoordinate constructor checks the size of the position vector, so that it can be in m (as demanded by GLAST) or km (as returned by the ephemeris). After it computes the actual altitude, it throws an exception if it is not in the range 500-600 km. I don't know what the actual range of validity is, but that should be OK for GLAST." This addresses a problem with processing some FT2 files in recent versions of the Science Tools; Analia Cillis and Thierry Reposeur independently came up against the problem, which Jim tracked to the astro package. The change was made May 12, and is only in LATEST builds so far.
Source Catalog
Met last week. No technical topics - at least not algorithms - were discussed.
Science Tools Working Group
We won't have a meeting this week this week either; there's not a lot of new news and the Catalog group will be putting the time slot to good use.
The current version of the Science Tools remains v9r5p2. Jim has made preliminary versions of the Pass 6 IRFs available in the LATEST builds of Science Tools (1.2334 and later); see his news item.
Data products: Examples of all of the LAT Science Data Products have been produced, sent via FastCopy to the GSSC, and accepted by the ingest system there. Don Horner patiently reviewed iterations of the various attempted deliveries. David Band patiently updated the File Format Document (Rev A) as circumstances required. The GSSC and ISOC were able to sign off on the delivery requirements for these.
The LAT source catalog (LS-008) needed the most adjustments to the specification of the format - mostly small, but some omissions were found. Jean has produced a summary page of the format, with a couple of proposed tweaks to the current format; these will be discussed at a future Catalog meeting.
Databases and related utilities
No news
Likelihood analysis
No news
GRB tools
No news
Pulsar tools
No news
Observation simulation
No direct news. Max is adding or updating environment variable definitions to control the output of log files by PulsarSpectrum
User interface and infrastructure (& utilities)
No direct news. In the LATEST builds the versions of tip and hoops have been incremented by James. These have not been propopagated to releases yet, and so are presumably still being worked on.
Source Catalog
Met last week. Ludovic and Jean presented some results on Nicola's test gtobssim run of the Big Run sky model (Pass 5 Transient). Some words were said about the status of making resampled backgrounds for Science Tools analyses. Toby briefly introduced spectral fits in pointlike, including how the spectra may be stored.