Science Tools Working Group
Did not have a VRVS meeting this week, and with the ISOC workshop next week we won't meet again for 2 weeks.
The points of agreement and plan for near-term work (i.e., Action Items) that came out of GSSC-LAT Science Tools Meeting at SLAC, April 2007 have been added to that page.
The current version of ScienceTools remains v8r0p3. v8r0p4 will be on its way this afternoon, containing (at least) updates to PulsarSpectrum and gtmaketime.
Data products: No news yet this week
Databases and related utilities
No news
Likelihood analysis
No news. Work is continuing on understanding the differences in results (parameter values and TS values) between binned and unbinned likelihood analyses. Jean and Jim gave presentations on this at the Catalog VRVS meeting yesterday. Florien Kraft and Vincent Le Biez are running more simulations at SLAC like those Jim has made - in fact, starting with Jim's script. In the catalog analysis of the SC1 data, Jean is finding that binned likelihood analysis produces greater TS values and less biased results for source fluxes. In the simulations of point sources on isotropic backgrounds that Jim, Florian, and Vincent have made, the TS values from binned analyses are found to depend fairly strongly on the bin size - not a big surprise. So far the latter studies are finding that the unbinned likelihood analysis (on average) gives correct results, but Jean's studies using the SC1 data (which are much more complicated - including structured diffuse emission and faint sources and soures with a variety of spectra) appaear to find that unbinned results are biased.
GRB tools
No news
Pulsar tools
From Masa: "In the pulsar tools area, now gtpsearch and gtpspec are capable of creating a FITS file that contains a search result. They are available in the latest tagged version."
Observation simulation
Max has found a workaround to the problem of PulsarSpectrum sometimes wanting to look outside the time range of an FT2 file. This can happen owing to the nature of the arrival time corrections needed for pulsar simulations. He had worked out a way to catch this circumstance - as described before. The new method should be more robust.
User interface and infrastructure (& utilities)
I'll put it here: Jim has fixed a time precision problem with how gtmaketime defined starting and ending times when it operated on an FT1 file.
Source Catalog
Met this week; see above.
Science Tools Working Group
Did not have a VRVS meeting this week. Dave Davis, James Peachey, Chris Shrader, Tom Stephens, and Eric Winter are visiting SLAC from GSSC, and Toby Burnett was at SLAC on Wednesday. Discussions will continue through the day on Thursday. The discussion topics and approximate schedule are here. As of Thursday morning, we are somewhat ahead of schedule, although today we add one or two topics not on the posted agenda, like a discussion about ISOC operations tools and the coming ISOC workshop.
The current version of ScienceTools is v8r0p3. See what is different from v8r0p2. The updates from v8r0p2 include some related to the PSF normalization and the default behavior of makeFT1 that Jim described in his presentation at Monday's C&A meeting.
Data products: See the comments from David Band, Anders Borgland, Jonathan Ormes, and Tom Stephens on the issues that I had posted for the pointing and livetime history.
Databases and related utilities
No news
Likelihood analysis
No news. See the release notes diff linked above to read about the update to gtfindsrc in the current version of the Science Tools.
GRB tools
From James: "James P. worked with Eric Winter to check in and tag a new evtbin that contains the test data they generated previously."
Pulsar tools
From James: "James P. and Masa added the capability to the periodSearch package to write detailed search output into a FITS output file similar to that produced by the Xronos/Ftool powspec. They added this capability to gtpspec and gtpsearch as well as to the periodSearch unit test code." I haven't asked what 'detailed' means, and I'm not familiar with powspec.
Observation simulation
Max is working on a new fix to the problem of PulsarSpectrum sometimes wanting to look outside the time range of an FT2 file - owing to the nature of barycenter decorrection, PulsarSpectrum could want to look forwards (or backward) in time by up to the ~9 minute Sun-earth light travel time. His former method for trapping the error condition of trying to look outside the time range for which the FT2 file was defined broke when the format of the error string from the astro package was changed. Max needed a way to determine (in the code) the start and end times of the FT2 file, and to make it robust against the use of the start_time parameter; Jim has suggested a way to do this.
Richard has realized that his microQuasar source needs 'persistency' in the definitions of its outbursts if a simulation is divided into more than one time range, as it is for the long Service Challenge simulations. The symptom was that some of the microquasar sources did not look like they should have in the recent SC simulation.
User interface and infrastructure (& utilities)
No news.
Source Catalog
Did not meet this week.
Science Tools Working Group
Did not meet this week. The next meeting will be face-to-face at SLAC on April 18-19.
The current version of ScienceTools is v8r0p2. See what is different from v8. The updates from v8 are mostly small but important. Many relate to handling simulations (in flux, genericSources, GRBtemplate, GRBobs, and microQuasar packages), but the rest are in tools (and handoff_response) that you use.
Data products: Join David B., Anders, and Tom S. and get your comments in by tomorrow on issues related to cleaning up the definition of FT2 have been posted for comment.
Databases and related utilities
No news
Likelihood analysis
No news. The developments related to Likelihood in v8r0p2 of the Science Tools have already been reported on.
GRB tools
No news
Pulsar tools
Masa reported (in an e-mail that I lost track of and unfortunately cannot quote directly from) that gtpspec has a bug that he and James are working on fixing. The problem arises if the cancelpdot option is used and the timing information is read from an ephemeris database file, and the particular pulsar has several entries (different parameters for different time ranges) in the ephemeris file. I am not sure I understood exactly what the problem was but I guess that it related to the changing pdot values.
Observation simulation
In case this has not been mentioned already, Jim tracked down the reason that the last 2 days of the SC2 sky simulation were not generating any gamma rays - it had to do with a miscommunication between GPS and PulsarSpectrum related to PulsarSpectrum trying to recognize when GPS was about to complain about a requested time being outside the range of the FT2 file. The trick that formerly worked stopped working when GPS was updated.
Otherwise I'm not aware of any simulation development news for this week.
User interface and infrastructure (& utilities)
No news.
Source Catalog
Met this week. Jean presented further investigations of the results from the catalog analysis of the SC1 data set, including results of a light curve analysis run on monthly (instead of weekly, to save time) time intervals and comparison of results from unbinned and binned likelihood analysis. We also discussed in detail plans for making more careful and detailed comparisons of source detection algorithms - Tom S. will lead this effort.
Science Tools Working Group
Met this week. The next meeting will be face-to-face at SLAC on April 18-19.
The current version of ScienceTools remains v8.
Data products: Some issues related to cleaning up the definition of FT2 have been posted for comment. We had some discussion about TT vs. UTC and how what comes down in the LAT science data telemetry gets related to TT. Also, Anders has been thinking about questions like how versions of processing will be kept track of in the L1 pipeline and recorded with events or runs (and eventually find its way to the FT1 files). This is surprisingly subtle; most likely the CALIB_VERSION and RECON_VERSION keywords now defined for FT1 will need modification.
Databases and related utilities
No news
Likelihood analysis
Jim has added the gtexposure tool to the Science Tools. Basically, it calculates exposure vs. time for a given direction on the sky. It is described in his What's New in ST since DC2? presentation last week at the collaboration meeting.
GRB tools
From James: "James worked with Eric Winter to regenerate test data for the evtbin packages using DC1 data. This was a "warm-up" project but useful to do because the previous test data was very old and out of date. Fortunately, the update uncovered no new bugs."
Pulsar tools
Masa reports that the first version of gtpspec is available and it seems to work. Significances of results for separate runs for different Pdot values can be interpreted by keeping track of the trials factors that each run reports individually.
From James: "James and Masa finished rationalizing the time corrections in gtpspec and gtpsearch. All times are now corrected for pdot cancellation and binary demodulation, thus resolving the subtle bug described [previously]. In addition, they discovered that the period search tools were using TSTART/TSTOP from the event header to set up the start and stop times of the search, but that using the GTIs (start time of first GTI to stop time of last GTI) is more correct, and results in better searches. The effect is particularly strong in gtpspec. With all these changes, they checked in and tagged the first version of the
pulsar tools to contain the gtpspec binary."
Observation simulation
Jim has fixed a bug that caused a small (1/2 pixel typically) offset between the positions of diffuse sources in simulations and corresponding DiffuseSource counterparts in likelihood models. Riccardo Rando and Luigi Tibaldo had noticed this problem.
Max has implemented a new way to simulate timing noise - one he likes much better than his previous versions - and this is in the latest version of PulsarSpectrum.
User interface and infrastructure (& utilities)
No news.
Source Catalog
Did not meet this week.