Science Tools Working Group
Did not this week. We will meet next week, on Nov. 1. The U.S. will be back on Standard Time, but as far as I can tell so will Europe.
The version of ScienceTools remains v7r5. Navid has made DC2-like distribution packages of v7r5 for all of the platforms except MacOS. The motivation for this was the GUC test mentioned last week. Navid has posted them on DC2 code releases; as these don't have anything in particular to do with DC2, eventually this probably should live in the Science Tools space.
Data products: No news.
Databases and related utilities
No news.
Likelihood analysis
No news. Analia will give a presentation at next week's Science Tools meeting on her Python script that makes TS contour maps. As far as I know the script is not yet in the CVS repository at SLAC.
GRB tools
No news.
Pulsar tools
From James: "James and Masa continued work on the A4 tool. This week they succeeded in getting the test code to perform FFTs of the test data sets with reasonable looking results." A4 is the tool for blind searches for periodicity.
Observation simulation
No news. The Service Challenges have a working group on sky simulation - that does not have a mailing list or Confluence page yet. The coordinators are Nicola and Alice Harding. For the first SC run - next week - only currently-existing source types will be used.
User interface and infrastructure
James reports that he is learning from Larry Brown, who I infer is leaving HEASARC and who you might recognize as the author of fv, how he has figured out how to build the Science Tools in HEASARC's hmake system. This is how the GSSC will distribute the tools to guest investigators.
Source Catalog
Did not meet this week and will not meet until Nov. 8.
Science Tools Working Group
We met this week. We may meet again next week, to get out of phase again with the Catalog group; Nov. 1 is a national holiday in France, and the Service Challenge Steering committee occasionally has designs on the time slot, so nothing is definite yet regarding the next Science Tools meeting.
At the meeting this week, Jim gave a overview of the charge and current activities of the IRF working group, as he has also been asked to present in this meeting.
The version of ScienceTools is now v7r5. This release will be what the GLAST Users Committee will use at Goddard on November 16-18 to 'beta test' the science tools. According to Chris, about half of that time will be devoted to hands-on use of the tools. Navid is preparing DC2-like distribution packages of v7r5 - probably he has done it already - and GUC members will use these (downloaded from somewhere on the GSSC Web site, I think).
Data products: No news.
Databases and related utilities
No news. Dan's Data Handling report will have any relevant news from SLAC.
Likelihood analysis
Jim fixed a TS calculation bug (reported by Jean) in pyLikelihood.
Analia is continuing to work on the python script described last week to produce TS contour maps for model parameters. The current plan is to not make these parameters be coordinates of sources - owing I think to the computational overhead. Analia may have the script in SLAC CVS next week. This script will not be used in the GUC Beta test.
GRB tools
James described a quick fix that he has implemented in the now-released version of gtrspgen. Artifacts in the response matrices that it generated were traced to the energy dispersions not being normalized for the DC2 response functions, i.e., the normalization was energy and angle dependent. The calculations in gtrspgen now enforce normalization.
Pulsar tools
Masa reports that development is continuing on the tool for blind searches for periodicity (A4). This will be via discrete FFTs, possibly segmented in time to limit memory requirements at the cost of some sensitivity. Pat may offer pointers about methods for large FFTs that do not require correspondingly large memory spaces.
Observation simulation
Max described his plans for further development of PulsarSpectrum. In addition to the MJDREF time system correction (checked in and in v7r5) he has implemented a method for describing phase-dependent spectra for the use of the simulator. It uses a 2-dimensional histogram. We didn't say it at the meeting but I'm fairly sure that this is a code word for ROOT file. He is also developing detailed specifications of these histograms for the bright EGRET pulsars. Also, Max has been researching timing noise, and consulting with Alice about it. He plans to implement timing noise as an option (actually 2 different options). This will be important for realistic studies of performance of, e.g., pulsar searches. As he makes the code available Max will also update the documentation for the Workbook.
User interface and infrastructure
No news. Dave, Chris, and James have been discussing implementing GUIs for the science tools via st_graph. Development is still on the horizon; everyone is busy.
Source Catalog
The catalog group met this week. Jean presented detailed results from likelihood analysis of the 'test pattern' files. You might be intersted to see how well likelihood did at resolving closely-spaced sources (with known positions), and how well the optimal filter algorithm did in finding them in the first place. We also continued the discussion about standardizing on counterpart catalogs to be used to evaluate candidate identifications for LAT point sources.
Science Tools Working Group
We did not meet this week. The tentative plan is to meet during our usual time slot next Wednesday. Jean has offered to push back the Catalog VRVS meeting by one hour.
The current version of ScienceTools remains v7r4p1. The next release will be at the end of the current 6-week build cycle, Oct. 13 - this Friday. Barring surprises, this release will be what the GLAST Users Committee will use in mid-November to 'beta test' the science tools. Two members of the GUC, Reshmi Mukherjee and Greg Stacy, will be trying to install and run the tools ahead of the GUC beta test.
Data products: No news. I've queried Jim and Toby for advice about the issue that David raised regarding the date strings in DATE-OBS and DATE-END keywords in FT1 and FT2 files. As David points out, the time system is supposed to be TT, but the strings claim to be in UTC. From looking at the code, I am not sure that the time system can be UTC; at least it does not seem to be aware of leap seconds.
Databases and related utilities
No news. Dan's Data Handling report will have any relevant news from SLAC.
Likelihood analysis
From Analia: "Analia Cillis wrote a python script to produce TS contour (confidence levels) plots in the parameter space. A first fit is generated using python likelihood, and the values of the model parameters are obtained along with the log likelihood value. Afterwards, the procedure to obtain the boundaries of the TS contour map in the parameter space consists in the following:
- One parameter is stepped (to the right and left side of the value obtained with the first fit). For each of these new values, which now are fixed, python likelihood is run again to produce a new fit until (eventually) a new maximum of log likelihood is found.
- From this new maximum and assuming a Chi2 distribution the value of this parameter is obtained until a given of confidence level. The same procedure is done for the second parameter.
- After the boundaries in both parameters are obtained, python likelihood runs several times in a nxn grid (with the boundaries obtained in the previous step) with the first and second parameter fixed.
- Finally the contour plot is produced with parameter 1 in x-axis, parameter 2 in y-axis and confidence levels in z-axis. This procedure may be repeated for each pair of parameter."
GRB tools
No news
Pulsar tools
From Masa: "In the pulsar tools development area, we finished revising the Workbook tutorials for the latest versions of the pulsar analysis tools. Then, we started detailed design and implementation of A4 tool [blind periodicity search] last week. My hope is to make it work soon and run it with DC2 data (or something equivalent) to see how realistic the tool is with typical GLAST observations."
Observation simulation
Max has implemented the correct specification of MJDREF (expressing it as the TT version of midnight Jan. 1, 2001) in PulsarSpectrum (v2r1p2). I do not know yet for sure whether this has been cross checked for consistency with the current versions of the pulsar tools.
Max's plans for further development for PulsarSpectrum may be inferred from the page on Service Challenge plans for the PSR/SNR group.
User interface and infrastructure
No news
Source Catalog
The catalog group did not meet this week. Last week's meeting included presentations by Jean on source detection in the DC2 test images and by Olaf and Juergen on catalogs for source identification. Regarding source detection in the DC2 'test pattern' images, Jean has run a likelihood analysis and finds the fluxes to be systematically underestimated relative to the model (typically by about 20-25%). In possibly-related news, a recent long run of gtobssim by Richard (DC2 sky model for 1-year period) produced about 30% fewer gamma rays than expected from a simple extrapolation of the number of class A+B gamma rays in DC2. These need diagnosing.