Science Tools Working Group
Did not meet this week, but we will meet next week.
The current version of ScienceTools is v9r2. Here are the differences from v9r1. Enough updates have accumulated so that if you are a v9r1 user you should definitely want to switch.
N.B. v9r2 has the first versions of the Pass 5 IRFs, P5_v0_transient, P5_v0_source, and P5_v0_diffuse. You'll note in their definitions that they differ only in the cut applied to the variable CTBClassLevel. That is, these response functions are nested and (unlike DC2 class A and class B) do not partition the CTBClassLevel 'space'.
So this means that only one of these response functions can be used at a time in the Science Tools. gtobssim will accept any one of the response function sets (by name). Jim has modified makeFT1 (which makes FT1 files from Merit files) to by default apply the CTBClassLevel cut for the 'transient' class but also to include the CTBClassLevel variable values as an additional column in the FT1 files. This is so that you can use an FTOOL like fcopy to make the event selection be for source or diffuse class afterward.
Last week, Chris and David (B or D) presented a proposal for the next beta test by the GUC to the GLAST Users Committee. The presentation and minutes are not yet online; see agenda.
Data products: No new news.
Databases and related utilities
No development news.
Likelihood analysis
No development news. Jim has updated the tutorial for likelihood analysis; in particular the 'Create a Source Model XML File' section has been brought up-to-date. If you cannot wait until it is in the Workbook, it is available in the latest build of the doxygen documentation.
GRB tools
No development news.
Pulsar tools
Masa and James report that they are continuing to work on adding barycentering on-the-fly to the pulsar tools.
Observation simulation
No news. Yesterday Tom G. noticed strange behavior of some data generation runs (using Gleam) for the Operations Simulation next month. The effect was like changing the run start time during a run; GRBobsmanager may be implicated.
User interface and infrastructure (& utilities)
From James: "James Peachey delivered Ape to the LAT team for incorporation in the IExternal package for use in the Science Tools. Ape supersedes the PIL library which is currently still used by the Science Tools. The next step would be to deliver a new version of Hoops that uses Ape instead of PIL. After this step, many of the JIRA issues concerned with PIL problems could be closed.
James Peachey continued work on porting the Science Tools into the HEADAS Windows environment. He ran into problems with the CMake utility, probably caused by changes to CMake since the last time it was used. Currently he is experimenting with generating Visual Studio project
files directly for each package."
Source Catalog
Did not meet this week, and will make way for the Science Tools meeting next week.
Science Tools Working Group
Did not meet this week.
The current version of ScienceTools is still v9r1p1. HEAD1.581 (built on September 7) is also still the current release candidate. It differs from HEAD1.580 only in the GRBobs package, which now records more information about the bursts it generates; both of these have preliminary versions of the Pass 5 IRFs, as previously noted. Here are the differences between the most recent HEAD versions of the packages and the latest LATEST build.
Data products: No new news.
Databases and related utilities
No development news.
Likelihood analysis
No development news.
GRB tools
No development news.
Pulsar tools
Masa and James report that they are continuing to work on adding barycentering on-the-fly to the pulsar tools.
Observation simulation
No news. Max's feature article in this month's LAT newsletter includes a description of the inputs to PulsarSpectrum for simulating phase-dependent spectra.
User interface and infrastructure (& utilities)
From James: "James Peachey and Eric Winter began working to extend work done previously by Larry Brown to port the ScienceTools into the HEADAS build system. Specifically James attempted to use Larry's method to create a Windows distribution, starting with the external libraries. James was able to build most of the libraries, but he had problems using a more recent version of Cmake to generate working Visual Studio project files for the latest HEADAS. He is currently pursuing a work-around that will allow HEADAS to build, and will then continue with the Science Tools themselves.
James Peachey made some changes to Ape to support Visual C++ 8, then delivered version v2r2 to Navid Golpayegani for inclusion in the Science Tools external libraries."
Leftover from last week: Do we have plans to make native 64-bit builds of the Science Tools. Not that I'm aware of, but I do want to ask.
Source Catalog
Met this week. Ludovic presented more results on the improvement of source position accuracy with Toby's pointfit. Toby gave a detailed presentation of his results for finding sources in the obssim2 data set - taking into account the diffuse emission model in his analysis has markedly decreased the number of spurious sources detected. Jean presented a study of setting upper limits on expected significances of sources in the obssim2 sky model.
Science Tools Working Group
We met this week.
The current version of ScienceTools is still v9r1p1. HEAD1.581 (built on September 7) is the current release candidate. It differs from HEAD1.580 only in the GRBobs package, which now records more information about the bursts it generates. The next release, tentatively v9r2, will include versions of the Pass 5 IRFs that are designed to be used for analyzing the forthcoming Interleave dataset.
The plan at the GSSC is still for the next beta test by the GLAST Users Committee to take place in early December. Chris will be presenting the plan to the GUC next Monday, who either will or won't have other ideas about what should be done. The plan for the GSSC is to refer the GUC users to the User Workbook for documentation - and to work on cleaning up and updating documentation between now and December.
Data products: No new news.
Databases and related utilities
No development news.
Likelihood analysis
A couple of weeks ago, Toby added the package diffuse to the Science Tools. This looks like it can make allsky images, and write FITS files of them. It doesn't build any applications but is used by Toby's pointlike package, which does build an application called pointfind. The latter is not an FTOOL and does not have documentation in an obvious (to me, as I write this) location, and I didn't see it coming, but it is certainly related to the work Toby has been doing on source detection and characterization; the Catalog pipeline is using pointlike for refining the positions of candidate sources before they are run through standard likelihood analysis.
GRB tools
No development news.
Pulsar tools
James reports that he and Masa "continued to work on adding barycentering on-the-fly to the pulsar tools. Specifically, this week they began work to make the C code by Arnold Rots available through the new interfaces used by the pulsar tools libraries."
Observation simulation
See note above about GRBbos
User interface and infrastructure (& utilities)
gtgraph is now being built with the LATEST builds. According to James' documentation, "Gtgraph is a rudimentary package for plotting arbitrary combinations of fits columns in 2 and 3 dimensional plots. This was originally a test application for st_graph. It is not under active development." I haven't tried it myself and didn't know it was coming, although a plotting utility that would make people less likely to hate FITS files would be welcome. [gtgraph has actually been included in the ST LATEST builds since January 2005 - JC]
James and Eric W. have started working on automating porting the Science Tools from CMT to the HEASARC's hmake build system. They want to have this working in advance of the next GUC beta test.
Eric asked whether we have plans to make native 64-bit builds of the Science Tools. Not that I'm aware of, but I do want to ask.
Source Catalog
Did not meet this week.
Science Tools Working Group
We did not meet this week.
The current version of ScienceTools is still v9r1p1, and with HEAD1.580 remains poised for the next release.
Data products: No new news.
Databases and related utilities
No development news.
Likelihood analysis
- bug-fixes
- gtmodel now correctly produces output maps with image datatypes of float when the input counts cube map has datatype of integer (LK-35)
- The NewMinuit interface to the C++ version of Minuit did not account for parameter name collisions for spectral components from different sources. This is now fixed in optimizers v2r11.
- new functionality
- The best-fit parameters that have been found are stored so that when optimizers exit with an exception, thereby often leaving the parameters at a non-optimal set, the user now has the option of restoring the best fit values that have been found so far. This is available in pyLikelihood v1r1p2.
GRB tools
No development news.
Pulsar tools
No development news.
Observation simulation
- bug-fixes
- Toby modified the SimpleSpectrum source in the flux package so that it can model sources with photon spectral indices of exactly zero.
User interface and infrastructure (& utilities)
No news.
Source Catalog
Met briefly this week. Jean posted some updates to his investigations of the TS, Flux and spectral index distributions produced by Likelihood fits of the obssim2 data.