Science Tools Working Group
Met this week.
The current version of ScienceTools remains v9r1p1. In the current LATEST builds by the Release Manager Jim has added an initial version of the Pass 5 response functions (designated Pass5_v0). This corresponds to Bill's 'Source' class; Jim has posted some details of the definition.
Chris et al. at the GSSC are starting to plan the next 'beta test' of the Science Tools by the GLAST Users Group (formerly GUC). This may be as soon as November.
Data products: No new news.
Databases and related utilities
No development news. Tom's server at the GSSC recently underwent an internal code review, and passed. Eric W. says that 3 nodes of the server/cluster are currently running; more will be turned on as power/cooling become available in the machine room at Goddard.
Likelihood analysis
No development news
GRB tools
No development news.
Pulsar tools
Masa and James have started to implement on-the-fly barycentering and are working out how to present that as an option to users of the pulsar tools. gtbary will still be available as a standalone tool in case anyone wants to export barycenter arrival time-corrected files.
Observation simulation
David mentioned that users (in the GRB group, I think) trying to fit simulated GRBs were having troubles that could be traced to misunderstandings about the simulations. Julie reported that Valerie (who will be at Goddard today for the latest GI workshop) is working on a revamp of the backgrounds for simulated GBM data.
User interface and infrastructure (& utilities)
James has fixed some bugs in APE that were found by Swift team members. In the next week or two he expects to be able to implement APE as a replacement for PIL in the Science Tools. This will have the side effect of clearing up a some old PIL-related JIRA issues.
Source Catalog
Met yesterday. Jean presented results that indicated that the DRMNGB optimizer has convergence problems (i.e., not necessarily reaching the maximum likelihood) in binned likelihood analysis. This is not a major problem - alternative optimizers are available in likelihood, and the convergence issues may be more related to the diffuse terms of the source models, which are not the focus of the catalog analysis.