Notes from Science Tools VRVS, December 13, 2006

[These notes are unedited and freely paraphrase the discussion; corrections are welcome.  SD]


C Patterson, D Band, J Ballet, S Digel, T Burnett, C Shrader, P Nolan, J Chiang, J McEnery, T Stephens, J Cohen-Tanugi, J Peachey, A Cillis, D Davis, J Carson, M Hirayama, W Focke (8:33)
 

News


SD - Posters for the GLAST Symposium - the GSSC is submitting some
CS - 4 technically-oriented posters - NRA overview (David), GSSC services (Shrader), archive and data operations (Don Horner), SAE overview (Dave Davis) - no duplication with LAT team
 

SD - From the ISOC we will have a poster on ASP (led by Jim), which will include the ASP products that will be available in the first year.  Rob will have an overall poster on the ISOC.


Likelihood


JC - A couple of minor items.  gtlikelihood in imaging likelihood - exception being thrown by code that James and Analia added (evaluating gamma function for Poisson fit checking) now made that optional.  Also David L wants to run PyLikelihood without CMT - re-enabled that - now easier because we have a central lib directory.  Could not get it to work reliably on Windows - but David L wants it on Linux

DB - Toby has created code for flux limit has a function of position - how strong a source gives a TS within a certain range of time in a certain direction.  We are in the process of getting this code here but I understand there's interest in this as a general tool.
TB - Benoit has now produced indepedently a ROOT macro which is supposed to do the same thing - I have not checked it yet.  One thing it does is produce a sensitivity map for the whole sky at once, which I had not really attacked.
DB - One thing we wanted to do was to give results for a point source on a flat background.  Toby has provided the link but

SD - gtexpmap
JC - gtexpmap is accounting for the exposures to the point sources - does indeed need to calculate the angular integrals
JB - I have not tried it yet - I ran into the LK-24 issue that I see you have solved but is still not on the standard version that we use.  gtexpmap ran on the wrong energy range.

JC - I've been working on the problem of optimizing the integrals for the angular integrals - will present something on Friday - not the answer but maybe people will have ideas


GRB
 

DB  - No news on tools but might interest people as another tool to help people prepare proposals and for general use we have an interface to Xspec to do simulations and have GBM files there.  We are going to have the capability to show the burst spectrum of different strengths - and to download sample response files for simulations.  And files with different backgrounds (different intensities).
DB - Uses the fakeit command of Xspec - input information on a Web interface - it runs fakeit in Xspec to generate simulated data and then fits it - Google 'Webspec' which has a general version for X-ray astronomy - several missions - we have created a GLAST version - not generally available yet - at the moment has 6 GBM cases and 3 LAT cases (different orientations for GBM, different backgrounds for the LAT) and different spectral models.  To run it now (development version) you have to be at Goddard. 
SD - Different background levels?
DB - For the LAT this is not necessarily just for GRBs - The response is not really for GRBs only - at least one of the matrices available - this is not really how people will analyze LAT data but will make a counts spectrum possible
SD - So backgrounds must be celestial.  I'm not thrilled with this approach - does not admit analyzing complicated (i.e., real) regions - no imaging
DB - Yes, quick and dirty - after all these years of denigrating Xspec I find it ok for quick and dirty - for Cycle 2 proposers we will encourage people to use gtobssim
JC - Who is working on this tool?
DB - The tool itself has existed for a long time - our input is response files and background files - I created using DC2 data.  For GBM I used the response functions that Marc Kippen created ages ago (nothing more recent is available).  You provide response files and background files
JC - I'm wondering whether we can generate the background files on the fly based on the regions that the user selects - e.g., 1 mo of 3C 279 - maybe have a 1-year canned simulation sitting there - filter it based on MC_SRC_ID - make it into background file - run fakeit, etc.
DB - That sounds feasible - frankly I wanted to take advantage of the existing software in terms of the resources that we have available.  I understand the caveats, but want to provide something that gives at least correct sense - gets complicated of course as get more real - including event classes, backgrounds.
TS - Related to that is the person doing that is not in the GSSC - no resources to change the interface
JC - In Webspec can the users provide their own background files?
CS - But users could download the response that we provide and run Xspec themselves
DB - Idea is to provide simple, Web (no software installation) tools for investigating performance - for cycle 1 we are not releasing the SAE anyway.


Pulsar tools
 

MH - Starting to respond to GUC issues - like overwriting time column with gtbary - problem downstream for confusing likelihood tool.  We have come up with a different solution, posted on my Web page - see also JIRA issue.  We also started looking into the detailed design of the pulsar ephemeris database - James and I and GSSC people involved with event data.  We have made a couple of bug fixes and will make a couple of minor interface tweaks in response to GUC comments.
SD - Barycentering on the fly would mean getting into the Chandra code that you wanted to leave alone
MH - Right - we don't want to change it heavily - maybe like a function.  Also we want to keep gtbary - so can have path to Xronos-type tools. 
JP - Most of the structure is there - we have a general way of representing time.  It does use the code that we got from Arnold Rots but does work with our formalism
SD - Database is something behind the scenes, not used by the pulsar tools, right?
MH - This would be something that controls new entries, replacing, logging.  Technical details about building the master pulsar database file that you already know.


Obs simulation
 

TB - Sun and moon reference frames are available for diffuse sources.  Idea is to have a user-defined source but photon directions are relative to the positions of the sun and moon.
JC - So it works?
TB - No it doesn't.  Assumes that the coordinates are in Galactic coordinates - so it converts them to the wrong place - an 0.5 deg pixel map is pretty crude.
DB - Sounds like this is a source that will be defined for simulation.  How about for likelihood analysis?
JB - Presumably you can analyze them by putting the photons in the coordinate system of the source - you need something to concentrate the photons to model the spectrum, then redistribute them over the track of the object in real space

SD - If modeling a source in a moving reference frame, you'd need to work out how to transform diffuse model into that frame.


Infrastructure
 

JP - I have reconciled ape with other parameter interfaces - some differences with Chandra tools.  I need to retrofit hoops to use APE on the backend - I have started that - need a couple of more days - I'd like to do it over the quiet holiday period
SD - What changes from perspective of a science tools user?
JP - Will be impossible to recompile without readline - also automatic handling of environment variables.  Also PIL command-line problems will be resolved.  A number of JIRA issues that are PIL bugs will be addressed by this - small annoyances will go away.
TB - Regarding TIP - I haven't gotten a chance to go back to see what went wrong when I upgraded to the new version but I noticed that the unit test is failing - have you noticed?
JP - Windows-only?
JC - Yes
JP - This is a bizarre problem that I cannot reproduce when I run the test myself.  Sounds like an environment issue - tries to delete keywords via a cfitsio call - I need to talk with Navid about the environment.  This does not sound like the problem you have describe but if you send me a test case I can try it
JC - What is failing?
TB - Reading a FITS table


User workbook
 

CP - No new contributions since our last meetings - workbook is current and no changes are in the pipeline

DB - Regarding Cicerone, Jim Buckley (GUC) handed me an enormous wad of printouts marked up, and it is in my briefcase for getting to.
 

Data products
 

SD - David has gently reminded me that some long-standing data products need to be cleaned up - like LS-007, which was the original concept of a transient alert.  I'd like to set up a modern counterpart to the original Data Products Working Group to thrash these out.  Some Science Operations products, not deliverables to the GSSC, will also benefit from attention from such a group.


DB - Renewed interest here in getting the data products finalized.  One thing we never pinned down is periodic reports on the status of the LAT - things like number of tiles dead, etc. We have talked about this and had preliminary lists and gone around and around - we should close that as soon as possible

CS - One note - we are in the process of compiling the feedback from the beta test into a more concise set of action items - and we had discussion at are our meeting yesterday.
SD - And Dave has plans to put a high-level summary in the science tools development directions page
DD - It is there now - although the format is not great yet - Confluence did not like my browser

Wrap-up


SD - This is obviously the last meeting of the year.  We'll meet at TBD date early next year