28 June '07: first rough sketch of what Bordeaux pulsar group could contribute to the upcoming collaboration meeting. Look for "Bordeaux" in the following draft agenda. ========================================================================== Monday, July 30 ========================================================================== Science Working Group Coordinators meeting SSAC meeting ========================================================================== Tuesday, July 31 ========================================================================== BORDEAUX: a topic not on the agenda -- update of hardware time tests. Valerie Connaughton told me (=Dave S) she aims to show the muon time coincidences between LAT & GBM. While essential for pulsars, it's already been shown, we're still waiting for the fix to the S/C FSW, and is a distraction from the analysis-focus of the collab meet. BORDEAUX: another topic not on the agenda -- where to point during L&EO. Dave S and Thierry R's work-in-progress will be presented at the Friday June 29 SO meeting, and is very pulsar-relevant. If it gets addressed at the collab meeting, it could simply be a slide or three in a talk given by e.g. Eric Grove, the L&EO czar. 8:30 am welcome 8:40 am LAT status 9:10 am Mission Status 9:30 am Publication and collaboration policy 10:00 am break 10:30 am Multiwavelength report 11:00 am Service Challenge status/overview 11:30 am Analysis Coordinator report 12:00 - 1:00 pm Lunch 1:00pm - 3:00 pm GRB, Transient analysis, and moving sources (including ASP and onboard analysis) (Francesco, Nicola) 3:00 - 3:30 pm Break 3:30 pm - 5:00 pm Variability studies (including ASP) [Gino, Richard] 5:00 - 5:30 Burst and Flare advocates discussion 5:30 pm Adjourn ========================================================================== Wednesday, Aug 1 ========================================================================== 8:30 am - 10:00am Periodic analysis (can include status of timing databases) [Alice, Patrizia] BORDEAUX: Lucas Guillemot is preparing a half-dozen slides to summarize pulsar Science Tools tests using Giant Radio Pulses (Crab, and B1937+21). We (me, he, and Denis D) are debating which other work would also be useful for a general audience. These slides can be presented by Dave S, or integrated into some other talk given by someone else. 10:00 am Break 10:30 - 12:00 Studies of extended sources (Jan, Olaf, Seth) BORDEAUX: no presentation, but Marianne is building a head of steam here, and will have a new thesis student on the subject in the Fall. 12:00 - 13:00 lunch 13:00-2:30 Spectral analyses (Isabelle, Jan) BORDEAUX: A topic that I (= Dave S) think is important for the pulsar group (for all groups!) is for the collaboration to agree on a tool and/or method to create SED's. At the end of this text I am appending a thread of e-mail messages on the subject. I don't expect to have any results to show... If asked I could give a short statement of the problem in the hope of stimulating a debate... Damien Parent has taken on two tasks for his thesis work, one of which is this, but he is just getting started. 2:30 - 3:00 Break 3:00 pm - 5:00pm Catalog analysis and systematic studies of likelihood (Gino, Jean, Robert) ========================================================================== Thursday Aug 2 ========================================================================== 8:30 - 10:30 Diffuse analysis and cosmic-rays (Troy, Philippe) 10:30 - 11:00 break 11:00 - 11:30 SPARE 11:30 - 12:00 Beamtest status and schedule for getting updates into MC. 12:00 - 1:30 Lunch 1:30 - 3:30 Instrument performance and sensitivity studies (Benoit, Luca) BORDEAUX A: Marianne has done substantial work to get IRF monitoring into the L1 pipeline. She uses Vela heavily. Her work will presumably be shown by the Perugia IRFmon accomplices, and will ressemble what Marianne will show at the SO meeting tomorrow (Friday 29 June). BORDEAUX B: Damien Parent's 2nd task is "IRF validation during L&EO" along the lines of what he showed at the November SC kickoff. Of course, Vela based and thus pulsar related. He is working towards having something to contribute to this session. 3:30 - 4:00 break 4:00 pm - 4:30 pm AGILE 4:30 - 5:00 Future simulation and analysis plans and summary (or spare) ========================================================================== Friday Aug 3 ========================================================================== 8:30 AM Start Additional analysis topics related to ISOC (to be posted soon) Other issues arising -- SO workshop I think maybe? 5:00 PM Adjourn ======================================================================= E-MAIL THREAD ABOUT BUILDING LAT SED'S (=Spectral Energy Distributions) From smith@cenbg.in2p3.fr Tue Jun 26 11:20:43 2007 Date: Tue, 26 Jun 2007 11:20:43 +0200 (CEST) From: D.A. Smith To: Grenier Isabelle , conrad@particle.kth.se, mcenery@milkyway.gsfc.nasa.gov, luca.latronico@pi.infn.it, B.Lott , digel@stanford.edu Cc: d.parent , M. Lemoine-Goumard , T. Reposeur Subject: IRF validation ; SED's ; collaboration meeting QUESTION 1 -- SED's Isabelle & Jan, You are listed as probable organizers of the "Spectral Analyses" session of the collab meet. We have a problem -- how do you convert the output of gtlikelihood into a spectral energy distribution? The functional fit is easy, but one also wants to superimpose the data points. We asked several people and have gotten three categories of answers, namely: i) XSPEC ii) cobbled up a tool of my own by hand iii) huh? We'd be willing to do some work in this direction for the collab meet, we'd like your advice to choose the right tools. Benoit Julie Seth maybe you have advice? QUESTION 2 -- IRF VALIDATION Benoit & Luca, You are listed as probable organizers for the "performance & sensitivity" session of the collab meet. At the November SC kickoff meet (link below), Damien & Lucas presented an idea of ours (link below). In a word -- apply std analysis cuts successively to e.g. Vela and compare MC to real efficiencies for each cut, as a validation step for the effective area. Damien plans to invest heavily in this study (when pass 5 is ready). We'd like your advice as to whether we're barking up the right tree. http://confluence.slac.stanford.edu/display/DC2/Draft+Service+Challenge+I+Workshop+Agenda http://confluence.slac.stanford.edu/download/attachments/21021/SCkickoff30Nov2006_v3_TR.pdf?version=1 THANKS, David et al. ===================== From lemoine@cenbg.in2p3.fr Fri Jun 22 12:34:16 2007 Date: Fri, 22 Jun 2007 12:53:10 +0200 From: Marianne Lemoine-Goumard To: komin nukri , d.parent , D.A. Smith Subject: Differential energy spectrum Dear Nukri, Here in Bordeaux, we are currently working on the spectral analysis of pulsars using the SC2 data. Until now, we were using the science tools to do that and especially the black box called "gtlikelihood". The problem is that the output is a number of counts for each energy interval, whereas we would like to obtain the points of the differential energy spectrum (as we do it for HESS for example). This should be feasible since we have the exposure maps and the IRFs. I think (but I'm not sure) that XSPEC has a routine that does this transformation from counts to differential spectrum points. Do you know how it works ? And if not, do you have any script/routine that does this conversion ? Thanks for all, Marianne ===================== From lemoine@cenbg.in2p3.fr Fri Jun 22 13:53:40 2007 Date: Fri, 22 Jun 2007 14:12:34 +0200 From: Marianne Lemoine-Goumard To: D.A. Smith , d.parent Subject: Spetcre différentiel Salut, Ca y est la réponse de Nukri m'a permis de me rappeler de XSPEC. Donc il y a bel et bien une fonction dans XSPEC qui passe de couts/cm2 s en photons /cm2s: elle s'appelle eeufspec. Il suffit donc soit de travailler complètement dans XSPEC, soit d'implémenter la fonction dans les science tools. Marianne ps: voici la réponse de Nukri: /Salut Marianne, Yes, it is posssible to do this kind of plots with xspec. Ones you loaded all the files and performed the fit you can do: plot eufspec or plot eeufspec I think the latter is what your looking for. You can also print the data points: tcloutr plot eeufspec x gives the energies. Using xerr you get the error in energy (i.e. the bin width) and you can get flux and its error with y and yerr. / ============================== From conrad@particle.kth.se Tue Jun 26 18:02:36 2007 Date: Tue, 26 Jun 2007 17:52:26 +0200 (CEST) From: Jan CONRAD To: D.A. Smith Cc: Grenier Isabelle , Julie McEnery , Luca LATRONICO , B.Lott , digel@stanford.edu, d.parent , M. Lemoine-Goumard , T. Reposeur Subject: Re: IRF validation ; SED's ; collaboration meeting Hi David, my personal answer would be iii). People in my institute use xspec for both the fit and the plot, people in the DMNP group use their own tool (based on python). I think developing a useable tool based on python would be my preferred choice. If you wish I can put you in contact with the people who are using this (I think Larry Wai developed it and Larry has left for greener fields, so that would be Ping Wang). cheers, Jan ========================================== From isabelle.grenier@cea.fr Wed Jun 27 00:59:46 2007 Date: Wed, 27 Jun 2007 00:49:37 +0200 From: Grenier Isabelle To: D.A. Smith , conrad@particle.kth.se, mcenery@milkyway.gsfc.nasa.gov, luca.latronico@pi.infn.it, B.Lott , digel@stanford.edu Cc: d.parent , M. Lemoine-Goumard , T. Reposeur Subject: RE: IRF validation ; SED's ; collaboration meeting Dear David, This is clearly a point to be discussed thoroughly at the meeting. One option is to use the likelihood to measure the flux and error in small energy bands. This process requires iterating between the power-law (or 2-power-law or log_parabola) spectrum chosen for the entire LAT band to compute the IRFs and the multi-bin results. This is clearly a pedestrian, CPU-hungry, inefficient way to plot SEDs. Gtlikelihood is not handling a 'staircase' multi-bin spectral fit. We should discuss its need at the meeting. The ASI guys have used XSPEC to fit spectra, but XSPEC does not show the resulting SED with error bars on nuFnu points. It shows the raw data and the fit to the data. It infuriates me in X rays each time I want to build an SED because you cannot show data points, only the best fit. Since XSPEC works from the model, convolves it with the IRFs and fits to the raw data, you are stuck as with the likelihood. Cheers Isabelle ==================================================== From luca.latronico@pi.infn.it Wed Jun 27 12:30:40 2007 Date: Wed, 27 Jun 2007 12:20:29 +0200 From: Luca Latronico To: D.A. Smith Cc: Grenier Isabelle , conrad@particle.kth.se, mcenery@milkyway.gsfc.nasa.gov, B.Lott , digel@stanford.edu, d.parent , M. Lemoine-Goumard , T. Reposeur , Philippe Bruel Subject: Re: IRF validation ; SED's ; collaboration meeting D.A. Smith ha scritto: Hi David, I am happy to organize this session with Benoit, and feel good for a first contribution to put in the agenda. Few comments: - cuts: we will not have pass5 in time for the collaboration meeting, but you should be able to use the so called interim pass5 variables and cuts within a week; you might consider using pass4 for the time being and refer to the following page for standard cuts: https://confluence.slac.stanford.edu/display/DC2/Summary+of+response+function+sets - data sets: it would be good to start making comparison now, well ahead real data comes in, comparing MC data generated with Gleam and those generated with obssim. We started discussing a lot such test to quantify IRF-induced systematics, I invite you to contribute on this in the C&A and IRF working group meeings Thanks, luca ============================================= From pingw@slac.stanford.edu Wed Jun 27 19:37:17 2007 Date: Wed, 27 Jun 2007 10:26:56 -0700 From: "Wang, Ping" To: D.A. Smith Subject: RE: IRF validation ; SED's ; collaboration meeting (fwd) Hi David, When you said SED, do you mean plotting the data and the flux from each component in the model? Like this example http://glast-ground.slac.stanford.edu/workbook/pages/sciTools_pyLikelihood_tutorial/pythonTutorial.htm Attachment is my example code to write out a file which can be read in and plotted by Hippodraw. This code is written based on the tutorial I listed above. Is this what you need? Regards, Ping ============================================= Date: Wed, 27 Jun 2007 10:34:30 -0700 From: James Chiang To: Damien Parent Cc: digel@razzle.stanford.edu, richard@slac.stanford.edu, jchiang@slac.stanford.edu Subject: Re: pyLikelihood Dear Damien, (...) Regarding SED plots, the issue is computing unfolded spectra from the counts data when you are fitting more than one source at a time. Xspec can do this since it is fitting the counts from just a single source (and finesses the issue of background subtraction). I can discuss the problem with doing this with likelihood at greater length later. -Jim > Date: Wed, 27 Jun 2007 19:19:19 +0200 > From: Damien Parent > MIME-Version: 1.0 > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Dear Jim, ( ... ) > > ii ) how do you convert the output of gtlikelihood into a spectral energy distribution? > The functional fit is easy, but one also wants to superimpose the data points. Can I use pyLikelihood to do that or ...? > > Thank you for your help, Damien.