1. Science Analysis Software
Gamma-ray Large Area Space Telescope
GLAST Large Area Telescope:
Science Analysis Software
Richard Dubois
Stanford Linear Accelerator Center

2. Outline Introduction to SAS
SAS Mission as defined by Level 3 Requirements and Milestones
Support of Engineering Tests
Level 1 Pipeline progress
High Level Science Tools development progress
Mission Ground Systems End-to-end testing
Preparation for LAT Ground System Peer Review and CDR
Cost and Schedule
Concerns
Summary

3. Level III Requirements Summary
Ref: LAT-SS-00020

4. Science Analysis Software Overview
Data Pipeline
Prompt processing of Level 0 data through to Level 1 event quantities
Providing near real time monitoring information to the IOC
Monitoring and updating instrument calibrations
Reprocessing of instrument data
Performing bulk production of Monte Carlo simulations
Higher Level Analysis
Creating high level science products from Level 1 for the PI team
Transient sources
Point source catalogue
Providing access to event and photon data for higher level data analysis
Interfacing with other sites (sharing data and analysis tool development)
mirror PI team site(s)
SSC
Supporting Engineering Model and Calibration tests
Supporting the collaboration for the use of the tools
Carbon-Fiber Wall

5. SAS Organization
Update science tools org under Seth

6. SAS in the Ground System
DPF is robotic backbone of IOC/SAS process handling –
Performs L1 & L2 processing
DPF server and database can handle
multiple arbitrary sequences of tasks:
L1 pipeline; reprocessing; MC; ….
Keep everything on disk

7. Level 1 Chain
3 GeV g
Real Data
Sim/Recon

8. Multiple Scattering in Converter Layers
100 MeV gammas
Mean angle: ~17 mr
Separation at next layer: ~550 mm
Strip pitch 228 mm
Barely resolvable into separate strip MeV!
MS blows up the opening angle significantly!
Mean angle: ~ 140 mr
Separation at next layer: ~4.5 mm
Easily resolvable
Note design:
Blue is “front” 12 3% X0 layers
Green is “back” 4 18% X0 layers
Last 2 have no radiator
To optimize interaction rate vs resolution
100 MeV g
vertical
x2 scale change!
Multiple scattering critical to tracking at low E
Use Kalman filter to account for large MS contributions
Apparent opening angle
T.Usher

9. Tracking Reconstruction Example 100 MeV Gamma
T.Usher

10. Sim/Recon Toolset applications - unique to GLAST Root, IDL – analysis
TkrRecon, CalRecon, AcdRecon, Astro sources
GEANT4 – simulation package
xml – geometry, parameters
Root – object I/O
Gaudi – code framework
doxygen – doc
VC++ – Windows IDE
gnu tools - Linux
vcmt – Windows, Linux gui
CMT – package version management
ssh – secure cvs access
cvsweb – www view of repo
cvs – file version management
utilities

11. Software Development Enable distributed development via cvs repository
Extensive use of electronic communications
Web conferencing (VRVS), Instant Messaging (icq)
CMT tool permits egalitarian development on Windows and Linux
Superior development environment on Windows; compute cycles on linux
Coding rules followed up with documentation and coding reviews
“Continuous integration”
Eliminate surprises for incoming code releases
Build code every night; alert owners to failures in build or running of unit tests. Results tracked in database.
Developing comprehensive system tests in multiple source configurations. Track results in database; web viewable.

12. Nightly Builds Display created from database query Performing builds
for Science Tools
now also
Display created from database query
Past release
Build status
Unit test status
Release in progress
Future release

13. System Tests
Comparison of
current to previous
release.

14. CU-Validated Sim/recon,
SAS Timeline GS
CDR GS
Peer Rev
LAT Cosmic
Ray Tests
MDC 1
TBD
LAT
CDR
GRT 4
2003
2004
2005
2006
GRT 1 EM
CU Beam
Test
MDC 2
FSW I-Sim MC
Sim/recon,
Proto pipeline
Sim/recon,
Proto SciTools,
Pipeline,
Data xfer to SSC
CU-Validated Sim/recon,
SciTools,
Final pipeline,
Data xfer to SSC

15. Engineering Tests Support – EM – mid 2003
See
LAT-MD – SVAC Plan
LAT-MD SVAC EM Tests spec, section 6.1
LAT-MD – I&T – SAS ICD for EM
LAT-TD – SAS Calibration Infrastructure
LAT-TD – Calibration Algorithms for EM
LAT-TD – EM Geometry for Simulations
SAS required to deliver
TKR, CAL subsystem calibration algorithms
Calibration infrastructure for time dependent parameters
Flexible geometry facility to describe EM unit
Reasonable fidelity simulation/reconstruction
Disk & CPU resources for simulation and analysis
Would like to run processing with the pipeline. Not required.

16. EM - 18 MeV on-axis photon (from VDG)
Engineering Model
Mini-Tower
(5 trays)

17. m m g g Cuts: TKR trigger Cuts: TKR trigger TKR - Number of TRACKS
TKR – number of CLUSTERS m m g g
Cuts: TKR trigger
Cuts: TKR trigger
Differential distribution
Differential distribution
Signal dominates
Signal dominates
Negative values are not shown
Negative values are not shown
I&T / E. do Couto e Silva

18. FSW MC Support for FE-Sim – late 2003
FSW has requested a full orbit’s worth of background to test the Front End Simulator
~50 Million events
~ secs per event
~500 GB output
Needed around Aug 2003
Must interface FSW code to output flight format data
Goal of using prototype pipeline to do the processing
MC/Sim already in place for this

19. Engineering Tests Support – CU – mid 2004
See
LAT-MD – SVAC Plan
LAT-MD SVAC EM Tests spec, section 6.1
LAT-MD – I&T – SAS ICD for CU
LAT-TD – Calibration Algorithms for CU
LAT-TD – CU Geometry for Simulations
SAS required to deliver (in addition to EM)
ACD subsystem calibration algorithms
Flexible geometry facility to describe CU
Good fidelity simulation/reconstruction
Disk & CPU resources for simulation and analysis
Processing Pipeline and Data Catalogue
Check these!

20. CU – 500 MeV angled electron (from test beam)
500 MeV e-

21. Functional Reqs in draft now
Level 1 Pipeline
Goal is to do early prototyping using EM and MC simulation runs as undemanding clients
Provide a general robot that can be configured to run any of the task chains we need
L1 processing
MC simulations
Data reprocessing
I&T/IOC tasks
Underlying database design complete
Evaluating STScI OPUS pipeline
Heritage from SLD experiment at SLAC
Then incremental improvements for more demanding clients like CU and Data Challenges
Flesh out the processing chain components
Design interfaces to make the pipeline portable
Generic database usage
Interfaces to submit processes to do the work
Docs:
database: LAT-TD-00553
server: LAT-TD-00773
diagnostics: LAT-TD-00876
Functional Reqs in draft now

22. Pipeline Server Layout

23. Working with Mission Ground Systems
Contact via biweekly GOWG meetings
Series of End-to-End tests being planned
SAS involved with GRT1 and GRT4
GRT1 (11/04)
First transmission of Level 0 data from MOC to IOCs
GRT4 (9/05)
Required Level 1 processing with transfer of results to SSC
Should already have been done in CU and MDC1
Support Mission GS PDR etc

24. Development of Science Tools
Extensive planning on which tools are needed to do science - and their requirements
One set of tools for all – “astronomy standard”
Had external review (9/2002) to see if we are on the right track
No major problems noted
In progress with the SSC
Joint oversight group
Sorted out technical basis (HEASARC standards; support of community; re-use of LAT developments)
Effort ramping up now
Have initial stab at Level 1 database technology
Looks like it will meet performance requirements
Starting to implement at GSFC and SLAC

25. Summary of Recent Developments
The most important policy decisions have been made regarding development of the software
Our tools will be FTOOLs and use the HEADAS libraries:
Data I/O through FITS files; existing types will be used as possible
IRAF parameter interface for prompting & specifying default parameter values
FITS I/O and IRAF parameter interface (an enhanced version of ISDC’s PIL) code will come from HEASARC-developed HEADAS
The LAT software development environment will be used:
CVS for storing the software
CMT for configuration and build management
DOXYGEN for documenting the code
C++ for new code
Support for Windows and Linux platforms
Scripting language: Python (probably)
Graphics (& GUI): Root (or plplot, with DS9)

26. Existing Software: ReUse Standards
We are NOT planning to reinvent all the wheels!
Existing FTOOLS (e.g., from the XRONOS suite) can do most of the pulsar analysis
Online access to astronomical catalog is provided at many data centers (e.g., CDS, HEASARC). HEASARC’s Browse may be the core of U9 and A2.
XSPEC can be used to fit GRB spectra binned in time and energy
Chandra’s SHERPA will be able to do GRB physical model fits
As already mentioned, graphics, image display (e.g., DS9), GUI interface, and scripting will of course come from existing software
Other tools will be considered.

27. Data Challenges Now traditional in HEP experiments
exercise the full analysis chain prior to needing it
involve the collaboration in science prep early
Doing planning now
Fall 2003
1 day’s data through full instrument simulation and first look at Science Tools
Fall 2004 – 1 month’s background/1 year signal
Test more Science Tools; improved Pipeline
Spring 2006
run up to flight – test it all!
DC1 Plans
Focus effort through Analysis Group (S.Ritz) and kickoff workshop in mid-summer
Including geometry and simulation validation
Sept collaboration meeting as milestone for start

28. Prep for GS Peer Review and CDR
SAS was baselined in PDR
Ground Systems CDR has been scheduled for 2/2004, with Peer Review in 11/2003
Expectations for Peer Review
Successful EM support
Level 1 Prototype operational
Functional requirements; Design documents ready
Science Tools
Major components understood, with schedule, manpower and milestones
Plan to schedule next external review to be coincident with Peer Review
ICD with SSC

29. 4.1.D Science Analysis Software
Cost/Schedule Summary

30. CCB Actions Affecting 4.1.D
Change Request #
Description
Status
LAT-XR
UW Manpower
Approved, $283K
LAT-XR
NRL Resource Leveling
Approved, $0K

31. Budget, Cost, Performance

32. Cost/Schedule Status Status as of February 28, 2003: Item In k$
Budget at Complete
3,611
Budgeted Cost for Work Scheduled (a)
1,170
Budgeted Cost for Work Performed (b)
1,160
Actual Cost for Work Performed
1,031
Cost Variance
129
11% of (b)
Schedule Variance -9
-1% of (a)

33. Manpower Plan

34. Concerns Manpower is the major concern No technological risks
“just” a matter of implementing and supporting the solutions we have designed for
Infrastructure group is thin, and hard to find people willing to do it.
SLAC, GSFC are providing much of that support
Mitigation
We concentrate on early starts to critical elements with incremental improvements over time.
Reuse appropriate software from other projects as much as possible
As much automation of repetitive tasks as possible

35. Summary SAS driven by Engineering Tests and LAT Integration
EM support in hand; CU looking good
Sim/Recon in good shape
Science Tools under development
In concert with the SSC
Drive schedule with Data Challenges
Level 1 Pipeline early start
Trying to have prototype in place for EM, FSW & DC1 support this year
End-to-end tests scheduled with Mission Ground Systems
Always need more people

36. Backups

37. Components of the Environment
User Interface aspects of the standard analysis environment, such as Image/plot display (UI2), Command line interface & scripting (UI4), and GUI & Web access (UI5) are not shown explicitly.
1 This tool also performs periodicity tests and the results can be used to refine ephemerides
2 These tools can also take as input binned data from other instruments, e.g., GBM; the corresponding DRMs must also be available.
Pulsar
ephem. (D4)
Level 1 (D1)
LAT Point source
catalog (D5)
Interstellar em.
model (U5)
Pointing/livetime
history (D2)
Astron.
catalogs (D6)
Level 0.5
IRFs (D3)
Alternative source for testing high-level analysis
Alternative for making additional cuts on already-retrieved event data
Pt.ing/livetime
simulator (O1)
Observation
simulator (O2)
extractor (U3)
Data sub-
selection (U2)
Data extract
(U1)
Exposure
calc. (U4)
Likelihood (A1)
Map gen (U6)
Src. ID (A2)
Event
display (UI1)
profiles (A3)1
Catalog
Access (U9)
Pulsar phase
assign (U12)
Pulsar period
search (A4)
GRB spectral-temporal
modeling (A10)
Source model
def. tool (U7)
Arrival time
correction (U10)
GRB temporal
analysis (A7)2
GRB LAT DRM
gen. (U14)
GRB spectral
analysis (A8)2
GRB event
binning (A5)
GRB unbinned
spectral analysis (A9)
GRB visual-
ization (U13)
IRF visual-
ization (U8)
Ephemeris
extract (U11)
GRB rebinning
(A6)2

38. Data Flow
Data recon + MC on disk. Abstract full-recon output into L1 DB for analysis
DPF
Italian mirror
French mirror MC
Recon
MOC
Calibs
IOC
L1 DB
Fully automated server, with RDB for data catalogue + processing state. Uses SLAC batch CPU and disk farms.
L2 DB
SSC
Parts of L2 processing also automated