High-Level LAT Simulations

1
High-Level LAT Simulations S. W.
Digel Stanford Linear Accelerator Center M.
Razzano INFN and University of Pisa
2
High-level LAT simulations

• This is short-hand for what some call a Fast
  Monte Carlo
• The science tools include an observation
  simulator (gtobssim, by Jim Chiang) that uses
  the instrument response functions (hence
  high-level) to generate simulated data, rather
  than instrument-level interaction simulation,
  reconstruction, and classification (Gleam)
• gtobssim uses the same event generator
  (scheduler) as Gleam, and can read the same
  source specifications (although it simulates only
  gamma rays).
• Can use the same pointing history, too (more
  later)
• Advantages of gtobssim
• Easier to use than Gleam much, much faster (DC2
  gamma rays 10 CPU hours in gtobssim, 500 hours
  in Gleam)
• Response functions are by construction perfect

Ask me why all the tools start with gt DC2
Gleam run including (resampled) background was
20,000 CPU hours
3
gtobssim as a science tool

• Well, the obvious application is to be able to
  make a data set for which you know the truth
• For DC2, consider following Francescos advice to
  simulate a region of the DC2 sky that you are
  studying
• Some applications include evaluating
  sensitivities running multiple trials of some
  simulation
• Or you could evaluate effects of different
  observing strategies, if you could make up your
  own FT2 file (although this is not a goal of DC2)

4
Defining a gtobssim simulation

• Basically, you need a source model and a
  pointing/attitude/livetime history
• The source models are defined as XML file(s) see
  later
• The pointing history is an FT2 file. gtobssim
  pays attention to the livetime of each interval
  and enforces those livetime fractions in the
  simulation
• gtobssim will run without a pointing history file
  
• In this case you get a default orbit of some kind
  with idealized rocking, (probably) no SAA
  outages, and a uniform livetime fraction that you
  specify
• The resulting FT2 equivalent is then also written
  to a file
• gtorbsim can do this for you if all you want is a
  phony FT2 file

5
Source model specifications

• First question Can the source model files be
  the same as gtlikelihood uses? No
• Example source specification for a point source
• N.B. this is not really Vela
• (The SourcePop source is available for defining
  point sources in bulk)
• General rule fluxes are integrated over the
  energy range specified and are in m-2 s-1
• This is a source from the Flux package, one of
  the originals by Toby. Most of the new sources
  are SpectrumClass (by Jim unless otherwise
  noted - in the genericSources package) and have a
  simpler but less readable specification

ltsource name"vela" flux"0.00928"gt    ltspectrum
escale"MeV"gt        ltparticle name"gamma"gt
ltpower_law emin"30.0" emax"100000.
gamma"1.62"/gt lt/particlegt            
ltcelestial_dir ra"128.73" dec"-45.2"/gt
        lt/spectrumgt     lt/sourcegt
If broken power-law, ebreak and gamma2 tags are
accepted
6
More on source model specifications

• Example generic source from the Workbook
• Want to know what those parameters are? See the
  workbook
• Options Extended sources
• GaussianSource (2-dim gaussian power-law)
• MapSource (2-dim image power-law spectrum)
• MapCube (3-dim image 2 spatial one spectral)
• Isotropic (power-law)

ltsource name"periodic_source"gt ltspectrum
escale"MeV"gt ltSpectrumClass
name"PeriodicSource" params"0.1, 2.1, 1e3, 1,
0.75, 30., 2e5"/gt ltgalactic_dir
l"0" b"0"/gt lt/spectrumgt lt/sourcegt
7
More on source model specifications (2)

• Options Time-varying sources
• SimpleTransient Point source with power-law
  spectrum and a finite on time
• Transient template Like above but light curve
  can be specified in detail
• SpectralTransient Point source with broken
  power-law spectrum and detailed light curve can
  handle multiple sources and can attenuate
  according to redshift and its own idea of the
  Extragalactic Background Light
• PeriodicSource Point source with power-law
  spectrum and sinusoidal variation of flux
• GRBobs (Nicola Omodei) GRB simulator (Valerie
  described how this tells GBM simulation software
  about a given burst)

8
More on time-varying sources (3)

• More time-varying sources
• Pulsar Specified p and pdot, lightcurve in an
  ASCII file, power-law spectrum does not
  decorrect arrival times
• PulsarSpectrum (Max Razzano) detailed pulsar
  simulator

9
And nowcreate your pulsar!(I)
Part A Adding the pulsar general parameters to
a Datalist
Every pulsar to be simulated must be inserted in
a DataList file, containing the pulsar general
parameters that not depend on specific model
(i.e. total flux, ephemerides, etc..)
All the DataList files that PulsarSpectrum can
manage must be specified in a default file named
PulsarDataList.txt, located in the /data
directory.
2 - Add a line with myDataList.txt in the
PulsarDataList.txt file
10
And nowcreate your pulsar!(II)
Part B Creating a pulsar entry in a XML file
The remaining pulsar parameters must be defined
in a XML file. Part of them depend on the model
you want to use.
Here En1E6, E08E6, g1.6, b1.7 (out of page)
Lightcurve option 2 random generated 3
takes from a profile (named NAMETimeProfile.txt
and located in the /data directory)
For more informations, please see at
http//glast.pi.infn.it/soft/Pulsar/psrpisa.htm
11
PulsarSpectrum footnote on /data directory

• This is in the ST/celestialSources/Pulsar/vXrYpZ/
  data directory, or C\Glast\ScienceTools-vXrYpZ\co
  de\celestialSources\Pulsar\v1r1p10\data
• Alternatively you may define an environment
  variable PULSARDATA that points to where the
  pulsar definition files are.
• If you want that output files are redirected to a
  directory different from the current one, you
• This will contain the various log and ROOT 2d
  histogram files that PulsarSpectrum writes
• Except for the file SimPulsars_Spin.txt, which
  has the timing information formatted so that
  gtpulsardb can ingest it this is how you go from
  simulation to analysis with pulsars
• To create a D4-formatted fits file
• Gtppulsardb SimPulsars_Spin.txt and enter a name
  for the output file.

12
About time

• As far as the time-dependent sources are
  concerned, time is expressed in MET, Mission
  Elapsed Time, which is the number of seconds
  elapsed since midnight, January 1, 2001
  (MJD51910).
• Right now is approximately 162985200
• Yes, MET is inconvenient to use directly
• The start_date parameter of gtobssim allows you
  to specify your own reference date, so that the
  source specifications can be made with
  conveniently small time offsets
• No, I am not entirely sure right now that we are
  handling UTC (earth rotation-based) vs. TT
  (Terrestrial Time, absolute time, constant offset
  from GPS time) conversion carefully enough at the
  leap second level

13
Walk through the parameter file (1)
xml_source_file,f,a,"none",,,"File of flux-style
source definitions" source_list,fr,a,"source_names
.txt",,,"File containing list of source
names" scfile,f,a,"none",,,"Pointing history
file" sctable,s,h,"SC_DATA",,,"Spacecraft data
extension" outfile_prefix,s,a,"test",,,"Prefix
for output files" evtable,s,h,"EVENTS",,,"Event
data extension" simulation_time,r,a,86400,1,4e7,"
Simulation time (seconds)" livetime_frac,r,h,0.9,,
,"Livetime fraction" start_time,r,h,0,,,Simulation
start time (seconds wrt start_date) use_as_numeve
nts,b,h,no,,,Use simulation time as number of
events max_simulation_time,r,h,3.155e8,,,Maximum
simulation time (seconds) start_date,s,a,"2001-01-
01 000000",,,"Simulation start date"
Any parameter with an h is hidden (not
prompted for)

• xml_source_file can be a file that is a list of
  other XML files, too none means use the
  default libraries
• source_list is the file that lists the sources to
  be included from the libraries
• scfile is the FT2 file
• If you want to you can simulate by of events
  (instead of time)

14
Walk through the parameter file (2)
use_acceptance_cone,b,a,no,,,"Apply acceptance
cone?" ra,r,a,0,-360,360,"RA of cone center
(degrees)" dec,r,a,0,-90,90,"Dec of cone center
(degrees)" radius,r,a,20,0,180,"Acceptance cone
radius (degrees)" emin,r,h,30,20,2e5,"Minimum
event energy (MeV)" emax,r,h,200000,20,2e5,"Maximu
m event energy (MeV)" rspfunc,s,a,"DC2",DC2DC1A
G25,,"Response functions" max_effarea,r,h,1.21,,,"
Maximum effective area value" max_numrows,i,h,200
000,,,"Maximum number of rows in FITS
files" random_seed,i,a,293049,,,"Random number
seed" chatter, i, h, 2, 0, 4, "Output
verbosity" clobber, b, h, yes, , ,
"Overwrite existing output files" debug,
b, h, no, , , "Activate debugging mode" gui,
b, h, no, , , "GUI mode activated" mode,
s, h, "ql", , , "Mode of automatic
parameters"

• An acceptance cone and energy range can be
  specified if you want to, and the maximum number
  of events per file
• Setting gui to yes will tell the science tools to
  prompt for the parameter files values via a GUI

15
Tips and tricks

• The output event summary (FT1) files from
  gtobssim have an additional column called
  MC_SRC_ID
• For each event, this identifies the particular
  source that produced it the mapping is in the
  ltprefixgt_srcIds.txt file that gtobssim writes
• Yes, we wont have this in real life, but it is
  extremely useful for studying algorithms
• Of course you can also filter on this column with
  the fselect FTOOL
• What is more, most (eventually all) of the
  science tools that use FT1 files as input will
  accept an ASCII file with a list of event files
  gtbin requires _at_ as the prefix
• So, for example, you can keep your large,
  relatively time consuming Galactic diffuse model
  FT1 files separate from an additional component
  that you are generating

16
More on using gtobssim

• Read the User Workbook
• The section with descriptions of each of the
  sources and their parameters will be updated
  soon. It currently does not include steady point
  sources and some other useful sources
• The workbook also contains a useful worked
  example
• Also, Francesco Longo has regenerated the
  Checkout 3 sky model using the DC2 response
  functions, and the input and output files will be
  made available

17
ObsSim GUI

• Also by Jim Chiang
• Nice interface for defining models and running
  gtobssim
• Can also communicate with ds9
• Not 100 working under Windows at the moment
• Most useful if you want to pick and choose within
  a library that you have already defined (or has
  already been defined for you)