Skip to content

Latest commit

 

History

History
172 lines (121 loc) · 7.25 KB

README.md

File metadata and controls

172 lines (121 loc) · 7.25 KB
Raw

Eventdisplay Analysis Scripts for CTA

DOI Codacy Badge

Run scripts for CTA. Allows to efficiently run all analysis steps starting from the raw MC files to sensitivities using Eventdisplay.

Installation

Directories

All scripts expect the following setup for directories. Any deviation from this will break the scripts.

Main directory for all analysis products:

${CTA_USER_DATA_DIR}/analysis/AnalysisData/${DSET}

Main directory for all software and auxilaury files plus all log files from the analysis:

${CTA_USER_WORK_DIR}/analysis/AnalysisData/${DSET}

$DSET is the name of the data set to be analysed, e.g. prod3b-paranal20deg_SCT-sq08-LL.

The following subdirectories are expected:

  1. ${CTA_USER_WORK_DIR}/analysis/AnalysisData/Eventdisplay_AnalysisScripts_CTA - the directory with this repository
  2. ${CTA_USER_DATA_DIR}/analysis/AnalysisData/FileList_prod3b - directory with the lists for the simulation files (see below)
  3. ${CTA_USER_DATA_DIR}/analysis/AnalysisData/{DSET} - the directory with all analysis products (several subdirectories will be prepared from the analysis scripts)
  4. ${CTA_USER_WORK_DIR}/analysis/AnalysisData/{DSET} - the directory with all and auxiliary files

Installation of scripts and Eventdisplay

Requires root to be installed and $ROOTSYS to be set (use ROOT versions >=6.20)

Install Eventdisplay_AnalysisScripts_CTA from github and select the corresponding branch to work with (e.g., prod5-v08):

git clone https://github.com/Eventdisplay/Eventdisplay_AnalysisScripts_CTA.git
cd Eventdisplay_AnalysisScripts_CTA
git checkout prod5-v08

Install and compile eventdisplay (expect all Eventdisplay repositories with same branch names):

cd install
./prepareProductionBinaries.sh <data set> prod5-v08

This installs the following packages:

  • hessioxx
  • sofa
  • Eventdisplay analysis files
  • Eventdisplay code

Note that only the data set name needs to be given here (e.g., prod3b-paranal20deg_SCT-sq08-LL), and not a full path name.

Path setting

Before running any scripts, the correct paths for all executables and libraries needs to be set.

cd ${CTA_USER_DATA_DIR}/analysis/AnalysisData/Eventdisplay_AnalysisScripts_CTA
source ./setSoftwarePaths.sh ${DSET}

List of MC files for the evndisp stage

For the first stage of the analysis, a list with file names (full paths) for all MC input files needs to be filled.

Expected directory structure (first level):

  1. prod3b: ${CTA_USER_DATA_DIR}/analysis/AnalysisData/FileList_prod3b/
  2. prod5: ${CTA_USER_DATA_DIR}/analysis/AnalysisData/FileList_prod5/

Second level directory structure expects a substring of ${DSET} (without the analysis version number). E.g. for the example of prod3b-paranal20deg_SCT-sq08-LL, it should be only prod3b-paranal20deg_SCT.

In this directory, the list of files for the different particle types are:

  • gamma_cone.list
  • gamma_onSource.list
  • electron.list
  • proton.list

e.g.,

/lustre/fs24/group/cta/prod3b/CTA-ProdX-Download-DESY/Prod3b_Paranal_20deg_HB9//electron/electron_20deg_0deg_run945___cta-prod3-sct_desert-2150m-Paranal-SCT.simtel.gz
/lustre/fs24/group/cta/prod3b/CTA-ProdX-Download-DESY/Prod3b_Paranal_20deg_HB9//electron/electron_20deg_0deg_run2962___cta-prod3-sct_desert-2150m-Paranal-SCT.simtel.gz
/lustre/fs24/group/cta/prod3b/CTA-ProdX-Download-DESY/Prod3b_Paranal_20deg_HB9//electron/electron_20deg_180deg_run2634___cta-prod3-sct_desert-2150m-Paranal-SCT.simtel.gz
/lustre/fs24/group/cta/prod3b/CTA-ProdX-Download-DESY/Prod3b_Paranal_20deg_HB9//electron/electron_20deg_180deg_run4722___cta-prod3-sct_desert-2150m-Paranal-SCT.simtel.gz
/lustre/fs24/group/cta/prod3b/CTA-ProdX-Download-DESY/Prod3b_Paranal_20deg_HB9//electron/electron_20deg_0deg_run2762___cta-prod3-sct_desert-2150m-Paranal-SCT.simtel.gz

Running the analysis

Central execution scripts are CTA.mainRunScriptsReduced.sh and CTA.runAnalysis.sh. In the best case, no changes are required to these scripts.

e.g., to run the first step of the analysis with evndisp, do

./CTA.mainRunScriptsReduced.sh prod6-Paranal-20deg-dark-sq10-LL EVNDISP

(or set any other data set, as outlined in ./CTA.mainRunScriptsReduced.sh)

To submit script, check the log file directory printed to the screen (the directory with the UUID) and then run:

./utilities/submit_scripts_to_htcondor.sh <log file directory> submit

Try this first without the submit argument and check the submit.txt file. This assumes the HTCondor job submission system. Gridengine will work after changing the variable SUBC from condor to qsub in the scripts analysis/*sub*.

The script ./CTA.mainRunScriptsReduced.sh does the following:

  • read a list of arrays from a subdirectory specificed for your data set in ./CTA.runAnalysis.sh (e.g., prod3b/subArray.prod3b.South-SCT.list)
  • execute scripts to submit jobs from the ./analysis directory
  • all output products are written to ${CTA_USER_DATA_DIR}/analysis/AnalysisData/${DSET}
  • for all telescope multiplicity dependent analysis, this is done for the multiplicities defined in NIM-South.dat and NIM-South-sub.dat.

On the list of arrays:

  • arrays are defined by the telescope numbering as defined during the simulations.
  • array layout definition files can be found in $$CTA_EVNDISP_AUX_DIR}/DetectorGeometry

For a complete analysis, one needs to cycle through all reconstruction steps in the following sequence:

  1. EVNDISP - calibration and image analysis
  2. MAKETABLES and DISPBDT - lookup table filling and disp BDT training (can be done in parallel)
  3. ANATABLES - stereo analysis using lookup tables and disp BDTs
  4. PREPARETMVA - write data products needed for BDT training
  5. TRAIN - train BDTs for gamma/hadron separation
  6. ANGRES - determine angular resolution for 40% signal efficiency
  7. QC - determine data rates after quality cuts (used for cut optimisation)
  8. CUTS - optimise gamma/hadron cuts and calculate instrument response functions
  9. PHYS - fill instrument response functions

Testing

Running the analysis is complex and involves the reading and creating of many files (possibly >100k files). Testing the results for consistent is important; please look into the testProduction directory for testing scripts.

Utilities

Set links to EVNDISP production directory:

cd utitilities
./linkEvndispProduction.sh prod5-LaPalma-20deg-EVNDISP prod5-LaPalma-20deg-v02-LL ../prod5/subArray.prod5.North-noHyper-N.list

Set links for hyper array analysis:

cd utiltities
./prepareHyperProduction.sh prod5-LaPalma-20deg-v01-LL prod5-LaPalma-20deg-h01-LL ../prod5/subArray.prod5.North-noHyper.list

Count number of files in production directories:

cd utilities
./countFilesinProduction.sh prod5-LaPalma-20deg-EVNDISP ../prod5/subArray.prod5.North-noHyper.list EVNDISP

e.g., cross checks that number of EVNDISP files is correct

Licence

License: BSD-3 (see LICENCE file)

Contact

Gernot Maier