Running SLIC Events in Pandora PFA New

Overview

This tutorial will show you how to process SLIC output files with the Pandora PFA New package, currently the best performing PFA algorithm, in order to produce Particle Flow Objects (PFOs) (also called ReconstructedParticles) for analysis. It covers building and installing a slic-specific frontend called slicPandora and all its dependencies. Then it will go through the steps for processing slic output using LCSim in order to prepare it for Pandora. Finally, it shows how to run slicPandora using these events. There is also a note on how to generate Pandora detector XML files from compact descriptions.

Quick Setup Instructions

These commands can be executed to produce a working slicPandora without monitoring enabled, using a shell such as bash.

cd /my/work/dir
svn co https://svnsrv.desy.de/public/ilctools/CMakeModules/trunk/ CMakeModules
export CMAKE_MODULE_PATH=`pwd`/CMakeModules
cvs -d :pserver:anonymous@cvs.freehep.org:/cvs/lcd co lcio
cd lcio
cmake .
make
export LCIO_HOME=`pwd`
cd ..
svn co https://svnsrv.desy.de/public/PandoraPFANew/PandoraPFANew/trunk PandoraPFANew
cd PandoraPFANew
cmake -DCMAKE_MODULE_PATH=$CMAKE_MODULE_PATH .
make
export PandoraPFANew_HOME=`pwd
cd ..
cvs -d :pserver:anonymous@cvs.freehep.org:/cvs/lcd co slicPandora
cd slicPandora
cmake -DROOT_HOME=$ROOTSYS -DCMAKE_MODULE_PATH=$CMAKE_MODULE_PATH -DPandoraPFANew_HOME=$PandoraPFANew_HOME .
make

You need to have a preexisting installation of ROOT, which is not covered here.

More detailed instructions are listed below.

Detailed Setup Instructions

Preliminaries

A Linux or Unix platform is assumed and the bash shell is used for all command-line instructions. The actual setup procedures have been tested on Redhat Enterprise Linux 5.5 (Tikanga).

The cmake tool must be installed for building several of the packages.

You will need the standard GNU tools installed, e.g. Make, gcc/g++, etc.

The cvs and svn command-line tools are also required for obtaining project source codes.

Dependencies

The slicPandora package currently has the following dependencies.

Installing Packages

LCIO

Checkout the LCIO project from cvs.

cd /workdir
cvs -d :pserver:anonymous@cvs.freehep.org:/cvs/lcd co lcio

Building this project requires the CMake tool, the installation of which is not covered here. It additionally requires command-line GNU Make.

cd lcio
cmake .
make

The cvs head or last release tag should both work.

Set the LCIO environment variable.

export LCIO=`pwd`

This variable will be used by subsequent packages to locate LCIO and make a setup script.

CMakeModules

Checkout the head of CMakeModules, required by PandoraPFANew.

svn co https://svnsrv.desy.de/public/ilctools/CMakeModules/trunk/ CMakeModules

Make an environment variable that will be referenced later.

cd CMakeModules
export CMAKE_MODULE_PATH=`pwd`

This package does not need to be compiled. It just provides some CMake macros.

PandoraPFANew

Check out the version of Pandora that will be used. The example below shows how to checkout the head version.

svn co https://svnsrv.desy.de/public/PandoraPFANew/PandoraPFANew/trunk PandoraPFANew

Now we need to build another package before configuring Pandora.

PandoraMonitoring

Checkout the head version of PandoraMonitoring.

svn co https://svnsrv.desy.de/public/PandoraPFANew/PandoraMonitoring/trunk PandoraMonitoring

Now configure the build using cmake.

cmake -DROOT_HOME=/path/to/root -DCMAKE_MODULE_PATH=$CMAKE_MODULE_PATH -DPandoraPFANew_HOME=/path/to/pandora .

Set the environment variable.

export PANDORAMONITORING=`pwd`

Technically, the monitoring package is optional, though it is quite useful for event and detector visualization.

PandoraPFANew Continued

Since Pandora has minimal external dependencies, it is straightforward to build. Only the CMakeModules must be located for the build and PandoraMonitoring, and the latter is optional.

cd PandoraPFANew
cmake -DCMAKE_MODULE_PATH=$CMAKE_MODULE_PATH -DPandoraMonitoring_Home=/path/to/PandoraMonitoring .
make

If monitoring is not being used, this CMake command can be used instead.

cmake -DCMAKE_MODULE_PATH=$CMAKE_MODULE_PATH -DCMAKE_INSTALL_PREFIX=`pwd` BUILD_SHARED_LIBS=on .

Also, the Framework library needs to be built.

cd PandoraPFANew/Framework
cmake .
make

Set the environment variable.

cd PandoraPFANew
export PandoraPFANew_HOME=`pwd`

This will be referenced in the build of the next package.

If the release compiles successfully, you're now ready to setup the project that interfaces SLIC to Pandora.

slicPandora

Checkout the slicPandora project from cvs.

cd /workdir
cvs -d :pserver:anonymous@cvs.freehep.org:/cvs/lcd co slicPandora
cd slicPandora

The following should generate a setup script called mysetup.sh for setting up your runtime and compilation environment later.

./make_setup_script.sh

This setup script looks for the following environment variables to determine your Pandora setup.

If ROOTSYS and PANDORAMONITORING are not defined in the environment, then visualization will be disabled.

Source the new setup script that was generated.

. mysetup.sh

Alternately, a Makefile can be generated with CMake, using the following command.

cmake -DPandoraPFANew_HOME=$PandoraPFANew_HOME -DLCIO_HOME=$LCIO_HOME CMAKE_MODULE_PATH=$CMAKE_MODULE_PATH -DROOT_HOME=$ROOTSYS

Now build the slicPandora project.

make

The slicPandora project should now be built successfully, if its dependencies are present and setup correctly. If compilation or linking errors occur, then check that LCIO and Pandora were both installed successfully and that the setup script is pointing to the correct locations of these packages. If monitoring has been enabled, check that ROOT and PandoraMonitoring are configured and built successfully and that their environment variables are set.

Running Events

Steps

The essential steps to generating slicPandora events are as follows.

1) Use SLIC to generate LCIO simulation data.
2) Add tracks to the event by running a tracking package such as LCSim's SeedTracker.
3) Add the TrackState collections so slicPandora knows the Track momenta. (can be combined with #2 into one LCSim job)
4) Run slicPandora with the LCSim output, a Pandora XML geometry generated by GeomConverter, and a Pandora settings XML file.

These steps are covered in more detail below.

Generating Tracks Using LCSim

Before the simulated LCIO events can be run through Pandora, the LCSim tracking needs to run in order to add a collection of tracks.

If you don't know how LCSim batch mode works, then review the LCSim XML instructions.

Tracking Steering File

Below is an example XML steering file for LCSim to generate the Tracks and TrackStates for the sidloi3 detector.

<lcsim xmlns:xs="http://www.w3.org/2001/XMLSchema-instance"
        xs:noNamespaceSchemaLocation="http://www.lcsim.org/schemas/lcsim/1.0/lcsim.xsd">
    <inputFiles>
        <file>./pi_Theta90_10GeV-0-1000_SLIC-v2r8p3_geant4-v9r3p1_QGSP_BERT_sidloi3.slcio</file>
    </inputFiles>
    <control>
        <numberOfEvents>1000</numberOfEvents>
        <verbose>true</verbose>
        <printSystemProperties>false</printSystemProperties>
    </control>
    <execute>
        <driver name="EventMarkerDriver"/>
        <driver name="CalInfoDriver"/>
        <driver name="MainTrackingDriver"/>
        <driver name="TrackStateDriver"/>
        <driver name="Writer"/>
    </execute>
    <drivers>
        <driver name="CalInfoDriver"   
                type="org.lcsim.recon.util.CalInfoDriver"/>
        <driver name="MainTrackingDriver"
                type="org.lcsim.recon.tracking.seedtracker.trackingdrivers.sidloi2.MainTrackingDriver"/>
        <driver name="EventMarkerDriver" 
                type="org.lcsim.job.EventMarkerDriver">
            <eventInterval>10</eventInterval>
        </driver>
        <driver name="TrackStateDriver"
                type="org.lcsim.recon.tracking.seedtracker.SeedTrackerTrackStateDriver"/>
        <driver name="Writer"
                type="org.lcsim.util.loop.LCIODriver">
            <outputFilePath>./pi_Theta90_10GeV-0-1000_SLIC-v2r8p3_geant4-v9r3p1_QGSP_BERT_sidloi3_lcsimTracking.slcio</outputFilePath>
        </driver>
    </drivers>
</lcsim>

The input files section needs to be changed to point to your local simulated SLIC events, and the outputFilePath would also be changed to have a name based on the input file. You can easily make a script to automate this.

More LCSim Tracking Details

The LCSim job must accomplish three tasks before the events can be read into Pandora, in this order.

1. Generation of a Tracks collection using an appropriate Seed Tracker Driver.
2. Adding TrackState collections using the SeedTrackerTrackState Driver.
3. Writing out the necessary LCIO collections to a data file to be read into Pandora.

To generate the Tracks, a top-level Driver should be run that covers subdetector setup, digitization, and track finding and fitting. This top-level Driver will likely be specific to a certain detector design.

For instance, this simple Driver definition is sufficient to generate tracks in the sidloi3 detector.

 <driver name="MainTrackingDriver"
         type="org.lcsim.recon.tracking.seedtracker.trackingdrivers.sidloi3.MainTrackingDriver"/>

Three TrackState collections need to be added to the LCIO output to provide Pandora with track information at the track origins, the ECal, and the end point.

The following Driver in the Seed Tracker package will add these necessary TrackState collections. This Driver also requires that another Driver be run beforehand to cache Calorimeter subdetector data.

<driver name="CalInfoDriver"   
        type="org.lcsim.recon.util.CalInfoDriver"/>
<driver name="TrackStateDriver"
        type="org.lcsim.recon.tracking.seedtracker.SeedTrackerTrackStateDriver"/>

Finally, an LCIODriver should be added to the end of the event processing to output the appropriate collections.

<driver name="Writer"
        type="org.lcsim.util.loop.LCIODriver">
            <outputFilePath>OUTPUT_FILE</outputFilePath>
</driver>

The OUTPUT_FILE argument needs to be replaced with the actual name of the LCIO output file to be fed to Pandora.

Once these Drivers are defined in the <driver> section of your LCSim XML file, the execution order should look like the following.

<execute>
    <driver name="CalInfoDriver"/>
    <driver name="MainTrackingDriver"/>
    <driver name="TrackStateDriver"/>
    <driver name="Writer"/>
</execute>

Now that tracks and track states have been added to the events, we are ready to use Pandora itself.

PandoraFrontend

The PandoraFrontend binary provides a simple frontend to slicPandora.

The syntax of this command is the following.

./bin/PandoraFrontend geometry.xml pandoraSettings.xml inputEvents.slcio reconOutput.slcio [nevents]

The nevents command is optional, but all other arguments are required and must be supplied in order.

For example, to repeat the results of the JobManagerTest using PandoraFrontend, execute this command.

./bin/PandoraFrontend ./examples/sidloi3_pandora.xml ./examples/PandoraSettingsRelease.xml ./input.slcio ./pandoraRecon.slcio 2

The above command requires that the input events were generated in the sidloi3 detector and that a symlink has been setup pointing from "input.slcio" to your input event file.

Using GeomConverter to Output the Pandora Geometry Format

The GeomConverter package can convert from compact detector descriptions to various output formats, including one for input to slicPandora. (Installation of GeomConverter is not covered here.)

Use this command from your GeomConverter root directory to generate a Pandora geometry file from a compact detector description.

java -jar ./target/GeomConverter-1.10-SNAPSHOT-bin.jar -o pandora ./myCompactDetector.xml ./myPandoraGeom.xml

This will write out a Pandora XML file to "myPandoraGeom.xml" from the compact description in "myCompactDetector.xml".