Run the CombinatorialKalmanFilter (CKF) tracking example


Acts must be build with activated examples and Pythia8 support (ACTS_BUILD_EXAMPLES_PYTHIA8=on) to enable the fast simulation. <build> is used to identify the path to the build directory.

We assume that you have generated a simulation dataset based on the TrackML detector as described in Run the FAst TRAck Simulation. A good example dataset would be e.g. a ttbar sample with pileup 200 in a 2T magnetic field. Suppose the generated ttbar sample is available at data/sim_trackML/ttbar_mu200.

Run the CKF tracking

Run the CKF tracking tool that reads the simulation output (truth hits and truth particles), creates smeared measurements from the true hits, creates seeds (i.e. starting track parameters) from the pre-selected truth particles, and run the CKF which will perform the track finding and track fitting simultaneously:

Currently, there are two configurable criteria to select compatible source links on a surface with track parameters in CKF:

  • Global maximum chi2 of Kalman filtering. This could be set up via --ckf-selection-chi2max

  • Global maximum number of measurements on a surface. This could be set up via --ckf-selection-nmax

The digitization of the truth hits must also be configured. Since the command-line configuration of this step can get unwieldy, an example json configuration file for the smearing digitizer is provided with the source code. The detector volumes and layers used in the space point maker are also configured using another example json file in the source code.

$ <build>/bin/ActsExampleCKFTracksGeneric \
    --input-dir=data/sim_trackML/ttbar_mu200 \
    --bf-constant-tesla=0:0:2 \
    --ckf-selection-chi2max 15 \
    --ckf-selection-nmax 10 \
    --output-dir=data/reco_trackML/ttbar_mu200 \
    --digi-config-file <source>/Examples/Algorithms/Digitization/share/default-smearing-config-generic.json \
    --geo-selection-config-file <source>/Examples/Algorithms/TrackFinding/share/geoSelection-genericDetector.json

In default, the starting track parameters estimated from reconstructed seeds by the seeding algorithm are used to steer the CKF. Alternative options are using properly smeared track parameters of truth particles and using track parameters estimated from truth tracks, which can be configured by the options --ckf-truth-smeared-seeds and --ckf-truth-estimated-seeds, respectively.

The magnetic field setup should be consistent between simulation and CKF tracking.

Look at the CKF tracking performance

The CKF tracking will generate a root file named performance_ckf.root (the name is currently not configurable via the command line) in the data/reco_trackML/ttbar_mu200. This file includes a few efficiency plots showing the CKF efficiency, fake rate, duplication rate and other plots showing detailed reconstruction info etc.

Example plots to show the CKF efficiency, fake rate and duplication rate for the ttbar sample generated above:

../_images/trackeff_vs_eta_ttbar_pu200.png ../_images/fakerate_vs_eta_ttbar_pu200.png ../_images/duplicationrate_vs_eta_ttbar_pu200.png