Analysis applications

The ACTS examples come with a certain variety of ROOT based validation and performance writers, whose output can be use to understand various aspects of the reconstruction in more detail.

The building of these applications can be switched on by setting ACTS_BUILD_ANALYSIS_APPS=On, which requires (on top of the Core dependencies) ROOT for the analysis code.

These analysis applications are steered via BOOST program options, hence a quick <APP> -h will quickly show the relevant options, they can be executed in silent mode, i.e. without opening a window, when specifying the -s option.

Material Composition Analysis

This analysis allows to inspect the output of the Geant4 based material recording, and split the contained material by atomic number in predefined detector regions. For the moment, the detector regions need to be identified and constraint in the rz plane.

*** Material Composition plotting
*** Usage::
-h [ --help ]                           Display this help message
-s [ --silent ]                         Silent mode (without X-window/display).
-i [ --input ] arg                      Input ROOT file containing the input TTree.
-t [ --tree ] arg (=material-tracks)    Input TTree name.
-o [ --output ] arg                     Output ROOT file with histograms
-b [ --bins ] arg (=60)                 Number of bins in eta/phi
-e [ --eta ] arg (=4)                   Eta range.
--sub-names arg                         Subdetector names.
--sub-rmin arg                          Minimal radial restrictions.
--sub-rmax arg                          Maximal radial restrictions.
--sub-zmin arg                          Minimal z radial restrictions
--sub-zmax arg                          Maximal z radial restrictions.

For the output of the material recording, of e.g. the OpenDateDetector, the following command will create a material composition plot of the entire detector:

$ <build>/bin/ActsAnalysisMaterialComposition \
    -i geant4_material_tracks.root \
    -t material-tracks \
    -o material_composition.root \
    --sub-names beampipe detector \
    --sub-rmin 0.:0. \
    --sub-rmax 30.:1100 \
    --sub-zmin -4000.:-4000. \
    --sub-zmax 4000.:4000.

The output file material_composition.root would then contain profile histograms versus pseudorapidity and azimuthal angle phi for the radiation length X0 and the nuclear interaction length L0, restricted to the given rz regions, in this case only for the beampipe and the entire detector.

The resulting file contains the following histograms:

KEY: TProfile beampipe_x0_vs_eta_all;1        X_{0} vs. #eta
KEY: TProfile beampipe_l0_vs_eta_all;1        L_{0} vs. #eta
KEY: TProfile beampipe_x0_vs_phi_all;1        X_{0} vs. #phi
KEY: TProfile beampipe_l0_vs_phi_all;1        L_{0} vs. #phi
KEY: TProfile beampipe_x0_vs_eta_A9;1 X_{0} vs. #eta
KEY: TProfile beampipe_l0_vs_eta_A9;1 L_{0} vs. #eta
KEY: TProfile beampipe_x0_vs_phi_A9;1 X_{0} vs. #phi
KEY: TProfile beampipe_l0_vs_phi_A9;1 L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_all;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_all;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_all;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_all;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A9;1 X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A9;1 L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A9;1 X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A9;1 L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A12;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A12;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A12;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A12;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A13;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A13;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A13;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A13;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A27;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A27;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A27;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A27;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A28;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A28;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A28;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A28;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A48;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A48;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A48;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A48;1        L_{0} vs. #phi
KEY: TProfile detector_x0_vs_eta_A64;1        X_{0} vs. #eta
KEY: TProfile detector_l0_vs_eta_A64;1        L_{0} vs. #eta
KEY: TProfile detector_x0_vs_phi_A64;1        X_{0} vs. #phi
KEY: TProfile detector_l0_vs_phi_A64;1        L_{0} vs. #phi

Only histograms with non-zero contribution are written out per specified region, the following shows a resulting stacked histogram showing different components:

../../_images/aa_mc_stacked_x0.gif

The source code for this application can be found in Examples/Scripts/MaterialMapping.

Tracking Performance Analysis

Two different applications are available for analysing the output of track fitting and track finding, sitting on top of the corresponding ROOT output writers from the Example applications.

Residuals and Pull analysis per layer

To investigate the per layer residual and pull distributions, one can use the ActsAnalysisResidualAndPulls application, which runs on top of the ROOT file produced by the RootTrackStatesWriter.

The following options are available:

*** ACTS Residual and Pull plotting
*** Usage::
-h [ --help ]                    Display this help message
-s [ --silent ]                  Silent mode (without X-window/display).
-i [ --input ] arg               Input ROOT file containing the input TTree.
-t [ --tree ] arg (=trackstates) Input TTree name.
-o [ --output ] arg              Output ROOT file with histograms
--predicted                      Analyze the predicted parameters.
--filtered                       Analyze the filtered parameters.
--smoothed                       Analyze the smoothed parameters.
--fit                            Fit the smoothed parameters.
--save arg (=png)                Output save format (to be interpreted by
                                   ROOT).

Again, this application is capable of running in silent mode (-s) without opening a dedicated screen window.

Originally designed for the Acts::KalmanFilter output, it is capable of producing histograms of the --predicted, --filtered and --smoothed track states (i.e. track parameters) and will do so per layer and volume.

On request (--fit) the resulting distributions can be fitted for the summary plots that are created, otherwise the RMS and its mean are taken.

The application will (by parsing the geometry id range) automatically determine the different layers and volumes and create detailed and summary plots for all of them.

As a example, the pull distributions for predicted, filtered and smoothed track states is shown below.

../../_images/aa_rp_layers.png

Track summary analysis

A higher level view of the track reconstruction performance is the ActsAnalysisTrackSummary application, which runs on top of the RootTrackSummaryWriter output of the examples code.

The following options are available:

*** ACTS Perigee parameters and Track summary plotting
*** Usage::
  -h [ --help ]                         Display this help message
  -s [ --silent ]                       Silent mode (without X-window/display).
  -n [ --events ] arg (=0)              (Optionally) limit number of events to
                                        be processed.
  -p [ --peak-events ] arg (=0)         (Optionally) limit number of events for
                                        the range peaking.
  -i [ --input ] arg                    Input ROOT file(s) containing the input
                                        TTree.
  -t [ --tree ] arg (=tracksummary)     Input TTree/TChain name.
  -o [ --output ] arg                   Output ROOT file with histograms
  --hist-bins arg (=61)                 Number of bins for the residual/pull
                                        histograms
  --pull-range arg (=5)                 Number of sigmas for the pull range.
  --eta-bins arg (=10)                  Number of bins in eta.
  --eta-range MIN:MAX (=-3:3)           Range for the eta bins.
  --phi-bins arg (=10)                  Number of bins in phi.
  --phi-range MIN:MAX (=-3.14159:3.14159)
                                        Range for the phi bins.
  --pt-borders arg                      Transverse momentum borders.
  --config-output arg                   (Optional) output histogram
                                        configuration json file.
  --config-input arg                    (Optional) input histogram
                                        configuration json file.
  --all                                 Process all residual/pull and auxiliary
                                        parameters
  --d0                                  Residual/pulls for d0
  --z0                                  Residual/pulls for z0
  --phi0                                Residual/pulls for phi0
  --theta0                              Residual/pulls for theta0
  --qop                                 Residual/pulls for qop
  --time                                Residual/pulls for time
  --pt                                  Residual/pulls for pt
  --chi2ndf                             Auxiliary information for chi2ndf
  --measurements                        Auxiliary information for measurements
  --holes                               Auxiliary information for holes
  --outliers                            Auxiliary information for outliers
  --shared                              Auxiliary information for shared

This application is highly configurable and produces residual and pull (regional, integral and summary) plots for the fitted perigee parameters of track fitting. It can be run in `eta,phi,pT` bins, and as the different histograms in the various bins will require different histogram ranges, these will be automatically determined.

However, this process is relatively slow and makes comparisons between runs difficult, thus the range configuration can be written out by specifying a --config-output JSON file, and successively re-using it with a --config-input flag in future analysis runs.

For very large files, the number of entries used for range calculation (peak entries) can be set using the --peak-events option.

Some example histograms (transverse impact parameter d0 distribution or a summary plot showing the number of detector hits, are added below).

../../_images/aa_ts_d0.png
../../_images/aa_ts_nhits.png

The source code for these applications can be found in Examples/Scripts/TrackingPerformance.