AliPhysics  6a0d37d (6a0d37d)
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
Fast simulations anchored in Runs

Christian Holm Christensen


The files in this archive represents a generic approach to running FAST simulations - that is - only generator level data (no swimming through ALICE). The idea is that the user/committer only specifies the run number for which an anchor run is needed, and possibly an event generator.



Here URL is a string specifying the job. The URL argument has the format


where PROTOCOL can be one of

  • local: Run a single-thread simulation on the local machine.
  • lite:: Run a multi-thread simulation on the local machine, using the ProofLite functionality
  • proof: Run a multi-thread simulation on a PROOF farm specified by HOST.

The HOST part only makes sense for PROTOCOL=proof.

OPTIONS is a list of options separated by ampersands &, and can contain

  • run=RUN: Specify which run to anchor in. Beam conditions (beam species, collision energy, and such is taken from GRP for that run).
  • events=NUMBER: Number of events to simulate
  • eg=NAME: The event generator to use. Note, if this is not specified, then a default EG based on the run parameters is chosen. Names are typically of the form


where name could be Hijing, Pythia, DpmJet, and so on, while sub-type and sub-sub-type specifies tunes or variants of the main model.

  • monitor=SECONDS: Specifies that monitor histograms should be updated every SECONDS. If 0 or smaller or not specified, then no monitoring is done.
  • b=RANGE: Specifies the impact parameter range.
  • override=LIST: Specifies a comma-separated (,) list of settings that should be used rather than the ones being read from GRP. The can contain. Each item in the list is a key, value pair separated by a colon (:)
    • beamEnergy:BEAMENERGY: Total beam energy (not collision energy) in GeV
    • energy=ENERGY: Collision energy in GeV
    • period=IDENTIFIER: ALICE running period (e.g., LHC15a)
    • run=NUMBER: Run number
    • beam1.a=ATOMIC_NUMBER: Atomic number of particles in beam1
    • beam1.z=CHARGE: Charge of particles in beam1
    • beam2.a=ATOMIC_NUMBER: Atomic number of particles in beam2
    • beam2.z=CHARGE: Charge of particles in beam2
  • save=MODE: Only relevant for Proof(Lite). Values can be
    • none: Do not retrieve the final galice.root and Kinematics.root files.
    • split: Return the final galice.root and Kinematics.root files - one for each worker. The files are moved to a sub-directory on the client machine, and an collection of TUrl objects is written to index.root. One can easily define a chain using this information.
    • merge: Does not work

OPTIONS can also contain options for the execution environment, such as workers=N for ProofLite.



will make 10000 events using the default generator (Hijing), anchored in run 138190 (LHC10h, PbPb @ 2.76TeV). The output file will be




where, again the URL argument specifies the input and execution environment. It has the format


where PROTOCOL and HOST is as above. The INPUT part specifies the input file to read. OPTIONS is again a list of options separated by ampersands. Valid options are

  • events=NUMBER: Number of events to analyze
  • type=NAME Type of analysis to perform. Defined types are
    • INEL
    • NSD
    • CENTV0M
    • CENTV0A
    • CENTV0C
    • MULTRefMult00d80
    • MULTRefMult00d50

OPTIONS can also contain options for the execution environment, such as workers=N for ProofLite.

The OUTPUT argument specifies which ROOT file to write the results to.


Using the output from the above example


will analyze 1000 events from the above run, and build dN/deta per centrality bin, using a simulated V0M centrality estimator.

Centrality selection

The centrality estimation is simulated by counting the number of charged particles in a given region - e.g., the eta region corresponding to the V0 acceptance. The distribution of that signal is then integrated from the top to the bottom and the incremental integral is written to a histogram. Furthermore, the estimator quantity is saved as a separate branch in the output tree.

All this happens during RunFast.C. When looping over the data with ProcessFast.C we retrieve the integral histogram from the input file, and for each event we use the stored estimator quantity to look up the centrality of that event.

In this way, we do not need an additional pass to extract the centrality information, and we are sure that the data is self-consistent because of we replay the estimator quantity using the stored value.


The fast simulation uses the same strategy as for full simulations. For more details see in this directory.