Merge branch 'feature-movesplit'

[u/mrichter/AliRoot.git] / PWGGA / PHOSTasks / PHOS_embedding / Readme

These are tasks to perform PHOS embedding. Steps are following:
1. Steering macro Embeding.C reads list of data ESDs to embed and calculates number of events
2. Performs simulation of the same number of events, using enclosed Config.C 
   (single pi0 per event with flat or Hagedorn spectrum)
   then simulated signals are reconstructed clusters and converted to AOD
   There are several special moments:
   - no noise is added to digits in simulation (special setting in sim.C)
   - all digits are copied to CaloCells (threshold is zero in cpecial RecoParams) 
3. Then class AliPHOSEmbedding is called
   - read data ESDs, extract PHOS calocells, convert them to PHOSDigits
   - run default reconstruction to avoid possible problems with different 
     OCDBs used now and in official reconstruction. Apply final calibration if necessary
   - merge digits with (re-calibrated if necessary) signal digits
   - reconstruct digits with embedding signal (with final calibration if necessary)
4. As a result of all these procedures AliAODout file is produced with
   - MC information about embedded pi0s
   - standard CaloClosters, CaloCells with final calibration
   - Embedded CaloClusters, embedded CaloCells with final calibration
5. Run analysis with
   - AliAnalysisTaskPi0Efficiency: Signal (nominator in efficiency calculation) is any cluster with 
     contribution from simulated pi0s. Only for cross-check, this appoach is not quite correct at high occupancy.
   - AliAnalysisTaskPi0DiffEfficiency: correct approach, Signal (nominator in efficiency calculation) is
     difference between Real inv. mass distribution after and before embedding.

Note that to perform MC simulation one needs either access to official OCDB (Why do I need TPC drift velocity to simulate PHOS????)
or local copy of OCDB. 

One can have improved calibration produced after official reconstruction.
In usual analysis this is hidden in PHOSTenderSupply, but here it can not be used.
So we use complicated apporach: first return from ESDCaloCells to PHOSDigits with calibration
used in official reconstruction and then use final calibration in reconstruction of step 3. 
Note, that in this case simulation should also be done with final calibration.
As a result histograms with new/old calibrations should be provided for PHOSEmbedding and special settings in simulations and 
reconstruction of simulated digits should be used.