CalculateImpacts

This action performs an impact calculation, computing the impacts of individual systematic uncertainty parameters on your POI.

There are two variations of the computation, the "normal" and the "inverted" one.

• In the "normal" way, three fits are run for each systematic uncertainty: a fully unconditional one with all NPs floating, and two that are partially conditional on this uncertainty (with the respective NPs fixed to their nominal values plus or minus 1 sigma). Then, the difference between the post-fit values of the POI between the two fits are computed.
• The "inverted" calculation works the same way, but fixes all the other NPs to the nominal values instead of letting them float.

Usage:

+CalculateImpacts {
 +Morphing {
    # compute impacts for the fit to asimov
    +asimov { 
      <datasetName="asimovData_SM">
      <invert=false>
    }   
    +asimovinvert {
      <datasetName="asimovData_SM">
      <invert=true>
    }   
    @ ? {
      # the "singles" will compute the breakdown for each individual NP matching the filter
      <singles.SYS = "alpha_ATLAS_*">
      <singles.THEO = "alpha_theo*">                                                                                                            
      <singles.MCSTAT = "gamma_*">
      <singles.NORM = "ATLAS_norm*">
      <singles.FAKE = "HWW_FakeFactor_*"> 
    }
    <fit.startingStrategy=1,fit.printLevel=1,fit.numRetry=3, fit.offset=true, fit.verbose=1> @ ?;
  }
}

As this performs a series of fits, it can take quite long to run. Hence, there is a batch-parallelized version of this action, WriteCalculateImpacts. A different method of assessing the effects of different systematical uncertainties also exists, see CalculateBreakdown.