ExportHistograms

Export histograms from a workspace.

Usage:

+ExportHistograms{
  +HWWRun2GGF {
    # export a set of postfit histograms
    +postfit {
      <snapshot="SnSh_NuisPOI_Unconditional_observed_muhat">
      <outputFile="workspaces/run2-ggf-$(fitLabel)/histograms-postFit.root">
      <dataName="obsData">
      <fitResult="HWWRun2GGF/Significance/observed/FitResults/observed_muhat">
    }
    # also export a set of prefit histograms
    +prefit {
      <snapshot="SnSh_AllVars_Nominal">
      <outputFile="workspaces/run2-ggf-$(fitLabel)/histograms-preFit.root">
      <dataName="obsData">
      <constPars={"*_norm_*","muGGF"}>
    }

    @?{
      <singleRegions=false>
      +CombinedSamples{
        +top {
          <Components={"ttbar*","wt*"}>
        }
        +higgs {
          <Components={"sig*","otherH*"}>
        }
        +VV {
          <Components={"Vgamma*","*NonWW*"}>
        }
        +WW {
          <Components={"ggWW*","*qqWW*"}>
        }
      }
      +CombinedRegions{
        +SR_0j_DF_all {
          <Components="SR_*0j*">
        }
        +SR_1j_DF_all {
          <Components="SR_*1j*">
        }
        +SR_DF_all {
          <Components="SR_*">
        }
        +all {
          <Components="*">
        }                
      }
    }
#    <normPattern="ATLAS_HWWlvlvGGF_sampleNorm_$(SAMPLENAME)">
     <cloneStyle="HWWRun2GGF/.Histograms/Nominal"> @?;

#    <export.decorrelate={"theo_*"}> @?;
#    <export.correlate={"*"}> @?;      
  }
}

Subfolders can be provided to the action to write more than one set of histograms. Each of these sets can be configured separately. The following configuration options are supported:

• snapshot: Give the name of the snapshot to be used. Uses the name of the subfolder as a default.
• outputFile: Output file to write the histograms to.
• fitResult: Define the fit result from which to take the correlation matrix. If not given, a diagonal matrix will be used.
• dataName: One or more names of datasets to be written to the output file.
• constPars: Set one or more parameters constant to have them removed from the uncertainty bands.
• export.decorrelate list one or more parameters to be decorrelated for the uncertainty band.
• export.correlate list one or more parameters to be correlated (with c=1) for the uncertainty band.
• normPattern: Optionally provide a normalization pattern (see below).
• cloneStyle: Provide a path to set of histograms in a model to clone histogram styles from. Furthermore, the sample and channel list can be manually chosen:
• singleSamples=false will disable all pre-defined samples from export.
• singleRegions=false will disable all pre-defined regions from export.
• Provide an additional subfolder CombinedSamples to list one or more combinations of samples in the form of a subfolder with a tag named Components.
• Provide an additional subfolder CombinedRegions to list one or more combinations of regions in the form of a subfolder with a tag named Components.

Normalization Pattern vs. selectComponents

This action can use one of two strategies to extract histograms from a workspace, either of which has its own limitations. In order to extract histograms, some components of the pdf need to be activated/deactivated selectively, such that the different contributions (samples) can be isolated. For this, two mechanisms exist.

• The selectComponentsmechanism is the default. It works in all cases and makes no specific assumptions about the workspace. However, it is not fully functional inROOT` versions prior to 6.12.
• The normPattern mechanism relies on specific normalization factors being present in the workspace, following a certain naming scheme. Each sample has to have an individual normalization factor. The wildcard $(SAMPLENAME) can and should be used as a placeholder for the sample name in the expression. For example <normPattern="ATLAS_HWWlvlvGGF_sampleNorm_$(SAMPLENAME)"> can be provided as a normalization pattern to identify these normalization factors. Prior, they can be added in the CreateModels step with an addition like this one: @? { # create normalization factors for each sample to allow easier extraction of postfit plots # we set these constant, so they make no difference for the fit +NormFactor.ATLAS_HWWlvlvGGF_sampleNorm_$(BASEFOLDERNAME) { <Val = 1., Low = 0., High = 50., Const = true> } }