Input to pyhf/cabinetry

SysVar provides native support for exporting templates and variations in the format expected by cabinetry. Users who prefer to work directly with pyhf (or any other template-fit framework) can also manually extract the variations and build the corresponding inputs themselves. The tempalate eigenvariations generated by SysVar incorporate both the normalisation and shape variations. These can be decoupled later manually by the analyst in the pyhf json file. file. If a user wants to follow the recommended workflow—i.e., pandas.DataFrame \(\to\) SysVar \(\to\) cabinetry \(\to\) pyhf — we refer them to the cabinetry documentation for instructions on preparing a cabinetry configuration file. As a minimal example, we provide below the cabinetry configuration file required to perform a fit following our SysVar 101 example.

- General:

    Measurement: SysVar 101
    POI: sig
    InputPath: test_output.root:{RegionPath}/{SamplePath}/{VariationPath}
    HistogramFolder: path/where/histograms/will/be/saved
    VariationPath: Nominal

- Samples:

  - Name: Signal
    SamplePath: sig

  - Name: Background
    SamplePath: bkg

- Regions:

  - Name: Channel 1
    RegionPaths: channel1

  - Name: Channel 2
    RegionPaths: channel2

- Systematics:

  - Name: charged_slow_pi_var_1
    Up:
      VariationPath: charged_slow_pi_var_1_up
    Down:
      VariationPath: charged_slow_pi_var_1_down
    Type: NormPlusShape