Binning explanation for 2D observables

[bonanomi]

As a first benchmark for 2D measurements try to produce binned pTH in nJets bins measurement.

In principle it's still a "1D" measurement, where we use as gen/reco observables a cut on both pTH and nJets.

This measurement is interesting because of its many possible interpretations.

E.g. see Fig. 10 in ATLAS differential paper.

[AlessandroTarabini]

Working on RunCoefficients.py and RunPlotCoefficients.py

[AlessandroTarabini]

RunCoefficients and RunPlotCoefficients updated for 2D measurements.
Three different formats of binning:

• |0|1|2|3|20| vs |0|10|20|45|90|250| / |0|10|20|80|250| / |0|20|90|250| / |0|25|250| (like njets vs pT4l)
• |50|80| vs |10|30| / |50|80| vs |30|60| / |80|110| vs |10|25| / |80|110| vs |25|30| (like mZ1 vs mZ2)
• |0|1|2|3|20| vs |0|10|20|45|90|250|

Name of the 2D variable: '(name_var_1) vs (name_var_2)'

The obs_bins variable is a dictionary whose index is the number of the bin, each entry of the dictionary is a list [obs_bins_low, obs_bins_high, obs_bins_low_2nd, obs_bins_high_2nd]. The first two entries are the lower and upper bound of the first variable. The second two entries are the lower and upper bound of the second variable

Working on RunTemplates and plot_templates

[AlessandroTarabini]

RunTemplates and plot_templates updated for 2d measurements.
The structure recalls the one used in RunCoefficients.
There is the tricky point, already present in RunCoefficients, of the length of obs_bins. For 1D measurement the number of bins is len(obs_bins)-1, instead for 2D measurement the number of bins is len(obs_bins). This is the source of the if statements to differentiate between 1D and 2D analysis.

Working on "fit" scripts

[AlessandroTarabini]

"fit" scripts updated for 2d measurements.
Found a wiser method to implement the difference between obs_bins in 1D measurement and 2D measurements. Define a varibale nBins=len(obs_bins) for 2D and nBins=len(obs_bins)-1 for 1D. Already implemented in "fit" scripts. With this method it is not necessary to provide bin boundaries in inputs (they are taken from inputs files).

To do: Implement nBins method in other scripts as well to optimize the framework

Working on producePlots

[bonanomi]

When ready, PR [to this branch] for testing before final release in the framework.

[AlessandroTarabini]

#24