Skip to content

YUMA (Yaml Plotting)#

YUMA (Yaml for pUMA) removes boilerplate required for making simple flavour tagging plots, instead moving everything required into yaml config files, and acting as a wrapper around the high-level plotting functionality included in puma. To make plots, two config files are required, a taggers.yaml and plot_cfg.yaml. Examples of these can be found in puma/examples.

To create plots, after modifying the two files, run

yuma --config examples/plot_cfg.yaml

Additional arguments can be included:

  • --plots [roc, scan, disc, prob, peff] Select one or more type of plots to produce.
  • --signals [bjets, cjets] what signals to plot
  • --num_jets [n] number of jets to load per tagger (before cuts are applied)

taggers.yaml#

The taggers.yaml file contains required information for taggers we wish to load. Under 'taggers' are the defined taggers. Each should be assigned a tag, which is used as the 'name' of the tagger and its settings. Within in tagger, arguments that can be parsed to the Tagger high-level class can be used. If you wish to plot the same tagger multiple times, for example with different fc values, you can also include a name:

GN2_fc0:
    name: GN2v01
    [args]
GN2_fc0p1:
    name: GN2v01
    [args]

In this case, both models will be the GN2v01 from whichever sample is specified. You should add GN2_fc0 and GN2_fc0p1 as the taggers in the plot config to plot both.

plot_cfg.yaml#

This file contains info on what taggers to load, what plots to make, and where to save them.

Global options#

  • plot_dir: - The base directory to write plots to. The pltos will be saved to a directory of the form plot_dir/plt_cfg.
  • timestamp: False - If True, will create a new directory each time the script is run, with a timestamp included in the name. If False, then will save to the default directory, and overwrite any files.
  • results_config: - Arguments to parse to the 'Results' class.
  • taggers_config: - Path to the taggers.yaml file.
  • taggers: - List of tagger names that we wish to plot.
  • reference_tagger: - Tagger name that shall be the 'reference' tagger. Any ratios by default are with respect to this tagger.
  • sample: - Information on the sample plotted, including the name, any cuts for the sample, and a sample string to include in the plot info.
  • all_flavours - A list which defines all the flavours available. If defined, then background flavours are defined as all_flavours - signal, if not defined, then default backgrounds are assigned (e.g, cjets and lightjets if signal=bjets) -label_var - What variable define the class labels. By default HadronConeExclTruthLabelID

Plots#

Each plot type should come under its own section. The allowed plots are:

  • roc_plots
  • fracscan_plots
  • disc_plots
  • prob_plots
  • eff_vs_var_plots

Within each section, a list of plots should be included. Each plot requires a 'signal'. Additional options that work for all plot types are:

  • suffix: Allows a custom suffix for the plot name.
  • include_taggers: Only include the specified taggers for this plot
  • exclude_taggers: Plot all taggers, except those in 'exclude_taggers'
  • plot_kwargs: arguments parsed to the base plot object, such as figsize.

Other specific plot types have additional arguments that can be included:

ROC plots#

  • x_range: The x-range of the signal efficiency

Fraction Scan plots#

  • frac_flav: The x in 'f_x', for example 'c' for standard f_c plots
  • efficiency: The signal efficiency to plot at
  • backgrounds: List of two backgrounds to plot the scan over.

Efficiency Vs Var plots#

  • peff_var: The variable to bin on the x-axis, default is pt bins: The bin edges for the x-axis, default for pT depends on if sample is ttbar/zprime