High quality HEP plotting

This code allows to produce, in a very flexible way, publication quality HEP distributions based on ROOT histograms.

1. Installation

git clone https://github.com/rmadar/hepplotting
cd hepplotting
pip install -e . --user

2. Usage

2.1 Needed inputs

1. Histograms for every processes and data, and total histogram (for the uncertainty)
2. Legend names and color definitions for every background
3. Possibly every systematic variations to be passed through the total histogram (after appropriate combination across systematics, which is not performed by this tool)

2.2 Simplest example

The simplest use you can do is to call the following function, once you get your histograms (with their color):

import hepplotting as plt
plt.make_nice_canvas(dictBkg,hTot,hData,plot_name='myplot')

where:

• dictBkg is a dictionnary made of background name and the corresponding histogram {bname:[TH1F,color,leg_name]}
• hTot is the total histogram with possibly larger uncertainty (to account for systematics)
• hData is the data histograms

A concrete example can be found in this notebook or this simple script which produce each this plot:

2.3 Description of the options

The point of this tool is to be able to highly tune many details on a plot-by-plot level, which is often required during the approval process of a publication.

Histogram properties

• dictSig [dict {sigName:[TH1,color,norm,legName]}] is dictionnary with name [string], histo [TH1], color [int], norm [float] and legName [string] of several signals
• m_size [float] is the marker size for data
• error_fill [int] is the filling style for the uncertainty band
• error_alpha [float] is the transparency for the uncertainty band (in [0,1])
• histo_border [int] is the border size of background histograms in the stacks

Axis properties

• xtitle [string] is x-axis title
• ytitle [string] is y-axis title
• xmin [float] lower x-axis value
• xmax [float] higher x-axis value
• ymin [float] lower y-axis value
• ymax [float] higher y-axis value
• r_ymin [float] lower y-axis value on the ratio plot
• r_ymax [float] higher y-axis value on the ratio plot
• xticksInt [bool] keep only integer values for x-axis ticks
• bin_label [list of string] to name bins (e.g plots with one region yield per bin)
• xlabel_size [float] size of the x-axis bin labels
• xlabel_offset [float] offset of the x-axis bin labels

Legend properties

• leg_pos [list of float] specify the legend position via bottom left (x1,y1) and top right (x2,y2) using [x1,y1,x2,y2]
• unc_leg [string] to tune the name of uncertainty (eg. stat-only or Stat #oplus Norm Syst.)
• leg_ncols [int] number of columns used for the legend
• leg_put_nevts [bool] to print events yields in the legend
• leg_textsize [float] size of the legend text (~0.030 to ~0.045)

Canvas properties

• canvas [TCanvas] on which to perform the plots
• plot_ratio [boolean] to plot or not the ratio panel
• is_logy [boolean] to plot in log scale or not
• can_ratio [float] specify the canvas size such as width=900/ratio and height=800
• can_scale [float] specify the canvas size without ratio change such as width=900*scale and height=800*scale
• plot_labels [list of string] given the labels printed below ATLAS and Lumi
• atlas_label [string] is 'Internal' by default but can be 'ATLAS', 'Preliminary', 'Simulation'

Output properties

• plotdir [string] is a directory where the plots will be stored (default is plots)

3 Technical comments

3.1 To-do list

• Add the possibility of having several legends with the position/number if entry for each: done via leg.SetNcolumns(ncols) method;
• Add sanity checks for the number of bins of each histograms and the size of the bin_label list

3.2 Dependencies

• ROOT