PlottingTool

Small tool to create nice plots from ROOT output files created by AnalysisTools

This tool is intended to be used with the output .root files as created by the AnalysisTools package. Its major goal is to provide publication-quality plots for your thesis, analysis note, or whatever, while remaining small and flexible. Please feel free to fork this repository and collaborate!

Instructions

Running the tool

To run the tool, just call
root runPlotting.cc+
on your machine. Be aware that the code must be compiled in root, thus the + at the end is mandatory.
As output the PlottingTool will create eps as well as pdf files for each individual plot.

Define what to plot

There are two files that define what is plotted, namely userConfig.h and defineSamplesAndPlots.h.

userConfig.h

In this file the user specifies general information about what to plot:

• infile: input root file
• ident: the identifier of the histograms within the root file (often myClassName_default_)
• isLumiScaled: bool to specify if input root file contains histograms that have been scaled to luminosity already

Some additional variables should be self-explanatory. If the boolean testPlotting is set, only one plot will be created, which can be helpful for testing purposes.

defineSamplesAndPlots.h

This is the file to specify which samples you want to include in your plots, and which variables you would like to plot.

In the function defineSamples() you can create samples using various constructors, and combine them using the += syntax. Which samples (or combination of samples) will be plotted by the tool is specified by adding them to the vector samples. Elements added first will be drawn first, and thus appear at the bottom of your plot, and vice versa.

The one thing you still need to specify is which variables to plot. This is done in the function definePlots(). You can create your plots using various constructors of the plotInfo struct, and add the individual plotInfo objects to the vector plots.

Special plots

In case you want to create individual, fancy plots that are not normal stacks of histograms, you can create those in the file runPlotting.cc directly. Make use of the various functions defined in plottingHelpers.h (and also add any new features to this collection as a function).

Understanding what's happening

The PlottingTool is based on structs (basically a thin class). We will take a quick tour through all the structs and helper files.

sample struct

The sample struct holds all information on one specific Monte-Carlo sample, which is to be plotted as a histogram. It is mandatory to define the following members of each sample object:

• color: Color to be used for this sample
• legName: Name of the sample to appear in the legend
• identifier: String used by the framework to identify the sample in the root file

In case your root input file is not already scaled by luminosity, one additional object is mandatory:

• mcScale: scale factor to normalize the sample to luminosity

The sample struct can hold information on various sub-samples. For example you can create a sample top which combines top-pair and single-top samples. Use the += syntax to combine samples. Information on various sub-samples is stored as elements of a vector in the sample struct; thus, the identifier and mcScale members (as well as some others) are vectors.

The file sample.h defines the struct and various constructors for it. You can specify the individual members by hand, but there are also some helper functions available. The function readSkimSummary will obtain the number of events before skimming of a given sample by its DataMCType from the SkimSummary.log.

plotInfo struct

The plotInfo struct holds all information for a plot that you want to create. The mandatory members for each object are:

• identifier: String used by the framework to identify the variable to plot
• unit: String which holds the unit of the variable

The additional members allow you to manipulate the histograms before plotting and should be self-explanatory.

configInfo struct

This struct holds information about the individual configuration. One instance of it, named conf, is created in userConfig.h and used throughout the framework. The struct takes care of opening the input root file and other important stuff.

Testing functionality

Each struct used in the framework defines a function testXYZ, which tests the setup of this struct and produces warning messages. The function testInputs calls these functions and creates a warning, if something in your configuration seems to be strange. The tool will not stop, so the crash message might give you some additional information what is happening.

plottingHelpers.h file

This file is a dump for all the functionality of the tool. It contains functionality to

• obtain histograms from the root file
• manipulate histograms
• create stacked histograms
• produce Data-MC comparison plots
• create legends
• draw plots

It also defines some colors (corporate design, Higgs legacy result) which can be used to create nice plots.

Additional helper files

There are some additional files present, namely

• CMS_lumi.h/.cc: Script to add luminosity information to plots
• tdrstyle.C: style of CMS plots

These are copies of the original files by CMS with some slight modifications to make them compatibel with the framework.