RDataframe or write the code?

RDataFrame is a ROOT class that does just one thing, but does it really, really well: It goes through the entries in a ROOT file and does… well… something with each entry.

Using RDataFrame, each of the main Exercises in this tutorial (2 through 9) can be written in a few lines of code. You don’t need to create event loops with macros or analysis skeletons; the RDataFrame class and its associated methods handle all of that.

If you want to just click through to The RDataframe Path and get started, go right ahead. However, my conscience demands that I offer you some reasons to take The Python Path or The C++ Path instead:

  • As I said, RDataFrame is designed to loop through every entry in an n-tuple. That describes a large portion of typical physics analysis tasks. The whole raison d’être of this tutorial is to teach you exactly that.

    However, that’s not the only analysis task you may be asked to do this summer. In fact, none of the Advanced Exercises or Expert Exercises can be done in this way. So you may to take go through the coding portions of this tutorial, in order to prepare you for more challenging tasks.

  • The RDataFrame class is a relatively new addition to ROOT.1 It’s possible your supervisor has never heard of it.

  • It’s easy to use to use RDataFrame… at first. There’s point at which you hit a “wall”: You suddenly have to understand about C++ functions and lambdas, even if you’re doing your work in Python.

    To give you an idea how complex using RDataFrame can get, consider these advanced examples in C++ or Python; they are from the ROOT dataframe tutorials and demonstrate Higgs-boson reconstruction. If you just glance at those examples, you’ll confirm for yourself that RDataFrame doesn’t keep you from learning how to code.

Now that I’ve scared you, let’s look at the reasons to use RDataFrame for this tutorial:

  • Some students have had difficulty getting through The Python Path or The C++ Path portions of the tutorial in the time we have available. If you do the The RDataframe Path, you’re almost guaranteed to complete the whole thing.

  • Although I only show examples using the n-tuple tree1, you can also use other file formats as input to dataframes; e.g., TTrees and CSV files.

  • It’s easy to make RDataFrame multi-threaded, which can greatly speed up the execution time of its operations.

If you’re ambitious, you might consider working through The Python Path or The C++ Path up to Exercise 10, then do The RDataframe Path. After doing Exercises 2 through 9 using code, doing the same Exercises using RDataFrame will take very little time.

xkcd data_trap

Figure 29: https://xkcd.com/2582/ by Randall Munroe

1

The current RDataFrame class was introduced in ROOT 6.14 (June 2018). From ROOT 6.10 to 6.12, the class was called ROOT::Experimental::TDataFrame. Prior to 6.10, you won’t find dataframes in ROOT at all. Since this is an actively evolving feature of ROOT, you’ll want to check which version of ROOT your collaboration uses.

The Nevis notebook server uses the latest stable version of ROOT, but collaborations often stick with a particular older ROOT version.