uproot 3.8.0

ROOT I/O in pure Python and Numpy.

uproot (originally μproot, for “micro-Python ROOT”) is a reader and a writer of the ROOT file format using only Python and Numpy. Unlike the standard C++ ROOT implementation, uproot is only an I/O library, primarily intended to stream data into machine learning libraries in Python. Unlike PyROOT and root_numpy, uproot does not depend on C++ ROOT. Instead, it uses Numpy to cast blocks of data from the ROOT file as Numpy arrays.

Python does not necessarily mean slow. As long as the data blocks (“baskets”) are large, this “array at a time” approach can even be faster than “event at a time” C++. Below, the rate of reading data into arrays with uproot is shown to be faster than C++ ROOT (left) and root_numpy (right), as long as the baskets are tens of kilobytes or larger (for a variable number of muons per event in an ensemble of different physics samples; higher is better).

uproot is not maintained by the ROOT project team, so post bug reports here as GitHub issues, not on a ROOT forum. Thanks!

Installation

Install uproot like any other Python package:

pip install uproot                       # maybe with sudo or --user, or in virtualenv

or install with conda:

conda config --add channels conda-forge  # if you haven't added conda-forge already
conda install uproot

The pip installer automatically installs strict dependencies; the conda installer also installs optional dependencies (except for Pandas).

Strict dependencies:

• numpy (1.13.1+)

• awkward-array (0.12.0+)

• uproot-methods (0.7.0+)

• cachetools

Optional dependencies:

• lz4 to read/write lz4-compressed ROOT files

• xxhash to write lz4-compressed ROOT files

• lzma to read/write lzma-compressed ROOT files in Python 2

• xrootd to access remote files through XRootD

• requests to access remote files through HTTP

• pandas to fill Pandas DataFrames instead of Numpy arrays

Reminder: you do not need C++ ROOT to run uproot.

Tutorial

See the project homepage for a tutorial.

Run that tutorial on Binder.

Tutorial contents:

• Introduction

• What is uproot?

• Exploring a file

  • Compressed objects in ROOT files

  • Exploring a TTree

  • Some terminology

• Reading arrays from a TTree

• Caching data

  • Automatically managed caches

  • Caching at all levels of abstraction

• Lazy arrays

  • Lazy array of many files

  • Lazy arrays with caching

  • Lazy arrays as lightweight skims

  • Lazy arrays in Dask

• Iteration

  • Filenames and entry numbers while iterating

  • Limiting the number of entries to be read

  • Controlling lazy chunk and iteration step sizes

  • Caching and iteration

• Changing the output container type

• Filling Pandas DataFrames

• Selecting and interpreting branches

  • TBranch interpretations

  • Reading data into a preexisting array

  • Passing many new interpretations in one call

  • Multiple values per event: fixed size arrays

  • Multiple values per event: leaf-lists

  • Multiple values per event: jagged arrays

  • Jagged array performance

  • Special physics objects: Lorentz vectors

  • Variable-width values: strings

  • Arbitrary objects in TTrees

  • Doubly nested jagged arrays (i.e. std::vector<std::vector<T>>)

• Parallel array reading

• Histograms, TProfiles, TGraphs, and others

• Creating and writing data to ROOT files

  • Writing histograms

  • Writing TTrees

Reference documentation

• Opening files

  • uproot.open

  • uproot.xrootd

  • uproot.http

  • uproot.iterate

  • uproot.pandas.iterate

  • uproot.lazyarray(s)

  • uproot.daskarray/daskframe

  • uproot.numentries

• ROOT I/O

  • uproot.rootio.ROOTDirectory

  • uproot.rootio.ROOTObject

  • uproot.rootio.ROOTStreamedObject

• TTree Handling

  • uproot.tree.TTreeMethods

  • uproot.tree.TBranchMethods

• Interpretation

• Caches

• Parallel I/O