Optimization Framework To HDF5 Converter

This converts ROOT files from the b-tagging optimization framework into HDF5 files for use with DL1 / IPRNN / whatever algorithm.

Setup

Assuming you have CVMFS, you can run the following to set up for the first time:

git clone git@github.com:dguest/btag2hdf5.git
cd btag2hdf5
. setup.sh
make

After building all the libraries should be permanently linked to CVMFS (thanks to some -rpath magic). There should be no need to run the setup script next time you log in.

If you don't have CVMFS

You'll need the following:

• ROOT: use master branch for 6, root5 branch for version 5
• boost: for program_options
• HDF5: with the C++ bindings installed

Running

The executables will be built as ./bin/btag-*. For more "standard" b-taggers, you can use btag-write-track-jets. For studies with fat jets, there's also btag-write-fat-jets. Note that this is more experimental.

Note On Selection

There is some standard pt, eta, and JVT selection applied to jets. For details look in src/select_jet.cxx.

How to read the output

The output files should be easy to read with h5py:

from h5py import File

file = File('the-output.h5') # <- load file
jets = file['jets']          # <- get jets dataset

# you can read in slices
slice = jets[:1000]          # <- get the first thousand jets

# or select variables
pt = jets['pt']              # <- select the pt array

This step doesn't depend on ROOT, so it should be easy to install everything you need via pip.