Preparing to use on LXPlus:
Use a recent root environment. To match the one expected, run:
lb-run LCG/84 $SHELL
where $SHELL is your shell of choice. If you are not in this environment for setup, the mkvirtualenv step will fail with a libpython error.
Recently, on LXPlus, I have also had to add LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/afs/cern.ch/sw/lcg/releases/LCG_87/Python/2.7.10/x86_64-slc6-gcc49-opt/lib to get things to work.
Make sure .local/bin is near the beginning of your path.
pip install --user --upgrade pip
pip install --user virtualenv virtualenvwrapper
Add the following to your ~/.bashrc file:
export WORKON_HOME=$HOME/.virtualenvs
export PROJECT_HOME=$HOME/Devel
source $HOME/.local/bin/virtualenvwrapper.sh
If using zsh, this should be your .zshrc file, or if using oh-my-zsh, this can be added with the virtualenvwrapper plugin.
Note on zsh: This should be a custom build, the one on LXPlus is buggy.
After reruning your bashrc, you'll need to make a k3pi environment:
mkvirtualenv k3pi
pip install numpy scipy matplotlib
pip install -r requirements.txt
Then you should be able to source ./setup.sh (be careful, if you don't specify the ./ it might not work due to a bad name choice on LXPlus).
To make the ROOT classes, try running make in CPP/shape_classes then in CPP. This creates the ROOTex library.
Basic names are stored in config.py, including the eos root path.
The modes as instances of mode_base, in modes/mode_base.py. Year, polarity, etc. are stored here, and it can produce a list of files using the files in filelists/*. Modes are created in the modes directory.
In the main __init__.py there is a MODE context manager that picks a file based on mode, and gcm which gets the current mode.
The options for the jobs use two python package directories, helpers and modes. These files use the modes defined previously in config_files, and mostly set up the adding or add the required descriptors. The files are run through ganga_collision (ganga) or davinci_collision (small test run). The completed files are prepared for download through make_lfn_file.
A command line program to apply a selection and merge files is in /k3pi_root/Tree_Splitter.py.
There are lots of utilities also. selective_load is a decorator that loads the current mode. Particle is a class that stores the info about a particle, such as title and daughters. parser is a utility for creating a parser. logger.py handles logging, variables.py defines a var class and creates a bunch of variables like pt with nice LaTeX representations and units. tex_compile.py contains tex compile commands. helpers.py reimplements things like progress bars (why not use Plumbum.cli or Click?). Several are obvious tools just with an added logger call, but there also are a few custom routines. variables_from_formulae parses a string for variables, make_cut_string builds a TCut, and get_mpv computes a most probable value using Gaussian kdes. Wrappers for loading and dumbing classifiers and BDTs are in bdt_utils.py. PlotConfig in plot_config is a class to help with plotting. There is also a base folder, /K3pi_plotting, that can pdf plotting tools.
This creates the ROOTex library, and uses PyBind11. It contains Leading, SecondLeading, etc. It has TreeSplitter.
Data can be downloaded from the grid with download_data.py.
A parser is implemented for bdt_studies.py and correlations.py. Data functions for BDT are in bdt_data.py. get_root_preselection.py will get the selections and uses ROOTex.