pyLCIO3 on LLR jupyterhub¶

Usage¶

In a terminal:
```
ssh -N -L20443:polui01.in2p3.fr:443 -L7001:polui.in2p3.fr:7001 kunath@llrgate01.in2p3.fr &
```
- If you are in the LLR network, this step is not necessary. You can directly point you web browser to: https://polui01.in2p3.fr.
Open in a web browser: https://localhost:20443.
Open a notebook. If you followed Setup, the pyLCIO kernel should appear under Kernel > Change kernel.

Setup¶

On polui, it should not actually be necessary for you to create the pyroot conda environment and to link it with LCIO. It suffices to use the reference installation at /home/llr/ilc/kunath/bin/LCIO/setup.sh.

The simplest way to expose the python environment/kernel to jupyter is:

source /home/llr/ilc/kunath/bin/LCIO/setup.sh  # To activate the conda environment.
python -m ipykernel install --user

The output of the last command tells you where the new kernelspec is installed. In may case, the folder name is /grid_mnt/vol_home/llr/ilc/kunath/.local/share/jupyter/kernels/python3. Then, a small change to the kernel.json in that folder is necessary:

{
 "argv": [
  "bash",
  "-c",
  "source /home/llr/ilc/kunath/bin/LCIO/setup.sh' && exec /home/llr/ilc/kunath/miniconda3/envs/pyroot/bin/python -m ipykernel_launcher -f '{connection_file}' "
 ],
 "display_name": "pyLCIO",
 "language": "python",
 "metadata": {
  "debugger": true
 }
}

Optionally, we can change the display_name All that we needed to change the argv field. By default, it contained:

 "argv": [
  "/home/llr/ilc/kunath/miniconda3/envs/pyroot/bin/python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "{connection_file}"
 ],

pyROOT¶

When working with ROOT, I prefer setting up a conda python environment. This requires having set up the conda package/environment management system.

mamba create -n pyroot -c conda-forge root cmake ipykernel numpy matplotlib pandas uproot

mamba: A faster depency solver system than default conda. Can be installed with: conda activate base; conda install -c conda-forge mamba. Alternatively, just replace mamba by conda in the line above.
root: Installs CERN ROOT with python bindings. Is much faster than an installation from source.
cmake: Is needed for the build of LCIO (The default cmake version on polui is too old).
ipykernel: Makes this environement available to jupyter
python=3.10: Could be added if you want a specific version of python.
numpy matplotlib pandas uproot: Some python packages that I already know that I will need for my analysis. More could be added here. Later, packages can be added via conda activate pyroot; conda install $PACKAGE_NAME.

LCIO¶

mkdir -p ~/bin
cd ~/bin
git clone https://github.com/iLCSoft/LCIO.git
cd LCIO
git checkout v02-17  #  Optional: Use a specific version.
conda activate pyroot
mkdir build
cd build
cmake -DBUILD_ROOTDICT=ON -D CMAKE_CXX_STANDARD=17 ..
make -j 4 install
cd ..

Two changes should be performed on ~/bin/LCIO/setup.sh

Add a line that activates the conda environment.
Change lib to lib64 in LD_LIBRARY_PATH.

The altered setup script then should look like this:

#!/bin/sh
# Next line is new.
source '/home/llr/ilc/kunath/miniconda3/bin/activate' '/home/llr/ilc/kunath/miniconda3/envs/pyroot'
export LCIO='/home/llr/ilc/kunath/bin/LCIO'
export PATH=$LCIO/bin:$LCIO/tools:$PATH
export LD_LIBRARY_PATH=$LCIO/lib64:$LD_LIBRARY_PATH  # Currently `lib` in LCIO. Must be changed to lib64.
export PYTHONPATH=$LCIO/src/python:$PYTHONPATH
alias pylcio='python $LCIO/src/python/pylcio.py'

and can be used as

source /home/llr/ilc/kunath/bin/LCIO/setup.sh

eehiq analysis

pyLCIO3 on LLR jupyterhub

Contents

pyLCIO3 on LLR jupyterhub¶

Usage¶

Setup¶

pyROOT¶

LCIO¶