Getting started with PlasmaNet

First create a virtual environment for PlasmaNet to avoid version issues and to have a clean start. Please use at least Python >= 3.7 . Make sure that the loaded python version contains the sqlite library, (for kraken load python/anaconda3.8)

python3 -m venv plasma_env

Once the environment is created, activate it and don’t forget to upgrade pip:

source /path/to/your/plasma_env/bin/activate
pip install --upgrade pip

Then, clone the PlasmaNet repository from the CERFACS public GitLab:

git clone https://gitlab.com/cerfacs/plasmanet.git

At this point, you’re all set to begin the journey!

Follow the instructions of the GitLab repo to install the packages. This will notably install PyTorch, our framework of choice and TensorBoard for the monitoring of training jobs (cf. the corresponding tutorial):

pip install -r requirements.txt
pip install -e .

You also need to define the following environmnet variables:

export ARCHS_DIR=path/to/plasmanet/NNet/archs
export POISSON_DIR=path/to/plasmanet/PoissonSolver/linsystem

Now, you can test your install by running the tests with PyTest in each test directory:

cd tests/nnet/operators
pytest
cd ../../poissonsolver
pytest