Skip to content

Latest commit

 

History

History
59 lines (40 loc) · 2.39 KB

setup_environment.md

File metadata and controls

59 lines (40 loc) · 2.39 KB

Environment Setup

SNELLIUS

  1. Clone the lie_learn package into your lib. This package needs to be manually installed, because pip install incorrectly processes the CPython files, leading to errors in the future.
cd $HOME/path-to-src/lib
git clone https://github.com/AMLab-Amsterdam/lie_learn.git

  1. Run the script DL2_setup_env.job, this will automatically install all required software in the correct order.

  2. IMPORTANT : An important .npy file must be manually moved into the created .venv folder $HOME/path-to-src/src/lib/lie_learn/lie_learn/representations/SO3/pinchon_hoggan/J_dense_0-150.npy. When this is moved into the corresponding folder in the .venv directory, the environment is ready to use.

MAC/PC

  1. Clone the lie_learn package and the escnn package into you lib folder.
cd /path-to-src/lib
git clone https://github.com/AMLab-Amsterdam/lie_learn.git
git clone https://github.com/QUVA-Lab/escnn.git
  1. Create a conda environment based on python 3.10.4, and afterwards activate your environment and install pip.
conda create -n DL2 python=3.10.4
conda activate DL2
conda install pip

  1. Determine the path within your system to the conda directory, and use that to manually pip install packages within your environment, will be something like /path-to-conda/envs/DL2/bin/pip.

  2. Make sure the package cython is installed, and again move to lie_learn and install that package first.

cd /path-to-src/lib/lie_learn
/path-to-conda/envs/DL2/bin/pip install cython
/path-to-conda/envs/DL2/bin/pip install .

  1. IMPORTANT: An important .npy file must be manually moved into the created environment folder in conda /path-to-src/src/lib/lie_learn/lie_learn/representations/SO3/pinchon_hoggan/J_dense_0-150.npy. When this is moved into the corresponding folder in the conda environment directory (within the site-packages folder), the environment is ready to use.

  2. Change directory to escnn, AND FOR MAC ARM USERS : remove py3nj from the requirements list within the setup.py file. This package caused installation problems, but is not a neccessary dependency of the escnn package. Afterwards install the escnn package.

cd /path-to-src/lib/escnn
/path-to-conda/envs/DL2/bin/pip install .
  1. Install the rest of the requirements.
/path-to-conda/envs/DL2/bin/pip install -r /path-to-project-folder/requirements.txt

DONE!