Getting started
The tutorials in this repository work best when executed from a Linux environment. If you are running Windows, you can make use of Windows Subsystems for Linux (WSL). The tutorials have been tested on Ubuntu 22.04.
First, install the following packages
sudo apt install python3 python3-dev python3-venv build-essential cmake
Next, create a virtual Python environment and activate it.
Note
Using python-venv provides a lightweight and efficient way to create
isolated environments, ensuring that dependencies for different projects
remain separate from system-wide packages. This prevents version conflicts,
as each virtual environment maintains its own set of installed libraries
without affecting or being affected by global Python packages. By using
python-venv, developers can work on multiple projects requiring
different package versions without interference, reducing the risk of
breaking dependencies across projects. Additionally, it enhances
reproducibility, as the environment can be easily recreated, making it
easier to share projects and deploy them consistently across different
systems.
python3 -m venv ~/.venv-perceptronium
source ~/.venv-perceptronium/bin/activate
Finally, install the following Python packages in the environment we just created.
pip install torch torchvision torchaudio pyyaml matplotlib ase tqdm
Important
When using python-venv, the user must ensure the environment is
activated before running any Python scripts that depend on it. Activation is
done using source ~/venv-perceptronium/bin/activate. However, this
activation is temporary: when the user logs out or closes the terminal, the
environment is automatically deactivated. To continue using it in a new
session, the user must reactivate the environment before running any
commands.