Installation

There are two main ways to install FluidGym: via PyPI or by downloading the source code from GitHub. Regardless of the installation method, it is recommended to set up a dedicated Python virtual environment using tools like venv or conda to avoid dependency conflicts.

Before installing FluidGym, ensure that you have Python 3.10 or higher installed on your system. To enable the GPU-accelerated solver, PyTorch with CUDA 12.8 is required. It can be installed using the following command:

pip install torch --index-url https://download.pytorch.org/whl/cu128

Then, follow one of the methods below to install FluidGym.

1. Using PyPI

This is the simplest way to install FluidGym. After setting up the python environment, just run the following command:

pip install fluidgym

2. Downloading from GitHub

This is the best way to install the latest version of FluidGym. Additionally, this method allows you to compile FluidGym from source on architectures/operating systems that are not supported by the PyPI binaries. First, clone the FluidGym repository from GitHub:

git clone https://github.com/safe-autonomous-systems/fluidgym.git
cd fluidgym

Then, install the package:

make install

Depending on whether you want to reproduce our experiments or develop new features, you can install FluidGym with different sets of dependencies:

To install FluidGym for development purposes, run:
```
make install-dev
```
To install FluidGym for reproducing experiments, run:
```
make install-exp
```