Usage

Installation

ADEPT uses uv. Clone the repository and sync the environment:

git clone https://github.com/ergodicio/adept.git
cd adept
uv sync --extra dev

This creates .venv/ and installs the exact versions pinned in uv.lock. Python 3.11+ is required. For an NVIDIA GPU, use uv sync --extra gpu instead, which pulls jax[cuda12].

To use ADEPT as a dependency of another project:

uv add "adept @ git+https://github.com/ergodicio/adept.git"

Run an Example

The most common use case for ADEPT is a simple forward simulation that can be run from the command line. For example, to run a 1D1V Vlasov simulation of a driven electron plasma wave:

uv run run.py --cfg configs/vlasov-1d/epw

The input parameters are provided in configs/vlasov-1d/epw.yaml. Note that --cfg takes the path without the .yaml extension.

Access the Output

The output will be saved and made accessible via MLFlow. To access it:

  1. Launch an mlflow server via running uv run mlflow ui from the command line

  2. Open a web browser and navigate to http://localhost:5000

  3. Click on the experiment name to see the results

Solver-Specific Guides

Each solver has a single page covering the equations it solves, its boundary conditions and forcing, what it writes out, and how to run it. They are listed under Available Solvers.

Shared initialization options — density profiles, tanh flat-tops, super-Gaussians — are documented once: