Skip to content

Dev Setup

Prerequisites

  • Python 3.12+
  • Docker Engine 24+ (for solver containers)
  • Git
  • NVIDIA GPU + Container Toolkit (optional, for XLB solver)

Clone and Install

git clone https://github.com/arxynehq/arxyne && cd arxyne
python3.12 -m venv arxyne-env && source arxyne-env/bin/activate
pip install -e ".[dev]"

The [dev] extra installs pytest, coverage, and linting tools.

Verify

arxyne --version
arxyne --list
python -m pytest tests/ -x --timeout=60

Solver Images

Build solver Docker images from the monorepo:

arxyne --pull openfoam      # CPU RANS (OpenFOAM v2312)
arxyne --pull xlb           # GPU LBM (needs NVIDIA GPU)
arxyne --pull chrono        # Vehicle dynamics (Project Chrono)
arxyne --pull all           # Pull all solver images

DoMINO NIM (AI surrogate) is not built locally — pull from NGC:

docker pull nvcr.io/nim/nvidia/domino-automotive-aero:2.1.0

LLM Setup (Optional)

For agent features (--ask, --brief):

# Option A: Local (requires Ollama)
ollama pull gemma4:e4b-it-q8_0

# Option B: Cloud API key
export ANTHROPIC_API_KEY=sk-ant-...
export DEEPSEEK_API_KEY=sk-...

Docs

pip install -e ".[docs]"
mkdocs serve           # localhost:8000
mkdocs build --strict  # verify build

Project Layout

arxyne_platform/     # Platform engine: CLI, API, gateway, gates, USD
solvers/             # Solver Docker images (monorepo)
│   ├── openfoam/    #   CPU RANS (OpenFOAM v2312)
│   ├── xlb/         #   GPU LBM (XLB + JAX/Warp)
│   ├── chrono/      #   Vehicle dynamics (Project Chrono)
│   └── domino/      #   Neural-operator surrogate (metadata only)
products/            # Default USD product definitions (ahmed, onera-m6, drivaer-notchback)
docs/                # This documentation (MkDocs Material)
tests/               # Test suite

See Also