siconos

Siconos is a software package for modeling and simulating nonsmooth dynamical systems, available through multiple interfaces (C++, Python) and a variety of components for mechanics, control, and optimization. This MCP server exposes a Docker-based environment that lets you start a JupyterLab session or terminal-backed access to Siconos tutorials and examples. You can run end-user demos (e.g., bouncing ball, contact dynamics) and experiment with the provided tutorials to understand how nonsmooth systems behave under impact, friction, and piecewise dynamics. The tooling emphasizes solver backends, time-stepping schemes, and complementary problem formulations such as LCPs, MLCPs, and NCPs, enabling you to explore both low-level numerical methods and higher-level simulations.

Prerequisites:

• Docker or an equivalent container runtime installed on your machine
• Sufficient disk space for images and datasets
• Basic familiarity with running containerized applications

Installation steps:

1. Install Docker: follow the official guide for your OS (https://docs.docker.com/get-dstarted/).
2. Pull or run the Siconos tutorial image directly via the MCP launcher (no local build required):
   • This will pull and start the container: docker run --rm -it gricad-registry.univ-grenoble-alpes.fr/nonsmooth/siconos-tutorials/siconoslab-master:latest
3. If you prefer to run a local build, clone the Siconos tutorials repo (optional) and start a container from that image using your own tag:
   • git clone https://github.com/siconos/siconos
   • docker build -t my-siconos-tutorials ./siconos-tutorials
   • docker run --rm -it my-siconos-tutorials
4. Access the environment:
   • The container will launch a JupyterLab session or a shell, depending on the image entrypoint. Follow the on-screen instructions to access the server (e.g., a browser URL for JupyterLab).

Prerequisites (alternative):

• If you plan to run Siconos from source, you will also need a C++ toolchain, Python, and CMake installed on your host or inside the container as needed.

Tips and notes:

• The Siconos project supports multiple backends for solving nonsmooth problems (LCP/MLCP/NCP) and offers both event-driven and time-stepping simulations. Use the tutorials to see which solver configurations fit your model.
• If you run into port or token issues with JupyterLab, check the container logs for the access URL and token. The provided Docker image often exposes a web interface at a local URL like http://127.0.0.1:8888/lab.
• When experimenting with parameters (gravity, friction, restitution, time step), use small increments and observe convergence and stability of the solver.
• The tutorials are designed to showcase end-user examples as well as low-level interfaces; you can switch between Python and C++ components within the same environment.
• If you need to run in headless mode or without a GUI, you can use a terminal-based workflow or a non-interactive Python script leveraging the siconos.kernel and related modules.