- preCICE is cross-plattform. While we mostly test on Linux, preCICE has also been succesfully used on Windows and macOS.
- Scalability for most features has been shown up to 10,000 cores.
- Minimally-invasive, high-level API in C, C++, Fortran 90/95, Fortran 2003, and Python
- Parallel or sequential coupling between two or more coupling participants, configurable at run-time
- Timestep size can be set by either participant or by preCICE.
- Fully parallel point-to-point communication, no central instance
- Based on either TCP/IP or MPI ports.
- Projection-based mapping: nearest neighbor or nearest projection
- Black-box mapping based on radial-basis functions
- All data mappings either in a consistent or in a conservative variant
- Static or dynamic (Aitken) underrelaxation
- Anderson or generalized Broyden acceleration techniques (also known as interface quasi-Newton schemes)
- Extrapolation in time
- Python action interface to modify coupling data at run-time
- Coupling mesh export to VTK
- Automatic XML reference generation
Missing a feature?
We describe our plans for the future in our Roadmap. If you don’t see what you need there, let us know!