RAMSES-CPP
RAMSES-CPP aims to provide a functional and physically consistent alternative to the original Fortran implementation while maintaining strict bit-perfect parity in data structures and I/O.
RAMSES-CPP Documentation
Welcome to the documentation for RAMSES-CPP, a high-performance, modern C++17 port of the renowned RAMSES-2025 Adaptive Mesh Refinement (AMR) code.
Overview
RAMSES-CPP aims to provide a functional and physically consistent alternative to the original Fortran implementation of RAMSES, while maintaining strict bit-perfect parity in data structures and I/O. By porting to C++, we open the door to modern software engineering practices, easier integration with C/C++ libraries, and potentially improved performance and maintainability.
🚀 Current Status: Stable Hydro Core
The project has achieved a major milestone in its core Adaptive Mesh Refinement (AMR) system. After a comprehensive restoration phase, the engine now supports full recursive refinement up to levelmax=10 with correct grid connectivity. The 1D and 3D Hydro solvers are fully functional and maintain bit-perfect parity with legacy RAMSES-2025 results.
Key Features
- Restored AMR Engine: $O(1)$ constant-time neighbor lookups using a pointer-based
nborsystem, enabling seamless refinement across grid boundaries. - Advanced Initializer: Fully supports complex namelist lists and multi-region setups with coordinate-accurate grid placement.
- High-Order Hydro: MUSCL-Hancock 2nd-order reconstruction with HLLC, HLL, and LLF Riemann solvers.
- Level-Wide Caching: Performance-optimized MUSCL implementation that caches interface states, reducing redundant computations.
- Magnetohydrodynamics (MHD): Ported HLLD/CT solver (currently undergoing synchronization with the new
AmrGridAPI). - Radiative Transfer (RT): M1 closure scheme with photo-ionization coupling (currently undergoing synchronization).
- Self-Gravity: Multigrid Poisson solver with iterative Gauss-Seidel smoothing.
- MPI Parallelization: Hilbert Space-Filling Curve domain decomposition (active development for full multi-process parity).
- Legacy Ecosystem Parity: 100% compatible with existing Fortran Namelists (
.nml) and visualization tools (visu_ramses.py).