DNDSR Documentation

DNDSR is a C++17 / Python CFD (Computational Fluid Dynamics) research code implementing Compact Finite Volume methods with MPI parallelism and optional CUDA GPU support.

• Guides
• Architecture
• Theory
• Unit Tests
• Development Notes

Core Modules

Module	Directory	Description
DNDS	src/DNDS	MPI arrays, serialization (JSON, HDF5), profiling, CUDA
Geom	src/Geom	Unstructured mesh, CGNS I/O, partitioning (Metis/ParMetis)
CFV	src/CFV	Compact Finite Volume, variational reconstruction
Euler	src/Euler	Compressible Navier-Stokes solvers (2D/3D, SA, k-omega)
EulerP	src/EulerP	Alternative evaluator with CUDA GPU support
Solver	src/Solver	ODE integrators and Krylov solvers (GMRES, PCG)

Quick Start

# 1. Build external dependencies
git submodule update --init --recursive --depth=1
cd external/cfd_externals && CC=mpicc CXX=mpicxx python cfd_externals_build.py && cd ../..
 
# 2. Build C++ solvers
cmake --preset release-test
cmake --build build -t euler -j32
 
# 3. Install Python package (editable)
python3.12 -m venv venv && source venv/bin/activate
pip install numpy scipy pytest pytest-mpi pytest-timeout mpi4py pybind11 pybind11-stubgen scikit-build-core ninja
CC=mpicc CXX=mpicxx CMAKE_BUILD_PARALLEL_LEVEL=32 pip install -e . --no-build-isolation
 
# 4. Run tests
ctest --test-dir build -R dnds_ --output-on-failure   # C++
pytest test/DNDS/test_basic.py -v                      # Python
 
# 5. Build and serve documentation
cmake --build build -t serve-docs