Editable and source installation

Fork pulserver/bartorch and clone the fork with its submodules:

git clone --recurse-submodules https://github.com/YOUR-USERNAME/bartorch.git
cd bartorch
git remote add upstream https://github.com/pulserver/bartorch.git

An existing clone is completed with git submodule update --init --recursive.

Editable installation

CC=clang CXX=clang++ python -m pip install -e '.[dev]'

compiles the library through scikit-build-core into build/<wheel tag>/ and installs the package in editable mode, with pytest, numpydoc and ruff. The pre-commit hooks, the spelling check and the documentation build need

python -m pip install pre-commit 'codespell[toml]' -r docs/requirements.txt

With CUDA:

python -m pip install -e '.[dev]' \
    --config-settings=cmake.define.BARTORCH_CUDA=ON

Building and testing without installing

./scripts/run_tests.sh builds the library into build/local, regenerates the checked-in generated files, and runs the suite against src/; arguments after its options are passed to pytest:

./scripts/run_tests.sh
./scripts/run_tests.sh tests/test_solve.py -k pics
BARTORCH_BUILD_DIR=$PWD/build-clang CC=clang CXX=clang++ ./scripts/run_tests.sh --rebuild
./scripts/run_tests.sh --cuda          # configure with BART's CUDA kernels

The same steps by hand:

cmake -S . -B build/local -DCMAKE_BUILD_TYPE=Release \
    -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++
cmake --build build/local -j
BARTORCH_LIBRARY=$PWD/build/local/libbartorch.so PYTHONPATH=src pytest tests/

On macOS the library is libbartorch.dylib, and on Windows libbartorch.dll. Keep builds with different compilers or CUDA settings in separate build directories. CUDA tests skip without a device; on a machine with one, python scripts/check_device.py checks the device path in dependency order.