tvbk

Computational kernels for tvb.

setup

My local dev setup is in VS Code w/ Python, C/C++ extensions, and a venv setup for incremental rebuilds like so

rm -rf build env
uv venv env
source env/bin/activate
uv pip install nanobind 'scikit-build-core[pyproject]' pytest pytest-benchmark numpy cibuildwheel scipy 
uv pip install --no-build-isolation -Ceditable.rebuild=true -ve .

following https://nanobind.readthedocs.io/en/latest/packaging.html#step-5-incremental-rebuilds. This enables editing and running the tests directly, with changes to the C++ automatically taken into account, just running

pytest

will rebuild the C++ if required. This also occurs on import in e.g. a Jupyter kernel.

next

• make first release to start integrating w/ TVB

• all the neural mass models

• add bold

• refactor buffers

• rm scipy dep for sparsity

• cuda/hip/webgpu or something