Our optional dependency scikit-sparse does not yet support numpy 2. This means that anyone who wants to use the more efficient CHOLMOD for Cholesky decomposition instead of LU will have to install numpy 1* for now.
Also Conda installs will not work with numpy 2 as our conda package builds with scikit-sparse.
Our optional dependency scikit-sparse does not yet support numpy 2. This means that anyone who wants to use the more efficient CHOLMOD for Cholesky decomposition instead of LU will have to install numpy 1* for now. Also Conda installs will not work with numpy 2 as our conda package builds with scikit-sparse.
For numpy 2 integration into scikit-sparse see: See https://github.com/scikit-sparse/scikit-sparse/pull/122