num::diagonal

Defined in header <linear/sparse/sparse.hpp>

template <class M>
requires repr::csr<M>
vec<real> diagonal(const M &A);

Extract the main diagonal up to the smaller matrix dimension.