pub fn eig(matrix: &Array2<f64>) -> Result<(Array1<f64>, Array2<f64>), String>Expand description
Real symmetric eigensolver using the cyclic Jacobi eigenvalue algorithm.
This is a pure-Rust fallback used when the scirs2 feature is disabled.
It assumes the input matrix is symmetric (e.g. covariance or Fisher
information matrices). For robustness the input is symmetrized as
(A + Aᵀ) / 2 before the iteration begins.
Returns (eigenvalues, eigenvectors) where the columns of the
eigenvector matrix are orthonormal. A non-square input is an honest error.