Function compute_evd_real_custom_epsilon

pub fn compute_evd_real_custom_epsilon<E>(
    matrix: MatRef<'_, E>,
    s_re: ColMut<'_, E>,
    s_im: ColMut<'_, E>,
    u: Option<MatMut<'_, E>>,
    epsilon: E,
    zero_threshold: E,
    parallelism: Parallelism<'_>,
    stack: PodStack<'_>,
    params: EvdParams,
)
where
    E: RealField,

See compute_evd_real.

This function takes an additional epsilon and zero_threshold parameters. epsilon represents the precision of the values in the matrix, and zero_threshold is the value below which the precision starts to deteriorate, e.g. due to denormalized numbers.

These values need to be provided manually for types that do not have a known precision at compile time, e.g. a dynamic multiprecision floating point type.