Function tsvd 

pub fn tsvd<T>(
    matrix: &Matrix<T, Dyn, Dyn, VecStorage<T, Dyn, Dyn>>,
    config: TSVDConfig<T>,
) -> Result<SVDResult<T>, Error>
where
    T: ComplexField + RealField + Copy + Debug + ToPrimitive + CustomNumeric,

Main TSVD function using QR + SVD approach

Computes the truncated SVD using the algorithm:

1. Apply QR decomposition to A to get Q and R
2. Compute SVD of R
3. Reconstruct final U and V matrices

Arguments

• matrix - Input matrix (m × n)
• config - TSVD configuration

Returns

• SVDResult - Truncated SVD result

Errors

• Error::EmptyInput if the matrix has no rows or no columns
• Error::InvalidParameter unless 0 < config.rtol < 1 (a NaN tolerance is rejected)
• Error::NonFiniteInput if an entry of the matrix is NaN or infinite
• Error::DecompositionFailed if the QR of the matrix overflows (its R factor has a non-finite entry) or the SVD iteration does not converge