randomized_svd

scirs2_linalg::scalable

Function randomized_svd 

Source 
pub fn randomized_svd<F>(
    matrix: &ArrayView2<'_, F>,
    rank: usize,
    config: &ScalableConfig,
) -> LinalgResult<(Array2<F>, Array1<F>, Array2<F>)>
where
    F: Float + NumAssign + Zero + One + Sum + ScalarOperand + Send + Sync + 'static,
Expand description

Randomized SVD for low-rank approximation

This algorithm is particularly effective for matrices with rapidly decaying singular values, providing significant speedup over classical SVD algorithms.

§Arguments

  • matrix - Input matrix
  • rank - Target rank for approximation
  • config - Configuration parameters

§Returns

  • (U, S, Vt) approximate SVD with rank columns/rows