Struct ndarray_linalg::least_squares::LeastSquaresResult

pub struct LeastSquaresResult<E: Scalar, I: Dimension> {
    pub singular_values: Array1<E::Real>,
    pub solution: Array<E, I>,
    pub rank: i32,
    pub residual_sum_of_squares: Option<Array<E::Real, I::Smaller>>,
}

Result of a LeastSquares computation

Takes two type parameters, E, the element type of the matrix (one of f32, f64, c32 or c64) and I, the dimension of b in the equation Ax = b (one of Ix1 or Ix2). If I is Ix1, the right-hand-side (RHS) is a n x 1 column vector and the solution is a m x 1 column vector. If I is Ix2, the RHS is a n x k matrix (which can be seen as solving Ax = b k times for different b) and the solution is a m x k matrix.

Fields

singular_values: Array1<E::Real>

The singular values of the matrix A in Ax = b

solution: Array<E, I>

The solution vector or matrix x which is the best solution to Ax = b, i.e. minimizing the 2-norm ||b - Ax||

rank: i32

The rank of the matrix A in Ax = b

residual_sum_of_squares: Option<Array<E::Real, I::Smaller>>

If n < m and rank(A) == n, the sum of squares If b is a (m x 1) vector, this is a 0-dimensional array (single value) If b is a (m x k) matrix, this is a (k x 1) column vector