`scipy.linalg._flapack:sgelsd`

Signature

def sgelsd(a, b, lwork, size_iwork, cond = -1.0, overwrite_a = 0, overwrite_b = 0)

Solve a least-squares problem by divide-and-conquer SVD (LAPACK sgelsd).

a : ndarray: Matrix of shape (m, n).
b : ndarray: Right-hand side(s) with max(m, n) rows.
lwork : int: Size of the workspace. Use gelsd_lwork for the required value.
size_iwork : int: Size of the integer workspace, as reported by gelsd_lwork.
cond : float, optional: Singular values below cond times the largest are treated as zero. Default is -1.0, which uses machine precision.
overwrite_a : int, optional: If nonzero, a may be overwritten in place. Default is 0.
overwrite_b : int, optional: If nonzero, b may be overwritten in place. Default is 0.

x : ndarray: Minimum-norm solution in the first n rows.
s : ndarray: Singular values of a, in descending order.
rank : int: Effective rank of a, determined using cond.
info : int: 0 on success; if negative, the -info-th argument had an illegal value; if positive, the algorithm failed to converge.