KLU Add-On for CVXOPT, Python Software for Convex Optimization
This repository is a fork of the CVXOPT with KLU fork by Uriel Sandoval (@sanurielf).
Q: "A must be square sparse matrix" error when calling klu.linsolve
A: This happens when you passed a scipy.sparse matrix to klu.linsolve, which only takes cvxopt.spmatrix.
Use the following wrapper:
def klu_linsolve(A, b):
"""
KLU wrapper function for linear system solve A x = b (author: @SanPen)
:param A: System matrix
:param b: right hand side
:return: solution
"""
A2 = A.tocoo()
A_cvxopt = cvxopt.spmatrix(A2.data, A2.row, A2.col, A2.shape, 'd')
x = cvxopt.matrix(b)
klu.linsolve(A_cvxopt, x)
return np.array(x)[:, 0]