scipy sparse matrix to cvxopt spmatrix?

Question

I need to convert a scipy sparse matrix to cvxopt's sparse matrix format, spmatrix, and haven't come across anything yet (the matrix is too big to be converted to dense, of course). Any ideas how to do this?

Answer 1

The more robust answer is a combination of hpaulj's answer and OferHelman's answer.

def scipy_sparse_to_spmatrix(A):
    coo = A.tocoo()
    SP = spmatrix(coo.data.tolist(), coo.row.tolist(), coo.col.tolist(), size=A.shape)
    return SP

Defining the shape variable preserves the dimensionality of A on SP. I found that any zero columns ending the scipy sparse matrix would be lost without this added step.

Answer 2

taken from http://maggotroot.blogspot.co.il/2013/11/constrained-linear-least-squares-in.html

coo = A.tocoo()
SP = spmatrix(coo.data, coo.row.tolist(), coo.col.tolist())

Answer 3

From http://cvxopt.org/userguide/matrices.html#sparse-matrices

cvxopt.spmatrix(x, I, J[, size[, tc]])

looks similar to the scipy.sparse

coo_matrix((data, (i, j)), [shape=(M, N)])

My guess is that if A is a matrix in coo format, that

cvxopt.spmatrix(A.data, A.row, A.col, A.shape)

would work. (I don't have cvxopt installed to test this.)