How to assign values to slices in Scipy sparse huge matrix

Question

• I have Huge matrix called Mat_A, when I tried to slice it and assign values, I had got the following Error:
  • Mat_A[:,0 ] = np.ravel(Mat_A.sum(axis=0)) TypeError: coo_matrix' object does not support item assignment
• How can I assign values in the same spirit to the dense matrixweights befor going to sparse space?

Mat_A = sparse.coo_matrix((weights, (i_indices_O, j_indices_O)), shape=(pixel_nb_O, pixel_nb_O))
Mat_A[:,0 ] = np.ravel(Mat_A.sum(axis=0))
Mat_A[:,1:] = 0
Mat_A = Mat_A.tocsr()

Answer 1

coo_matrix's sparse format comes with some disadvantages, which are well mentioned in the docs:

does not directly support:

• arithmetic operations
• slicing

COO is a fast format for constructing sparse matrices, though for arithmetic operations you should switch to CSR or CSC.