A sparse matrix stores only its nonzero entries, saving memory when most elements are zero. Using NumPy together with SciPy's scipy.sparse module:

from scipy import sparse
import numpy as np

F = np.eye(3, k=1)          # 3x3 matrix with ones on the first superdiagonal
G = np.mat(np.identity(2))  # 2x2 identity matrix
C[C > 0.5] = 0              # Zero out entries so the array becomes sparse

H = sparse.csr_matrix(C)    # Compressed Sparse Row (CSR) matrix
I = sparse.csc_matrix(D)    # Compressed Sparse Column (CSC) matrix
J = sparse.dok_matrix(A)    # Dictionary Of Keys (DOK) matrix

E.todense()                 # Convert a sparse matrix back to a dense (full) matrix
sparse.isspmatrix_csc(A)    # Check whether a matrix is stored in CSC format