def test_sparse_dcm_rd(self):
A = np.array([[0, 1, 1, 0, 0], [1, 0, 1, 0, 0], [0, 0, 0, 1, 0],
[1, 1, 0, 0, 1], [0, 0, 1, 0, 0]])
sA = scipy.sparse.csr_matrix(A)
sol, step, diff = iterative_solver(sA,
max_steps=300,
eps=0.01,
method='dcm_rd')
# output convergence
# print('steps = {}'.format(step))
# print('diff = {}'.format(diff))
# epectation degree vs actual degree
d = sample.scalability_classes(A, method='dcm_rd')
expected_out_it = sample.expected_out_degree(sol, 'dcm_rd', d)
expected_in_it = sample.expected_in_degree(sol, 'dcm_rd', d)
expected_k = np.concatenate((expected_out_it, expected_in_it))
k_out = sample.out_degree(sA)
k_in = sample.in_degree(sA)
k = np.concatenate((k_out, k_in))
# debug check
# print(d)
# print(k)
# print(expected_k)
# print(np.linalg.norm(k- expected_k))
self.assertTrue(np.allclose(expected_k, k, atol=1e-02, rtol=1e-02))
def test_dcm_rd(self):
A = np.array([[0, 1, 1, 0, 0], [1, 0, 1, 0, 0], [0, 0, 0, 1, 0],
[1, 1, 0, 0, 1], [0, 0, 1, 0, 0]])
right_d = {(2, 2): [0, 1], (1, 3): [2], (3, 1): [3], (1, 1): [4]}
k_out = sample.out_degree(A)
k_in = sample.in_degree(A)
k = np.concatenate((k_out, k_in))
d = sample.scalability_classes(A, 'dcm_rd')
self.assertTrue(
np.alltrue(
sample.rd2full_dcm_rd(np.array([2, 1, 3, 1, 2, 3, 1, 1]), d) ==
k))
def tester(A, verbose=False):
t = time.time()
tic = time.time()
k_out = sample.out_degree(A)
k_in = sample.in_degree(A)
s_out = sample.out_strength(A)
s_in = sample.in_strength(A)
k = np.concatenate((k_out, k_in, s_out, s_in))
toc = time.time() - tic
if len(k) < 30:
print("\n\nExperiment matrix: \n {}".format(A))
print("\n degree sequence:\n{}".format(k))
print('\ndegree sequence computation time = {}'.format(toc))
# dianati iterative function for dcm
tic = time.time()
n = len(k_in)
par, x0 = sample.setup(A, method='decm')
print('initial guess = {}'.format(x0))
sol, step, diff = sample.iterative_solver(A,
max_steps=300,
eps=1e-4,
method='decm',
verbose=verbose)
toc = time.time() - tic
print('\nsolver exectution time = {}'.format(toc))
# test results: degree reconstruction
tic = time.time()
k_in_sol = sample.expected_in_degree_decm(sol)
k_out_sol = sample.expected_out_degree_decm(sol)
s_in_sol = sample.expected_in_strength_decm(sol)
s_out_sol = sample.expected_out_strength_decm(sol)
k_sol = np.concatenate((k_out_sol, k_in_sol, s_out_sol, s_in_sol))
toc = time.time() - tic
print('\n number of steps = {}'.format(step))
print('\nsolution = {}'.format(sol))
print('\ndegree reconstruction time = {}'.format(toc))
print('\n{}reconstruction error = {}{}'.format(bcolors.WARNING,
np.linalg.norm(k - k_sol),
bcolors.ENDC))
if len(k) < 30:
print("\n Original degree sequence:\n{}".format(k))
print("\n Reconstructed degree sequence\n{}".format(k_sol))
toc = time.time() - t
print('\ntotal test time = {}'.format(toc))
def test_array_dcm(self):
A = np.array([[0, 1, 1, 0], [0, 0, 1, 1], [0, 0, 0, 0], [0, 0, 1, 0]])
sol, step, diff = iterative_solver(A, max_steps=300, eps=0.01)
# output convergence
# print('steps = {}'.format(step))
# print('diff = {}'.format(diff))
# epectation degree vs actual degree
expected_out_it = sample.expected_out_degree(sol, 'dcm')
expected_in_it = sample.expected_in_degree(sol, 'dcm')
expected_k = np.concatenate((expected_out_it, expected_in_it))
k_out = sample.out_degree(A)
k_in = sample.in_degree(A)
k = np.concatenate((k_out, k_in))
# debug check
# print(k)
# print(expected_k)
# print(np.linalg.norm(k- expected_k))
self.assertTrue(np.allclose(expected_k, k, atol=1e-02, rtol=1e-02))
def tester(A, verbose=False):
t = time.time()
tic = time.time()
k_out = sample.out_degree(A)
k_in = sample.in_degree(A)
k = np.concatenate((k_out, k_in))
toc = time.time() - tic
if len(k) < 12:
print("\n\nExperiment matrix: \n {}".format(A))
print("\n degree sequence:\n{}".format(k))
print('\ndegree sequence computation time = {}'.format(toc))
# dianati iterative function for dcm
tic = time.time()
n = len(k_in)
x0 = np.array([ki / np.sqrt(n) for ki in k]) # initialial point
sol, step, diff = sample.iterative_solver(A,
max_steps=300,
eps=0.01,
method='dcm')
toc = time.time() - tic
print('\nsolver exectution time = {}'.format(toc))
# test results: degree reconstruction
tic = time.time()
k_in_sol = sample.expected_in_degree_dcm(sol)
k_out_sol = sample.expected_out_degree_dcm(sol)
k_sol = np.concatenate((k_out_sol, k_in_sol))
toc = time.time() - tic
print('\n number of steps = {}'.format(step))
print('\nsolution = {}'.format(sol))
print('\ndegree reconstruction time = {}'.format(toc))
print('\n{}reconstruction error = {}{}'.format(bcolors.WARNING,
np.linalg.norm(k - k_sol),
bcolors.ENDC))
if len(k) < 12:
print("\n Original degree sequence:\n{}".format(k))
print("\n Reconstructed degree sequence\n{}".format(k_sol))
toc = time.time() - t
print('\ntotal test time = {}'.format(toc))
def test_array_inin(self):
A = np.array([[0, 0, 1, 0, 1],
[1, 0, 0, 1, 0],
[1, 1, 0, 0, 0],
[0, 1, 1, 0, 1],
[1, 0, 0, 1, 0]])
A_nx = nx.convert_matrix.from_numpy_array(A, create_using=nx.DiGraph())
k = sample.in_degree(A)
knn_sample = sample.nearest_neighbour_degree_inin(A)
# knn_nx = nx.degree_assortativity_coefficient(A_nx)
knn_nx = nx.k_nearest_neighbors(A_nx, source='in', target='in')
# debug check
"""
print(k)
print('\n')
print(knn_sample)
print(knn_nx)
"""
self.assertTrue(knn_sample == knn_nx)
G = nx.read_graphml(db_path + week)
# A = nx.to_numpy_array(G, dtype=int)
A = nx.to_scipy_sparse_matrix(G)
# iterative solver
tic = time.time()
sol, step, diff = sample.iterative_solver(A, eps=1e-04, method='dcm_rd')
t = time.time() - tic
# test solution
d = sample.scalability_classes(A, 'dcm_rd')
expected_out_it = sample.expected_out_degree(sol, 'dcm_rd', d)
expected_in_it = sample.expected_in_degree(sol, 'dcm_rd', d)
expected_k = np.concatenate((expected_out_it, expected_in_it))
k_out = sample.out_degree(A)
k_in = sample.in_degree(A)
k = np.concatenate((k_out, k_in))
error = max(abs(k - expected_k))
result = np.allclose(expected_k, k, atol=1e-02, rtol=1e-02)
# output
print('experiment success: {}'.format(result))
if not result:
print('error = {}'.format(error))
print('diff = {}'.format(diff) )
print('steps = {}'.format(step) )
print('time = {}'.format(t) )
# save results
csvfile = 'df_btc_dcmrd.csv'
# if the csv file already exists, do not write the header
