def test_generate_cover_coverage():
for graph_size in range(1, 40):
g = generate_weighted_connected_graph(graph_size)
for requested_cover_size in range(1, graph_size):
cover = generate_cover(g, requested_cover_size)
check_cover_size(cover, requested_cover_size)
check_cover_correctness(cover, g.nodes())
def cover_size_experiment(experiment_type='normal'):
sampled_graphs = EXPERIMENTS_PARAMS['sampled graphs'][experiment_type]
k_integers = EXPERIMENTS_PARAMS['k integers'][experiment_type]
cover_types = [(5, 30, 0), (10, 20, 0), (20, 10, 30), (30, 5, 20),
(40, 5, 15), (50, 0, 10)]
results = np.zeros((len(cover_types), NUM_OF_RESULTS_PARAMS))
for _ in range(sampled_graphs):
g = generate_weighted_connected_graph(NUM_OF_NODES, p=0.5)
for i, (cover_size, min_cluster_size,
max_cluster_size) in enumerate(cover_types):
cover = generate_cover(g,
cover_size,
min_cluster_size=min_cluster_size,
max_cluster_size=max_cluster_size)
cover_radius, cover_degree = get_collection_data(g, cover)
for k in k_integers:
coarsening_cover = max_cover(cover, k)
coarsening_radius, coarsening_degree = get_collection_data(
g, coarsening_cover)
results[
i] += cover_radius, coarsening_radius, cover_degree, coarsening_degree
results /= (sampled_graphs * len(k_integers))
x_axis = [cover_type[0] for cover_type in cover_types]
plot_experiment_results(results,
x_axis,
'cover size',
title='Cover Size Experiment')
def k_integer_experiment(experiment_type='normal'):
sampled_graphs = EXPERIMENTS_PARAMS['sampled graphs'][experiment_type]
sampled_covers = EXPERIMENTS_PARAMS['sampled covers'][experiment_type]
k_limit = EXPERIMENTS_PARAMS['k limit'][experiment_type]
cover_size = 20
max_cluster_size = 15
results = np.zeros((k_limit, NUM_OF_RESULTS_PARAMS))
for _ in range(sampled_graphs):
g = generate_weighted_connected_graph(NUM_OF_NODES)
for _ in range(sampled_covers):
cover = generate_cover(g,
cover_size,
max_cluster_size=max_cluster_size)
cover_radius, cover_degree = get_collection_data(g, cover)
for k in range(1, k_limit + 1):
coarsening_cover = max_cover(cover, k)
coarsening_radius, coarsening_degree = get_collection_data(
g, coarsening_cover)
results[
k -
1] += cover_radius, coarsening_radius, cover_degree, coarsening_degree
results /= (sampled_graphs * sampled_covers)
x_axis = [k for k in range(1, k_limit + 1)]
plot_experiment_results(results, x_axis, 'k', title='K Integer Experiment')
def test_max_cover():
for size in range(1, 101, 20):
g = generate_weighted_connected_graph(size)
cover = generate_cover(g, 1)
for k in range(1, 10):
t = max_cover(cover, k)
check_t_radius(g, cover, t, k)
check_t_degree(cover, t, k)
check_coarsening(cover, t)
def test_generate_random_cluster_min_max_size():
for size in range(1, 30):
g = generate_weighted_connected_graph(size)
for i in range(1, size):
for j in range(i, size):
c = generate_random_cluster(g,
min_cluster_size=i,
max_cluster_size=j)
check_cluster_connectivity(g, c)
def test_procedure_cover():
for size in range(1, 101, 20):
g = generate_weighted_connected_graph(size)
cover = generate_cover(g, 1)
for k in range(1, 10):
dr, dt = procedure_cover(cover, k)
check_coarsening(dr, dt)
check_dt_disjointness(dt)
check_dr_bound(dr, cover, k)
check_dt_radius(g, dt, cover, k)
def test_generate_cover_min_max_cluster_size():
for graph_size in range(1, 20):
g = generate_weighted_connected_graph(graph_size)
for i in range(1, graph_size):
for j in range(i, graph_size):
cover = generate_cover(g,
1,
min_cluster_size=i,
max_cluster_size=j)
check_cover_size(cover, 1)
check_cover_correctness(cover, g.nodes())
check_cover_clusters_size(cover, i, j)
def graph_density_experiment(experiment_type='normal'):
sampled_graphs = EXPERIMENTS_PARAMS['sampled graphs'][experiment_type]
sampled_covers = EXPERIMENTS_PARAMS['sampled covers'][experiment_type]
k_integers = EXPERIMENTS_PARAMS['k integers'][experiment_type]
probability_step = EXPERIMENTS_PARAMS['probability step'][experiment_type]
cover_size = 20
max_cluster_size = 15
probabilities = [
0.01 + probability_step * i for i in range(int(1 / probability_step))
]
if DEBUG:
print(probabilities)
results = np.zeros((len(probabilities), NUM_OF_RESULTS_PARAMS))
for i, p in enumerate(probabilities):
for _ in range(sampled_graphs):
g = generate_weighted_connected_graph(NUM_OF_NODES, p=p)
for _ in range(sampled_covers):
cover = generate_cover(g,
cover_size,
max_cluster_size=max_cluster_size)
cover_radius, cover_degree = get_collection_data(g, cover)
for k in k_integers:
coarsening_cover = max_cover(cover, k)
coarsening_radius, coarsening_degree = get_collection_data(
g, coarsening_cover)
results[
i] += cover_radius, coarsening_radius, cover_degree, coarsening_degree
results[i] /= (sampled_graphs * sampled_covers * len(k_integers))
plot_experiment_results(results,
probabilities,
'edge probability',
title='Graph Density Experiment')
def test_generate_random_cluster_connectivity():
for size in range(1, 100):
g = generate_weighted_connected_graph(size)
for _ in range(10):
c = generate_random_cluster(g)
check_cluster_connectivity(g, c)
def test_generate_weighted_connected_graph_weights():
for size in range(1, 101, 20):
for max_weight in range(50):
g = generate_weighted_connected_graph(size, max_weight=max_weight)
check_graph_weights(g, max_weight)
def test_generate_weighted_connected_graph_connectivity():
for size in range(1, 100):
g = generate_weighted_connected_graph(size)
check_graph_connectivity(g)
