def generateLargeGraphs(M, N):
startTotal = time.time()
print("Generating %d graphs of size %d..." % (M, N), end=" ")
# Generate graphs using barabasi-albert
graphs = []
num_of_edges = min(100, N // 10)
for m in range(0, M):
graphs.append(barabasi_albert_graph(N, num_of_edges))
m += 1
endTotal = int(time.time())
print("Duration: %d minutes and %d seconds" % secondsToMinSec(endTotal - startTotal))
return graphs
def generateSmallGraphs(N):
startTotal = time.time()
print("Importing graphs of size %d..." % N, end=" ")
if not (5 < N < 10):
raise Exception('The size for small graphs should be between 6 and 10')
exit(0)
graphs = []
with open('./graphs/non-isomorphs/graph' + str(N) + 'c.g6') as graphs_file:
graph_line = graphs_file.readline()
while graph_line:
graph_line = graph_line.rstrip().encode('ascii')
graph = from_graph6_bytes(graph_line)
graphs.append(graph)
graph_line = graphs_file.readline()
endTotal = int(time.time())
print("Duration: %d minutes and %d seconds" % secondsToMinSec(endTotal - startTotal))
return graphs
def getCentralityValuesDict(graphs, centralities):
startTotal = time.time()
print("Building centrality dictionaries...", end=" ")
result = OrderedDict()
for c in centralities:
result[c] = OrderedDict()
if "betweenness" in centralities:
for g in graphs:
result["betweenness"][g] = betweenness_centrality(g,
k=len(g.nodes))
if "closeness" in centralities:
for g in graphs:
result["closeness"][g] = closeness_centrality(g)
if "katz" in centralities:
for g in graphs:
result["katz"][g] = katz_centrality_numpy(g)
if "eigenvector" in centralities:
for g in graphs:
result["eigenvector"][g] = eigenvector_centrality_numpy(g)
if "pagerank" in centralities:
for g in graphs:
result["pagerank"][g] = pagerank_numpy(g)
if "degree" in centralities:
for g in graphs:
result["degree"][g] = degree_centrality(g) # is normalized
endTotal = int(time.time())
print("Duration: %d minutes and %d seconds" %
secondsToMinSec(endTotal - startTotal))
return result
def buildDataSetPar(graphs, centralities_dict, spread_param, iterations):
X, y = [], []
startTotal = time.time()
processors = cpu_count()
bs = batches(graphs, floor(int(len(graphs) / processors)))
print("Building dataset from", len(bs), "batches of", len(bs[0]), "graphs")
results = Parallel(n_jobs=-1, verbose=0)(delayed(buildDataSetWorker)(
batch, centralities_dict, spread_param, iterations) for batch in bs)
for _X, _y in results:
for __X in _X:
X.append(__X)
y.extend(_y)
Xn, yn = np.array(X), np.array(y)
endTotal = int(time.time())
print("Duration: %d minutes and %d seconds" %
secondsToMinSec(endTotal - startTotal))
return Xn, y
def buildDataSet(graphs, centralities_dict, spread_param, iterations):
X, y = [], []
startTotal = time.time()
print("Building data set...", end=" ")
for graph in graphs:
for seed in graph.nodes:
temp_centralities = []
for centr_key in centralities_dict:
temp_centralities.append(
centralities_dict[centr_key][graph][seed])
X.append(temp_centralities)
if spread_param is not None:
if concurrent and iterations >= 1500 and len(
graphs[0].nodes) > 50:
spread = independentCascadePar(graph, seed, spread_param,
iterations)
else:
spread = independentCascade(graph, seed, spread_param,
iterations)
else:
spread = weightedCascade(graph, seed, iterations)
y.append(spread)
Xn, yn = np.array(X), np.array(y)
endTotal = int(time.time())
print("Duration: %d minutes and %d seconds" %
secondsToMinSec(endTotal - startTotal))
return Xn, y