def compute_scores(self): A = self.get_adj() r = list(range(self.num_nodes)) eigval = hf.scipy_eigsh(A) scores = np.zeros((self.num_nodes, self.num_nodes)) for i in range(1,self.num_nodes): for j in range(i): if A[i,j] < 1: first = hf.partition(A,np.array([i,j]),np.array([i,j])) comp = hf.complement(r,[i,j]) second = hf.partition(A,np.array([i,j]),comp) temp = hf.partition(A, comp, comp) third = np.linalg.inv(hf.partition(A, comp, comp)-eigval*np.identity(self.num_nodes-2)) fourth = hf.partition(A, comp,np.array([i,j])) #print(first.shape, second.shape, third.shape, fourth.shape) okay = second @ third @ fourth this = first-okay scores[i,j] = this[0,1] return scores
def par(fname):
rank = 0
size = 20
start = time.time()
filename = 'data/' + fname + '.txt'
G, to_predict, fullG = load_data(filename)
directed = False
k = len(to_predict)
min_links = k
score_tolerance = 1e-10
score_inf = 1e18
force_exact = False
A = nx.adjacency_matrix(G)
num_nodes = A.shape[0]
node_list = G.nodes()
degrees = np.sum(A, axis=1)
M = A / degrees
r = list(range(num_nodes))
eigval = hf.scipy_eigsh(M)
for i in range(num_nodes):
for j in range(rank, num_nodes, size):
if A[i, j] == 0 and i != j:
first = hf.partition(M, np.array([i, j]), np.array([i, j]))
comp = hf.complement(r, [i, j])
second = hf.partition(M, np.array([i, j]), comp)
temp = hf.partition(M, comp, comp)
third = np.linalg.inv(
hf.partition(M, comp, comp) -
eigval * np.identity(num_nodes - 2))
fourth = hf.partition(M, comp, np.array([i, j]))
okay = second @ third @ fourth
this = first - okay
send = this[0, 1]
return time.time() - start
score_inf = 1e18
force_exact = False
start = time.time()
A = nx.adjacency_matrix(G)
num_nodes = A.shape[0]
node_list = G.nodes()
degrees = np.sum(A, axis=1)
M = A / degrees
r = list(range(num_nodes))
eigval = hf.scipy_eigsh(M)
scores = np.zeros((num_nodes, num_nodes))
for i in range(num_nodes):
for j in range(num_nodes):
if A[i, j] == 0 and i != j:
first = hf.partition(M, np.array([i, j]), np.array([i, j]))
comp = hf.complement(r, [i, j])
second = hf.partition(M, np.array([i, j]), comp)
temp = hf.partition(M, comp, comp)
third = np.linalg.inv(
hf.partition(M, comp, comp) -
eigval * np.identity(num_nodes - 2))
fourth = hf.partition(M, comp, np.array([i, j]))
okay = second @ third @ fourth
this = first - okay
scores[i, j] = this[0, 1]
print(time.time() - start)