def test_ising_simulated_annealing_sample_quality(self):
# because simulated annealing has randomness, we cannot
# really test that it finds the solution. Instead we
# note that it should return better-than-average solutions,
# so if we test the returned energy against the energy of
# 100 random samples, it should do better than the average
nV = 100 # number of variables in h,J
nS = 100 # number of samples
h = {v: random.uniform(-2, 2) for v in range(nV)}
J = {}
for u, v in itertools.combinations(h, 2):
if random.random() < .05:
J[(u, v)] = random.uniform(-1, 1)
random_energies = [
dimod.ising_energy({v: random.choice((-1, 1))
for v in h}, h, J) for __ in range(nS)
]
average_energy = sum(random_energies) / float(nS)
sample, energy = ising_simulated_annealing(h, J)
self.assertLess(energy, average_energy)
def test_ising_simulated_annealing_empty_J(self):
h = {0: -1, 1: 1, 2: -.5}
J = {}
sample, energy = ising_simulated_annealing(h, J)
self.assertIsInstance(sample, dict)
self.assertIsInstance(energy, float)
# make sure all of the nodes are present in sample
for v in range(3):
self.assertIn(v, sample)
def test_ising_simulated_annealing_basic(self):
# AND gate
h = {0: -.5, 1: 0, 2: 1, 3: -.5}
J = {(0, 2): -1, (1, 2): -1, (0, 3): .5, (1, 3): -1}
sample, energy = ising_simulated_annealing(h, J)
self.assertIsInstance(sample, dict)
self.assertIsInstance(energy, float)
# make sure all of the nodes are present in sample
for v in range(4):
self.assertIn(v, sample)
