def test_simple_schedule_more_machines(self):
jobs = {"j0": [(0, 1)], "j1": [(1, 1)], "j2": [(2, 1)]}
max_time = 3
# Get JSS BQM
scheduler = JobShopScheduler(jobs, max_time)
bqm = scheduler.get_bqm()
# Expected solution
expected = {"j0_0,0": 1, "j1_0,0": 1, "j2_0,0": 1}
fill_with_zeros(expected, jobs, max_time)
expected_energy = get_energy(expected, bqm)
# Sampled solution
# response = EmbeddingComposite(DWaveSampler()).sample(bqm, num_reads=10000)
# response_sample, sample_energy, _, chain_break_fraction = next(response.data())
# print("Chain Break Fraction: ", chain_break_fraction)
# response = SimulatedAnnealingSampler().sample(bqm, num_reads=2000, beta_range=[0.01, 10])
response = ExactSolver().sample(bqm)
response_sample, sample_energy, _ = next(response.data())
# Print response
self.assertTrue(scheduler.csp.check(response_sample))
self.assertEqual(expected_energy, sample_energy)
self.compare(response_sample, expected)
def test_no_bounds(self):
bqm = BinaryQuadraticModel({
0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0
}, {
(0, 1): -2.0,
(0, 2): -5.0,
(0, 3): -2.0,
(0, 4): -2.0,
(1, 2): -2.0,
(1, 3): -2.0,
(1, 4): 4.0,
(2, 3): -3.0,
(2, 4): -5.0,
(3, 4): -4.0
}, 0, dimod.SPIN)
with self.assertWarns(DeprecationWarning):
sampler = ClipComposite(ExactSolver())
solver = ExactSolver()
response = sampler.sample(bqm)
response_exact = solver.sample(bqm)
self.assertEqual(response.first.sample,
response_exact.first.sample)
self.assertAlmostEqual(response.first.energy,
response_exact.first.energy)
def test_vertex_cover_weighted(self):
weight = 'weight'
G = nx.path_graph(6)
# favor even nodes
nx.set_node_attributes(G, {node: node % 2 + 1 for node in G}, weight)
cover = dnx.min_weighted_vertex_cover(G, weight, ExactSolver())
self.assertEqual(set(cover), {0, 2, 4})
# favor odd nodes
nx.set_node_attributes(G, {node: (node + 1) % 2 + 1
for node in G}, weight)
cover = dnx.min_weighted_vertex_cover(G, weight, ExactSolver())
self.assertEqual(set(cover), {1, 3, 5})
# make nodes 1 and 4 unlikely
nx.set_node_attributes(G, {0: 1, 1: 3, 2: 1, 3: 1, 4: 3, 5: 1}, weight)
cover = dnx.min_weighted_vertex_cover(G, weight, ExactSolver())
self.assertEqual(set(cover), {0, 2, 3, 5})
for __ in range(10):
G = nx.gnp_random_graph(5, .5)
nx.set_node_attributes(G, {node: random.random()
for node in G}, weight)
cover = dnx.min_weighted_vertex_cover(G, weight, ExactSolver())
self.vertex_cover_check(G, cover)
def test_with_labels(self):
bqm = BinaryQuadraticModel(
{
'a': 0.0,
'b': 0.0,
'c': 0.0,
'd': 0.0,
'e': 0.0
}, {
('a', 'b'): -2.0,
('a', 'c'): -5.0,
('a', 'd'): -2.0,
('a', 'e'): -2.0,
('b', 'c'): -2.0,
('b', 'd'): -2.0,
('b', 'e'): 4.0,
('c', 'd'): -3.0,
('c', 'e'): -5.0,
('d', 'e'): -4.0
}, 0, dimod.SPIN)
sampler = ClipComposite(ExactSolver())
solver = ExactSolver()
response = sampler.sample(bqm, lower_bound=-1, upper_bound=1)
response_exact = solver.sample(bqm)
self.assertEqual(response.first.sample, response_exact.first.sample)
self.assertAlmostEqual(response.first.energy,
response_exact.first.energy)
def check_generated_ising_model(self, feasible_configurations, decision_variables,
linear, quadratic, ground_energy, infeasible_gap):
"""Check that the given Ising model has the correct energy levels"""
if not feasible_configurations:
return
from dimod import ExactSolver
samples = ExactSolver().sample_ising(linear, quadratic)
# samples are returned in order of energy
sample, ground = next(iter(samples.items()))
gap = float('inf')
self.assertIn(tuple(sample[v] for v in decision_variables), feasible_configurations)
seen_configs = set()
for sample, energy in samples.items():
config = tuple(sample[v] for v in decision_variables)
# we want the minimum energy for each config of the decisison variables,
# so once we've seen it once we can skip
if config in seen_configs:
continue
if config in feasible_configurations:
self.assertAlmostEqual(energy, ground)
seen_configs.add(config)
else:
gap = min(gap, energy - ground)
self.assertAlmostEqual(ground_energy, ground)
self.assertAlmostEqual(gap, infeasible_gap)
def test_sample_hising_quadratic_range(self):
linear = {'a': -4.0, 'b': -4.0, 'c': -4.0}
quadratic = {('a', 'b', 'c'): 3.2}
offset = 5
ignored_variables, ignored_interactions = _check_params(None, None)
comp_parameters = dict(ignored_interactions=ignored_interactions,
ignored_variables=ignored_variables,
penalty_strength=5.,
quadratic_range=(-1, 2)
)
sampler = ScaleComposite(
ScalingChecker(HigherOrderComposite(ExactSolver()),
h=linear,
J=quadratic, offset=offset,
**comp_parameters))
response = sampler.sample_ising(linear, quadratic, offset=offset,
**comp_parameters)
self.assertAlmostEqual(response.first.energy, -3.8)
comp_parameters = dict(ignored_interactions=ignored_interactions,
ignored_variables=ignored_variables,
penalty_strength=5.,
quadratic_range=(-1, 0.4)
)
sampler = ScaleComposite(
ScalingChecker(HigherOrderComposite(ExactSolver()),
h=linear,
J=quadratic, offset=offset,
**comp_parameters))
response = sampler.sample_ising(linear, quadratic, offset=offset,
**comp_parameters)
self.assertAlmostEqual(response.first.energy, -3.8)
def test_ub_only(self):
bqm = BinaryQuadraticModel({
0: 0.0,
1: 0.0,
2: 0.0,
3: 0.0,
4: 0.0
}, {
(0, 1): -2.0,
(0, 2): -5.0,
(0, 3): -2.0,
(0, 4): -2.0,
(1, 2): -2.0,
(1, 3): -2.0,
(1, 4): 4.0,
(2, 3): -3.0,
(2, 4): -5.0,
(3, 4): -4.0
}, 0, dimod.SPIN)
sampler = ClipComposite(ExactSolver())
solver = ExactSolver()
response = sampler.sample(bqm, upper_bound=1)
response_exact = solver.sample(bqm)
self.assertEqual(response.first.sample, response_exact.first.sample)
self.assertAlmostEqual(response.first.energy,
response_exact.first.energy)
def test_largerSchedule(self):
jobs = {
'small1': [(1, 1)],
'small2': [(2, 2)],
'longJob': [(0, 1), (1, 1), (2, 1)]
}
max_time = 4
# Get exact sample from Job Shop Scheduler BQM
jss = JobShopScheduler(jobs, max_time)
bqm = jss.get_bqm()
response = ExactSolver().sample(bqm)
response_sample = next(response.samples())
# Verify that response_sample obeys constraints
self.assertTrue(jss.csp.check(response_sample))
# Create expected solution
expected = {
"small1_0,0": 1,
"small2_0,0": 1,
"longJob_0,0": 1,
"longJob_1,1": 1,
"longJob_2,2": 1
}
# Compare variable values
self.compare(response_sample, expected)
def test_dimod_vs_list(self):
G = nx.path_graph(5)
matching = dnx.min_maximal_matching(G, ExactSolver())
matching = dnx.algorithms.matching.maximal_matching(G, ExactSolver())
matching = dnx.min_maximal_matching(G, SimulatedAnnealingSampler())
matching = dnx.algorithms.matching.maximal_matching(
G, SimulatedAnnealingSampler())
def test_edge_cases(self):
# get the empty graph
G = nx.Graph()
S = dnx.maximum_cut(G, ExactSolver())
self.assertTrue(len(S) == 0)
S = dnx.weighted_maximum_cut(G, ExactSolver())
self.assertTrue(len(S) == 0)
def test_maximal_matching_typical(self):
G = nx.complete_graph(5)
matching = dnx.algorithms.matching.maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
for __ in range(10):
G = nx.gnp_random_graph(7, .5)
matching = dnx.algorithms.matching.maximal_matching(
G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
def test_min_maximal_matching_typical(self):
G = nx.complete_graph(5)
matching = dnx.min_maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
for __ in range(10):
G = nx.gnp_random_graph(7, .5)
matching = dnx.min_maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching),
"nodes: {}\nedges:{}".format(G.nodes(), G.edges()))
def test_vertex_color_basic(self):
# all small enough for an exact solver to handle them reasonably
G = dnx.chimera_graph(1, 2, 2)
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
G = nx.path_graph(5)
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
for __ in range(10):
G = nx.gnp_random_graph(5, .5)
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
def test_vertex_cover_basic(self):
G = dnx.chimera_graph(1, 2, 2)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
G = nx.path_graph(5)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
for __ in range(10):
G = nx.gnp_random_graph(5, .5)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
def test_path_graph(self):
G = nx.path_graph(10)
matching = dnx.algorithms.matching.maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
matching = dnx.min_maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
G.add_edge(0, 9)
matching = dnx.algorithms.matching.maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
matching = dnx.min_maximal_matching(G, ExactSolver())
self.assertTrue(dnx.is_maximal_matching(G, matching))
def test_vertex_color_disconnected_graph(self):
"""One edge and one disconnected node"""
G = nx.path_graph(2)
G.add_node(3)
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
def _get_sampler(self, profile, solver):
"Return a dimod.Sampler object."
# Handle built-in software samplers as special cases.
info = {}
if solver != None:
info["solver_name"] = solver
if solver == "exact":
return ExactSolver(), info
elif solver == "neal":
return SimulatedAnnealingSampler(), info
elif solver == "tabu":
return TabuSampler(), info
# In the common case, read the configuration file, either the
# default or the one named by the DWAVE_CONFIG_FILE environment
# variable.
if profile != None:
info["profile"] = profile
try:
with Client.from_config(profile=profile) as client:
if solver == None:
solver = client.default_solver
else:
solver = {"name": solver}
sampler = DWaveSampler(profile=profile, solver=solver)
info = {
"solver_name": sampler.solver.name,
"endpoint": client.endpoint
}
return sampler, info
except Exception as err:
self.qmasm.abend("Failed to construct a sampler (%s)" % str(err))
class TestExactSolver(unittest.TestCase, TestSamplerAPI):
def setUp(self):
self.sampler = ExactSolver()
def test_all_samples(self):
"""Check that every sample is included and that they all have the correct energy."""
n = 10
# create a qubo
Q = {(v, v): (v % 3) for v in range(n)}
Q[(0, n - 1)] = 1.3
Q[(3, n - 2)] = -.26666666
response = self.sampler.sample_qubo(Q)
self.assertEqual(len(response), 2**n,
"incorrect number of samples returned")
sample_tuples = set()
for sample in response.samples():
stpl = tuple(sample[v] for v in range(n))
sample_tuples.add(stpl)
for tpl in itertools.product((0, 1), repeat=n):
self.assertIn(tpl, sample_tuples)
# let's also double check the enegy
for sample, energy in response.items():
self.assertTrue(abs(energy - qubo_energy(Q, sample)) < .000001)
def test_vertex_color_no_edge_graph(self):
"""Graph with many nodes but no edges, should be caught before QUBO"""
# this should get eliminated in software so be fast to run
G = nx.Graph()
G.add_nodes_from(range(100))
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
def test_vertex_cover_basic(self):
"""Runs the function on some small and simple graphs, just to make
sure it works in basic functionality.
"""
G = dnx.chimera_graph(1, 2, 2)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
G = nx.path_graph(5)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
for __ in range(10):
G = nx.gnp_random_graph(5, .5)
cover = dnx.min_vertex_cover(G, ExactSolver())
self.vertex_cover_check(G, cover)
def test_structural_imbalance_basic(self):
sampler = ExactSolver()
blueteam = ['Alice', 'Bob', 'Carol']
redteam0 = ['Eve']
redteam1 = ['Mallory', 'Trudy']
S = nx.Graph()
for p0, p1 in itertools.combinations(blueteam, 2):
S.add_edge(p0, p1, sign=1)
S.add_edge(*redteam1, sign=1)
for p0 in blueteam:
for p1 in redteam0:
S.add_edge(p0, p1, sign=-1)
for p1 in redteam1:
S.add_edge(p0, p1, sign=-1)
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.check_bicolor(colors)
greenteam = ['Ted']
for p0 in set(S.nodes):
for p1 in greenteam:
S.add_edge(p0, p1, sign=1)
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.check_bicolor(colors)
def test_roof_duality_3path(self):
sampler = FixVariablesComposite(ExactSolver(),
algorithm='roof_duality')
sampleset = sampler.sample_ising({'a': 10}, {'ab': -1, 'bc': 1})
# all should be fixed, so should just see one
self.assertEqual(len(sampleset), 1)
self.assertEqual(set(sampleset.variables), set('abc'))
def test_maximum_independent_set_basic(self):
"""Runs the function on some small and simple graphs, just to make
sure it works in basic functionality.
"""
G = dnx.chimera_graph(1, 2, 2)
indep_set = dnx.maximum_independent_set(G, ExactSolver())
self.set_independence_check(G, indep_set)
G = nx.path_graph(5)
indep_set = dnx.maximum_independent_set(G, ExactSolver())
self.set_independence_check(G, indep_set)
for __ in range(10):
G = nx.gnp_random_graph(5, .5)
indep_set = dnx.maximum_independent_set(G, ExactSolver())
self.set_independence_check(G, indep_set)
def test_sampleset_trim(self):
h = {'a': -4.0, 'b': -4.0, 'c': 0}
J = {('a', 'b'): 3.2}
sampler = TruncateComposite(ExactSolver(), 6)
samples = sampler.sample_ising(h, J)
self.assertEqual(len(samples), 6)
def test_invalid_graph(self):
"""should throw an error with a graph without sign attribute"""
S = nx.Graph()
S.add_edge('Alice', 'Bob', sign=1)
S.add_edge('Alice', 'Eve', sign=-1)
S.add_edge('Bob', 'Eve') # invalid edge
with self.assertRaises(ValueError):
frustrated_edges, colors = dnx.structural_imbalance(S, ExactSolver())
def test_vertex_color_almost_complete(self):
# this should get eliminated in software so be fast to run
G = nx.complete_graph(10)
mapping = dict(zip(G.nodes(), "abcdefghijklmnopqrstuvwxyz"))
G = nx.relabel_nodes(G, mapping)
n0, n1 = next(iter(G.edges()))
G.remove_edge(n0, n1)
coloring = dnx.min_vertex_coloring(G, ExactSolver())
self.assertTrue(dnx.is_vertex_coloring(G, coloring))
def test_tinySchedule(self):
jobs = {"a": [(1, 1), (2, 1)], "b": [(2, 1)]}
max_time = 2
# Get exact sample from Job Shop Scheduler BQM
jss = JobShopScheduler(jobs, max_time)
bqm = jss.get_bqm()
response = ExactSolver().sample(bqm)
response_sample = next(response.samples())
# Verify that response_sample obeys constraints
self.assertTrue(jss.csp.check(response_sample))
# Create expected solution
expected = {"a_0,0": 1, "a_1,1": 1, "b_0,0": 1}
# Compare variable values
self.compare(response_sample, expected)
def test_typical_cases(self):
G = nx.complete_graph(10)
S = dnx.maximum_cut(G, ExactSolver())
self.assertTrue(len(S) == 5) # half of the nodes
with self.assertRaises(dnx.DWaveNetworkXException):
S = dnx.weighted_maximum_cut(G, ExactSolver())
nx.set_edge_attributes(G, 1, 'weight')
S = dnx.weighted_maximum_cut(G, ExactSolver())
self.assertTrue(len(S) == 5) # half of the nodes
G = nx.Graph()
G.add_edges_from([(0, 1), (0, 2), (1, 2), (1, 3), (3, 4), (2, 4)])
S = dnx.maximum_cut(G, ExactSolver())
self.assertTrue(len(S) in (2, 3))
def test_default_sampler(self):
G = nx.complete_graph(5)
dnx.set_default_sampler(ExactSolver())
self.assertIsNot(dnx.get_default_sampler(), None)
cover = dnx.min_vertex_cover(G)
dnx.unset_default_sampler()
self.assertEqual(dnx.get_default_sampler(), None,
"sampler did not unset correctly")
def test_sampleset_trim(self):
h = {'a': -4.0, 'b': -4.0, 'c': 0}
J = {('a', 'b'): 3.2}
sampler = PolyTruncateComposite(HigherOrderComposite(ExactSolver()), 6)
with self.assertWarns(DeprecationWarning):
samples = sampler.sample_ising(h, J)
self.assertEqual(len(samples), 6)
