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_structural_imbalance_docstring_example(self):
sampler = ExactSolver()
S = nx.Graph()
S.add_edge('Alice', 'Bob', sign=1) # Alice and Bob are friendly
S.add_edge('Alice', 'Eve', sign=-1) # Alice and Eve are hostile
S.add_edge('Bob', 'Eve', sign=-1) # Bob and Eve are hostile
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.check_bicolor(colors)
self.assertEqual(frustrated_edges, {})
S.add_edge('Ted', 'Bob', sign=1) # Ted is friendly with all
S.add_edge('Ted', 'Alice', sign=1)
S.add_edge('Ted', 'Eve', sign=1)
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.check_bicolor(colors)
self.assertTrue(frustrated_edges == {('Ted', 'Eve'): {'sign': 1}} or
frustrated_edges == {('Eve', 'Ted'): {'sign': 1}})
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_default_sampler(self):
S = nx.Graph()
S.add_edge('Alice', 'Bob', sign=1) # Alice and Bob are friendly
S.add_edge('Alice', 'Eve', sign=-1) # Alice and Eve are hostile
S.add_edge('Bob', 'Eve', sign=-1) # Bob and Eve are hostile
dnx.set_default_sampler(ExactSolver())
self.assertIsNot(dnx.get_default_sampler(), None)
frustrated_edges, colors = dnx.structural_imbalance(S)
dnx.unset_default_sampler()
self.assertEqual(dnx.get_default_sampler(), None, "sampler did not unset correctly")
def test_frustrated_hostile_edge(self):
"""Set up a graph where the frustrated edge should be hostile"""
sampler = ExactSolver()
S = nx.florentine_families_graph()
# set all hostile
nx.set_edge_attributes(S, -1, 'sign')
# smoke test
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.check_bicolor(colors)
def test_sign_zero(self):
"""though not documentented, agents with no relation can have sign 0.
This is less performant than just not having an edge in most cases."""
sampler = ExactSolver()
S = nx.Graph()
S.add_edge('Alice', 'Bob', sign=1) # Alice and Bob are friendly
S.add_edge('Alice', 'Eve', sign=-1) # Alice and Eve are hostile
S.add_edge('Bob', 'Eve', sign=0) # Bob and Eve have no relation
frustrated_edges, colors = dnx.structural_imbalance(S, sampler)
self.assertEqual(frustrated_edges, {}) # should be no frustration
def main(sampler_type, region, show, inspect):
if sampler_type is None:
print("No solver selected, defaulting to hybrid")
sampler_type = 'hybrid'
# get the appropriate signed social network
G = global_signed_social_network(region=region)
# choose solver and any tuning parameters needed
if sampler_type == 'cpu':
params = dict(num_reads=100)
sampler = SimulatedAnnealingSampler()
elif sampler_type == 'hybrid':
params = dict()
sampler = LeapHybridSampler()
elif sampler_type == 'qpu':
params = dict(
num_reads=100,
chain_strength=2.0,
)
sampler = dimod.TrackingComposite(EmbeddingComposite(DWaveSampler()))
else:
raise RuntimeError("unknown solver type")
# use the chosen sampler (passing in the parameters)
edges, colors = dnx.structural_imbalance(
G, sampler, label='Example - Structural Imbalance', **params)
if inspect and sampler_type == 'qpu':
dwave.inspector.show(sampler.output)
print("Found", len(edges), 'violations out of', len(G.edges), 'edges')
draw_social_network(G, colors)
if show:
plt.show()
else:
filename = 'structural imbalance {} {}.png'.format(
sampler_type, region)
plt.savefig(filename, facecolor='white')
plt.clf()
def main(sampler_type, region, show):
if sampler_type is None:
print("No solver selected, defaulting to hybrid")
sampler_type = 'hybrid'
if region == 'global' and sampler_type == 'qpu':
print("Given region is too large for the QPU, please choose another "
"region or use hybrid.")
# get the appropriate signed social network
G = global_signed_social_network(region=region)
# choose solver and any tuning parameters needed
if sampler_type == 'cpu':
params = dict(num_reads=100)
sampler = SimulatedAnnealingSampler()
elif sampler_type == 'hybrid':
params = dict()
sampler = LeapHybridSampler()
elif sampler_type == 'qpu':
params = dict(
num_reads=100,
chain_strength=2.0,
)
sampler = EmbeddingComposite(DWaveSampler())
else:
raise RuntimeError("unknown solver type")
# use the chosen sampler (passing in the parameters)
edges, colors = dnx.structural_imbalance(G, sampler, **params)
print("Found", len(edges), 'violations out of', len(G.edges), 'edges')
draw_social_network(G, colors)
if show:
plt.show()
else:
filename = 'stuctural imbalance {} {}.png'.format(sampler_type, region)
plt.savefig(filename, facecolor='white', dpi=500)
plt.clf()
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import networkx as nx
import dwave_networkx as dnx
import dwave_micro_client_dimod as micro
import dwave_qbsolv
urlc = 'https://cloud.dwavesys.com/sapi'
tokenc = 'SE-bb7f104b4a99cf9a10eeb9637f0806761c9fcedc'
solver_namec = 'DW_2000Q_1'
structured_samplerc = micro.DWaveSampler(solver_namec, urlc, tokenc)
samplerc = micro.EmbeddingComposite(structured_samplerc)
samplerq = dwave_qbsolv.QBSolv()
cloudsi = dnx.structural_imbalance(G, samplerc , num_reads=10000)
qbsolvsi = dnx.structural_imbalance (G, samplerq , solver= samplerc)
h = {v: node_values [v] for v in G.nodes }
J = {(u, v): eval for u, v in G.edges }
response = samplerc.sample_ising (h, J, num_reads=10000)
from dwave.system.samplers import DWaveSampler
from dwave.system.composites import EmbeddingComposite
sampler = EmbeddingComposite(DWaveSampler(
solver={'qpu': True
})) # Some accounts need to replace this line with the next:
# sampler = EmbeddingComposite(DWaveSampler(solver='paste missing solver name here', token='paste missing API token here'))
# ### Solving the Problem
# Next, the *structural_imbalance()* algorithm, from Ocean's [dwave_networkx](https://docs.ocean.dwavesys.com/projects/dwave-networkx/en/latest/) extension of the NetworkX graphic package, submits the Ising model we formulated in the previous section to a D-Wave system. It returns a partition of our social network into two colored sets and the frustrated edges.
# In[ ]:
# Return a good partition (minimal structural imbalance) and its frustrated edges
import dwave_networkx as dnx
imbalance, bicoloring = dnx.structural_imbalance(G, sampler)
# Mark the returned frustrated edges and node set (color) on the graph
for edge in G.edges:
G.edges[edge]['frustrated'] = edge in imbalance
for node in G.nodes:
G.nodes[node]['color'] = bicoloring[node]
# Print the relationships for the returned partition
print('Yellow set: ',
[person for (person, color) in bicoloring.items() if (color == 0)])
print('Blue set: ',
[person for (person, color) in bicoloring.items() if (color == 1)])
print('\nFrustrated relationships: ', list(imbalance.keys()))
# Display the solution using a *draw()* function that represents friendly interactions as green lines, hostile interactions as red lines, and frustration as dashed lines.
def diagramDateRange(directory_name,
start,
end,
graph,
sampler,
subsolver,
subarea=None,
subarea_name=None):
# Create directories
directory_path = os.path.join('Results', directory_name)
if not os.path.exists(directory_path):
os.makedirs(directory_path)
# Write .csv header line
csv = open(os.path.join(directory_path, 'Structural Imbalance.csv'), 'w')
csv.write(
'year,total groups,total edges,imbalanced edges,structural imbalance')
if subarea is not None and subarea_name is not None:
csv.write(
',%s groups,%s edges,%s imbalanced edges,%s structural imbalance' %
tuple([subarea_name] * 4))
csv.write('\n')
for year in range(start, end):
# Compute structural imbalance up to and including year
filtered_edges = ((u, v) for u, v, a in graph.edges(data=True)
if a['event_date'].year <= year)
subrange = graph.edge_subgraph(filtered_edges)
imbalance, bicoloring = dnx.structural_imbalance(subrange,
sampler,
solver=subsolver)
# Write stats to .csv
num_nodes = len(subrange.nodes)
num_edges = len(subrange.edges)
num_imbalanced = len(imbalance)
ratio = num_imbalanced / num_edges
csv.write('%s,%s,%s,%s,%s' %
(year, num_nodes, num_edges, num_imbalanced, ratio))
if subarea is not None and subarea_name is not None:
num_nodes = len(
set.intersection(set(subrange.nodes), set(subarea.nodes)))
num_edges = len(
set.intersection(set(subrange.edges), set(subarea.edges)))
num_imbalanced = len(
set.intersection(set(imbalance), set(subarea.edges)))
if num_edges != 0:
ratio = num_imbalanced / num_edges
else:
ratio = '-'
csv.write(',%s,%s,%s,%s' %
(num_nodes, num_edges, num_imbalanced, ratio))
csv.write('\n')
# Draw graph
file_name = 'Structural Imbalance %s.png' % year
file_path = os.path.join(directory_path, file_name)
sbdemo.draw(file_path, subrange, imbalance, bicoloring)
print('output %s' % file_path)
csv.close()
sampler = qbsolv.QBSolv()
if _sapi:
print("Running on the QPU")
subsolver = sapi.EmbeddingComposite(
sapi.SAPISampler(solver_name, url, token))
else:
print("Running classically")
subsolver = None
# get the graphs
Global, Syria, Iraq = maps()
# calculate the imbalance of Global
imbalance, bicoloring = dnx.structural_imbalance(Global,
sampler,
solver=subsolver)
# draw the Global graph
sbdemo.draw('syria_imbalance.png', Global, imbalance, bicoloring)
# Images of the structural imbalance in the local Syrian SSN
# for years 2010-2016 showing frustrated and unfrustrated edges.
diagramDateRange('Syrian Theatre', 2010, 2016 + 1, Syria, sampler,
subsolver)
# Images of the structural imbalance in the world SSN for
# years 2007-2016 showing frustrated and unfrustrated edges.
diagramDateRange('World Network', 2007, 2016 + 1, Global, sampler,
subsolver, Syria, 'Syria')
from dwave.system import LeapHybridSampler
import dwave_networkx as dnx
import networkx as nx
import random
# Create a hybrid solver
solver = LeapHybridSampler()
# Create a random graph where nodes are connected based on distance and set the weights to 1 or -1
problem_node_count = 30
G = nx.random_geometric_graph(problem_node_count,
radius=0.0005 * problem_node_count)
G.add_edges_from([(u, v, {
'sign': random.choice((-1, 1))
}) for u, v in G.edges])
imbalance, bicoloring = dnx.structural_imbalance(G, solver)
set1 = int(sum(list(bicoloring.values())))
print("One set has {} nodes; the other has {} nodes.".format(
set1, problem_node_count - set1))
print("The network has {} frustrated relationships.".format(
len(list(imbalance.keys()))))
