Define a DQM model for graph partitioning

dqm_graph_partitioning.py

import networkx as nx
import dimod
from dwave.system import LeapHybridDQMSampler

...

dqm = dimod.DiscreteQuadraticModel()
num_nodes = G.number_of_nodes()

for i in G.nodes:
    dqm.add_variable(num_partitions, label=i)

for i in G.nodes:
    linear_term = ...
    dqm.set_linear(i, linear_term)

# Quadratic term for node pairs that do *not* share edges
for p0, p1 in nx.non_edges(G):
    quadratic_term = ...
    dqm.set_quadratic(p0, p1, {(c, c): quadratic_term for c in range(num_partitions)})

# Quadratic term for node pairs which have edges between them
for p0, p1 in G.edges:
    quadratic_term = ...
    dqm.set_quadratic(p0, p1, {(c, c): quadratic_term for c in range(num_partitions)})

# solve!
sampler = LeapHybridDQMSampler()
sampleset = sampler.sample_dqm(dqm, label='Example - Graph Partitioning DQM')

# do something with the lowest-energy solution:
sample = sampleset.first.sample