def _analyze(graph):
# Get the walk object
w_obj = polygraph.walk_classes(graph)
# Solve the linear system involving the eig_matrix
res = polygraph.positive_linear_system_check(w_obj)
# Check to see if x is positive
return np.all(res.x >= 0)
}
pyramid = gen.pyramid_prism(4, 0)
pwff = []
acff = []
classes = []
for graph_type, generator in GRAPHS.items():
pwrow = [graph_type]
acrow = [graph_type]
crow = [graph_type]
for i in range(MIN, MAX_1):
graph = nx.cartesian_product(pyramid, generator(i))
w_obj = polygraph.walk_classes(graph)
w = w_obj['eig_matrix']
pwrow.append(polygraph.pair_wise_flip_flopping(w))
acrow.append(polygraph.average_condition_flip_flopping(w))
crow.append(w_obj['num_classes'])
pwff.append(pwrow)
acff.append(acrow)
classes.append(crow)
print('Number of Walk Classes')
print(tabulate(
classes,
tablefmt='grid',
headers=['Graph', *[i for i in range(MIN, MAX_1)]]
))
print('Pair-Wise Flip-Flopping')
for name, generator in graphs_generators:
logger.info('Analyzing graph {}'.format(name))
# Construct graph
logger.info('Generating graph')
g = generator()
# Analyze walk classes
logger.info('Analyzing walk classes')
if type(g) is dict:
graph = g['graph']
w_obj = polygraph.spider_torus_walk_classes(g)
else:
graph = g
w_obj = polygraph.walk_classes(g, max_power=MAX_POWER)
# Generate solutions to nonnegative linear system
logger.info('Performing nonnegative system check')
res = polygraph.nonnegative_linear_system_check(w_obj)
# Check for success
if not res.success:
logger.warning('Failed nonnegative check on {} with {}'.format(
name,
res.message
))
else:
# Take solutions as coefficients
coefficients = res.x[::-1]