def simulate_and_dump_result(algorithm, graph_filename, out_filename):
# parsing valid algorithm
if algorithm == 'sun':
opt_class = Sun
config = dict(log_level=LOG_LEVEL,
force_update=True, # update graph even if swap is not done
max_trials=1000, # and continue even if swap is done
Nt=6, Ng=10)
n_steps = 30
elif algorithm == 'ichinose':
opt_class = IchinoseSatotani
config = dict(log_level=LOG_LEVEL,
force_update=False) # does not accept if swap is not done
n_steps = 500
elif algorithm == 'ichinose_greedy':
opt_class = IchinoseSatotani
config = dict(log_level=LOG_LEVEL,
greedy=True, # greedy swapping
force_update=False) # same as above
n_steps = 500
else: raise ValueError('invalid algorithm: ', algorithm)
# number of attacks (to calculate R)
n_attacks = 10
# load graph
graph = nx.read_gpickle(graph_filename)
# define columns in csv
columns = [
('R', lambda G: R(G, n_attacks)),
('r', nx.degree_pearson_correlation_coefficient)
]
# use TimeTracker to track R and r
tracker = TimeTracker(opt_class, graph, columns, **config)
tracker.track(steps=n_steps)
tracker.dump(out_filename) # in this time, algorithm$i.ba.csv
G1 = scale_free_network()
R1 = R(G1, n=10) # calculate 10 times to get more accurate result
# use of optimizer
print('----- optimizer test -----')
print('original R =', R1)
for opt_cls in [Schneider, WuHolme, IchinoseSatotani]:
# still there are more options
# see robust_graph/optimize document
optimizer = opt_cls(G1, log_level=LOG_LEVEL, max_trials=10)
G2 = optimizer.optimize(steps=100)
R2 = R(G2, n=10)
print('optimized by', opt_cls.__name__, 'R =', R2)
# use of tracker
print('----- time tracker test -----')
# measurements used in timetracker
cols = [('R', lambda G: R(G, n=10)), # R robustness
('r', nx.degree_pearson_correlation_coefficient), # r degree correlation
]
# WuHolme is not iterative algorithm
for opt_cls in [Schneider, IchinoseSatotani]:
# options are given after columns(measurements)
tracker = TimeTracker(opt_cls, G1, cols=cols,
log_level=LOG_LEVEL, max_trials=5)
tracker.track(steps=2)
print('tracked by', opt_cls.__name__, 'algorithm...')
print(tracker.dumps())
# tracker.dump('test.csv') # also, tracker can dump to file
import networkx as nx
from robust_graph import R
from robust_graph import LOG_LEVEL_INFO as LOG_LEVEL
from robust_graph import TimeTracker
# an algorithm
from robust_graph import IchinoseSatotani
# function to load US airline network
from robust_graph import load_us
# some important constants
n_attacks = 10
n_steps = 100
filename = 'Randr.csv'
graph = load_us()
# define columns in csv
columns = [
# R robustness
('R', lambda G: R(G, n_attacks)),
# r degree correlation
('r', nx.degree_pearson_correlation_coefficient)
]
# use TimeTracker to track R and r
tracker = TimeTracker(IchinoseSatotani, graph, columns,
log_level=LOG_LEVEL, force_update=False, greedy=True)
tracker.track(steps=n_steps)
tracker.dump(filename)
