def test_generate_and_parse_sparse6(self):
for i in list(range(13)) + [31, 47, 62, 63, 64, 72]:
m = min(2 * i, i * i // 2)
g = nx.random_graphs.gnm_random_graph(i, m, seed=i)
gstr = nx.generate_sparse6(g, header=False)
g2 = nx.parse_sparse6(gstr)
assert_equal(g2.order(), g.order())
assert_equal(sorted(g2.edges()), sorted(g.edges()))
def test_generate_and_parse_sparse6(self):
for i in list(range(13)) + [31, 47, 62, 63, 64, 72]:
m = min(2 * i, i * i // 2)
g = nx.random_graphs.gnm_random_graph(i, m, seed=i)
gstr = nx.generate_sparse6(g, header=False)
g2 = nx.parse_sparse6(gstr)
assert_equal(g2.order(), g.order())
assert_equal(sorted(g2.edges()), sorted(g.edges()))
def directed_error(self):
nx.generate_sparse6(nx.DiGraph())
def test_generate_sparse6(self):
# Checked against sage encoder
assert_equal(nx.generate_sparse6(nx.empty_graph(0)), '>>sparse6<<:?')
assert_equal(nx.generate_sparse6(nx.empty_graph(1)), '>>sparse6<<:@')
assert_equal(nx.generate_sparse6(nx.empty_graph(5)), '>>sparse6<<:D')
assert_equal(nx.generate_sparse6(nx.empty_graph(68)),
'>>sparse6<<:~?@C')
assert_equal(nx.generate_sparse6(nx.empty_graph(258049)),
'>>sparse6<<:~~???~?@')
G1 = nx.complete_graph(4)
assert_equal(nx.generate_sparse6(G1, header=True), '>>sparse6<<:CcKI')
assert_equal(nx.generate_sparse6(G1, header=False), ':CcKI')
# Padding testing
assert_equal(nx.generate_sparse6(nx.path_graph(4), header=False),
':Cdv')
assert_equal(nx.generate_sparse6(nx.path_graph(5), header=False),
':DaYn')
assert_equal(nx.generate_sparse6(nx.path_graph(6), header=False),
':EaYnN')
assert_equal(nx.generate_sparse6(nx.path_graph(7), header=False),
':FaYnL')
assert_equal(nx.generate_sparse6(nx.path_graph(8), header=False),
':GaYnLz')
def test_generate_sparse6(self):
# Checked against sage encoder
assert_equal(nx.generate_sparse6(nx.empty_graph(0)), '>>sparse6<<:?')
assert_equal(nx.generate_sparse6(nx.empty_graph(1)), '>>sparse6<<:@')
assert_equal(nx.generate_sparse6(nx.empty_graph(5)), '>>sparse6<<:D')
assert_equal(nx.generate_sparse6(nx.empty_graph(68)),
'>>sparse6<<:~?@C')
assert_equal(nx.generate_sparse6(nx.empty_graph(258049)),
'>>sparse6<<:~~???~?@')
G1 = nx.complete_graph(4)
assert_equal(nx.generate_sparse6(G1, header=True),
'>>sparse6<<:CcKI')
assert_equal(nx.generate_sparse6(G1, header=False), ':CcKI')
# Padding testing
assert_equal(nx.generate_sparse6(nx.path_graph(4), header=False),
':Cdv')
assert_equal(nx.generate_sparse6(nx.path_graph(5), header=False),
':DaYn')
assert_equal(nx.generate_sparse6(nx.path_graph(6), header=False),
':EaYnN')
assert_equal(nx.generate_sparse6(nx.path_graph(7), header=False),
':FaYnL')
assert_equal(nx.generate_sparse6(nx.path_graph(8), header=False),
':GaYnLz')
def directed_raises(self):
nx.generate_sparse6(nx.DiGraph())
def toyBatch(tg,
td,
outfile,
irts=Irts({}),
fitinfo=Fitinfo({}),
start_seed=0,
methods=[
'rw', 'fe', 'uinvite', 'uinvite_irt', 'uinvite_prior',
'uinvite_irt_prior'
],
header=1,
debug=False):
np.random.seed(start_seed)
# break out of function if using unknown method
for method in methods:
if method not in [
'rw', 'fe', 'uinvite', 'uinvite_irt', 'uinvite_prior',
'uinvite_irt_prior', 'goni', 'goni_valid', 'threshold',
'threshold_valid'
]:
raise ValueError(
'ERROR: Trying to fit graph with unknown method: ', method)
# flag if using a prior method
if ('uinvite_prior' in methods) or ('uinvite_irt_prior' in methods):
use_prior = 1
else:
use_prior = 0
# flag if using an IRT method
if ('uinvite_irt' in methods) or ('uinvite_irt_prior' in methods):
use_irt = 1
else:
use_irt = 0
if use_prior:
prior = genSWPrior(tg, fitinfo.prior_samplesize)
# stuff to write to file
globalvals = ['numedges', 'graph_seed', 'x_seed', 'truegraph',
'll_tg'] # same across all methods, updates with each seed
methodvals = [
'method', 'cost', 'll', 'time', 'bestgraph', 'hit', 'miss', 'fa', 'cr'
] # differ per method
f = open(outfile, 'a', 0) # write/append to file with no buffering
if header == 1:
objs = [tg, td, irts, fitinfo]
header_towrite = [
str(i) for var in objs for i in var.keys() if i != 'data'
] + globalvals + methodvals
f.write(','.join(header_towrite) + '\n')
# write all parameters to file except irts.data (too long for csv file! if they're toy irts, they can be re-generated from seed)
objs = [tg, td, irts, fitinfo]
params_towrite = [
str(var[i]) for var in objs for i in var.keys() if i != 'data'
]
# how many graphs to run?
seed_param = start_seed
last_seed = start_seed + tg.numgraphs
while seed_param < last_seed:
# generate toy graph and data
# give up if it's hard to generate data that cover full graph
# generate toy data
graph_seed = seed_param
g, a = genG(tg, seed=graph_seed)
# ugly code -- who writes this shit? oh wait, it's me
tries = 0
while True:
x_seed = seed_param
[Xs, irts.data, alter_graph] = genX(g, td, seed=x_seed)
# generate IRTs if using IRT model
if use_irt: irts.data = stepsToIRT(irts, seed=x_seed)
if alter_graph == 0: # only use data that covers the entire graph
break
else:
tries = tries + 1
seed_param = seed_param + 1
last_seed = last_seed + 1 # if data is unusable (doesn't cover whole graph), add another seed
if tries >= 1000:
raise ValueError(
"Data doesn't cover full graph... Increase 'trim' or 'numx' (or change graph)"
)
numedges = nx.number_of_edges(g)
truegraph = nx.generate_sparse6(g, header=False) # to write to file
for method in methods:
recordname = "record_" + str(graph_seed) + "_" + str(
x_seed) + "_" + method + ".csv"
if debug: print "SEED: ", seed_param, "method: ", method
# Find best graph! (and log time)
ll_tg = "" # only record TG LL for U-INVITE models; otherwise it's ambiguous whether it's using IRT/prior/etc
starttime = datetime.now()
if method == 'uinvite':
bestgraph, ll = uinvite(Xs,
td,
tg.numnodes,
debug=debug,
fitinfo=fitinfo,
recordname=recordname)
ll_tg = probX(Xs, a, td)[0]
if method == 'uinvite_prior':
bestgraph, ll = uinvite(Xs,
td,
tg.numnodes,
prior=prior,
debug=debug,
fitinfo=fitinfo,
recordname=recordname)
ll_tg = probX(Xs, a, td, prior=prior)[0]
if method == 'uinvite_irt':
bestgraph, ll = uinvite(Xs,
td,
tg.numnodes,
irts=irts,
debug=debug,
fitinfo=fitinfo,
recordname=recordname)
ll_tg = probX(Xs, a, td, irts=irts)[0]
if method == 'uinvite_irt_prior':
bestgraph, ll = uinvite(Xs,
td,
tg.numnodes,
irts=irts,
prior=prior,
debug=debug,
fitinfo=fitinfo,
recordname=recordname)
ll_tg = probX(Xs, a, td, prior=prior, irts=irts)[0]
if method == 'goni':
bestgraph = goni(Xs, tg.numnodes, fitinfo=fitinfo)
ll = probX(Xs, bestgraph, td)[0]
if method == 'goni_valid':
bestgraph = goni(Xs,
tg.numnodes,
td=td,
valid=1,
fitinfo=fitinfo)
ll = probX(Xs, bestgraph, td)[0]
if method == 'threshold':
bestgraph = goni(Xs, tg.numnodes, fitinfo=fitinfo, c=1)
ll = probX(Xs, bestgraph, td)[0]
if method == 'threshold_valid':
bestgraph = goni(Xs,
tg.numnodes,
td=td,
valid=1,
fitinfo=fitinfo,
c=1)
ll = probX(Xs, bestgraph, td)[0]
if method == 'rw':
bestgraph = noHidden(Xs, tg.numnodes)
ll = probX(Xs, bestgraph, td)[0]
if method == 'fe':
bestgraph = firstEdge(Xs, tg.numnodes)
ll = probX(Xs, bestgraph, td)[0]
elapsedtime = str(datetime.now() - starttime)
if debug:
print elapsedtime
print "COST: ", cost(bestgraph, a)
print nx.generate_sparse6(nx.to_networkx_graph(bestgraph),
header=False)
# Record cost, time elapsed, LL of best graph, hash of best graph, and SDT measures
graphcost = cost(bestgraph, a)
hit, miss, fa, cr = costSDT(bestgraph, a)
graphhash = nx.generate_sparse6(nx.to_networkx_graph(bestgraph),
header=False)
global_towrite = [
str(i)
for i in [numedges, graph_seed, x_seed, truegraph, ll_tg]
]
method_towrite = [
str(i) for i in [
method, graphcost, ll, elapsedtime, graphhash, hit, miss,
fa, cr
]
]
# log stuff here
f.write(','.join(params_towrite) + ',')
f.write(','.join(global_towrite) + ',')
f.write(','.join(method_towrite) + '\n')
seed_param = seed_param + 1
f.close()
def directed_raises(self):
nx.generate_sparse6(nx.DiGraph())
def directed_error(self):
nx.generate_sparse6(nx.DiGraph())
