def test_degree(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=5)
g.add_path([4, 5, 6, 7, 8], t=6)
ng = g.degree(4)
self.assertEqual(ng, 2)
ng = g.degree(4, 5)
self.assertEqual(ng, 1)
ng = g.degree(4, 6)
self.assertEqual(ng, 1)
ng = g.degree(4, 0)
self.assertEqual(ng, 0)
def test_stream_interactions(self):
g = dn.DynGraph()
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
g.add_interactions_from([(1, 3), (1, 5)], t=2, e=3)
sres = list(g.stream_interactions())
cres = [(1, 2, '+', 2), (1, 3, '+', 2), (1, 5, '+', 2), (1, 3, '-', 3),
(1, 5, '-', 3), (1, 2, '-', 6), (1, 2, '+', 7),
(1, 2, '-', 15), (1, 2, '+', 18)]
self.assertEquals(sorted(sres), sorted(cres))
def test_number_of_nodes(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=5)
g.add_path([4, 5, 6, 7, 8], t=6)
nn = g.number_of_nodes()
self.assertEqual(nn, 9)
nn = g.number_of_nodes(t=5)
self.assertEqual(nn, 5)
nn = g.number_of_nodes(t=0)
self.assertEqual(nn, 0)
avg = g.avg_number_of_nodes()
self.assertEqual(avg, 5)
def test_neighbores(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=5)
g.add_path([4, 5, 6, 7, 8], t=6)
ng = len(g.neighbors(0))
self.assertEqual(ng, 1)
ng = len(g.neighbors(0, 5))
self.assertEqual(ng, 1)
ng = len(g.neighbors(0, 6))
self.assertEqual(ng, 0)
ng = len(g.neighbors(0, 0))
self.assertEqual(ng, 0)
def test_inter_event_time(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=2)
g.add_path([4, 5, 6, 7, 8], t=3)
ivt = g.inter_event_time_distribution()
self.assertDictEqual(ivt, {0: 6, 1: 1})
ivt = g.inter_event_time_distribution(4)
self.assertDictEqual(ivt, {1: 1})
ivt = g.inter_event_time_distribution(0)
self.assertDictEqual(ivt, {})
ivt = g.inter_event_time_distribution(0, 1)
self.assertDictEqual(ivt, {})
def test_snapshots_interactions(self):
g = dn.DynGraph()
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
dn.write_snapshots(g, "test.txt", delimiter=" ")
h = dn.read_snapshots("test.txt", nodetype=int, timestamptype=int)
self.assertEqual(g.number_of_interactions(), h.number_of_interactions())
self.assertEqual(list(g.stream_interactions()), list(h.stream_interactions()))
dn.write_interactions(h, "test.txt", delimiter=" ")
h = dn.read_interactions("test.txt", nodetype=int, timestamptype=int, keys=True)
dn.write_snapshots(h, "test.txt", delimiter=" ")
os.remove("test.txt")
def test_interaction_graph_directed(self):
g = dn.DynGraph()
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
dn.write_interactions(g, "test2.txt", delimiter=" ")
h = dn.read_interactions("test2.txt",
directed=True,
nodetype=int,
timestamptype=int)
self.assertEqual(list(g.stream_interactions()),
list(h.stream_interactions()))
self.assertEqual(g.number_of_interactions(),
h.number_of_interactions())
os.remove("test2.txt")
def test_accumulative_growth(self):
g = dn.DynGraph(edge_removal=False)
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
g.add_interactions_from([(1, 3), (1, 5)], t=2, e=3)
sres = list(g.stream_interactions())
cres = [(1, 2, '+', 2), (1, 5, '+', 2), (1, 3, '+', 2)]
self.assertEqual(sorted(sres), sorted(cres))
self.assertEqual(g.has_interaction(1, 2, 18), True)
self.assertEqual(g.has_interaction(1, 2, 40), False)
try:
g.add_interaction(2, 1, 7)
except:
pass
def test_DynSI(self):
dg = dn.DynGraph()
for t in past.builtins.xrange(0, 3):
g = nx.erdos_renyi_graph(200, 0.05)
dg.add_interactions_from(g.edges(), t)
model = si.DynSIModel(dg)
config = mc.Configuration()
config.add_model_parameter('beta', 0.1)
config.add_model_parameter("percentage_infected", 0.1)
model.set_initial_status(config)
iterations = model.execute_snapshots()
self.assertEqual(len(iterations), 3)
iterations = model.execute_iterations()
trends = model.build_trends(iterations)
self.assertEqual(len(trends[0]['trends']['status_delta'][1]),
len([x for x in dg.stream_interactions() if x[2] == "+"]))
def test_visualize_dynamic(self):
dg = dn.DynGraph()
for t in past.builtins.xrange(0, 4):
g = nx.erdos_renyi_graph(200, 0.05)
dg.add_interactions_from(g.edges(), t)
model = dsi.DynSIModel(dg)
config = mc.Configuration()
config.add_model_parameter('beta', 0.1)
config.add_model_parameter("percentage_infected", 0.1)
model.set_initial_status(config)
iterations = model.execute_snapshots()
trends = model.build_trends(iterations)
# Visualization
viz = DiffusionPrevalence(model, trends)
viz.plot("prevd.pdf")
os.remove("prevd.pdf")
def test_number_of_interactions(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=5)
g.add_path([4, 5, 6, 7, 8], t=6)
its = g.number_of_interactions()
self.assertEqual(its, 8)
its = g.number_of_interactions(0)
self.assertEqual(its, None)
its = g.number_of_interactions(0, 1)
self.assertEqual(its, 1)
its = g.number_of_interactions(0, 1, 5)
self.assertEqual(its, 1)
its = g.number_of_interactions(0, 1, 6)
self.assertEqual(its, 0)
def test_nodes(self):
g = dn.DynGraph()
g.add_star([0, 1, 2, 3, 4], t=5)
nds = len(g.nodes())
self.assertEqual(nds, 5)
g.add_star([5, 1, 2, 3, 4], t=6)
nds = len(g.nodes())
self.assertEqual(nds, 6)
nds = len(g.nodes(t=6))
self.assertEqual(nds, 5)
nds = len(g.nodes(t=9))
self.assertEqual(nds, 0)
self.assertEqual(g.has_node(0), True)
self.assertEqual(g.has_node(0, 5), True)
self.assertEqual(g.has_node(0, 6), False)
self.assertEqual(g.has_node(0, 0), False)
def test_DynProfile(self):
dg = dn.DynGraph()
for t in past.builtins.xrange(0, 3):
g = nx.erdos_renyi_graph(200, 0.05)
dg.add_interactions_from(g.edges(), t)
model = pro.DynProfileModel(dg)
config = mc.Configuration()
config.add_model_parameter("percentage_infected", 0.1)
config.add_model_parameter("blocked", 0.1)
config.add_model_parameter("adopter_rate", 0.001)
profile = 0.1
for i in g.nodes():
config.add_node_configuration("profile", i, profile)
model.set_initial_status(config)
model.set_initial_status(config)
iterations = model.execute_snapshots()
self.assertEqual(len(iterations), 3)
def test_time_slice(self):
g = dn.DynGraph()
g.add_path([0, 1, 2, 3, 4], t=5)
g.add_path([4, 5, 6, 7, 8], t=6)
h = g.time_slice(5)
self.assertIsInstance(h, dn.DynGraph)
self.assertEqual(h.number_of_nodes(), 5)
self.assertEqual(h.number_of_interactions(), 4)
h = g.time_slice(5, 5)
self.assertIsInstance(h, dn.DynGraph)
self.assertEqual(h.number_of_nodes(), 5)
self.assertEqual(h.number_of_interactions(), 4)
h = g.time_slice(5, 6)
self.assertIsInstance(h, dn.DynGraph)
self.assertEqual(h.number_of_nodes(), 9)
self.assertEqual(h.number_of_interactions(), 8)
h = g.time_slice(0)
self.assertIsInstance(h, dn.DynGraph)
self.assertEqual(h.number_of_nodes(), 0)
self.assertEqual(h.number_of_interactions(), 0)
def test_dyngraph_add_interaction(self):
g = dn.DynGraph()
self.assertIsInstance(g, dn.DynGraph)
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
its = g.interactions()
self.assertEqual(len(its), 1)
g.add_interactions_from([(1, 3), (1, 5)], t=2)
its = g.interactions()
self.assertEqual(len(its), 3)
its = g.interactions(t=18)
self.assertEqual(len(its), 1)
its = g.interactions(t=20)
self.assertEqual(len(its), 0)
self.assertEqual(len(list(g.neighbors_iter(1))), 3)
self.assertEqual(len(list(g.neighbors_iter(1, 7))), 1)
self.assertEqual(len(list(g.neighbors_iter(1, 0))), 0)
self.assertEqual(g.order(), len(g.nodes()))
self.assertEqual(g.has_node(42), False)
self.assertEqual(g.has_node(42, 3), False)
g.add_cycle([3, 4, 5, 6], t=34)
try:
g.time_slice(2, 1)
except:
pass
g.interactions_iter([1, 2])
try:
g.add_interaction(1, 5)
except:
pass
try:
g.add_interactions_from([(1, 4), (3, 6)])
except:
pass
try:
g.remove_edge(1, 2)
except:
pass
try:
g.remove_edges_from([(1, 2)])
except:
pass
try:
g.remove_node(1)
except:
pass
try:
g.remove_nodes_from([1, 2])
except:
pass
self.assertEqual(g.number_of_interactions(1, 90), 0)
def test_functions(self):
g = dn.DynGraph()
g.add_interaction(1, 2, 2)
g.add_interaction(1, 2, 2, e=6)
g.add_interaction(1, 2, 7, e=11)
g.add_interaction(1, 2, 8, e=15)
g.add_interaction(1, 2, 18)
g.add_interaction(1, 2, 19)
self.assertEqual(len(dn.nodes(g)), 2)
self.assertEqual(len(dn.nodes(g, t=0)), 0)
self.assertEqual(len(dn.nodes(g, t=2)), 2)
self.assertEqual(len(dn.interactions(g)), 1)
self.assertEqual(len(dn.interactions(g, t=0)), 0)
self.assertEqual(len(dn.interactions(g, t=2)), 1)
self.assertDictEqual(dn.degree(g), {1: 1, 2: 1})
self.assertDictEqual(dn.degree(g, t=0), {1: 0, 2: 0})
self.assertDictEqual(dn.degree(g, [1, 2]), {1: 1, 2: 1})
self.assertEqual(len(dn.neighbors(g, 1)), 1)
self.assertEqual(len(dn.neighbors(g, 1, t=2)), 1)
self.assertEqual(len(dn.neighbors(g, 1, t=0)), 0)
self.assertEqual(dn.number_of_nodes(g), 2)
self.assertEqual(dn.number_of_nodes(g, t=0), 0)
self.assertEqual(dn.number_of_nodes(g, t=2), 2)
self.assertEqual(dn.number_of_interactions(g, t=0), 0)
self.assertEqual(dn.number_of_interactions(g, t=2), 1)
self.assertEqual(dn.number_of_interactions(g), 1)
self.assertEqual(dn.density(g), 1.0)
self.assertEqual(dn.density(g, t=0), 0)
self.assertEqual(dn.density(g, t=2), 0)
self.assertEqual(dn.degree_histogram(g), [0, 2])
self.assertEqual(dn.degree_histogram(g, t=0), [2])
self.assertEqual(dn.degree_histogram(g, t=2), [0, 2])
self.assertEqual(dn.is_directed(g), False)
dn.add_cycle(g, [1, 2, 3, 4], t=30)
dn.add_path(g, [4, 6, 7, 8], t=40)
dn.add_star(g, [1, 2, 3, 4], t=50)
dn.subgraph(g, [1, 3, 4])
dn.create_empty_copy(g)
self.assertEqual(len(dn.all_neighbors(g, 1, t=0)), 0)
self.assertEqual(len(dn.all_neighbors(g, 1)), 3)
self.assertEqual(len(dn.all_neighbors(g, 1, t=2)), 1)
self.assertEqual(len(list(dn.non_neighbors(g, 1, t=0))), 6)
self.assertEqual(len(list(dn.non_neighbors(g, 1))), 3)
self.assertEqual(len(list(dn.non_neighbors(g, 1, t=2))), 5)
dn.non_interactions(g, 2)
dn.non_interactions(g, 0)
dn.non_interactions(g)
self.assertEqual(dn.is_empty(g), False)
h = dn.DynGraph()
self.assertEqual(dn.is_empty(h), True)
h = dn.time_slice(g, 2, 4)
self.assertEqual(len(h.nodes()), 2)
dn.stream_interactions(g)
dn.temporal_snapshots_ids(g)
dn.interactions_per_snapshots(g, 0)
dn.interactions_per_snapshots(g, 2)
dn.interactions_per_snapshots(g)
dn.inter_event_time_distribution(g)
dn.inter_event_time_distribution(g, 1)
dn.inter_event_time_distribution(g, 1, 2)
dn.set_node_attributes(g,
values={n: 0
for n in g.nodes()},
name="test")
try:
dn.set_node_attributes(g, values={90000: 0}, name="test")
except:
pass
try:
dn.set_node_attributes(g, "test1", name="test")
except:
pass
dn.set_node_attributes(g, values={n: {"a": 3} for n in g.nodes()})
try:
dn.set_node_attributes(g, values={9000: {"a": 3}})
except:
pass
dn.get_node_attributes(g, name="test")
try:
dn.set_edge_attributes(g, 3, "dog")
except:
pass
try:
dn.get_edge_attributes(g, "test")
except:
pass
dn.freeze(g)
self.assertEqual(dn.is_frozen(g), True)
h = g.to_directed()
dn.all_neighbors(h, 1)
self.assertEqual(len(list(dn.non_interactions(g))), 13)
import networkx as nx
import numpy as np
import matplotlib
import dynetx as dn
import matplotlib.pyplot as plt
plt.close("all")
'''
plt.figure()
G= nx.Graph()
G.add_edge('a','b',timestamp='t1')
G.add_node('c',timestamp='t1')
G.add_edge('a','b',timestamp='t2')
G.add_edge('a','c',timestamp='t2')
G.add_edge('b','c',timestamp='t2')
'''
plt.figure()
G = dn.DynGraph(edge_removal=True)
G.add_path([0, 1, 2], 0)
dn.nodes(G, t=0)
G.add_edge(2, 3, t=1)
dn.interactions(G, t=0)
#G.add_interaction(u=1, v=2, t=0, e=3)
nx.draw(G)
plt.show()
'node_color': 'cyan',
'node_size': 1,
'line_color': 'red',
'linewidths': 0,
'width': 0.1,
}
nx.draw_circular(G, **options)
plt.show()
G=nx.read_edgelist("/Users/jennie/Documents/analytics/xanax/xanshort.csv", delimiter=",")
import networkx as nx
import dynetx as dn
import matplotlib.pyplot as plt
g = dn.DynGraph(edge_removal=False)
g = dn.read_snapshots("/Users/jennie/Documents/analytics/xanax/lyrics_ngrams.txt", timestamptype=int)
pos=nx.spring_layout(g,scale=20) # double distance between all nodes
plt.figure(3,figsize=(10,10))
nx.draw(g,pos,node_size=10, font_size=8, with_labels=True)
list(g.stream_interactions())
import networkx as nx
import ndlib.models.ModelConfig as mc
import ndlib.models.opinions.AlgorithmicBiasModel as ab
# Network topology
def test_dyn_edge_stochastic(self):
# Fixed Threshold
dg = dn.DynGraph()
for t in past.builtins.xrange(0, 3):
g = nx.erdos_renyi_graph(200, 0.05)
dg.add_interactions_from(g.edges(), t)
model = gc.DynamicCompositeModel(dg)
model.add_status("Susceptible")
model.add_status("Infected")
c1 = cpm.EdgeStochastic(0.1, triggering_status="Infected")
model.add_rule("Susceptible", "Infected", c1)
config = mc.Configuration()
config.add_model_parameter('percentage_infected', 0.1)
model.set_initial_status(config)
iterations = model.execute_snapshots()
self.assertEqual(len(iterations), 3)
iterations = model.execute_iterations()
trends = model.build_trends(iterations)
self.assertEqual(
len(trends[0]['trends']['status_delta'][1]),
len([x for x in dg.stream_interactions() if x[2] == "+"]))
# Ad-hoc Threshold
model = gc.DynamicCompositeModel(dg)
model.add_status("Susceptible")
model.add_status("Infected")
c1 = cpm.EdgeStochastic(triggering_status="Infected")
model.add_rule("Susceptible", "Infected", c1)
config = mc.Configuration()
for e in g.edges():
config.add_edge_configuration("threshold", e,
np.random.random_sample())
config.add_model_parameter('percentage_infected', 0.1)
model.set_initial_status(config)
iterations = model.execute_snapshots()
self.assertEqual(len(iterations), 3)
iterations = model.execute_iterations()
trends = model.build_trends(iterations)
self.assertEqual(
len(trends[0]['trends']['status_delta'][1]),
len([x for x in dg.stream_interactions() if x[2] == "+"]))
# Predefined threshold 1/N
model = gc.DynamicCompositeModel(dg)
model.add_status("Susceptible")
model.add_status("Infected")
c1 = cpm.EdgeStochastic(triggering_status="Infected")
model.add_rule("Susceptible", "Infected", c1)
config = mc.Configuration()
config.add_model_parameter('percentage_infected', 0.1)
model.set_initial_status(config)
iterations = model.execute_snapshots()
self.assertEqual(len(iterations), 3)
iterations = model.execute_iterations()
trends = model.build_trends(iterations)
self.assertEqual(
len(trends[0]['trends']['status_delta'][1]),
len([x for x in dg.stream_interactions() if x[2] == "+"]))
def RDyn(
size: int = 300,
iterations: int = 5,
avg_deg: int = 15,
sigma: float = 0.6,
lambdad: float = 1,
alpha: float = 2.5,
paction: float = 1,
prenewal: float = 0.8,
quality_threshold: float = 0.5,
new_node: float = 0.0,
del_node: float = 0.0,
max_evts: int = 1,
simplified: bool = True,
) -> [dn.DynGraph, object]:
"""
RDyn is a syntetic dynamic network generator with time-dependent ground-truth partitions having tunable quality (in terms of conductance).
Communities' ids are aligned across time and a predefined number of merge/plit events are planted in between consecutive stable iterations.
:param size: Number of nodes
:param iterations: Number of stable iterations
:param avg_deg: Average node degree. Int, default 15
:param sigma: Percentage of node's edges within a community. Float, default .6
:param lambdad: Community size distribution exponent. Float, default 1
:param alpha: Degree distribution exponent. Float, default 2.5
:param paction: Probability of node action. Float, default 1
:param prenewal: Probability of edge renewal. Float, default, .8
:param quality_threshold: Conductance quality threshold for stable iteration. Float, default .5
:param new_node: Probability of node appearance. Float, default 0
:param del_node: Probability of node vanishing. Float, default 0
:param max_evts: Max number of community events for stable iteration. Int, default 1
:param simplified: Simplified execution. Boolean, default True. (NB: when True an approximation of the original process is executed - some network characteristics can deviate from the expected ones)
:return: A dynetx DynGraph, the TemporalClustering object
:Example:
>>> from cdlib.benchmark import RDyn
>>> G, coms = RDyn(n=300)
:References:
Rossetti, Giulio. "RDyn: graph benchmark handling community dynamics." Journal of Complex Networks 5.6 (2017): 893-912.
.. note:: Reference implementation: https://github.com/GiulioRossetti/RDyn
"""
tc = TemporalClustering()
dg = dn.DynGraph()
rdb = rdyn.RDynV2(
size=size,
iterations=iterations,
avg_deg=avg_deg,
quality_threshold=quality_threshold,
sigma=sigma,
lambdad=lambdad,
alpha=alpha,
paction=paction,
prenewal=prenewal,
new_node=new_node,
del_node=del_node,
max_evts=max_evts,
)
t = 0
for g, communities in rdb.execute(simplified=simplified):
dg.add_interactions_from(g.edges(), t=t, e=t)
sg = dg.time_slice(t)
nc = NamedClustering(
communities,
sg,
"RDyn",
{
"size": size,
"iterations": iterations,
"avg_deg": avg_deg,
"quality_threshold": quality_threshold,
"sigma": sigma,
"lambdad": lambdad,
"alpha": alpha,
"paction": paction,
"prenewal": prenewal,
"new_node": new_node,
"del_node": del_node,
"max_evts": max_evts,
"simplified": simplified,
},
overlap=False,
)
tc.add_clustering(nc, time=t)
t += 1
return dg, tc