forked from tsinghua-fib-lab/DiskNet
-
Notifications
You must be signed in to change notification settings - Fork 0
/
simluator.py
194 lines (157 loc) · 9.26 KB
/
simluator.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
import os
import pickle
import numpy as np
import urllib.request
import zipfile
import networkx as nx
from sdeint import itoEuler
from utils import HyperbolicEmbedding
from dynamics import *
sites = {
'Drosophila': 'https://nrvis.com/download/data/bn/bn-fly-drosophila_medulla_1.zip',
'PowerGrid': 'https://nrvis.com/download/data/power/power-bcspwr10.zip',
'Social': 'https://nrvis.com/download/data/soc/fb-pages-tvshow.zip',
'Web': 'https://nrvis.com/download/data/web/web-EPA.zip',
'Airport': 'https://nrvis.com/download/data/inf/inf-openflights.zip',
}
class NetworkSimulator(object):
def __init__(self, args: dict):
self.args = args
self.G = None
self.HyperbolicG = None
def buildNetwork(self):
try:
with open(f'{self.args.data_dir}/graph.pkl', 'rb') as file:
graph = pickle.load(file)
self.G = graph.G
print(f'Load {self.args.graph_type} graph with {self.G.number_of_nodes()} nodes and {self.G.number_of_edges()} edges')
except:
os.makedirs(self.args.data_dir, exist_ok=True)
# Create graph
if self.args.graph_type == 'BA':
self.G = nx.barabasi_albert_graph(self.args.node_num, self.args.edge_num, seed=self.args.seed)
elif self.args.graph_type == 'WS':
self.G = nx.watts_strogatz_graph(self.args.node_num, self.args.ring_lattice_k, self.args.rewiring_prob, seed=self.args.seed)
elif self.args.graph_type == 'Drosophila':
self._downloadNetwork()
self.G = nx.read_edgelist(f'{self.args.data_dir}/bn/bn-fly-drosophila_medulla_1.edges', create_using=nx.DiGraph)
self.G = self.G.to_undirected()
elif self.args.graph_type == 'PowerGrid':
self._downloadNetwork()
self.G = nx.read_edgelist(f'{self.args.data_dir}/power-bcspwr10.mtx', create_using=nx.DiGraph)
self.G = self.G.to_undirected()
elif self.args.graph_type == 'Social':
self._downloadNetwork()
self.G = nx.read_edgelist(f'{self.args.data_dir}/fb-pages-tvshow.edges', create_using=nx.DiGraph)
self.G = self.G.to_undirected()
elif self.args.graph_type == 'Web':
self._downloadNetwork()
self.G = nx.read_edgelist(f'{self.args.data_dir}/web-EPA.edges', create_using=nx.DiGraph)
self.G = self.G.to_undirected()
elif self.args.graph_type == 'Airport':
self._downloadNetwork()
self.G = nx.read_edgelist(f'{self.args.data_dir}/inf-openflights.edges', create_using=nx.DiGraph)
self.G = self.G.to_undirected()
else:
raise Exception(f'Invalid graph type: {self.args.graph_type}')
# delete isolated nodes
self.G.remove_nodes_from(list(nx.isolates(self.G)))
# delete self-loop edges
self.G.remove_edges_from(nx.selfloop_edges(self.G))
# only keep the largest connected component
self.G = self.G.subgraph(max(nx.connected_components(self.G), key=len))
# relabel nodes
self.G = nx.convert_node_labels_to_integers(self.G, first_label=0)
# keep even number of nodes for static RG model
if self.G.number_of_nodes() % 2 == 1:
print(self.G.number_of_nodes())
self.G.remove_node(np.random.choice(list(self.G.nodes)))
print(self.G.number_of_nodes())
# relabel nodes
self.G = nx.convert_node_labels_to_integers(self.G, first_label=0)
with open(f'{self.args.data_dir}/graph.pkl', 'wb') as file:
pickle.dump(self, file)
nx.write_edgelist(self.G, f'{self.args.data_dir}/graph.txt', data=False)
print(f'Save {self.args.graph_type} graph with {self.G.number_of_nodes()} nodes and {self.G.number_of_edges()} edges')
return self.G, nx.to_numpy_array(self.G)
def getHyperbolicEmbedding(self):
he = HyperbolicEmbedding(args=self.args)
s1_kappa, s1_angular, h1_radius, mu, beta, radius_s1 = he.fit_transform()
exp_degree, he_degree = np.mean([d for n, d in nx.degree(self.G)]), 0.1
# while abs(exp_degree - he_degree) > 0.1:
p_matrix = self._connectivity_probability_matrix(s1_kappa, s1_angular, mu, beta, radius_s1)
sampled_A = np.random.binomial(1, p_matrix)
self.HyperbolicG = nx.from_numpy_array(sampled_A)
he_degree = np.mean([d for n, d in nx.degree(self.HyperbolicG)])
mu = mu * (exp_degree / he_degree)
nx.write_edgelist(self.HyperbolicG, f'{self.args.log_dir}/HE/hyperbolic_graph.txt', data=False)
print(f'clustering coeficient: {nx.average_clustering(self.G):.3f}-->{nx.average_clustering(self.HyperbolicG):.3f}')
print(f'degree: {np.mean([d for n, d in nx.degree(self.G)]):.3f}-->{np.mean([d for n, d in nx.degree(self.HyperbolicG)]):.3f}')
return s1_kappa, s1_angular, h1_radius, mu, beta, radius_s1
def getSimTraj(self):
try:
with np.load(f'{self.args.data_dir}/{self.args.dynamics}/dynamics.npz') as data:
X = data['X']
print(f'Load {self.args.dynamics} dynamics with {self.args.node_num} nodes and {self.args[self.args.dynamics].total_t} time steps')
except:
dim = self.args[self.args.dynamics].dim
if self.args.dynamics == 'HindmarshRose':
sde = HindmarshRose(args=self.args, A=nx.to_numpy_array(self.G))
x0_1 = np.random.uniform(-1, 0, size=self.args.node_num)
x0_2 = np.random.uniform(-5, 0, size=self.args.node_num)
x0_3 = np.random.uniform(3, 3.5, size=self.args.node_num)
x0 = np.concatenate((x0_1, x0_2, x0_3))
elif self.args.dynamics == 'FitzHughNagumo':
sde = FitzHughNagumo(args=self.args, A=nx.to_numpy_array(self.G))
x0 = np.random.uniform(-1, 1, size=self.args.node_num*dim)
elif self.args.dynamics == 'CoupledRossler':
sde = CoupledRossler(args=self.args, A=nx.to_numpy_array(self.G))
x0 = np.random.uniform(-0.05, 0.05, size=self.args.node_num*dim)
tspan = np.arange(0, self.args[self.args.dynamics].total_t, self.args[self.args.dynamics].sim_dt)
sol = itoEuler(sde.f, sde.g, x0, tspan) # (total_t, node_num*feature_dim)
# downsample
ratio = int(self.args[self.args.dynamics].dt / self.args[self.args.dynamics].sim_dt)
sol = sol[::ratio]
X = np.zeros((sol.shape[0], self.args.node_num, dim))
for i in range(dim):
X[:, :, i] = sol[:, i*self.args.node_num:(i+1)*self.args.node_num]
if self.args.dynamics == 'CoupledKuramoto':
X = np.sin(X)
os.makedirs(f'{self.args.data_dir}/{self.args.dynamics}', exist_ok=True)
np.savez(f'{self.args.data_dir}/{self.args.dynamics}/dynamics.npz', X=X)
print(f'Save {self.args.dynamics} dynamics with {self.args.node_num} nodes and {self.args[self.args.dynamics].total_t} time steps')
return X
def _downloadNetwork(self):
url = sites[self.args.graph_type]
file_name = f"{self.args.data_dir}/download.zip"
urllib.request.urlretrieve(url, file_name)
with zipfile.ZipFile(file_name, "r") as zip_ref:
zip_ref.extractall(self.args.data_dir)
os.remove(file_name)
if self.args.graph_type == 'PowerGrid':
# delete header
with open(f'{self.args.data_dir}/power-bcspwr10.mtx', 'r') as f:
lines = f.readlines()
with open(f'{self.args.data_dir}/power-bcspwr10.mtx', 'w') as f:
f.writelines(lines[14:])
elif self.args.graph_type == 'Social':
# replace ',' with ' '
with open(f'{self.args.data_dir}/fb-pages-tvshow.edges', 'r') as f:
lines = f.readlines()
with open(f'{self.args.data_dir}/fb-pages-tvshow.edges', 'w') as f:
for line in lines:
f.write(line.replace(',', ' '))
elif self.args.graph_type == 'Airport':
# delete header
with open(f'{self.args.data_dir}/inf-openflights.edges', 'r') as f:
lines = f.readlines()
with open(f'{self.args.data_dir}/inf-openflights.edges', 'w') as f:
f.writelines(lines[2:])
def _connectivity_probability_matrix(self, kappa, angular, mu, beta, radius):
"""p_ij = 1 / (1 + (radius*(delta_angular) / (mu*kappa_i*kappa_j))**beta)"""
abs_delta_angular = np.abs(angular.reshape(-1, 1) - angular.reshape(1, -1))
delta_angular = np.minimum(abs_delta_angular, 2 * np.pi - abs_delta_angular)
kappa_mul = kappa.reshape(-1, 1) * kappa.reshape(1, -1)
rescaled_dist = radius * delta_angular / (mu * kappa_mul)
p_matrix = 1 / (1 + rescaled_dist ** beta) - np.eye(self.args.node_num)
return p_matrix