networkx官方文档里的代码运行报错，求指导

龖釋

1. <blockquote>import networkx as nx

复制代码

运行时报错：

1. b = seed.choice(list(G.neighbors(a)))
   
2. AttributeError: 'NoneType' object has no attribute 'choice'

复制代码

代码是networkx官方文档里的代码，求指导怎么解决，官方文档链接https://networkx.org/documentation/latest/_modules/networkx/algorithms/smallworld.html#random_reference

龖釋

1. import networkx as nx
   
2. import numpy as np
   
3. 
   
4. def random_reference(G, niter=1, connectivity=True, seed=None):
   
5.     if G.is_directed():
   
6.         msg = "random_reference() not defined for directed graphs."
   
7.         raise nx.NetworkXError(msg)
   
8.     if len(G) < 4:
   
9.         raise nx.NetworkXError("Graph has less than four nodes.")
   
10. 
    
11.     from networkx.utils import cumulative_distribution, discrete_sequence
    
12. 
    
13.     local_conn = nx.connectivity.local_edge_connectivity
    
14. 
    
15.     G = G.copy()
    
16.     keys, degrees = zip(*G.degree())  # keys, degree
    
17.     cdf = cumulative_distribution(degrees)  # cdf of degree
    
18.     nnodes = len(G)
    
19.     nedges = nx.number_of_edges(G)
    
20.     niter = niter * nedges
    
21.     ntries = int(nnodes * nedges / (nnodes * (nnodes - 1) / 2))
    
22.     swapcount = 0
    
23. 
    
24.     for i in range(niter):
    
25.         n = 0
    
26.         while n < ntries:
    
27.             # pick two random edges without creating edge list
    
28.             # choose source node indices from discrete distribution
    
29.             (ai, ci) = discrete_sequence(2, cdistribution=cdf, seed=seed)
    
30.             if ai == ci:
    
31.                 continue  # same source, skip
    
32.             a = keys[ai]  # convert index to label
    
33.             c = keys[ci]
    
34.             # choose target uniformly from neighbors
    
35.             b = seed.choice(list(G.neighbors(a)))
    
36.             d = seed.choice(list(G.neighbors(c)))
    
37.             bi = keys.index(b)
    
38.             di = keys.index(d)
    
39.             if b in [a, c, d] or d in [a, b, c]:
    
40.                 continue  # all vertices should be different
    
41. 
    
42.             # don't create parallel edges
    
43.             if (d not in G[a]) and (b not in G[c]):
    
44.                 G.add_edge(a, d)
    
45.                 G.add_edge(c, b)
    
46.                 G.remove_edge(a, b)
    
47.                 G.remove_edge(c, d)
    
48. 
    
49.                 # Check if the graph is still connected
    
50.                 if connectivity and local_conn(G, a, b) == 0:
    
51.                     # Not connected, revert the swap
    
52.                     G.remove_edge(a, d)
    
53.                     G.remove_edge(c, b)
    
54.                     G.add_edge(a, b)
    
55.                     G.add_edge(c, d)
    
56.                 else:
    
57.                     swapcount += 1
    
58.                     break
    
59.             n += 1
    
60.     return G
    
61. 
    
62. def sigma(G, niter=100, nrand=10, seed=None):
    
63.     randMetrics = {"C": [], "L": []}
    
64.     for i in range(nrand):
    
65.         Gr = random_reference(G, niter=niter, seed=seed)
    
66.         randMetrics["C"].append(nx.transitivity(Gr))
    
67.         randMetrics["L"].append(nx.average_shortest_path_length(Gr))
    
68. 
    
69.     C = nx.transitivity(G)
    
70.     L = nx.average_shortest_path_length(G)
    
71.     Cr = np.mean(randMetrics["C"])
    
72.     Lr = np.mean(randMetrics["L"])
    
73. 
    
74.     sigma = (C / Cr) / (L / Lr)
    
75. 
    
76.     return sigma
    
77. 
    
78. if __name__ == "__main__":
    
79.     print(sigma(G))

复制代码

suchocolate

代码不全，G没有实际值，seed也没传入实际内容。
给的链接也只是码源，并没有demo，最后按官方demo来。