Python笔记——nextworkx有向图

环境：ubuntu16.04，python2.7.12

版本：networkx==2.2, matplotlib==1.5.1, numpy==1.11.0

官方文档：https://networkx.github.io/documentation/networkx-2.2/

参考：https://www.cnblogs.com/minglex/p/9205160.html

创建图

import networkx as nx
graph = nx.DiGraph()

---

节点

添加节点：

graph.add_node(1)
graph.add_node(2, value=float('inf'))

查看节点属性：

graph.node[2]

返回{'value': inf}，为dict类型。

修改/添加节点属性：

graph.node[2]['value'] = 7

遍历节点： 

graph.node(data=True)
graph.nodes(data=True)

结果均为：

[(1, {}), (2, {'value': 7})]

---

边

添加边：

graph.add_edge(1, 2, weight=9.0)

查看边属性：

graph[1][2]['weight']

修改/添加节点属性：

graph[1][2]['length'] = 10

遍历边：

graph.edges(data=True)

结果为

[(1, 2, {'length': 10, 'weight': 9.0})]

---

可视化

若指定节点位置，需首先给出位置dict：

keys = [node for index, node in enumerate(graph.nodes)]
positions = np.array([(1,1), (2,2)])
pos = dict(zip(keys, positions))

其次：

edge_width = 1.0

nx.draw_networkx_nodes(graph, pos)
nx.draw_networkx_labels(graph, pos)
nx.draw_networkx_edges(graph, pos, width=edge_width, arrowstyle='->')

plt.show()

效果：

• 不指定节点位置：

pos = nx.spring_layout(graph)

其中，spring_layout可替换。

• 根据边权重改变边的粗细：

edge_width = [float(d['weight']*1.0) for (u,v,d) in graph.edges(data=True)]

• 显示边的属性：

edge_labels = dict([((u,v,),d['weight']) for u,v,d in graph.edges(data=True)])
nx.draw_networkx_edge_labels(graph, pos, edge_labels=edge_labels)

效果：

---

附（源码）

import networkx as nx
import matplotlib.pyplot as plt
import numpy as np


graph = nx.DiGraph()

graph.add_node(1)
graph.add_node(2, value=float('inf'))
graph.node[2]['value'] = 7

graph.add_edge(1, 2, weight=9.0)
graph[1][2]['length'] = 10


keys = [node for index, node in enumerate(graph.nodes)]
positions = np.array([(1,1), (2,2)])
pos = dict(zip(keys, positions))

edge_width = 1.0

nx.draw_networkx_nodes(graph, pos)
nx.draw_networkx_labels(graph, pos)
nx.draw_networkx_edges(graph, pos, width=edge_width, arrowstyle='->')

edge_labels = dict([((u,v,),d['weight']) for u,v,d in graph.edges(data=True)])
nx.draw_networkx_edge_labels(graph, pos, edge_labels=edge_labels)

plt.show()