Draw a graph from networkx centered on a basemap position

Question

I'm searching to plot multiple sub-graphes on a map, each sub-graph would be centered in one geographical position (or one coordinate of a plot). Nodes don't have position by themselves (or they all belong to a city), but each subgraph corresponds to a local situation.

• I tried to assign a position to only one node per subgraph, leaving the remaining of the graph to be plotted by defaults using "centered" option

• I tried to inspire from https://stackoverflow.com/a/29597209/839119 to plot a hierarchically graph on a position but withou success

  # -*- coding: utf-8 -*-
  import networkx as nx
  import pygraphviz
  import matplotlib.pyplot as plt
  from mpl_toolkits.basemap import Basemap as Basemap
  
  G1 = nx.Graph()
  G1.add_edge('a', 'b', weight=0.6)
  G1.add_edge('a', 'c', weight=0.2)
  G1.add_edge('c', 'd', weight=0.1)
  G1.add_edge('c', 'e', weight=0.7)
  G1.add_edge('c', 'f', weight=0.9)
  G1.add_edge('a', 'd', weight=0.3)
  
  G2 = nx.Graph()
  G2.add_edge('a', 'b', weight=0.9)
  G2.add_edge('a', 'f', weight=0.5)
  G2.add_edge('c', 'd', weight=0.1)
  G2.add_edge('c', 'e', weight=0.4)
  G2.add_edge('c', 'f', weight=0.2)
  G2.add_edge('a', 'd', weight=0.1)
  
  edges = G.edges()
  weights = [G[u][v]['weight'] for u,v in edges] # liste des poids des edges
  
  fig = plt.figure(figsize=(8, 8))
  m = Basemap(projection='npstere',boundinglat=48,lon_0=270,resolution='l')
  m.etopo(scale=0.5, alpha=0.5)
  mx1,my1=m(-6.266155,53.350140) #would be long, lat coordinates of city 1
  mx2,my2=m(-21.827774, 64.128288) #would be long, lat coordinates of city 2
  
  nx.draw_networkx(G1,center=(mx1,my1),pos=nx.spring_layout(G1),node_size=200,node_color='green')
  nx.draw_networkx(G2,center=(mx2,my2),pos=nx.spring_layout(G2),node_size=200,node_color='red')
  plt.title("North Polar Stereographic Projection")
  plt.show()
  

Answer 1

By computing nodes' location separately, I can plot the networks quite nicely. You can try again using this snippet code in place of the relevant parts in yours.

# (other code above this line)
#
import numpy as np
# compute the positions here
# proper scaling (500000) is applied to the original positions ..
# .. obtained from xxx_layout() to get good spreading
pos1 = nx.spring_layout(G1)
for ea in pos1.keys():
    pos1[ea] =  np.array([mx1, my1]) + pos1[ea]*500000

pos2 = nx.circular_layout(G2)
for ea in pos2.keys():
    pos2[ea] =  np.array([mx2, my2]) + pos2[ea]*500000

nx.draw_networkx(G1, pos=pos1, node_size=100, node_color='green')
nx.draw_networkx(G2, pos=pos2, node_size=100, node_color='red')
#
# (more code below this line)

The output plot:

Edit

Alternate version:

import numpy as np
# compute the positions here
# proper scaling (500000) is applied to the original positions ..
# .. obtained from xxx_layout() to get good spreading

pos1 = nx.spring_layout(G1, scale=500000, center=[mx1, my1])
pos2 = nx.circular_layout(G2, scale=500000, center=[mx2, my2])

nx.draw_networkx(G1, pos=pos1, node_size=100, node_color='green')
nx.draw_networkx(G2, pos=pos2, node_size=100, node_color='red')
#
# (more code below this line)