NetworkX: plotting a graph in a for loop returns the wrong graph

Question

I am working with a regular network (grid) that I create as:

import networkx as nx
N=100
def graph_creating(N):
    G=nx.grid_2d_graph(N,N)
    pos = dict( (n, n) for n in G.nodes() ) #Dictionary of all positions
return G, pos

I have two ways of iterating the code. In both I remove nodes from the network and attempt to draw it. Depending on whether I create the network initially or within the loop, I get a different behavior when I plot it.

My problem: I want to plot the grid after stage 1 and after stage 2 , so to make comparisons with the unaltered grid/graph. I fail to do it properly because:

• If the original grid is created outside the for loop, I correctly get the first plots but later plots are empty as the graph is never restored back to its unaltered statu;
• If the original grid is created inside the for loop, I always end up having the unaltered grid plotted, as if the removal had no effect on it.

Where else should the graph-creating block be placed, in order to be able to plot the graph right after stage 1 and 2?

version 1 the graph is created outside the for loop:

G, pos = graph_creating(N)
nodelist = G.nodes()
for counter in range(5):
    G1 = nodelist[2*counter:2*counter+1]
    G.remove_nodes_from(G1)
    nx.draw_networkx(G, pos = pos)
    figurename = 'file{0}.png'.format(counter)
    plt.savefig(figurename)

    G2=nodelist[2*counter+1:2*counter+2]
    G.remove_nodes_from(G2)
    nx.draw_networkx(G,pos=pos)
    #it's not clear from your original question if you save this figure or not

Result: only the first iteration produces correct plots. Later plots are empty as the graph is never restored back to its unaltered status.

version 2 the graph is created inside the for loop:

for counter in range(5):
    G, pos = graph_creating(N)
    nodelist = G.nodes()
    G1 = nodelist[2*counter:2*counter+1]
    G.remove_nodes_from(G1)
    nx.draw_networkx(G, pos = pos)
    figurename = 'file{0}.png'.format(counter)
    plt.savefig(figurename)

    G2=nodelist[2*counter+1:2*counter+2]
    G.remove_nodes_from(G2)
    nx.draw_networkx(G,pos=pos)
    #it's not clear from your original question if you save this figure or not

Result: the calls to nx.draw_networkx result in the unaltered graph being plotted for each iteration. I wonder if the problem is in the way I call that function, as it always plots the graph with no failed nodes. Why do I have this plotting problem? .

Answer 1

• Do your graph creating block outside the for loop.
• Inside the for-loop create a copy of the graph using tmp_G = G.copy()
• when using draw_netwokx or remove_nodes_from use tmp_G instead of G
• When you want to compare the altered graphs to the original one call the draw_networkx function using graph G