Adding (float) coordinates to graph nodes in Networkx - IndexError

Question

I have a dataset in form of.txt which i want to plot as a graph. This txt provides data as follows:

68  57
65  86
67  83
105 156

etc, so it's an edgelist.

import networkx as nx
import numpy as py
import copy
import matplotlib.pyplot as plt

network0=nx.read_edgelist(r'C:\Users\alexl\Documents\new_python\init_edgelist.txt',create_using=nx.DiGraph)

nx.draw(network0)
plt.show()

After these, i can see a plotted digraph which, as I supposed, has wrong topology. Nodes ans edges are correctly connected, but there is no location information. I tried to import in this another.txt with 3 columns (node xy) for example:

2   478909.145  4204244.629

with this:

coordinates=py.loadtxt(r'C:\Users\alexl\Documents\new_python\nodes_coordinates.txt')
pos=coordinates
nx.draw(network0,pos=pos,with_labels=True)
plt.show()

but doesn't seem to be working. IndexError: only integers, slices ( : ), ellipsis ( ... ), numpy.newaxis ( None ) and integer or boolean arrays are valid indices. <<< That's what it shows in the console when I try to run this. I suppose it has to do with the fact that coordinates are float?? Or that given by txt they make up a list and we want a dictionary? Afterwards I want to add weights (0,1,2) to the edges but first of all I want to see my data plotted "properly" in terms of topology.

Answer 1

nx.draw expects pos to be a dictionary when they keys are the nodes and the values are the positions.

Assuming you want to read your input as a numpy array, what you can do is

pos = dict(zip(coordinates[:,0].astype(int), coordinates[:,1:]))
nx.draw(network0, pos=pos, with_labels=True)

Otherwise you could just iterate over the lines of the file and build the dictionary directly.

pos = {}
with open('coords.txt') as f:
    for line in f:
        node, x, y = line.split()
        pos[node] = float(x), float(y)