I'm using the networkx in python. I tested the function adjacency_matrix but I could not understand some of the results. For example:
import networkx as nx
import numpy as np
g = nx.Graph()
g.add_nodes_from([1,2,3])
g.add_edges_from([[1,2],[2,3],[1,3]])
adj = nx.adjacency_matrix(g)
print adj.todense()
# [[0 1 1]
# [1 0 1]
# [1 1 0]]
print adj.__dict__
# {'indices': array([1, 2, 0, 2, 0, 1]), 'indptr': array([0, 2, 4, 6]), 'maxprint'
# : 50, '_shape': (3, 3), 'data': array([1, 1, 1, 1, 1, 1])}
In the result of print adj.__dict__, what does the indices and indptr stand for? I think they are the key information that is used to reconstruct the adjacency matrix, together with the data attribute. But I cannot figure out how it is achieved.
Thank you all for helping me!
adj which is given by nx.adjacency_matrix is a compressed sparse row matrix format of your adjacency matrix.
A compressed format,
represents a matrix M by three (one-dimensional) arrays, that respectively contain [the] nonzero values, the extents of rows, and column indices.
In your case adj.indices is a one-dimensional (numpy) array with the indices of the nonzero values of the adjacency matrix. Together with adj.indptr you can know the exact location of nonzero values in your matrix. By definition adj.indptr[0] == 0, and adj.indptr[i] == adj.indptr[i-1] + number of nonzero values at row i. The data attribute refers to the nonzero values, in your case they are all 1's.