Check for symmetry in _graph_from_weighted_sparse_matrix

[psl-schaefer]

If one creates a weighted, undirected graph from an adjacency matrix, wouldn't it make sense to check if the adjacency matrix is symmetric? This check was also introduced in rigraph I think, see igraph/rigraph#182.

I am referring to this function:

python-igraph/src/igraph/sparse_matrix.py

Line 154 in 4c82c9d

def _graph_from_weighted_sparse_matrix(

A simple check could be something like symmetric = not np.any((matrix!=matrix.T).data) (here assuming that it is scipy.sparse.csr_matrix).

Here is a simple example (not the most efficient way to check the behavior though):

import numpy as np
from scipy.sparse import csr_matrix
import igraph as ig
rng = np.random.default_rng(0)
n = int(1e3)
sparse_adj = rng.choice(np.array([0, 1, 2]),
                        size=(n**2), replace=True, 
                        p=np.array([0.9, 0.05, 0.05])).reshape(n, n).astype(np.float64)
sparse_adj = csr_matrix(sparse_adj)
symmetric = not np.any((sparse_adj!=sparse_adj.T).data)
print(f"{symmetric=}")
g = ig.Graph.Weighted_Adjacency(sparse_adj, 
                                mode="undirected",
                                attr="weight")

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 14 days if no further activity occurs. Thank you for your contributions.