networkx将加权有向图改为无向

Question

Is a built-in function in networkx could change weighted directed graph to undirected? The functionality should change two directed edges (n1,n2,5) and (n2,n1,7) to one (n1,n2,12).

I searched a long time and didn't fund out one.

Answer 1

Networkx has a to_undirected function but it doesn't sum weights, it is just updating weight with the last found edge weight from the original graph:

If edges in both directions (u, v) and (v, u) exist in the graph, attributes for the new undirected edge will be a combination of the attributes of the directed edges. The edge data is updated in the (arbitrary) order that the edges are encountered. For more customized control of the edge attributes use add_edge().

You should do it manually like this:

G = nx.DiGraph()
G.add_weighted_edges_from([
    (1,2,3),
    (1,3,4),
    (2,1,5),
    (2,3,1),
    (3,2,2)
])
UG = G.to_undirected()
for node in G:
    for ngbr in nx.neighbors(G, node):
        if node in nx.neighbors(G, ngbr):
            UG.edges[node, ngbr]['weight'] = (
                G.edges[node, ngbr]['weight'] + G.edges[ngbr, node]['weight']
            )
UG.edges.data('weight')

will return summarized weights:

EdgeDataView([(1, 2, 8), (1, 3, 4), (2, 3, 3)])