Is there any API about getting Khop neighbor graph of a given node or node list?

Questions
#1

Is there any API about getting Khop neighbor graph of a given node or node list?
Thanks

1 Like
#2

I think currently we only allow you to compute k-hop graph for the whole graph. I’m not sure if this is enough for your case.

#3

I have noticed this API,but it is not very clear.
Can you give an example about the return value of the function?
Thank you! mufei

#5

Does this example below make sense to you?

import dgl

g = dgl.DGLGraph()
g.add_nodes(3)
g.add_edges([0, 1], [1, 2]) # 0->1, 1->2
g_2 = dgl.transform.khop_graph(g, 2)
print(g_2.edges()) # 0->2

Also note that I just found a bug for this API and fixed it in PR #1433. You will need to install from source to get the latest version.

#6

why only one edge in g_2?

#7

It’s equivalent to raise the adjacency matrix to the power of 2, where a nonzero entry indicates that we can go from node i to node j with two steps.

1 Like
#8

:+1:
thank you mufei

#9

How could I get the subgraph construbuted by khop neighbors of one specific node and itself?
Is there an API like sub_graph(graph, nid, k)?
Now I try to get the 1-hop subgraph of one certain node is

sub_graph = graph.subgraph(graph.in_edges(nid)[0])
1 Like
#10

You can find the subgraph by iteratively finding the 1st, 2nd, 3rd hop neighbors:

nids = [nid]
for _ in range(k):
    nid = graph.in_edges(nid)[0].unique()
    nids.append(new_nid)
nids = torch.cat(nids).unique()
subgraph = g.subgraph(nids)