pytorch_cluster
pytorch_cluster copied to clipboard
Question: Need some help about Edge Index
trafficstars
Thank you for your incredible work! I've learned a lot from it.
May I ask if I have the edge index computed using CUDA, how can I efficiently find the connected components in an arbitrary-degree graph?
Alternatively, is there a method to group nodes using an adjacency matrix with CUDA?
I've tested loading it to the CPU and using NetworkX to handle the subgraph, but it's a bit slow.