What's wrong with new_edge_index, edge_mask = g.subgraph(node_indices, data.edge_index, relabel_nodes=True, num_nodes=data.num_nodes) ? What's the correct way of calling graph.subgraph() ?

[raphael10-collab]

Hi!

Based on the docs: https://pytorch-geometric.readthedocs.io/en/latest/modules/utils.html#torch_geometric.utils.subgraph

            new_edge_index, edge_mask = subgraph(node_indices, data.edge_index, edge_attr= None, relabel_nodes=True, num_nodes=data.num_nodes)  

(taken from here: https://github.com/keitabroadwater/gnns_in_action/blob/master/chapter_4/GNN_GAT_part3.ipynb )

should be correct

But I get the error:

TypeError: Graph.subgraph() got an unexpected keyword argument 'relabel_nodes'

Removing relabel_nodes :

new_edge_index, edge_mask = g.subgraph(node_indices, data.edge_index, num_nodes=data.num_nodes)

I get :

Graph.subgraph() got an unexpected keyword argument 'num_nodes

Removing num_nodes

new_edge_index, edge_mask = g.subgraph(node_indices, data.edge_index)

I get:

Graph.subgraph() takes 2 positional arguments but 3 were given

Removing data.edge_index :

new_edge_index, edge_mask = g.subgraph(node_indices)

I get:

    new_edge_index, edge_mask = g.subgraph(node_indices)
    ^^^^^^^^^^^^^^^^^^^^^^^^^
    ValueError: not enough values to unpack (expected 2, got 0)

What's the correct way of calling graph.subgraph() ?

How to make it work?