How do I create an homogeneous graph like an heterogeneous graph?

[ppsmk388]

It is very easy to create nodes as well as edges when create the heterogeneous graph.
https://pytorch-geometric.readthedocs.io/en/latest/notes/heterogeneous.html#creating-heterogeneous-graphs

Is there a similar way to create homogeneous graph like create the heterogeneous graph?
I don't see a similar method in the documentation
Thanks!

[EdisonLeeeee]

A homogeneous graph in PyG is denoted by torch_geometric.data.Data

from torch_geometric.data import Data
data = Data(x=..., edge_index=..., edge_attr=...)

[EdisonLeeeee]

There should exist a y after data.to_homogeneous() if it already existed in the original heterogeneous graph. Could you specify?

[ppsmk388]

Thank you for your reply!

In this example the label of dataset disappear when use the to_homogeneous() .

https://pytorch-geometric.readthedocs.io/en/latest/notes/heterogeneous.html#utility-functions

[EdisonLeeeee]

I've run the code and got the results:

homogeneous_data = data.to_homogeneous()
print(homogeneous_data)

Data(edge_index=[2, 13605929], x=[1879778, 128], year=[1879778], y=[1879778], train_mask=[1879778], val_mask=[1879778], test_mask=[1879778], node_type=[1879778], edge_type=[13605929])

The goal of this tutorial is to give an illustration on some utility functions of torch_geometric.data.HeteroData class, and it only displays some important details for the sake of clarity. Sorry for the confusion, do you think we should update the tutorial?

[rusty1s]

Yes, we would need to update the tutorial since support for this just got recently added.

[ppsmk388]

Thanyk you!