magnet.graph_utils.normalized_cut

magnet.graph_utils.normalized_cut(Y: Tensor, edge_index: Tensor)

Compute normalized cut for a bipartition.

Parameters:

• Y (torch.Tensor) – Tensor of shape (num_nodes, 2) whose values are a one-hot encoding of the bipartition of the graph.

• edge_index (torch.Tensor) – Edge index tensor of the graph.

Returns:

The value of the loss function.

Return type:

torch.Tensor

Notes

If Y is a tensor of probabilities, this is the expected normalized cut.