magnet.graph_utils

losses module.

Contains loss function definitions for GNNs.

Functions

ENC_multiple(Y, graph)	Compute the expected normalized cut for more than 2 partitions.
align_to_x_axis(coords_sample)	Rotate the data so to align it with the x-axis.
cut(Y, edge_index)	Compute the cut of a bipartite graph.
expected_normalized_cut(Y, graph)	Compute the expected normalized cut for a bipartition.
loss_heterogeneous_domains(Y, graph)	Compute loss function for heterogeneous domains.
neighbours(graph, node)	Get the neighbours of a node.
normalized_cut(Y, edge_index)	Compute normalized cut for a bipartition.
randomrotate(coords_sample)	Randomly rotate the mesh.
volumes(Y, edge_index)	Compute the volumes of a partitioned graph.