pygmtools.utils

Utility functions: problem formulating, data processing, and beyond.

Functions

build_aff_mat	Build affinity matrix for graph matching from input node/edge features.
build_aff_mat_from_graphml	Convert networkx object to affinity matrix
build_aff_mat_from_networkx	Convert networkx object to affinity matrix
build_aff_mat_from_pyg	Convert torch_geometric.data.Data object to affinity matrix
build_batch	Build a batched tensor from a list of tensors.
compute_affinity_score	Compute the affinity score of graph matching.
dense_to_sparse	Convert a dense connectivity/adjacency matrix to a sparse connectivity/adjacency matrix and an edge weight tensor.
download	Check if content exits.
from_graphml	Convert graphml object to adjacency matrix
from_networkx	Convert networkx object to adjacency matrix
from_numpy	Convert a numpy ndarray to a tensor.
from_pyg	Convert torch_geometric.data.Data object to adjacency matrix
gaussian_aff_fn	Gaussian kernel affinity function.
generate_isomorphic_graphs	Generate a set of isomorphic graphs, for testing purposes and examples.
get_network	Get the network object of a neural network solver.
inner_prod_aff_fn	Inner product affinity function.
permutation_loss	Binary cross entropy loss between two permutations, also known as "permutation loss".
set_backend	Set the default backend.
to_graphml	Write an adjacency matrix to a GraphML file
to_networkx	Convert adjacency matrix to NetworkX object
to_numpy	Convert a tensor to a numpy ndarray.
to_pyg	Convert adjacency matrix to torch_geometric.data.Data object

Classes

MultiMatchingResult

A memory-efficient class for multi-graph matching results.