nip.graph_isomorphism.dataset_generation.GraphIsomorphicDatasetConfig

class nip.graph_isomorphism.dataset_generation.GraphIsomorphicDatasetConfig(num_samples: int, graph_sizes: list[int] = <factory>, edge_probabilities: list[float] = <factory>, max_iterations: int = 5, prop_non_isomorphic: float = 0.5, non_iso_prop_score_1: float = 0.1, non_iso_prop_score_2: float = 0.2, iso_prop_from_non_iso: float = 0.5)

Configuration for a graph isomorphism dataset.

Graphs are generated using the Erdős-Rényi model. The dataset is generated in three steps:

1. Generate prop_non_isomorphic non-isomorphic graphs. The pairs are divided equally between the different graph sizes and edge probabilities. The number of graphs with a score of 1, 2 and greater than 2 are divided according to the proportions non_iso_prop_score_1 and non_iso_prop_score_2.

2. Generate (1 - prop_non_isomorphic) * iso_prop_from_non_iso isomorphic graphs, by sampling from the non-isomorphic graph pairs and shuffling the nodes.

3. Generate the remaining (1 - prop_non_isomorphic) * (1 - iso_prop_from_non_iso) isomorphic graphs, by generating new graphs and shuffling the nodes.

Parameters:

• num_samples (int) – The number of graph pairs to generate.

• graph_sizes (list[int], default=[7, 8, 9, 10, 11]) – The sizes of the graphs to generate.

• edge_probabilities (list[float], default=[0.2, 0.4, 0.6, 0.8]) – The probabilities of an edge between two nodes.

• max_iterations (int, default=5) – The maximum number of iterations of the Weisfeiler-Lehman algorithm to run to compute the score.

• prop_non_isomorphic (float, default=0.5) – The proportion of graph pairs which are non-isomorphic.

• non_iso_prop_score_1 (float, default=0.1) – The proportion of non-isomorphic graph pairs which have a score of 1.

• non_iso_prop_score_2 (float, default=0.2) – The proportion of non-isomorphic graph pairs which have a score of 2.

• iso_prop_from_non_iso (float, default=0.5) – The proportion of graph pairs which are isomorphic and are generated from the non-isomorphic graph pairs.

Methods Summary

__eq__(other)	Return self==value.
__init__(num_samples[, graph_sizes, ...])
__repr__()	Return repr(self).

Attributes

iso_prop_from_non_iso
max_iterations
non_iso_prop_score_1
non_iso_prop_score_2
prop_non_isomorphic
num_samples
graph_sizes
edge_probabilities

Methods

__eq__(other)

Return self==value.

__init__(num_samples: int, graph_sizes: list[int] = <factory>, edge_probabilities: list[float] = <factory>, max_iterations: int = 5, prop_non_isomorphic: float = 0.5, non_iso_prop_score_1: float = 0.1, non_iso_prop_score_2: float = 0.2, iso_prop_from_non_iso: float = 0.5) → None

__repr__()

Return repr(self).