MutableGraph

The MutableGraph is an abstract base class, that allows modifications and additions to the graph. It is used to build graphs. SizeBoundedGraph is a subclass.

class deglib.graph.MutableGraph(graph_cpp: SearchGraph)

Bases: SearchGraph, ABC

abstract add_vertex(external_label: int, feature_vector: ndarray) → int

Add a new vertex. The neighbor indices will be prefilled with a self-loop, the weights will be 0.

Parameters:

  • external_label – The label for the new vertex

  • feature_vector – The feature vector to add. This numpy array should be c-contiguous, otherwise it has to be reallocated.

Returns:

the internal index of the new vertex

abstract remove_vertex(external_label: int)

Remove an existing vertex.

Parameters:

external_label – The external label of the vertex that should be removed.

abstract change_edge(internal_index: int, from_neighbor_index: int, to_neighbor_index: int, to_neighbor_weight: float) → bool

Swap a neighbor with another neighbor and its weight.

Parameters:

  • internal_index – vertex index which neighbors should be changed

  • from_neighbor_index – neighbor index to remove

  • to_neighbor_index – neighbor index to add

  • to_neighbor_weight – weight of the neighbor to add

Returns:

True if the from_neighbor_index was found and changed

abstract change_edges(internal_index: int, neighbor_indices: ndarray, neighbor_weights: ndarray)

Change all edges of a vertex. The neighbor indices/weights and feature vectors will be copied. The neighbor array need to have enough neighbors to match the edge-per-vertex count of the graph. The indices in the neighbor_indices array must be sorted.

Parameters:

  • internal_index – The index of the vertex for which edges should change

  • neighbor_indices – These neighbors will be set as the new neighbors of the specified vertex

  • neighbor_weights – These weights will be set as the new weights for the neighbors.

abstract get_neighbor_weights(internal_index: int, copy: bool = False) → ndarray

Get weights for each neighbor of the vertex defined by the given index.

Parameters:

  • internal_index – The index that specifies the vertex

  • copy – If True the returned neighbor weights are copied, otherwise they reference internal graph data.

Returns:

The weights of the neighbors

abstract get_edge_weight(from_neighbor_index: int, to_neighbor_index: int) → float

Get the weight from vertex to another vertex. If start vertex is not a neighbor of end vertex, -1.0 is returned.

Parameters:

  • from_neighbor_index – Internal index of the start vertex

  • to_neighbor_index – Internal index of the target vertex

Returns:

If present the weight between start and target vertex, -1.0 otherwise

abstract save_graph(path: str | Path) → bool

Save graph to specified file. Creates necessary directories.

Parameters:

path – The path where to save the file.

remove_non_mrng_edges()

Remove all edges which are not MRNG conform.