Network¶

Network is the first-class input builder. Construct it from a graph library with a from_* constructor, or build it incrementally with the add_* verbs, then partition it with infomap.run() or Network.run().

class infomap.Network(core: Core | None = None)¶

Bases: _NetworkWritersMixin

A first-class Infomap network input builder.

Build a network with the fluent add_*/set_* verbs (each returns self), then run it via run() or the functional infomap.run(). Both return an immutable Result.

Network is an in-memory builder: it runs silently and does not write output files (its engine is created with --silent --no-file-output, and that cannot be turned off per run). Read results off the returned Result. If you need Infomap to write .tree/.clu output files, use Infomap instead.

Examples

Build a small network and run it directly:

>>> from infomap import Network
>>> net = (
...     Network()
...     .add_link(1, 2)
...     .add_link(1, 3)
...     .add_link(2, 3)
...     .add_link(3, 4)
...     .add_link(4, 5)
...     .add_link(4, 6)
...     .add_link(5, 6)
... )
>>> result = net.run()
>>> result.num_top_modules
2
>>> for node_id, module_id in sorted(result.modules().items()):
...     print(node_id, module_id)
1 1
2 1
3 1
4 2
5 2
6 2

A Network can also be handed to infomap.run():

>>> from infomap import run
>>> result = run(net)
>>> result.num_top_modules
2

Parameters:: core (optional) – Internal engine handle to build onto. If None, a default silent, no-file-output Core is created and owned by this Network. When composed by infomap.Infomap, the shared, options-configured Core is passed in.

classmethod from_edge_index(edge_index: Any, *, edge_weight: Any = None, num_nodes: int | None = None, directed: bool = True, node_ids: Any = None) → Network¶

Build a Network from a PyG-style edge index.

Loads edge_index via the same adapter Infomap.add_edge_index() uses. The {internal_id: label} mapping is stored on node_id_to_label.

directed defaults to True: a PyG edge_index is directed by convention – each column is a source -> target edge, and undirected graphs store both directions explicitly. This per-format default differs across the constructors – from_networkx() and from_igraph() infer directedness from the graph, while from_scipy_sparse_matrix() defaults to directed=False.

The edge_weight parameter – a 1-D array with one value per edge, or None for unit weights – matches PyTorch Geometric’s edge_weight name. Each sibling constructor names its weight parameter after its own source library, so the spelling differs by design: from_networkx() weight, from_igraph() edge_weights / vertex_weights, from_scipy_sparse_matrix() weighted.

classmethod from_file(path: str | PathLike[str], *, accumulate: bool = True) → Network¶

Build a Network by reading a network file.

Parameters:

path (str or os.PathLike) – Path to an Infomap-readable network file.
accumulate (bool, optional) – Accumulate onto already added nodes and links. Default True.

classmethod from_igraph(g: igraph.Graph, *, edge_weights: Any = None, vertex_weights: Any = None, node_id: str = 'node_id', layer_id: str = 'layer_id', multilayer_inter_intra_format: bool = True, meta_attribute: str | None = None) → Network¶

Build a Network from a python-igraph graph.

Loads g via the same adapter Infomap.add_igraph_graph() uses. The {internal_id: label} mapping is stored on node_id_to_label. Pass meta_attribute to use a vertex attribute as Infomap metadata (values are encoded to integers, so string categories work).

The edge_weights / vertex_weights parameters match python-igraph’s own Graph.community_infomap(edge_weights, vertex_weights) names. Each sibling constructor names its weight parameter after its own source library, so the spelling differs by design: from_networkx() weight, from_scipy_sparse_matrix() weighted, from_edge_index() edge_weight.

classmethod from_networkx(g: networkx.Graph, *, weight: str | None = 'weight', node_id: str = 'node_id', layer_id: str = 'layer_id', multilayer_inter_intra_format: bool = True, meta_attribute: str | None = None) → Network¶

Build a Network from a NetworkX graph.

Loads g via the same adapter Infomap.add_networkx_graph() uses. The {internal_id: label} mapping is stored on node_id_to_label. Pass meta_attribute to use a node attribute as Infomap metadata (values are encoded to integers, so string categories work).

The weight parameter – the edge-data key to read, or None to treat every edge as unit weight – matches networkx’s own weight argument. Each sibling constructor names its weight parameter after its own source library, so the spelling differs by design: from_igraph() edge_weights / vertex_weights, from_scipy_sparse_matrix() weighted, from_edge_index() edge_weight.

classmethod from_scipy_sparse_matrix(A: Any, *, directed: bool = False, weighted: bool = True, node_ids: Any = None) → Network¶

Build a Network from a SciPy sparse adjacency matrix.

Loads A via the same adapter Infomap.add_scipy_sparse_matrix() uses. The {internal_id: label} mapping is stored on node_id_to_label.

directed defaults to False: a symmetric adjacency is the common undirected case, so A[i, j] and A[j, i] are folded together. This per-format default differs across the constructors – from_networkx() and from_igraph() infer directedness from the graph, while from_edge_index() defaults to directed=True.

weighted – use the matrix values as weights, else treat every stored entry as weight 1 – is a local name: scipy.sparse.csgraph has no such flag. Each sibling constructor names its weight parameter after its own source library, so the spelling differs by design: from_networkx() weight, from_igraph() edge_weights / vertex_weights, from_edge_index() edge_weight.

add_link(source_id: int, target_id: int, weight: float = 1.0) → Network¶

Add a link.

Parameters:

source_id (int)
target_id (int)
weight (float, optional)

add_links(links: Any) → Network¶

Add several links.

Parameters:: links (iterable of tuples or numpy.ndarray) – Iterable of tuples of int of the form (source_id, target_id, [weight]). NumPy arrays must be 2-dimensional with 2 or 3 columns.

add_multilayer_inter_link(source_layer_id: int, node_id: int, target_layer_id: int, weight: float = 1.0) → Network¶

Add an inter-layer link.

Parameters:

source_layer_id (int)
node_id (int)
target_layer_id (int)
weight (float, optional)

add_multilayer_inter_links(links: Any) → Network¶

Add several inter-layer links.

Parameters:: links (iterable of tuples) – Iterable of tuples of the form (source_layer_id, node_id, target_layer_id, [weight]). NumPy arrays must be 2-dimensional with 3 or 4 columns.

add_multilayer_intra_link(layer_id: int, source_node_id: int, target_node_id: int, weight: float = 1.0) → Network¶

Add an intra-layer link.

Parameters:

layer_id (int)
source_node_id (int)
target_node_id (int)
weight (float, optional)

add_multilayer_intra_links(links: Any) → Network¶

Add several intra-layer links.

Parameters:: links (iterable of tuples) – Iterable of tuples of the form (layer_id, source_node_id, target_node_id, [weight]). NumPy arrays must be 2-dimensional with 3 or 4 columns.

add_multilayer_link(source_multilayer_node: Any, target_multilayer_node: Any, weight: float = 1.0) → Network¶

Add a multilayer link.

Parameters:

source_multilayer_node (tuple of int, or MultilayerNode) – (layer_id, node_id).
target_multilayer_node (tuple of int, or MultilayerNode) – (layer_id, node_id).
weight (float, optional)

add_multilayer_links(links: Any) → Network¶

Add several multilayer links.

Parameters:: links (iterable of tuples) – Iterable of tuples of the form (source_node, target_node, [weight]). NumPy arrays must be 2-dimensional with 4 or 5 columns of the form (source_layer_id, source_node_id, target_layer_id, target_node_id, [weight]).

add_node(node_id: int, name: str | None = None, teleportation_weight: float | None = None) → Network¶

Add a node.

Parameters:

node_id (int)
name (str, optional)
teleportation_weight (float, optional) – Used for teleporting between layers in multilayer networks.

add_nodes(nodes: Any) → Network¶

Add nodes.

Parameters:: nodes (iterable of tuples or iterable of int or dict) – Iterable of tuples on the form (node_id, [name], [teleportation_weight]).

add_state_node(state_id: int, node_id: int, name: str | None = None) → Network¶

Add a state node.

Parameters:

state_id (int)
node_id (int) – Id of the physical node the state node should be added to.
name (str, optional)

add_state_nodes(state_nodes: Any) → Network¶

Add state nodes.

Parameters:: state_nodes (iterable of tuples or dict of int: int) – Iterable of tuples of the form (state_id, node_id) or (state_id, node_id, name), or dict of the form {state_id: node_id}.

read_file(filename: str, accumulate: bool = True) → Network¶

Read network data from file.

Parameters:

filename (str)
accumulate (bool, optional) – If the network data should be accumulated to already added nodes and links. Default True.

Raises:

NetworkParseError – If the file cannot be opened or its content cannot be parsed.

remove_link(source_id: int, target_id: int) → Network¶

Remove a link.

Parameters:

source_id (int)
target_id (int)

remove_links(links: Any) → Network¶

Remove several links.

Parameters:: links (iterable of tuples) – Iterable of tuples of the form (source_id, target_id).

run(*, options: Options | Mapping[str, Any] | None=None, seed: int = <unset>, num_trials: int = <unset>, two_level: bool = <unset>, directed: bool | None = <unset>, markov_time: float = <unset>, args: str | None = None, initial_partition: Mapping[Any, Any] | None=None, **overrides: Any) → Result¶

Run Infomap on this network and return a Result.

This is a thin convenience wrapper: net.run(**kw) is equivalent to infomap.run(net, **kw), the canonical entry point. It accepts the same keywords.

Parameters:

options (Options, mapping, or None, optional) – Configuration rendered to Infomap CLI flags for this run.
seed (optional) – The five common-tier engine options, accepted directly here to match infomap.run() and infomap.Infomap.run(). A supplied value overrides the options carrier.
num_trials (optional) – The five common-tier engine options, accepted directly here to match infomap.run() and infomap.Infomap.run(). A supplied value overrides the options carrier.
two_level (optional) – The five common-tier engine options, accepted directly here to match infomap.run() and infomap.Infomap.run(). A supplied value overrides the options carrier.
directed (optional) – The five common-tier engine options, accepted directly here to match infomap.run() and infomap.Infomap.run(). A supplied value overrides the options carrier.
markov_time (optional) – The five common-tier engine options, accepted directly here to match infomap.run() and infomap.Infomap.run(). A supplied value overrides the options carrier.
args (str, optional) – Raw Infomap arguments prepended before the rendered options.
initial_partition (mapping, optional) – Initial module assignment for this run only, keyed by internal node id ({node_or_state_id: module_id}) or, for a multilayer network, {(layer_id, node_id): module_id}. The internal ids are the ones you built the network with (the values of node_id_to_label).
**overrides – Any other Infomap engine option, as a keyword argument, forwarded to Options and matching infomap.run(). Convenient for a one-off; for a reusable or validated configuration prefer options=Options(...), the canonical carrier and full reference.

Returns:

An immutable snapshot of the run, bound to this Network.

Return type:

Result

Notes

A Network engine is silent for its whole lifetime; silent=False cannot re-enable the engine log here and warns instead. For the log, run through the stateful Infomap (constructed with silent=False) or pass the input directly to infomap.run().

set_bipartite_start_id(start_id: int) → Network¶

Set the bipartite start id, returning self for fluent chaining.

The chainable form of the bipartite_start_id setter, so a bipartite network can be built in a single expression alongside the add_* verbs.

set_meta_data(node_id: Any, meta_category: int | None = None) → Network¶

Set integer metadata for one node, or for many at once.

Parameters:

node_id (int or mapping) – A node id, or a {node_id: meta_category} mapping to assign metadata to several nodes in one call.
meta_category (int, optional) – The meta category, when node_id is a single node id (ignored when node_id is a mapping).

set_name(node_id: int, name: str | None) → Network¶

Set the name of a node.

Parameters:

node_id (int)
name (str)

set_names(names: Any) → Network¶

Set names to several nodes at once.

Parameters:: names (iterable of tuples or dict of int: str) – Iterable of tuples on the form (node_id, name) or dict of the form {node_id: name}.

write_pajek(filename: str | PathLike[str], flow: bool = False) → None¶

Write network to a Pajek file.

An existing file at filename is overwritten.