network

Represents the network of interest. This is the primary object of PyPhi and the context of all \(\varphi\) and \(\Phi\) computation.

class pyphi.network.Network(tpm, cm=None, node_labels=None, purview_cache=None)

A network of nodes.

Represents the network under analysis and holds auxilary data about it.

Parameters:

tpm (np.ndarray) –

The transition probability matrix of the network.

The TPM can be provided in any of three forms: state-by-state, state-by-node, or multidimensional state-by-node form. In the state-by-node forms, row indices must follow the little-endian convention (see Little-endian convention). In state-by-state form, column indices must also follow the little-endian convention.

If the TPM is given in state-by-node form, it can be either 2-dimensional, so that tpm[i] gives the probabilities of each node being ON if the previous state is encoded by \(i\) according to the little-endian convention, or in multidimensional form, so that tpm[(0, 0, 1)] gives the probabilities of each node being ON if the previous state is \(N_0 = 0, N_1 = 0, N_2 = 1\).

The shape of the 2-dimensional form of a state-by-node TPM must be (s, n), and the shape of the multidimensional form of the TPM must be [2] * n + [n], where s is the number of states and n is the number of nodes in the network.

Keyword Arguments:
 
  • cm (np.ndarray) – A square binary adjacency matrix indicating the connections between nodes in the network. cm[i][j] == 1 means that node \(i\) is connected to node \(j\) (see Connectivity matrix conventions). If no connectivity matrix is given, PyPhi assumes that every node is connected to every node (including itself).
  • node_labels (tuple[str] or NodeLabels) – Human-readable labels for each node in the network.

Example

In a 3-node network, the_network.tpm[(0, 0, 1)] gives the transition probabilities for each node at \(t\) given that state at \(t-1\) was \(N_0 = 0, N_1 = 0, N_2 = 1\).

tpm

The network’s transition probability matrix, in multidimensional form.

Type:np.ndarray
cm

The network’s connectivity matrix.

A square binary adjacency matrix indicating the connections between nodes in the network.

Type:np.ndarray
connectivity_matrix

Alias for cm.

Type:np.ndarray
causally_significant_nodes

See pyphi.connectivity.causally_significant_nodes().

size

The number of nodes in the network.

Type:int
num_states

The number of possible states of the network.

Type:int
node_indices

The indices of nodes in the network.

This is equivalent to tuple(range(network.size)).

Type:tuple[int]
node_labels

The labels of nodes in the network.

Type:tuple[str]
potential_purviews(direction, mechanism)

All purviews which are not clearly reducible for mechanism.

Parameters:
  • direction (Direction) – CAUSE or EFFECT.
  • mechanism (tuple[int]) – The mechanism which all purviews are checked for reducibility over.
Returns:

All purviews which are irreducible over mechanism.

Return type:

list[tuple[int]]

__len__()

int: The number of nodes in the network.

__eq__(other)

Return whether this network equals the other object.

Networks are equal if they have the same TPM and CM.

to_json()

Return a JSON-serializable representation.

classmethod from_json(json_dict)

Return a Network object from a JSON dictionary representation.

pyphi.network.irreducible_purviews(cm, direction, mechanism, purviews)

Return all purviews which are irreducible for the mechanism.

Parameters:
  • cm (np.ndarray) – An \(N \times N\) connectivity matrix.
  • direction (Direction) – CAUSE or EFFECT.
  • purviews (list[tuple[int]]) – The purviews to check.
  • mechanism (tuple[int]) – The mechanism in question.
Returns:

All purviews in purviews which are not reducible over mechanism.

Return type:

list[tuple[int]]

Raises:

ValueError – If direction is invalid.

pyphi.network.from_json(filename)

Convert a JSON network to a PyPhi network.

Parameters:filename (str) – A path to a JSON file representing a network.
Returns:The corresponding PyPhi network object.
Return type:Network