PBN module documentation¶

Module containing the PBN class and helpers.

class bang.core.pbn.pbn.PBN(n_nodes: int, n_functions: list[int], n_variables: list[int], functions: list[list[bool]], parent_variable_indices: list[list[int]], function_probabilities: list[list[float]], perturbation_rate: float, non_perturbed_nodes: list[int], n_parallel: int = 512, update_type: Literal['asynchronous_random_order', 'asynchronous_one_random', 'synchronous'] = 'asynchronous_one_random', save_history: bool = True, steps_batch_size=100000)¶

Class representing the PBN and the execution of simulations on it.

Parameters:

n_nodes (int) – The number of nodes.
n_functions (list[int]) – The size of each node.
n_variables (list[int]) – The size of each node’s truth table.
functions (list[list[bool]]) – The truth table of each node.
parent_variable_indices (list[list[int]]) – The index of each node’s truth table.
function_probabilities (list[list[float]]) – The selection probability of each function.
perturbation_rate (float) – The probability of perturbation at each step.
non_perturbed_nodes (list[int]) – Index of nodes without perturbation.
n_parallel (int, optional) – The number of parallel simulations. Defaults to 512.
history (np.ndarray) – The execution history of the PBN, tracking the states of all trajectories.
latest_state (np.ndarray) – The last encountered state of the PBN’s trajectories.
previous_simulations (list[np.ndarray]) – list of previous simulations.
update_type (str) – The type of update to use. The possible values are “asynchronous_one_random”, “asynchronous_random_order”, “synchronous”
save_history (bool) – Whether to save the history of the PBN.
steps_batch_size (int) – The size of the batch of the maximum number of steps executed in a single kernel invocation.

block_graph(filename: str | None = None, format: Literal['pdf', 'png', 'svg'] = 'svg', number_from_one: bool = False) → Digraph¶

Plot the blocks of a Probabilistic Boolean Network (PBN).

This function creates a directed graph where each node represents a block in the PBN, and each edge represents a transition between blocks.

Parameters:

filename (str, optional) – The filename to save the graph. If None, the graph is not saved.
format (Literal['pdf', 'png', 'svg']) – The format to save the graph in. Default is ‘svg’.
number_from_one (bool) – If True, the nodes are numbered starting from 1. If False, they are numbered starting from 0. Default is False.

Returns:

A graphviz.Digraph object representing the block graph.

Return type:

graphviz.Digraph

blocks_detect_attractors(repr: Literal['int']) → list[list[int]]¶

blocks_detect_attractors(repr: Literal['bool']) → list[list[list[bool]]]

Detects attractors in the system using a divide-and-conquer block-based approach.

Returns¶

attractor_stateslist[list[list[bool]]] or list[list[int]]: list of attractors where attractors are coded as lists of lists of bools, lists of bools representing the states.

blocks_detect_attractors_parallel() → list[numpy.ndarray[Any, numpy.dtype[numpy.uint32]]]¶

Detects attractors in the system using a divide-and-conquer block-based approach parallelized on the cpu as well as gpu.

Returns¶

attractor_statesnumpy.NDArray[np.uint32]: list of attractors where attractors are coded as 2D lists od 32 bit unsigned integers.

Creates a clone of the PBN with the specified parameters. If a parameter is not provided, the corresponding attribute of the original PBN is used.

Parameters:

n (int, optional) – The number of nodes.
nf (list[int], optional) – The size of each node.
nv (list[int], optional) – The size of each node’s truth table.
F (list[list[bool]], optional) – The truth table of each node.
varFInt (list[list[int]], optional) – The index of each node’s truth table.
cij (list[list[float]], optional) – The selection probability of each node.
perturbation (float, optional) – The perturbation rate.
npNode (list[int], optional) – Index of nodes without perturbation.
n_parallel (int, optional) – The number of parallel simulations.
update_type (str, optional) – The type of update to use. The possible values are “asynchronous_one_random”, “asynchronous_random_order”, “synchronous”
save_history (bool, optional) – Whether to save the history of the PBN.
steps_batch_size (int, optional) – The size of the batch of the maximum number of steps executed in a single kernel invocation.

Returns:

A new PBN object with the specified parameters.

Return type:

PBN

dependency_graph(filename: str | None = None, format: Literal['pdf', 'png', 'svg'] = 'svg', number_from_one: bool = False) → Digraph¶

Plot the dependency graph of a Probabilistic Boolean Network (PBN).

This function creates a directed graph where each node represents a variable in the PBN, and each edge represents a dependency between variables. An edge from node $i$ to node $j$ indicates that the value of $i$ influences the value of $j$.

Parameters:

filename (str, optional) – The filename to save the graph as a PNG image. If None, the graph is not saved.
format (Literal['pdf', 'png', 'svg']) – The format to save the graph in. Default is ‘svg’.
number_from_one (bool) – If True, the nodes are numbered starting from 1. If False, they are numbered starting from 0. Default is False.

Returns:

A graphviz.Digraph object representing the dependency graph.

Return type:

graphviz.Digraph

get_blocks(repr: Literal['int']) → list[list[int]]¶

get_blocks(repr: Literal['bool']) → list[list[list[bool]]]

get_blocks() → list[list[list[bool]]]

Returns the blocks of the PBN.

Parameters:: repr (str, optional) – The representation type. Can be “bool” or “int”. Defaults to “bool”.
Returns:: The blocks of the PBN.
Return type:: list[list[bool]] or list[list[int]]

monolithic_detect_attractors(initial_states, repr: Literal['bool']) → list[list[list[bool]]]¶

monolithic_detect_attractors(initial_states, repr: Literal['int']) → list[list[int]]

monolithic_detect_attractors(initial_states) → list[list[list[bool]]]

Detects all atractor states in PBN

Parameters¶

initial_stateslist[list[Bool]]: list of investigated states.

Returns¶

attractor_stateslist[list[int]]: list of attractors where attractors are coded as lists of ints, ints representing the states.

monte_carlo_detect_attractors(trajectory_length: int, attractor_length: int, repr='bool')¶

Detects attractors in the system by running multiple trajectories and checking for repetitions.

Parameters¶

trajectory_lengthint: Length after which we assume each trajectory is in attractor.
initial_trajectory_lengthint, optional: Length of trajectory from which we read attractors.

Returns¶

attractor_stateslist[list[list[bool]]] or list[list[int]]: list of attractors where attractors are coded as lists of lists of bools, lists of bools representing the states.

reduce_truthtables(states: list[list[int]]) → tuple¶

Reduces truth tables of PBN by removing states that do not change.

Parameters:: states (list[list[int]]) – list of investigated states. States are lists of int with length n where i-th index represents i-th variable. 0 represents False and 1 represents True.
Returns:: Tuple containing list of indices of variables that change between states and truth tables with removed constant variables.
Return type:: tuple

save_last_state(filename: str)¶

Saves the last encountered state of the PBN’s trajectories to a CSV file.

Parameters:: filename (str) – The name of the file to save the last state.

save_trajectories(filename: str)¶

Saves the execution history of the PBN to a CSV file.

Parameters:: filename (str) – The name of the file to save the history.

set_states(states: list[list[bool]] | ndarray[Any, dtype[uint32]], reset_history: bool = False, stream=None)¶

Sets the initial states of the PBN. If the number of trajectories is different than the number of previous trajectories, the history will be pushed into self.previous_simulations and the active history will be reset.

Parameters:

states (list[list[bool]]) – list of states to be set.
reset_history (bool, optional) – If True, the history of the PBN will be reset. Defaults to False.

simple_steps(n_steps: int, actions: ndarray[Any, dtype[uint64]] | None = None, device: Literal['cuda', 'cpu'] = 'cuda')¶

Simulates the PBN for a given number of steps.

Parameters:

n_steps (int) – Number of steps to simulate.
actions (npt.NDArray[np.uint], optional) – Array of actions to be performed on the PBN. Defaults to None.

Raises:

ValueError – If the initial state is not set before simulation.

trajectory_graph(index: int, filename: str | None = None, format: Literal['pdf', 'png', 'svg'] = 'svg', show_labels: bool = True) → Digraph¶

Plot the trajectory of a Probabilistic Boolean Network (PBN).

This function creates a directed graph where each node represents a state in the trajectory, and each edge represents a transition between states.

Parameters:

index (int) – The index of the trajectory to plot.
filename (str, optional) – The filename to save the graph. If None, the graph is not saved.
format (Literal['pdf', 'png', 'svg']) – The format to save the graph in. Default is ‘svg’.
show_labels (bool) – Whether to show labels on the nodes. Default is True. If set to False, the nodes are represented as points.
number_from_one (bool) – If True, the nodes are numbered starting from 1. If False, they are numbered starting from 0. Default is False.

Returns:

A graphviz.Digraph object representing the trajectory graph.

Return type:

graphviz.Digraph

property cum_extra_functions: list[int]¶

Returns a list of cumulative extra functions.

Returns:: list of cumulative extra functions.
Return type:: list[int]

property cum_n_functions: ndarray¶

Returns the cumulative sum of all elements in nf.

Returns:: Cumulative sum of all elements in nf.
Return type:: np.ndarray

property cum_n_variables: ndarray¶

Returns the cumulative sum of all elements in nv.

Returns:: Cumulative sum of all elements in nv.
Return type:: np.ndarray

property extra_function_index: list[int]¶

Returns a list of extra function indices.

Returns:: list of extra function indices.
Return type:: list[int]

property extra_functions: list[int]¶

Returns a list of extra functions.

Returns:: list of extra functions.
Return type:: list[int]

property function_probabilities: list[list[float]]¶

Returns the selection probability of each function per node.

Returns:: The selection probability of each function per node.
Return type:: list[list[float]]

property functions: list[list[bool]]¶

Returns the truth table of each node.

Returns:: The truth table of each node.
Return type:: list[list[bool]]

property history: ndarray¶

Returns the execution history of the PBN, tracking the states of all trajectories.

Returns:: The execution history of the PBN.
Return type:: np.ndarray

property history_bool: list[list[list[bool]]]¶

Returns the execution history of the PBN in boolean representation.

Returns:: The execution history of the PBN in boolean representation.
Return type:: list[list[list[bool]]]

property integer_functions: list[int]¶

Returns the integer representation of the truth table for each node.

Returns:: Integer representation of the truth table for each node.
Return type:: list[int]

property last_state: ndarray¶

Returns the last encountered state of the PBN’s trajectories.

Returns:: The last encountered state of the PBN’s trajectories.
Return type:: np.ndarray

property last_state_bool: list[list[bool]]¶

Returns the last encountered state of the PBN’s trajectories in boolean representation.

Returns:: The last encountered state of the PBN’s trajectories in boolean representation.
Return type:: list[list[bool]]

property n_extra_function_index: int¶

Returns the number of extra function indices.

Returns:: The number of extra function indices.
Return type:: int

property n_extra_functions: int¶

Returns the number of extra functions.

Returns:: The number of extra functions.
Return type:: int

property n_functions: list[int]¶

Returns the size of each node.

Returns:: The size of each node.
Return type:: list[int]

property n_nodes: int¶

Returns the number of nodes.

Returns:: The number of nodes.
Return type:: int

property n_variables: list[int]¶

Returns the size of each node’s truth table.

Returns:: The size of each node’s truth table.
Return type:: list[int]

property non_perturbed_nodes: list[int]¶

Returns the indices of nodes without perturbation.

Returns:: The indices of nodes without perturbation.
Return type:: list[int]

property parent_variable_indices: list[list[int]]¶

Returns the indices of the parent nodes for every boolean function.

Returns:: The indices of the parent nodes by function.
Return type:: list[list[int]]

property perturbation_rate: float¶

Returns the perturbation rate.

Returns:: The perturbation rate.
Return type:: float

property previous_simulations: list[numpy.ndarray]¶

Returns the list of previous simulation histories. A new simulation history is added every time the number of trajectories changes.

Returns:: The list of previous simulations.
Return type:: list[np.ndarray]

property state_size: int¶

Returns the number of 32-bit integers needed to store all variables.

Returns:: Number of 32-bit integers needed to store all variables.
Return type:: int

bang.core.pbn.pbn.load_from_file(path: str, format: str = 'sbml', n_parallel=512) → PBN¶

Loads a PBN from files of format .pbn or .sbml.

Parameters:

path (str) – Path to the file of format .pbn.
format (str, optional) – Choose the format. Can be either ‘sbml’ for files with .sbml format or ‘assa’ for files with .pbn format. Defaults to ‘sbml’.

Returns:

PBN object representing the network from the file.

Return type:

PBN

Raises:

ValueError – If the format is invalid.

PBN module documentation¶

Returns¶

Returns¶

Parameters¶

Returns¶

Parameters¶

Returns¶

bang

Navigation

Related Topics