1
0
Fork 0
Old engine for Continuous Time Bayesian Networks. Superseded by reCTBN. 🐍 https://github.com/madlabunimib/PyCTBN
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
This repo is archived. You can view files and clone it, but cannot push or open issues/pull-requests.
PyCTBN/PyCTBN/classes/set_of_cims.py

94 lines
4.0 KiB

import typing
import numpy as np
from .conditional_intensity_matrix import ConditionalIntensityMatrix
class SetOfCims(object):
"""Aggregates all the CIMS of the node identified by the label _node_id.
:param node_id: the node label
:type node_ind: string
:param parents_states_number: the cardinalities of the parents
:type parents_states_number: List
:param node_states_number: the caridinality of the node
:type node_states_number: int
:param p_combs: the p_comb structure bound to this node
:type p_combs: numpy.ndArray
:_state_residence_time: matrix containing all the state residence time vectors for the node
:_transition_matrices: matrix containing all the transition matrices for the node
:_actual_cims: the cims of the node
"""
def __init__(self, node_id: str, parents_states_number: typing.List, node_states_number: int, p_combs: np.ndarray):
"""Constructor Method
"""
self._node_id = node_id
self._parents_states_number = parents_states_number
self._node_states_number = node_states_number
self._actual_cims = []
self._state_residence_times = None
self._transition_matrices = None
self._p_combs = p_combs
self.build_times_and_transitions_structures()
def build_times_and_transitions_structures(self) -> None:
"""Initializes at the correct dimensions the state residence times matrix and the state transition matrices.
"""
if not self._parents_states_number:
self._state_residence_times = np.zeros((1, self._node_states_number), dtype=np.float)
self._transition_matrices = np.zeros((1, self._node_states_number, self._node_states_number), dtype=np.int)
else:
self._state_residence_times = \
np.zeros((np.prod(self._parents_states_number), self._node_states_number), dtype=np.float)
self._transition_matrices = np.zeros([np.prod(self._parents_states_number), self._node_states_number,
self._node_states_number], dtype=np.int)
def build_cims(self, state_res_times: np.ndarray, transition_matrices: np.ndarray) -> None:
"""Build the ``ConditionalIntensityMatrix`` objects given the state residence times and transitions matrices.
Compute the cim coefficients.The class member ``_actual_cims`` will contain the computed cims.
:param state_res_times: the state residence times matrix
:type state_res_times: numpy.ndArray
:param transition_matrices: the transition matrices
:type transition_matrices: numpy.ndArray
"""
for state_res_time_vector, transition_matrix in zip(state_res_times, transition_matrices):
cim_to_add = ConditionalIntensityMatrix(state_res_time_vector, transition_matrix)
cim_to_add.compute_cim_coefficients()
self._actual_cims.append(cim_to_add)
self._actual_cims = np.array(self._actual_cims)
self._transition_matrices = None
self._state_residence_times = None
def filter_cims_with_mask(self, mask_arr: np.ndarray, comb: typing.List) -> np.ndarray:
"""Filter the cims contained in the array ``_actual_cims`` given the boolean mask ``mask_arr`` and the index
``comb``.
:param mask_arr: the boolean mask that indicates which parent to consider
:type mask_arr: numpy.array
:param comb: the state/s of the filtered parents
:type comb: numpy.array
:return: Array of ``ConditionalIntensityMatrix`` objects
:rtype: numpy.array
"""
if mask_arr.size <= 1:
return self._actual_cims
else:
flat_indxs = np.argwhere(np.all(self._p_combs[:, mask_arr] == comb, axis=1)).ravel()
return self._actual_cims[flat_indxs]
@property
def actual_cims(self) -> np.ndarray:
return self._actual_cims
@property
def p_combs(self) -> np.ndarray:
return self._p_combs
def get_cims_number(self):
return len(self._actual_cims)