def renormalize_beliefs_multiproc(belief_space, normalization_factor):
belief_space.num_idxs_with_belief_above_threshold = 0
slicing_indices = generate_slicing_indices(belief_space.beliefs)
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
max_belief = -1
max_eles = []
num_idxs_with_belief_above_threshold = 0
return_tuples = parallel(
delayed(belief_space.renormalize_beliefs_common)
(normalization_factor, slicing_indices[i - 1], slicing_indices[i])
for i in range(1, len(slicing_indices)))
for return_tuple in return_tuples:
returned_max_eles, returned_max_belief, returned_beliefs, returned_num_idxs_with_belief_above_threshold, starting_index, ending_index = (
return_tuple)
# copy beliefs from results
belief_space.beliefs[
starting_index:ending_index] = returned_beliefs
num_idxs_with_belief_above_threshold += (
returned_num_idxs_with_belief_above_threshold)
# find max chains
if returned_max_belief > max_belief:
max_eles = returned_max_eles
max_belief = returned_max_belief
# find actual max in tuple returned
return max_eles, max_belief, num_idxs_with_belief_above_threshold
def update_bottom_up_beliefs_multiproc(
top_down_bottom_up_structure_belief_space_wrapper, attribute_order, trial_name,
):
slicing_indices = generate_slicing_indices(
top_down_bottom_up_structure_belief_space_wrapper.bottom_up_belief_space.beliefs
)
with Parallel(
n_jobs=multiprocessing.cpu_count(), verbose=5, max_nbytes=PARALLEL_MAX_NBYTES
) as parallel:
return_tuples = parallel(
delayed(
top_down_bottom_up_structure_belief_space_wrapper.update_bottom_up_beliefs_common
)(attribute_order, trial_name, slicing_indices[i - 1], slicing_indices[i])
for i in range(1, len(slicing_indices))
)
for return_tuple in return_tuples:
returned_beliefs, starting_index, ending_index = return_tuple
# copy beliefs from results
top_down_bottom_up_structure_belief_space_wrapper.bottom_up_belief_space.beliefs[
starting_index:ending_index
] = returned_beliefs
return (
top_down_bottom_up_structure_belief_space_wrapper.bottom_up_belief_space.beliefs
)
def instantiate_schemas_multiproc(
abstract_schema_space,
causal_chain_structure_space,
instantiation_index_mappings,
solutions_executed,
excluded_chain_idxs,
n_chains_in_schema,
):
slicing_indices = generate_slicing_indices(
abstract_schema_space.structure_space)
instantiated_schemas = []
instantiated_schema_beliefs = []
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
return_tuples = parallel(
delayed(abstract_schema_space.instantiate_schemas_common)(
causal_chain_structure_space,
instantiation_index_mappings,
solutions_executed,
excluded_chain_idxs,
n_chains_in_schema,
slicing_indices[i - 1],
slicing_indices[i],
) for i in range(1, len(slicing_indices)))
for return_tuple in return_tuples:
returned_schemas, returned_beliefs, starting_index, ending_index = (
return_tuple)
# copy beliefs from results
instantiated_schemas.extend(returned_schemas)
instantiated_schema_beliefs.extend(returned_beliefs)
return instantiated_schemas, instantiated_schema_beliefs
def prune_chains_from_initial_observation_multiproc(chain_pruner,
causal_chain_space,
position_dict, trial_count,
attempt_count):
slicing_indices = generate_slicing_indices(
causal_chain_space.structure_space.causal_chains)
chain_idxs_pruned = set()
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
chain_space_tuples = parallel(
delayed(chain_pruner.prune_chains_from_initial_observation_common)(
causal_chain_space,
position_dict,
trial_count,
attempt_count,
slicing_indices[i - 1],
slicing_indices[i],
) for i in range(1, len(slicing_indices)))
for chain_space_tuple in chain_space_tuples:
returned_beliefs, returned_chain_idxs_pruned, starting_index, ending_index = (
chain_space_tuple)
causal_chain_space.bottom_up_belief_space.beliefs[
starting_index:ending_index] = returned_beliefs
chain_idxs_pruned.update(returned_chain_idxs_pruned)
return causal_chain_space.bottom_up_belief_space.beliefs, chain_idxs_pruned
def sum_beliefs_multiproc(belief_space):
slicing_indices = generate_slicing_indices(belief_space.beliefs)
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
sums = parallel(
delayed(belief_space.sum_beliefs_common)(slicing_indices[i - 1],
slicing_indices[i])
for i in range(1, len(slicing_indices)))
return sum(sums)
def set_num_idxs_with_belief_above_threshold_multiproc(belief_space):
slicing_indices = generate_slicing_indices(belief_space)
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
summations = parallel(
delayed(belief_space.
compute_num_idxs_with_belief_above_threshold_common)(
slicing_indices[i - 1], slicing_indices[i])
for i in range(1, len(slicing_indices)))
summation = sum(summations)
return summation
def set_uniform_belief_for_idx_with_belief_above_threshold_multiproc(
belief_space,
threshold,
):
slicing_indices = generate_slicing_indices(belief_space.beliefs)
with Parallel(n_jobs=multiprocessing.cpu_count(),
verbose=5,
max_nbytes=PARALLEL_MAX_NBYTES) as parallel:
returned_tuples = parallel(
delayed(
belief_space.
set_uniform_belief_for_idx_with_belief_above_threshold_common)(
slicing_indices[i - 1],
slicing_indices[i],
threshold,
) for i in range(1, len(slicing_indices)))
for returned_tuple in returned_tuples:
returned_beliefs, starting_index, ending_index = returned_tuple
belief_space.beliefs[
starting_index:ending_index] = returned_beliefs
return belief_space.beliefs
