def bees_data():
dtmc_filepath = 'data/prism/bee_multiparam_synchronous_3.pm'
bscc_filepath = 'data/prism/bee_multiparam_synchronous_3.txt'
model = BeesModel.from_files(dtmc_filepath, bscc_filepath)
p_true = [0.2, 0.5, 0.7]
m, d = model.sample(chain_params=p_true, trials_count=10000)
return (m, d)
def mcmc():
dtmc_filepath = 'data/prism/bee_multiparam_synchronous_3.pm'
bscc_filepath = 'data/prism/bee_multiparam_synchronous_3.txt'
model = BeesModel.from_files(dtmc_filepath, bscc_filepath)
mcmc = BayesianMcmc(model)
assert model.bscc_eval_mode == BeesModel.BSCC_MODE_CHAIN_RUN
return mcmc
def small_batch():
logging.info('SMALL BATCH, TEST FOR CORRECT MODEL PROCESSING')
model_files = [
# 3 bees
('models/prism/bee_multiparam_synchronous_3.pm',
'models/prism/bee_multiparam_synchronous_3.txt'),
# 5 bees
('models/prism/bee_multiparam_synchronous_5.pm',
'models/prism/bee_multiparam_synchronous_5.txt')
]
mh_chain_length = [100, 200]
mh_chain_run_factor = [100, 200]
synthetic_data_trials = [100, 200]
for setting in itertools.product(model_files, synthetic_data_trials, mh_chain_length):
files, trials_count, mh_chainLength = setting
logging.info('{}'.format('-' * 80))
logging.info('MODEL INFO')
logging.info('Model DTMC file {}'.format(files[0]))
logging.info('Model BSCC file {}'.format(files[1]))
model = BeesModel.from_files(*files)
logging.info('#params {}, # BSCCs{}'.format(
model.get_params_count(), model.get_bscc_count()))
p_true = rand_increasing_sequence(model.get_params_count())
logging.info('True parameters: {}'.format(p_true))
(s, m, f) = model.sample(params=p_true, trials_count=trials_count)
summarize_data((s, m, f))
do_experiment_rational(model, m, mh_chainLength)
for chain_run_factor in mh_chain_run_factor:
do_experiment_chainrun(model, m, mh_chainLength, chain_run_factor)
def test_mm10params():
logging.info('SINGLE TEST, 15 BEES')
dtmc_filepath = 'models/prism/bee_multiparam_synchronous_10.pm'
bscc_filepath = 'models/prism/bee_multiparam_synchronous_10.txt'
model = BeesModel.from_files(dtmc_filepath, bscc_filepath)
p_true = rand_increasing_sequence(model.get_params_count())
do_experiment(model, p_true, 1000, 500, 100)
def test_chain_run():
print('Sampling with chain run')
dtmc_filepath = 'models/prism/bee_multiparam_synchronous_3.pm'
bscc_filepath = 'models/prism/bee_multiparam_synchronous_3.txt'
bees_model = BeesModel.from_files(dtmc_filepath, bscc_filepath)
r = [0.1, 0.2, 0.3]
start_time = timeit.default_timer()
h = bees_model.sample_run_chain(r, max_trials=1000)
stop_time = timeit.default_timer()
print('Finished in {} seconds, chain length {}'.format(
stop_time - start_time, 1000))
assert (sum(h) - 1.0 < 1e-5)
print('\tSample frequency:', h)
def synthesize_data():
logging.info('DATA SYNTHESIZING')
logging.info('Note that this will run only once')
"""
('models/prism/bee_multiparam_synchronous_3.pm',
'models/prism/bee_multiparam_synchronous_3.txt'),
('models/prism/bee_multiparam_synchronous_5.pm',
'models/prism/bee_multiparam_synchronous_5.txt'),
('models/prism/bee_multiparam_synchronous_10.pm',
'models/prism/bee_multiparam_synchronous_10.txt'),
('models/prism/bee_multiparam_synchronous_15.pm',
'models/prism/bee_multiparam_synchronous_15.txt')
"""
bscc_file = 'models/prism/bee_multiparam_synchronous_10.pm'
model_file = 'models/prism/bee_multiparam_synchronous_10.txt'
bees_model = BeesModel.from_files(model_file, bscc_file)
def evaluation_time():
logging.info('LARGE BATCH, SETTINGS TEST')
model_files = [
# 3 bees
('models/prism/bee_multiparam_synchronous_3.pm',
'models/prism/bee_multiparam_synchronous_3.txt'),
# 5 bees
('models/prism/bee_multiparam_synchronous_5.pm',
'models/prism/bee_multiparam_synchronous_5.txt'),
# 10 bees
('models/prism/bee_multiparam_synchronous_10.pm',
'models/prism/bee_multiparam_synchronous_10.txt'),
# 15 bees
('models/prism/bee_multiparam_synchronous_15.pm',
'models/prism/bee_multiparam_synchronous_15.txt')
]
mh_chain_run_factor = [100, 500, 1000]
max_trials = [100, 500, 1000, 10000]
for file in model_files:
logging.info('{}'.format('-' * 80))
logging.info('EVALUATION TIME')
logging.info('Model DTMC file {}'.format(file[0]))
logging.info('Model BSCC file {}'.format(file[1]))
model = BeesModel.from_files(*file)
logging.info('#params={}, #BSCCs={}'.format(
model.get_params_count(), model.get_bscc_count()))
p_true = rand_increasing_sequence(model.get_params_count())
logging.info('True parameters: {}'.format(p_true))
logging.info('Sample with BSCC Evaluation')
start_time = timeit.default_timer()
(s, m, f) = model.sample(params=p_true, trials_count=1)
stop_time = timeit.default_timer()
elapsed_time = stop_time - start_time
logging.info('Finished in {} seconds'.format(elapsed_time))
logging.info('...')
for factor in mh_chain_run_factor:
trials_count = factor * model.get_bscc_count()
logging.info('Sample with {} chain run'.format(trials_count))
start_time = timeit.default_timer()
h = model.sample_run_chain(p_true, max_trials=trials_count)
stop_time = timeit.default_timer()
elapsed_time = stop_time - start_time
logging.info('Finished in {} seconds'.format(elapsed_time))
def bees_model():
dtmc_filepath = 'models/prism/bee_multiparam_synchronous_3.pm'
bscc_filepath = 'models/prism/bee_multiparam_synchronous_3.txt'
bees_model = BeesModel.from_files(dtmc_filepath, bscc_filepath)
return bees_model
def load_files(self, ):
self.data_model = BeesModel.from_files(self.model_file, self.bscc_file)
assert self.data_model.bscc_eval_mode == BeesModel.BSCC_MODE_CHAIN_RUN
self.print_log("Loaded model file: {}".format(self.model_file))
self.print_log('Loaded BSCCs file: {}'.format(self.bscc_file))
def load_files(self, ):
self.data_model = BeesModel.from_model_file(self.model_file)
assert self.data_model.bscc_eval_mode == BeesModel.BSCC_MODE_CHAIN_RUN
self.print_log("Loaded model file: {}".format(self.model_file))
self.print_log(len(self.data_model.state_labels))
self.print_log(len(self.data_model.bscc_labels))