def run(self):
self.sample = test_uniform_threaded(
self.sample_size, self.wait_time)
def _generate_samples_in_parallel(self,
n_task=10,
n_workers=None,
wait_time=0,
parallelization_type='multiprocessing'):
"""This function generates samples of uniform distribution in parallel
Parameters
----------
n_task : `int`
Number of task that will be run
n_workers : `int`
Number of workers that will be used.
If it is set to None the maximum number of workers will be used.
Otherwise this number of processes will be created (in case of
multiprocessing) or work will be executed sequestially (in case
of threading)
wait_time : `float`
Time that will be waited on each thread after random number
simulation
parallelization_type : `string`
How work will be parallelized : either using multiprocessing or
using threading
"""
sample_size = self.stat_size
args = [(sample_size, wait_time) for _ in range(n_task)]
if parallelization_type == 'multiprocessing':
pool = Pool(processes=n_workers)
samples = pool.starmap(test_uniform_threaded, args)
samples = np.array(samples)
elif parallelization_type == 'threading':
class UniformThread(threading.Thread):
def __init__(self, sample_size, wait_time):
super(UniformThread, self).__init__()
self.sample_size = sample_size
self.wait_time = wait_time
self.sample = None
def run(self):
self.sample = test_uniform_threaded(
self.sample_size, self.wait_time)
if n_workers is None:
threads = []
for arg in args:
threads.append(UniformThread(*arg))
[t.start() for t in threads]
[t.join() for t in threads]
samples = np.array([t.sample for t in threads])
else:
samples = np.array(
[test_uniform_threaded(*arg) for arg in args])
else:
raise ValueError('Unknown thread type')
return samples