Random Number Generation

The bulk of random number generation in PyBrOpS is provided by NumPy. Random numbers in Python may also be generated via the random module. PyBrOpS synchronizes both of these random number generators underneath a single seed which can be set at the beginning of a simulation.

Seeding PyBrOpS

Seeding a pseudorandom number generator (PRNG) generator is essential to research reproducibility. The seed function within PyBrOpS may be used to seed both the base Python random module and the NumPy numpy.random module using a single seed. These random number generators remain independent despite a shared seeding mechanism.

The example below illustrates seeding:

# import seeding function
from pybrops.core.random.prng import seed

# seed random number generators used by PyBrOpS
seed(48823)

Creating Multiple Pseudorandom Number Generators

In some instances, one may desire to create multiple independent PRNGs. The spawn function within PyBrOpS may be used to initialize multiple PRNG streams.

The example below illustrates the creation of multiple PRNG streams.

# import stream spawning function
from pybrops.core.random.prng import spawn

# create five new PRNG streams
prng_list = spawn(5)