How to Set a Seed

To ensure reproducibility of results users can set a seed for all the random number streams that HCiw uses. This can be done using the HCiw functon hciw.seed:

>>> import hciw
>>> hciw.seed(5)

Note that due to sampling on initialisation, the seed will need to be set before the hciw.Simulation object is created.

As an example, take the following network:

>>> N = hciw.create_network(
...     arrival_distributions=[hciw.dists.Exponential(rate=5)],
...     service_distributions=[hciw.dists.Exponential(rate=10)],
...     number_of_servers=[1]
... )

Now let's run the system for 20 time units, using a seed of 1, and get the average waiting time:

>>> hciw.seed(1)
>>> Q = hciw.Simulation(N)
>>> Q.simulate_until_max_time(20)
>>> waits = [r.waiting_time for r in Q.get_all_records()]
>>> sum(waits)/len(waits)
0.0824058654563...

Using the same seed again, the exact same average waiting time result will occur:

>>> hciw.seed(1)
>>> Q = hciw.Simulation(N)
>>> Q.simulate_until_max_time(20)
>>> waits = [r.waiting_time for r in Q.get_all_records()]
>>> sum(waits)/len(waits)
0.0824058654563...

Now using a different seed, a different result will occur:

>>> hciw.seed(2)
>>> Q = hciw.Simulation(N)
>>> Q.simulate_until_max_time(20)
>>> waits = [r.waiting_time for r in Q.get_all_records()]
>>> sum(waits)/len(waits)
0.1691349404558...