Random generator parametrization

The seed of the pseudo random generator is initialized to 0 when the module is imported.

It means the same script will yield the same results across several replications.

It is also possible to initialize the random generator differently:

• by setting a seed

• by setting the complete generator state

import openturns as ot
import os
import time

ot.Log.Show(ot.Log.NONE)

Example 0: using a fixed seed

ot.RandomGenerator.SetSeed(77)

Example 1: using the python process id

ot.RandomGenerator.SetSeed(os.getpid())

Example 2: using the time in milliseconds

ot.RandomGenerator.SetSeed(int(1000 * time.time() % 1e9))

Example 3: using a previously saved generator state

# use the random generator
ot.Normal().getSample(100)
# save the generator state
particularState = ot.RandomGenerator.GetState()
ot.Normal().getRealization()

class=Point name=Unnamed dimension=1 values=[-0.452865]

load the generator state

ot.RandomGenerator.SetState(particularState)
ot.Normal().getRealization()

class=Point name=Unnamed dimension=1 values=[-0.452865]