Algorithm to generate pseudorandom numbers serie, reproducible knowing initial seed

Question

I have to write a lightweight algorithm to generate pseudorandom infinite numbers series, and strongly dependant by initial seed.

In python it should be like

seed = 3345                 // Common number
generator = numgen(seed)
while True:
    generator.getNext()     // With the same seed it produce same numbers

As I wrote, with same seed it has to produce the same number series, even in different machines and in different times. Is there a standard patten for this or do I have to implement my own algorithm?

Answer 1

With the given requirements, you can go for the constant series

generator = itertools.repeat(seed)

This is

• lightweight
• not necessarily random
• strongly dependant on the initial seed
• produces the same series even on a different machine

Edit: To make this an actual answer, here is how to use Python's Mersenne Twister to generate a series of random numbers:

seed = 3345
maximum = 1000000
generator = random.Random(seed)
while True:
    print generator.random(maximum)

Creating your own random.Random instance ensures that no calls to the random number generator from elsewhere interfere with your random state.