Controlling randomness

@pseudo_random(*, seed=0, evolutive=True, input_dependent=False, loop=0, make_key=repr)

Decorator Factory to control randomness of decorated function. The random seed of random and np.random are set depending on the function name, it's input arguments, and the number of time it has been called. Equality of calls is computed using repr. Inputs with different representations will be considered different. When called on a class method, it must be decorated with @method decorator included in the package. The parameters to pseudo_random() are:

• evolutive (bool): random seed is incremented each time the function is called. The counter is different for each set of arguments.
• input_dependent (bool): initial random seed depends on function's input.
• loop (int): get repeating input every loop iteration when evolutive is True. Else, loop argument is ignored.
• make_key: a function to compute the key based on the function name and it's arguments: (f.__name__, args, kwargs). Default is the builtin repr.

Example

Basic usage:

@pseudo_random(input_dependent=input_dependent, evolutive=evolutive)
def get_random_number(*args):
    return random.randint(0,9)
argument = '0'
print(" ".join([str(get_random_number(argument)) for _ in range(5)]))

Called 5 times with 5 different inputs will yield:

evolutive: True      input_dependent: True      (5 iterations)
-----------------------------------------------------------------
with argument='0':	3 0 4 2 5
with argument='1':	0 7 3 3 4
with argument='2':	8 3 3 4 9
with argument='3':	1 6 8 0 2
with argument='4':	4 0 2 9 1

evolutive: False     input_dependent: True      (5 iterations)
-----------------------------------------------------------------
with argument='0':	3 3 3 3 3
with argument='1':	0 0 0 0 0
with argument='2':	8 8 8 8 8
with argument='3':	1 1 1 1 1
with argument='4':	4 4 4 4 4

evolutive: True      input_dependent: False     (5 iterations)
-----------------------------------------------------------------
with argument='0':	6 2 0 3 3
with argument='1':	6 2 0 3 3
with argument='2':	6 2 0 3 3
with argument='3':	6 2 0 3 3
with argument='4':	6 2 0 3 3

evolutive: False     input_dependent: False     (5 iterations)
-----------------------------------------------------------------
with argument='0':	6 6 6 6 6
with argument='1':	6 6 6 6 6
with argument='2':	6 6 6 6 6
with argument='3':	6 6 6 6 6
with argument='4':	6 6 6 6 6

Citation

If you use this repository, please consider citing my work.