cma.fitness_transformations.NoisyFitness

class documentation

class NoisyFitness(Function):

Constructor: NoisyFitness(fitness_function, rel_noise, abs_noise)

apply noise via f += rel_noise(dim) * f + abs_noise(dim)

Method	`__call__`	Undocumented
Method	`__init__`	attach relative and absolution noise to `fitness_function`.
Instance Variable	`abs_noise`	Undocumented
Instance Variable	`rel_noise`	Undocumented

Inherited from Function:

Method	`initialize`	initialization of `Function`
Instance Variable	`evaluations`	Undocumented
Instance Variable	`ftarget`	Undocumented
Instance Variable	`target_hit_at`	Undocumented
Property	`function_names_to_evaluate_first_found`	attributes which are searched for to be called if no function was given to `__init__`.
Class Variable	`_function_names_to_evaluate_first_found`	Undocumented
Instance Variable	`__callable`	Undocumented
Instance Variable	`__initialized`	Undocumented

def __call__(self, x, *args): ¶

overrides cma.fitness_transformations.Function.__call__

Undocumented

def __init__(self, fitness_function, rel_noise=(lambda dim: 1.1 * np.random.randn() / dim), abs_noise=(lambda dim: 1.1 * np.random.randn())): ¶

overrides cma.fitness_transformations.Function.__init__

attach relative and absolution noise to fitness_function.

Relative noise is by default computed using the length of the input argument to fitness_function. Both noise functions take dimension as input.

>>> import cma
>>> from cma.fitness_transformations import NoisyFitness
>>> fn = NoisyFitness(cma.ff.elli)
>>> assert fn([1, 2]) != cma.ff.elli([1, 2])
>>> assert fn.evaluations == 1

abs_noise = ¶

Undocumented

rel_noise = ¶

Undocumented