alogos._optimization.shared.evaluation

Classes

RobustCallable

Wrapper to make a function robust against failure and invalid return values.

Functions

default_evaluator(function, args)	Evaluate a given function on each item of a list.
get_robust_fitness_function(objective_function, objective)	Wrap a given function to make it robust.

---

Detailed object descriptions

alogos._optimization.shared.evaluation.default_evaluator(function, args)

Evaluate a given function on each item of a list.

alogos._optimization.shared.evaluation.get_robust_fitness_function(objective_function, objective)

Wrap a given function to make it robust.

class alogos._optimization.shared.evaluation.RobustCallable(func, def_fit, def_det)

Wrapper to make a function robust against failure and invalid return values.

It provides following guarantees:

• The return value is a tuple with two entries: (fitness, details)

• fitness is ensured to be of type float, allowing +Inf and -Inf, but excluding NaN.

• details can be any user-chosen type.

Notes

• Simply wrapping the function into another function would cause pickling errors.

• If the robust fitness function cannot be pickled, most parallel and distributed evaluation solutions cannot be applied since they rely on some pickling function.

References

• http://gael-varoquaux.info/programming/decoration-in-python-done-right-decorating-and-pickling.html

__init__(self, func, def_fit, def_det)

Create a robust callable object from a given function.