#59 changed obj function to only return float

[jkriwet]

The objective functions of the calibrator returns a tuple: (total_res, unweighted objective).

ebcpy needs the obj function to return only a float, because the optimization algorithms need this.

Interestingly, this is almost not noticable. See the following code as an example:

from scipy.optimize import differential_evolution

def black_box_function(x, *args):
    """Function with unknown internals we wish to maximize.

    This is just serving as an example, for all intents and
    purposes think of the internals of this function, i.e.: the process
    which generates its output values, as unknown.
    """
    obj = -x[0] ** 2 - (x[1] - 1) ** 2 + 1
    print(f"Objective Value: {obj}")
    return (obj, 'Test')


# Bounded region of parameter space
pbounds = [(2, 4), (-3, 3)]

res = differential_evolution(
            func=black_box_function,
            bounds=pbounds,
            strategy="best1bin")

print(res)

The "black_box_function" is the equivalent of the obj function. When beeing evaluated during the optimization, this evaluates correctly (Observable prints of the values). Only at a specific time (in this case when a new population is to be created) there will be an error:

RuntimeError: func(x, *args) must return a scalar value

It is though noticable when implementing different optimization algorithms, which immediately cant handle tuple returns of the obj function. Therefore this has to be changed.

Closes #59

[FWuellhorst]

Thanks for noticing. You need to change the validation part as well. Maybe call the obj function using an arg with a new parameter verbose with the default being False? If this does not work maybe a class instance argument

[jkriwet]

Yes, this works.