BUG: average_func returns incorrect frames

[kay-ro]

Description:

If nr=Noneˋ ornr>available frames` then one frame is missing.

Code for reproduction:

load amep
traj=amep.load.traj("test", skip=0.99999)
data=np.arange(0,100,5)
amep.utils.average_func(lambda t:t, data, nr=10)

Error message:

No response

Python and AMEP versions:

1.0.2

Additional information:

Solution:

if (nr==None or nr>...) then:
    Nr=int(N-skip*N+1)

How did you install AMEP?

None

[kay-ro]

proposed solution to be reviewed

    N = len(data)   # number of time steps

    if(nr == None or nr > N - skip * N):
        nr = max(1,int(N-skip*N))

    evaluated_indices = np.ceil(np.linspace(skip*N, N-1, nr,))
    func_result = [func(x, **kwargs) for x in tqdm(data[evaluated_indices])]
    evaluated = np.array(func_result)
    if indices:
        return evaluated, np.mean(evaluated, axis=0), evaluated_indices
    return evaluated, np.mean(evaluated, axis=0)

[bemde42]

thanks, there is a small additional fix necessary:

    evaluated_indices = np.ceil(np.linspace(skip*N, N-1, nr,))

must be changed to

evaluated_indices = np.array(np.ceil(np.linspace(skip*N, N-1, nr,)), dtype=int)