However the probabilities are often output as scalars, or as vectors only in E.
I suggest making the output universal: it should always be an array of shape (len(t), len(E)) (or scalar if both are scalars), even if the probability doesn't depend on E or t. This will make it easier for vectorized calculations, and more clear to the user.
Currently
FlavorTransformation.prob_eeand similar methods take time and energy as arguments. These arguments can be arrays of values: https://github.com/SNEWS2/snewpy/blob/08c3b84b7ef58d1830f52bf50275011b91301173/python/snewpy/flavor_transformation.py#L39-L52However the probabilities are often output as scalars, or as vectors only in
E.I suggest making the output universal: it should always be an array of shape
(len(t), len(E))(or scalar if both are scalars), even if the probability doesn't depend onEort. This will make it easier for vectorized calculations, and more clear to the user.