I read the question above in that for a given ZDD, a set representation is to be found, and then a RMNF should be derived for the represented formula. If b means that we should interpret the decision diagram as a FDD, then we do no longer speak about the same formula Phi, right?