Hi!  Is this a bug:

>>> np.multiply.accumulate([2, 3, 5], dtype='f')
array([  2.,   6.,  30.], dtype=float32)

>>> np.multiply.accumulate([2, 3, 5], dtype='g')
array([0.0, -2, -3.1050362e+231], dtype=float96)

>>> np.multiply.accumulate([2, 3, 5], dtype='F')
array([  2.+0.j,   6.+0.j,  30.+0.j], dtype=complex64)

>>> np.multiply.accumulate([2, 3, 5], dtype='G')
array([0.0+0.0j, -2+0.0j, -3.1050362e+231+0.0j], dtype=complex192)

If not, why not (i.e., what am I missing)?