fixed issue 18

Author: oleksandr-pavlyk

mkl_fft/_numpy_fft.py

@@ -1039,10 +1039,10 @@ def rfftn(a, s=None, axes=None, norm=None):
     if unitary:
         a = asarray(a)
         s, axes = _cook_nd_args(a, s, axes)
-        n_tot = numpy.prod([ s[ai] for ai in axes])
 
     output = mkl_fft.rfftn_numpy(a, s, axes)
     if unitary:
+        n_tot = prod(asarray(s, dtype=output.dtype))
         output *= 1 / sqrt(n_tot)
     return output