signal.ss2tf doesn't handle zero-order state-space models

[arvoelke]

Take a scalar (zero-order) transfer function and map it to state-space:

tf = (2, 1)
ss = scipy.signal.tf2ss(*tf)
print ss

We get just a pass-through, as expected:

(array([], dtype=float64), array([], dtype=float64), array([], dtype=float64), array([ 2.]))

However, if we now try to map this back to a transfer function, we get an error due to the oddly shaped A, B, C matrices.

tf_check = scipy.signal.ss2tf(*ss)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-6-4db87808b9b6> in <module>()
      5 print ss
      6 
----> 7 scipy.signal.ss2tf(*ss)

c:\users\aaron\appdata\local\enthought\canopy\user\lib\site-packages\scipy\signal\ltisys.pyc in ss2tf(A, B, C, D, input)
    272 
    273     # Check consistency and make them all rank-2 arrays
--> 274     A, B, C, D = abcd_normalize(A, B, C, D)
    275 
    276     nout, nin = D.shape

c:\users\aaron\appdata\local\enthought\canopy\user\lib\site-packages\scipy\signal\ltisys.pyc in abcd_normalize(A, B, C, D)
    206 
    207     A, B, C, D = map(_none_to_empty_2d, (A, B, C, D))
--> 208     A = _restore(A, (p, p))
    209     B = _restore(B, (p, q))
    210     C = _restore(C, (r, p))

c:\users\aaron\appdata\local\enthought\canopy\user\lib\site-packages\scipy\signal\ltisys.pyc in _restore(M, shape)
    163     else:
    164         if M.shape != shape:
--> 165             raise ValueError("The input arrays have incompatible shapes.")
    166         return M
    167 

ValueError: The input arrays have incompatible shapes.