Reshape function broadcasting from wrong numpy shape

Question

I am trying to reshape a numpy.array from shape (4,4) to shape (2,2,2,2). The error I am receiving is:

ValueError: could not broadcast input array from shape (2,2,2,2) into shape (4,4).

This makes me think I have the shapes backwards, but after checking that is not the case.

I have a user defined function that uses np.reshape to reshape an array to a certain shape if it is not already that shape. I have tried eliminating the user defined function and using only np.reshape, but it returned the same error. What am I missing?

The reshape function:

def reshape(matrix, ports, modes):
    shape = (ports, ports, modes, modes)
    if(np.shape(matrix) != shape): #reshape if necessary
        return np.reshape(matrix, shape)
    else:
        return matrix

Where I call this function:


def plot(S, F, ports, modes, x_range, y_range, title, f_units, 
         multi_modal = True):
    data = {} #create dictionary to store S-parameters
    if(not multi_modal): #if we want average
        for f in range(0, len(F)): #iterate through frequencies
            print(np.shape(S[f]))
            S[f] = reshape(S[f], ports, modes)

In this instance, ports = 2 and modes = 2.

S[f] is a np.array of shape (4,4):

[[ 1.00000000e+00+0.00000000e+00j -1.02728868e-19+1.64952184e-22j
  -1.37762998e-20+2.40441793e-24j -4.18063430e-24-1.18287261e-21j]
 [ 0.00000000e+00-0.00000000e+00j -1.00000000e+00+1.22464680e-16j
   3.03393173e-26-1.77961140e-24j  1.57277027e-25+2.06062998e-23j]
 [-1.95100984e-27+3.66506948e-24j  2.38762635e-25+1.48052807e-22j
   1.00000000e+00+0.00000000e+00j  2.90518731e-20+1.33913685e-17j]
 [-3.47614015e-25-4.08540212e-23j -3.30653510e-21+2.87402660e-23j
   1.77338192e-21+2.27000073e-19j -1.00000000e+00+1.22464680e-16j]]

Why is it returning the error:

ValueError: could not broadcast input array from shape (2,2,2,2) into shape (4,4)

when it should be reshaping from (4,4) to (2,2,2,2)?

your problem is in the way you try to reassign the new shape to the old array — Gio

Gio Gio · Accepted Answer · 2019-04-09T13:43:05

This would do

def plot(S, F, ports, modes, x_range, y_range, title, f_units, 
         multi_modal = True):
    data = {} #create dictionary to store S-parameters
    if(not multi_modal): #if we want average
        for f in range(0, len(F)): #iterate through frequencies
            print(np.shape(S[f]))
            S_reshaped = reshape(S[f], ports, modes)

If you need to store your results you can create an empty list and then append the reshaped array to it.

def plot(S, F, ports, modes, x_range, y_range, title, f_units, 
         multi_modal = True):
    data = {} #create dictionary to store S-parameters
    S_reshaped=[]
    if(not multi_modal): #if we want average
        for f in range(0, len(F)): #iterate through frequencies
            print(np.shape(S[f]))
            S_reshaped.append(reshape(S[f], ports, modes))

Reshape function broadcasting from wrong numpy shape

1 Answers