Question

Say I make a weird little array:

>>> a = np.array([[[1,2,3],4],[[4,5,6],5]])
>>> a
array([[[1, 2, 3], 4],
       [[4, 5, 6], 5]], dtype=object)

And then take a the first column as a slice:

>>> b = a[:,0]
>>> b
array([[1, 2, 3], [4, 5, 6]], dtype=object)
>>> b.shape
(2,)

Say I now want to reshape b so that its shape is (2,3):

>>> b.reshape((-1,3))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: total size of new array must be unchanged

I presume that numpy is treating each array in b as an object rather than an array in and of itself. The question is, is there a good way of doing the desired resize?

Answer 1

In your particular example, you could use numpy.vstack :

import numpy as np


a = np.array([[[1,2,3],4],[[4,5,6],5]])
b = a[:,0]

c = np.vstack(b)
print c.shape # (2,3)

EDIT : Since your array a is not a real matrix but a collection of arrays (as pointed by wim ), you can also do the following :

   b = np.array([ line for line in a[:,0]])
   print b.shape #(2,3)