I'm currently padding a multi-dimensional array (to allow for transposition among other things), and while I've found a range of ways of doing this with loops (np.pad and row by row assignment to a size controlled empty array) I can't seem to vectorized this operation.
MWE
a = np.arange(1,3)
b = np.arange(3,7)
c = np.arange(7,10)
source = np.array([a,b,c])
This has the form:
>>> array([array([1, 2]), array([3, 4, 5, 6]), array([7, 8, 9])], dtype=object)
My desired output is:
>>> array([[ 1., 2., 1., 1.],
[ 3., 4., 5., 6.],
[ 7., 8., 9., 1.]])
In this trivial case I used:
desired = np.array([np.concatenate([a,np.ones(2)]),b,np.concatenate([c,np.ones(1)])])
(my belief is that the above wouldn't scale well as you would have to store the length of each array - which is also what stopped me using np.pad)
