With Numpy, I thought I could create a matrix this way
z = np.array( [np.linspace(0, 1, 2), np.ones((1, 2)), np.ones((1, 2))] )
which, however, gives me:
array([array([ 0., 1.]), array([[ 1., 1.]]), array([[ 1., 1.]])], dtype=object)
then checking its shape:
>>> z.shape
(3,)
The column dimension does not exist. So I think I got an array with three array objects.
How can I have the 3 x 2 matrix using linespace() and ones() here?
If the inputs are 1-dimensional, you could use numpy.array:
np.array([np.linspace(0, 1, 2), np.ones((2,)), np.ones((2,))])
yields
array([[ 0., 1.],
[ 1., 1.],
[ 1., 1.]])
Note that np.ones((1,2)) has a 2-dimensional shape (1,2), while np.linspace(0, 1, 2) has a 1-dimensional shape (2,). To create the desired result with np.array, the arrays need to have compatible shapes -- in this case, that means using the 1-dimensional array np.ones(2,) instead of np.ones((1,2)).
If you must use np.ones((1,2)), then you could instead make np.linspace(0, 1, 2) 2-dimensional as well, and then use numpy.vstack:
np.vstack([np.linspace(0, 1, 2)[np.newaxis, :], np.ones((1,2,)), np.ones((1,2,))])
yields
array([[ 0., 1.],
[ 1., 1.],
[ 1., 1.]])
(2,)and(1,2)) that produced the 1d object array.