Splicing NumPy arrays

Question

I am having a problem splicing together two arrays. Let's assume I have two arrays:

a = array([1,2,3])
b = array([4,5,6])

When I do vstack((a,b)) I get

[[1,2,3],[4,5,6]]

and if I do hstack((a,b)) I get:

[1,2,3,4,5,6]

But what I really want is:

[[1,4],[2,5],[3,6]]

How do I accomplish this without using for loops (it needs to be fast)?

possible duplicate of [How do I iterate over the tuples of the items of two or more lists in Python?](http://stackoverflow.com/questions/1210396/how-do-i-iterate-over-the-tuples-of-the-items-of-two-or-more-lists-in-python) [how can I iterate through two lists in parallel in Python?](http://stackoverflow.com/questions/1663807/how-can-i-iterate-through-two-lists-in-parallel-in-python-closed) — Nathan Fellman, Jul 21 '10 at 18:24
I'm curious why you accepted Philipp's answer, when it was identical to, and submitted later than, Amber's answer? — mtrw, Jul 21 '10 at 18:53
@mtrw Because Amber modified his/her answer. He/she suggested vstack first — NullUserException, Jul 21 '10 at 19:17
@NullUser - Looking at the edit history, the edit only added the `vstack().T` option. — mtrw, Jul 21 '10 at 19:26
@mtrw That's because the answer was edited within the 5 minute "grace period." I saw the original. — NullUserException, Jul 21 '10 at 20:02
@NullUser - That would definitely explain it. Thanks, and I withdraw my question. — mtrw, Jul 21 '10 at 20:14
I actually suggested `dstack` first, not vstack; but then remembered that dstack adds in the 3rd dimension (even if stripping off one dimension then does give the correct answer). — Amber, Jul 21 '10 at 20:44

Amber · Answer 1 · 2010-07-21T18:30:05.800

7

Try column_stack()?

http://docs.scipy.org/doc/numpy/reference/generated/numpy.column_stack.html

Alternatively,

vstack((a,b)).T

edited Jul 21 '10 at 18:30

answered Jul 21 '10 at 18:24

Amber

507,862
82
626
550

This creates a three-dimensional array. – Philipp Jul 21 '10 at 18:29

score 4 · Accepted Answer · answered Jul 21 '10 at 18:27

4

column_stack.

answered Jul 21 '10 at 18:27

Philipp

48,066
12
84
109

score 0 · Answer 3 · edited Jun 18 '14 at 22:59

0

I forgot how to transpose NumPy arrays, but you could do:

at = transpose(a)
bt = transpose(b)

result = vstack((a,b))

edited Jun 18 '14 at 22:59

Peter Mortensen

30,738
21
105
131

answered Jul 21 '10 at 18:26

Nathan Fellman

122,701
101
260
319

Transposing one-dimensional arrays is an identity operation. – Philipp Jul 21 '10 at 18:29
Feel free to fix the syntax. Like I said, I forgot how to transpose a numpy array. – Nathan Fellman Jul 22 '10 at 19:00

score 0 · Answer 4 · answered Jul 21 '10 at 18:28

0

>>> c = [list(x) for x in zip(a,b)]
>>> c
[[1, 4], [2, 5], [3, 6]]

or

>>> c = np.array([list(x) for x in zip(a,b)])
>>> c
array([[1, 4],
       [2, 5],
       [3, 6]])

depending on what you're looking for.

answered Jul 21 '10 at 18:28

mtrw

34,200
7
63
71

1

This question is about NumPy arrays, not lists. – Amber Jul 21 '10 at 18:31
Yes, but it will most likely be slower than the proper NumPy functions. – Amber Jul 21 '10 at 19:33
@Amber - True. I had never stumbled across `column_stack` before, your answer is definitely better. – mtrw Jul 21 '10 at 19:39

score 0 · Answer 5 · answered Jul 21 '10 at 18:36

0

numpy.vstack((a, b)).T

answered Jul 21 '10 at 18:36

Cheery

24,645
16
59
83

score -1 · Answer 6 · edited Dec 15 '22 at 21:00

-1

You are probably looking for shape manipulation of the array. You can look in the "Tentative NumPy Tutorial, Array Creation".

edited Dec 15 '22 at 21:00

Glorfindel

21,988
13
81
109

answered Jul 21 '10 at 18:27

anijhaw

8,954
7
35
36

Splicing NumPy arrays

6 Answers6