Indexing in multidimensional arrays in numpy library

Question

>>> y = np.arange(35).reshape(5,7)
>>> y[1:5:2,::3]
array([[ 7, 10, 13],
       [21, 24, 27]])

What does y[1:5:2,::3] mean? In detail.

Stack Overflow is not a replacement for reading the tutorial — Mad Physicist, Apr 17 '20 at 18:58
No, but it is [here](https://numpy.org/devdocs/user/basics.indexing.html). — roganjosh, Apr 17 '20 at 19:01
But, FWIW, I find it difficult to unpack what it's doing from just viewing it. Hopefully the code has good reason for using this — roganjosh, Apr 17 '20 at 19:03
It's take every other row starting with `1`, e.g. 1 and 3. And every third column, 0,3,6. — hpaulj, Apr 17 '20 at 20:07

azro · Accepted Answer · 2020-04-17T18:57:50.387

You can find the details of Python slicing notation here.

Your case combines slicing notation with numpy notation : y[1:5:2,::3] is leans slicing 1:5:2 in 1st dimension, and ::3 in 2nd dimension

# Initial array
[[ 0  1  2  3  4  5  6]
 [ 7  8  9 10 11 12 13]
 [14 15 16 17 18 19 20]
 [21 22 23 24 25 26 27]
 [28 29 30 31 32 33 34]]

1:5:2 take from values [1;5[ and one over 2, this, in the first dimension, so it keeps values 1 and 3 (you can say rows)
```
[[ 7  8  9 10 11 12 13]
 [21 22 23 24 25 26 27]]
```
::3 takes all elements are the 2 first values are not provided, but just one over three, in the second dimension
```
[ 7  8  9 10 11 12 13] => [ 7 10 13] # one over 3
```

I have no idea what this means. I think the language needs to be a lot clearer — roganjosh, Apr 17 '20 at 18:57

1 Answers1