How can numpy array indexing techniques can give different outputs for same input code?

Question

#Code

import numpy as np    
np.random.seed(124)
x_2d = np.random.randint(1,50,(3,4,5))
print("I am getting wrong output => {}".format(x_2d[0][:][1]))
print("This is what I want => {} ".format(x_2d[0,:,1]))

# Code Ended

# Output for above code

I am getting wrong output => [42  1 21 29 15]
This is what I want => [29  1 22 49]

I am new to NumPy so I was just experimenting on numpy array selection techniques. I came to know that we can use square bracket method or comma method. But I encountered a problem. I am trying to extract column index 1 of array 0. But I am getting different outputs when I use both techniques. I have attached code snippets and outputs. Can anyone guide me where did I go wrong ?

Mephy · Accepted Answer · 2021-04-15 16:10:36Z

2

x[0][:][1] is equivalent to x[0, 1, :], which is also equivalent to x[0][1].

Reason being [:] alone basically means "copy this array", both in basic Python as well as Numpy. So you can read x[0][:][1] as "take the first element of array x, make a copy of it, then take the second element of the result."

[:] does not mean "skip a dimension".

answered Apr 15, 2021 at 16:10

Mephy

3,0214 gold badges27 silver badges32 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

bananaquit · Accepted Answer · 2021-04-15 18:03:35Z

When you call x_2d[0][:][1], first you get the first matrix x_2d[0]:

>>> first = x_2d[0]
>>> first
array([[15, 29, 18,  8,  3],
       [42,  1, 21, 29, 15],
       [22, 22, 28, 10, 31],
       [47, 49, 41, 10, 10]])

When you call first[:] you gonna receive exactly first, since you asking for all lines in first:

>>> second = first[:]
>>> second
array([[15, 29, 18,  8,  3],
       [42,  1, 21, 29, 15],
       [22, 22, 28, 10, 31],
       [47, 49, 41, 10, 10]])

So when you get second[1] you get the line of index 1

>>> third = second[1]
>>> third
array([42,  1, 21, 29, 15])

However when you ask for x_2d[0,:,1], numpy interprets it as:

"From matrix 0 give me column 1 from the matrix composed by lines 0,1,2,3)"

So if you ask for x_2d[1,0:2,3] the result will be [18, 2].

Conclusion: numpy does not interpret x[0,:,1] in the same way of x[0][:][1]. You can read more in the NumPy documentation here.

Collectives™ on Stack Overflow

How can numpy array indexing techniques can give different outputs for same input code?

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related