changing array value one at a time

Question

I am trying to flip a binary array one at a time.

import numpy as np
k = np.array([0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1])

for example my out put should be like this;

[1, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1] # 1st output
[0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1] # 2nd output
[0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1] # 3rd output
[0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1] # 4th output
[0, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1] # 5th output

In the first output, I want to flip only the first element of the array (other elements do not change), in the second output, the second element should change (the 1st and the remaining elements should not change).. etc. Could anyone tell me how can flip one at a time? Thank you

• IanQ: What have you tried so far?

Answer 1

What you're describing is flipping the diagonal of a tiled version of your array. By stacking your array, you can operate on the entire array at once, using vectorized operations, rather than operating on each row individually.

Setup

arr = np.tile(k, 5).reshape(-1, k.shape[0])

array([[0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1]])

---

Using numpy.diag_indices:

x, y = np.diag_indices(arr.shape[0])
arr[x, y] = 1 - arr[x, y]

array([[1, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1],
 [0, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1]])

• Mass17: Thanks @user3483203. Very helpful technique!!

Answer 2

You can use a generator to save memory and even time on big arrays :

k=np.array([0, 0, 1])
def flip_one(k):
 k[0]=1-k[0]
 yield k
 for i in range(len(k)):
 k[i:i+2]=1-k[i:i+2]
 yield k
for f in flip_one(k) : 
 print (f) # or other useful things!
#[1 0 1]
#[0 1 1]
#[0 0 0]
#[0 0 1]

k is reset at the end of the loop.

• Mass17: Thanks @B. M.. very helpful!!