Strange assignment in numpy arrays

Question

I have a numpy array A with n rows of size 3. Each row is composed by three integers, each one is a integer which refers to another position inside the numpy array. For example If I want the rows refered by N[4], I use N[N[4]]. Visually:

N = np.array([[2, 3, 6], [12, 6, 9], [3, 10, 7], [8, 5, 6], [3, 1, 0] ... ])
N[4] = [3, 1 ,0]
N[N[4]] = [[8, 5, 6]
           [12, 6, 9]
           [2, 3, 6]]

I am building a function that modifies N, and I need to modify N[N[x]] for some specified x which is a parameter too (4 in the example). I want to change all the 6 in the subarray for another number (let's say 0), so I use numpy.where to find the indexes, which are

where_is_6 = np.where(N[N[4]] == 6)

Now, if I replace directly like N[N[4]][where_is_6] = 0 there is no change. If I make a previous reference like var = N[N[4]] and then var[where_is_6] the change is done but locally to the function and N is not changed globally. What can I do in this case? or what am I doing wrong?

The indexing with another indexing won't work with the way you have defined your list N. You need to have N converted to numpy.array() first. — Pacific Stickler
– Pacific Stickler, Commented Sep 10, 2014 at 21:45
I said it was a numpy array, the code although doesn't reflect that, editing right now... — Alejandro Sazo
– Alejandro Sazo, Commented Sep 10, 2014 at 21:46

perimosocordiae · Accepted Answer · 2014-09-10 21:38:37Z

6

Sounds like you just need to convert the indices to the original N's coordinates:

row_idxs = N[4]
r,c = np.where(N[row_idxs] == 6)
N[row_idxs[r],c] = 0

answered Sep 10, 2014 at 21:38

perimosocordiae

17.9k14 gold badges64 silver badges76 bronze badges

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

1 Comment

Alejandro Sazo Over a year ago

Yes! Thanks! I'll be thinking in that way next time I need to modify things.

Saullo G. P. Castro · Accepted Answer · 2014-09-10 21:35:25Z

4

The problem is that N[N[4]] is a new array, which you can check doing:

print(N[N[4]].flags)

C_CONTIGUOUS : True
F_CONTIGUOUS : False
OWNDATA : True
WRITEABLE : True
ALIGNED : True
UPDATEIFCOPY : False

where OWNDATA shows this fact.

answered Sep 10, 2014 at 21:35

Saullo G. P. Castro

59.4k28 gold badges191 silver badges244 bronze badges

2 Comments

Alejandro Sazo Over a year ago

Thanks for the .flags tip! Then what can I do to make the change?

Saullo G. P. Castro Over a year ago

@AlejandroSazo well... perimosocordiae's answer seems to be a good solution

Collectives™ on Stack Overflow

Strange assignment in numpy arrays

2 Answers 2

1 Comment

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related