DeepCopy the values of an numpy array inside a for loop

Question

I have a numpy array called initial_matrix and inside a for-loop in each step of the loop I want to randomly change one of the indexes of that array (every time a different one) and the process my matrix.

init_matrix = deepcopy(my_matrix)
for i in range(0, 30):

       new_mat, rep_index = replace_index(my_matrix, value)
       ...// proceed with my process

The function replace_index, randomly reaplces one index of the matreix with the value. I noticed that the results of the replacement it does not occur only to new_mat but also to my_matrix. However, I want in each step of the loop to use my initial matrix and not the one after the replacement of the array. How can I do so?

And? Did my solution solve your problem?

JE_Muc
– JE_Muc

2018-06-08 14:31:59 +00:00
Commented Jun 8, 2018 at 14:31 — JE_Muc
– JE_Muc, Commented Jun 8, 2018 at 14:31

JE_Muc · Accepted Answer · 2018-06-08 13:29:22Z

1

If you do not want to modify your current function (my_matrix, value), you can just pass a copy to this function with:

new_mat, rep_index = replace_index(my_matrix.copy(), value)

A short explanation:
This issue is generally occurring in python, whenever you pass a mutable into a function. Mutables, like np.ndarray, list or dict will be passed to a function by reference. This means all operations performed on them can be seen in the outer scope, as long as you don't rebind the reference. Rebinding operations are typically operations which change the memory address of the object, like appending, concatenating, copying, simply overwriting with a new object etc..
For an array a = np.zeros(3) also an operation like a = [1., 2., 3.] rebinds the reference (and the new type will obviously be a list) , while a[:] = [1., 2., 3.] does not, since a[:] = accesses the underlying memory scope for assigning the values, whereas a = ... does rebind the reference of a.

edited Jun 8, 2018 at 13:29

answered Jun 8, 2018 at 13:08

JE_Muc

5,8323 gold badges30 silver badges49 bronze badges

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

3 Comments

konstantin Over a year ago

Is there any tutorial that I can dig into that issue? Why exactly the values are changing?

JE_Muc Over a year ago

Yeah, there are lots of tutorials for that issue, in general here in the Scipy Cookbook and the numpy slicing and indexing documentation or more specific on views/slicing in numpy.ndarray.view.

JE_Muc Over a year ago

Oh btw.: These documentations are specifically for arrays, but this issue is occurring whenever you pass a mutable to a function!

mukul puspam · Accepted Answer · 2018-06-08 12:51:02Z

1

In replace_index() make a copy of my_matrix and use the copy.

answered Jun 8, 2018 at 12:51

mukul puspam

739 bronze badges

1 Comment

konstantin Over a year ago

Is there any tutorial that I can dig into that issue? Why exactly the values are changing?

Collectives™ on Stack Overflow

DeepCopy the values of an numpy array inside a for loop

2 Answers 2

3 Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related