Can't replace an array inside a tuple with another array of same size

I have a list of tuples:

I want to replace the second np.array of the first tuple with an array of the same length but filled with zeros:

I have tried the following:

Coeff[0][1] =np.zeros(len(Coeff[0][1]))

But I get the:

'tuple' object does not support item assignment

Any idea how to complete the replacement process ?

CodePudding user response：

Tuples are immutable in Python, you cannot change values stored in it.

You can do list(map(list, list_of_tuples)) for a quick casting into list, or [list(x) for x in list_of_tuples] if you want to use list comprehension.

I've done some ugly %timeit benchmark on my computer, both seems merely equivalent in speed.

CodePudding user response：

If you do not need to keep the original values in the arrays, then you could use numpy.nan_to_num() with copy=False to modify the second array in the tuple in-place. This has the advantage of not creating new arrays.

The following example replaces the second array of every tuple in Coeff with zeros.

import numpy as np

Coeff = [
    tuple(np.full((5, ), np.nan) for i in range(2))
    for j in range(5)
]
print(Coeff)

nil = [np.nan_to_num(tup[1], copy=False) for tup in Coeff]
print(Coeff)

The output is

[
    (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan])),
    (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan])),
    (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan])),
    (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan])), 
    (array([nan, nan, nan, nan, nan]), array([nan, nan, nan, nan, nan]))
]
[
    (array([nan, nan, nan, nan, nan]), array([0., 0., 0., 0., 0.])),
    (array([nan, nan, nan, nan, nan]), array([0., 0., 0., 0., 0.])),
    (array([nan, nan, nan, nan, nan]), array([0., 0., 0., 0., 0.])),
    (array([nan, nan, nan, nan, nan]), array([0., 0., 0., 0., 0.])),
    (array([nan, nan, nan, nan, nan]), array([0., 0., 0., 0., 0.]))
]

If you only need the work with the first tuple, you could use nil = np.nan_to_num(Coeff[0][1], copy=False). Note that I use nil to indicate you do not need the result of the function.