[FIXED] Setting a new value for a numpy array item saves a much smaller value. How to just set the value?

Issue

Code:

import numpy as np
data = '12345\n54321\n13542\n12354\n53124'
n = data.find('\n')
mat = np.array(list(data.replace('\n','')), dtype=np.uint8).reshape(-1, n)
mat

Out:

array([[1, 2, 3, 4, 5],
       [5, 4, 3, 2, 1],
       [1, 3, 5, 4, 2],
       [1, 2, 3, 5, 4],
       [5, 3, 1, 2, 4]], dtype=uint8)

Code:

mat2 = np.full_like(mat, 0)
mat2[0,0] = 9999999
mat2[0,1] = 8888
mat2[0,2] = 77777

Out:

array([[127, 184, 209,   0,   0],
       [  0,   0,   0,   0,   0],
       [  0,   0,   0,   0,   0],
       [  0,   0,   0,   0,   0],
       [  0,   0,   0,   0,   0]], dtype=uint8)

Thus, Python does not set the needed numpy array item value, only a random and tiny value gets saved instead for a wide range of number sizes. How to fix this?

Solution

Sharing this since I did not know for some time what was going on, even though it seems so clear at first sight if you look at a small example like this. In a bigger code, such things can hide for a while.

You need to change the data type from uint8 (--> unsigned int, and 8 Bits is the least you need to choose for an array, that is why I chose it for some small values in the first place) to a data type that can save the larger values. For example:

mat2 = np.full_like(mat, 0).astype('uint32')
mat2[0,2] = 77777
mat2

Out:

array([[    0,     0, 77777,     0,     0],
       [    0,     0,     0,     0,     0],
       [    0,     0,     0,     0,     0],
       [    0,     0,     0,     0,     0],
       [    0,     0,     0,     0,     0]], dtype=uint32)

Answered By - questionto42