python: how to insert elements in numpy array in a loop

Question

I have a dictionary of the following format:

{1: 7, 2: 1, 4: 2, 5: 1, ...}

The dictionary can be potentially very large, so I decided to use numpy library for more efficient work with vectors.

Now I need to insert values from the above dictionary in a numpy vector (1-dimensional array) at the key index. So tried naively the following:

import numpy as np

d = {1: 7, 2: 1, 4: 2, 5: 1}
T = np.zeros(shape=(n), dtype=np.int16)

for k, v in d.items():
    N = np.insert(T, k, v)

print(N)

However insert method always returns new array, and overwrites N in my example. So what is the right way to change array elements, just as I'd do in C, e.g. T[0] = 1.

Looks like numpy does not allow this?

Thanks.

UPDATE As suggested by Divakar, the solution is to assign multiple values at once:

T[list(d.keys())] = list(d.values())

And this does not require a loop.

@Divakar, thanks a lot. This is indeed very elegant solution, Where does it say that numpy array object can take a list as parameter? — Mark
– Mark, Commented Nov 29, 2019 at 16:30
Not sure how you were lead to numpy array object taking a list as parameter. It's a simple integer based indexding - docs.scipy.org/doc/numpy/reference/… — Divakar
– Divakar, Commented Nov 29, 2019 at 16:32
@Divakar, I'm newbie in Python. In your comment you did list(d.keys()) which returns a list of keys from dictionary, and then submit it as index to T object. — Mark
– Mark, Commented Nov 29, 2019 at 16:44

hpaulj · Accepted Answer · 2019-11-29 17:29:13Z

In numpy np.insert makes a new array by splitting the original array and inserting one or more values someplace in the middle.

In [193]: arr = np.arange(6)                                                    
In [194]: arr                                                                   
Out[194]: array([0, 1, 2, 3, 4, 5])
In [195]: np.insert(arr, 2, 100)                                                
Out[195]: array([  0,   1, 100,   2,   3,   4,   5])

Sometimes people talk about adding a value to an array, but `+' is elementwise numeric addition

In [196]: arr + 10                                                              
Out[196]: array([10, 11, 12, 13, 14, 15])
In [197]: 'one '+'two'        # string 'addition'                                                          
Out[197]: 'one two'

'assign a value(s)' is probably the clearest description of what you want to do. You want to modify an element of the original array, without changing its size.

In [198]: arr[3] = 100                                                          
In [199]: arr                                                                   
Out[199]: array([  0,   1,   2, 100,   4,   5])

@Divakar's idea assigns multiple values at once:

In [200]: arr[[1,2]] = [100,200]                                                
In [201]: arr                                                                   
Out[201]: array([  0, 100, 200, 100,   4,   5])

Collectives™ on Stack Overflow

python: how to insert elements in numpy array in a loop

1 Answer 1

Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Related