ValueError: could not broadcast input array from shape (3) into shape (2) simple solution

user_11692124 0
Viewed 5890

I have this np array and trying to add a number to one of just one of lines(trying to make asymmetric array if possible and if not a 100*3 array is also ok)

a=np.arange(100*2).reshape(-1,2)
a[40]=np.append(a[40],6)

note that a=np.arange(100*2).reshape(-1,2) is just simplified example and not real code that I wanna manipulate. and I receive this error

ValueError: could not broadcast input array from shape (3) into shape (2)

is there any simple solution(except making new array and filling it with loop with previous value then adding 6)?

numpy python
2 Answers

Would that be a solution to your problem?

import numpy as np

a = np.zeros((100, 3))
a[:,0:2] = np.arange(100*2).reshape(-1,2)

a[40,2]=6
user_86644160

The closest thing in numpy to a ragged array is an object dtype array:

In [475]: a = np.empty(2, object)                                                                    
In [476]: a                                                                                          
Out[476]: array([None, None], dtype=object)

If an element is a list, you can use its append to add a value:

In [477]: a[0] = [1,2]                                                                               
In [478]: a[1] = [2,3]                                                                               
In [479]: a[1].append(4)                                                                             
In [480]: a                                                                                          
Out[480]: array([list([1, 2]), list([2, 3, 4])], dtype=object)

But it's questionable whether such an array is any better that a list

In [481]: a.tolist()                                                                                 
Out[481]: [[1, 2], [2, 3, 4]]
user_9019250
Related