Numpy Append 数据到空数组

Question

I'm trying to write a code that will add 2 arrays(element by element) and store them in a 3rd array.

Basic Logic:

arr3[i] = arr1[i] + arr2[i]

For this, I have created two arrays arr1 and arr2 . The result of the sum of arr1 and arr2 is getting appended in an empty array arr3 .

code:

from numpy import append, array, int8


arr1 = array([1,2,3,4,5])

arr2 = array([2,4,6,8,10])

len = max(arr1.size,arr2.size)

arr3 = array([],dtype=int8)

for i in range(len):
    append(arr3,arr1[i]+arr2[i])
    print(arr1[i]+arr2[i])
    print(arr3[i])

print(arr3)

In this code, I'm able to refer to elements of arr1 and arr2 and add them, but I'm not able to append the data to arr3.

Can anyone please help me to understand, what is the mistake in the code due to which I'm not able to store the data to arr3?

Answer 1

You can simply use

arr3 = arr1 + arr2

The reason why your code doesn't work is that append doesn't mutate the array, but returns a new one. You can simply modify your code like this:

for i in range(len):
    arr3 = append(arr3,arr1[i]+arr2[i])

Answer 2

Use arr3.append .

from numpy import append, array, int8


arr1 = array([1,2,3,4,5])

arr2 = array([2,4,6,8,10])

len = max(arr1.size,arr2.size)

arr3 = []

for i in range(len):
    arr3.append(arr1[i]+arr2[i])
    print(arr1[i]+arr2[i])
    print(arr3[i])


print(arr3)

Answer 3

First things first

Do Not Use built-in function name as variable .

len is a built-in function in python.

@sagi's answer is right. Writing the for loop would mean your code is not time-optimized.

But if you still want to understand where your code went wrong, check array shape

import numpy as np
arr3 = np.array([],dtype=int8)
print (arr3.shape)
>>> (0,) 

Maybe you can create an empty array of the same shape as arr1 or arr2 . Seems like for your problem they have same dimension.

arr3 = np.empty(arr1.shape, dtype=arr1.dtype)
arr3[:] = arr1 + arr2

If you are still persisting to use the dreaded for loop and numpy.append then use this--

arr3 = np.empty(arr1.shape, dtype=arr1.dtype)
for i in range(len(arr1)):
   np.append(arr3, arr1[i] + arr2[i])
print(arr3)
>>> array([ 3,  6,  9, 12, 15])

Cheers, good luck!!

Answer 4

This could give indexing errors:

max(arr1.size,arr2.size)

if the arrays differ, range over this would produce index values that are too large for the smaller array.

The straight forward way of summing the 2 arrays is

In [79]: arr1 = np.array([1,2,3,4,5])
    ...: arr2 = np.array([2,4,6,8,10])
In [80]: arr1+arr2
Out[80]: array([ 3,  6,  9, 12, 15])

It makes optimal use of the numpy array definitions, and is fastest.

If you must iterate (for example because you want to learn from your mistakes), use something like (which is actually better if the inputs are lists, not arrays):

In [86]: alist = []
    ...: for x,y in zip(arr1,arr2):
    ...:     alist.append(x+y)
    ...: 
In [87]: alist
Out[87]: [3, 6, 9, 12, 15]

or better yet as a list comprehension

In [88]: [x+y for x,y in zip(arr1,arr2)]
Out[88]: [3, 6, 9, 12, 15]

I'm using zip instead of the arr1[i] types of range indexing. It's more concise, and less likely to produce errors.

np.append , despite the name, is not a list append clone. Read, if necessary reread, the np.append docs.

append : ndarray
        A copy of `arr` with `values` appended to `axis`.  Note that
        `append` does not occur in-place: a new array is allocated and
        filled.  If `axis` is None, `out` is a flattened array.

This does work, but is slower:

In [90]: arr3 = np.array([])
    ...: for x,y in zip(arr1,arr2):
    ...:     arr3 = np.append(arr3,x+y)
    ...: 
In [91]: arr3
Out[91]: array([ 3.,  6.,  9., 12., 15.])

I would like to remove np.append , since it misleads far too many beginners.

Iteration like this is great for lists, but best avoided when working with numpy arrays. Learn to use the defined numpy operators and methods, and use elementwise iteration only as last resort.