CODEWITHSUNDEEP
pythonarrayspython-3.xnumpymatrix
Tymur Dadsahov
Tymur Dadsahov

Reputation: 83

Add elements of array to each column of matrix using numpy

I have a matrix 3x3 and array with 3 values and I want to add each value of my array to each column of the matrix, so for example, if I have matrix like:

[[1, 2, 3],
 [4, 5, 6],
 [7, 8, 9]]

and array [1, 2, 3] I want to get a result like

[[2, 3, 4],
 [6, 7, 8],
 [10, 11, 12]]

But now when I'm trying to add my array to matrix it adds it by columns, so I get this:

[[2, 4, 6],
 [5, 7, 9],
 [8, 10, 12]]

And I can not change the axis to add operation or find method to do such calculation. Maybe I need to do it in a few steps? Or I just missed something?

Upvotes: 2

Views: 1398

Answers (4)

Sayandip Dutta
Sayandip Dutta

Reputation: 15872

Use transpose:

>>> arr1 = np.array([[1, 2, 3],
                    [4, 5, 6],
                    [7, 8, 9]])
>>> arr2 = np.array([1, 2, 3])

>>> (arr1.T + arr2).T
array([[ 2,  3,  4],
       [ 6,  7,  8],
       [10, 11, 12]])

Or make arr2 a column matrix by expanding dimension:

>>> arr1 + arr2[:,None]
array([[ 2,  3,  4],
       [ 6,  7,  8],
       [10, 11, 12]])

If both are lists:

>>> arr1 = [[1, 2, 3],
            [4, 5, 6],
            [7, 8, 9]]
>>> arr2 = [1, 2, 3]
np.add(arr1, np.expand_dims(arr2, 1))

References:

  1. Transpose: np.ndarray.T
  2. Expand dimensions: np.expand_dims
  3. np.add

Upvotes: 2

Michael
Michael

Reputation: 2367

You can use the numpy.newaxis operator, as shown in Code 1(explained in a grate book - Python Data Science Handbook by Jake VanderPlas) to achieve the desired result:

Code 1:

import numpy as np


A = np.array(
    [[1, 2, 3],
     [4, 5, 6],
     [7, 8, 9]]
)
x = np.array([1, 2, 3])
A + x[np.newaxis:]

Output:

Out:
array([[ 2,  4,  6],
       [ 5,  7,  9],
       [ 8, 10, 12]])

It also will be the most efficient way to accomplish this task in terms of speed and memory consumption, if you are dealing with numerical data in large datasets, because numpy is much more efficient than python.list object, due to numpys' type awareness.

Cheers.

Upvotes: 1

Hammad Ahmed
Hammad Ahmed

Reputation: 885

an elegant solution would be

A

>>>  [[1, 2, 3],
      [4, 5, 6],
      [7, 8, 9]]

B

>>>  [1, 2, 3]

ans = A + B[:, None]
ans

>>> [[ 2,  3,  4],
     [ 6,  7,  8],
     [10, 11, 12]]

Upvotes: 1

Simon Fredrich
Simon Fredrich

Reputation: 95

one way to do it would be the following.

array1 = [[1, 2, 3],
         [4, 5, 6],
         [7, 8, 9]]

array2 = [1, 2, 3]

for i in range(0, len(array2)):
    for j in range(0, len(array1[i-1])):
        array1[i][j] += array2[i]


print(array1)

Upvotes: -1

Related Questions