CODEWITHSUNDEEP
pythonnumpy
efirvida
efirvida

Reputation: 4855

numpy - issue with creating a 2D matrix

I have this function that returns a numpy array with two vectors that represent x and y coordinates respectively, and I want to turn them into (x;y) pairs.

For example:

import numpy as np

def rotate(coords, angle, center=(0, 0,)):
    rotated = np.array([np.cos(np.radians(angle)) * (coords[:,0] - center[0]) - np.sin(np.radians(angle)) * (coords[:,1] - center[1]) + center[0],
                np.sin(np.radians(angle)) * (coords[:,0] - center[0]) + np.cos(np.radians(angle)) * (coords[:,1] - center[1]) + center[1]])
    return rotated

xy = np.array([[1,2],[3,4],[5,6],[6,7]])
a = rotate(xy,20)
print a

This gives me:

[[ 0.25565233  1.45099729  2.64634224  3.24401472]
 [ 2.22140538  4.78483091  7.34825644  8.62996921]]

However, I actually want this output:

[[0.25565233, 2.22140538] 
 [1.45099729, 4.78483091]
 [2.64634224, 7.34825644]
 [3.24401472, 8.62996921]]

Upvotes: 1

Views: 109

Answers (2)

rayryeng
rayryeng

Reputation: 104464

Simply take the transpose. You can use numpy.transpose or you can use the overloaded operator .T:

xy = np.array([[1,2],[3,4],[5,6],[6,7]]);
a = rotate(xy,20).T # <--- Modification here

Doing this thus gives:

>>> print a
[[ 0.25565233  2.22140538]
 [ 1.45099729  4.78483091]
 [ 2.64634224  7.34825644]
 [ 3.24401472  8.62996921]]

If you want to avoid transposing the output, you can let the function do that. Just transpose what is returned inside your rotate function:

def rotate(coords, angle, center=(0, 0,)):
    rotated = np.array([np.cos(np.radians(angle)) * (coords[:,0] - center[0]) - np.sin(np.radians(angle)) * (coords[:,1] - center[1]) + center[0],
                np.sin(np.radians(angle)) * (coords[:,0] - center[0]) + np.cos(np.radians(angle)) * (coords[:,1] - center[1]) + center[1]])
    return rotated.T #<-- Modification here

You can thus do what you did before without thinking about it further:

xy = np.array([[1,2],[3,4],[5,6],[6,7]]);
a = rotate(xy,20)

... which gives us:

>>> print a
[[ 0.25565233  2.22140538]
 [ 1.45099729  4.78483091]
 [ 2.64634224  7.34825644]
 [ 3.24401472  8.62996921]]

Upvotes: 3

Anand S Kumar
Anand S Kumar

Reputation: 90869

You want to transpose it -

In [3]: n = np.array([[ 0.25565233,  1.45099729,  2.64634224 , 3.24401472],
   ...:  [ 2.22140538 , 4.78483091 , 7.34825644 , 8.62996921]])

In [4]: n
Out[4]:
array([[ 0.25565233,  1.45099729,  2.64634224,  3.24401472],
       [ 2.22140538,  4.78483091,  7.34825644,  8.62996921]])

In [6]: n.T
Out[6]:
array([[ 0.25565233,  2.22140538],
       [ 1.45099729,  4.78483091],
       [ 2.64634224,  7.34825644],
       [ 3.24401472,  8.62996921]])

In your case -

def rotate(coords, angle, center=(0, 0,)):
        rotated = np.array([np.cos(np.radians(angle)) * (coords[:,0] - center[0]) - np.sin(np.radians(angle)) * (coords[:,1] - center[1]) + center[0],
                    np.sin(np.radians(angle)) * (coords[:,0] - center[0]) + np.cos(np.radians(angle)) * (coords[:,1] - center[1]) + center[1]])
        return rotated


xy = np.array([[1,2],[3,4],[5,6],[6,7]])

a = rotate(xy,20)

print a.T
>> [[ 0.25565233  2.22140538]
 [ 1.45099729  4.78483091]
 [ 2.64634224  7.34825644]
 [ 3.24401472  8.62996921]]

Upvotes: 1

Related Questions