using arrays of different sizes within a function

Question

I'm trying to write a function that will take a set of arguments from the rows of a 2d array and use them in conjunction with all the elements of a longer 1d array:

x = np.linspace(-10,10,100)
abc = np.array([[1,2,1],
      [1,3,5],
      [12.5,-6.4,-1.25],
      [4,2,1]])


def quadratic(a, b, c, x):
    return a*(x ** 2) + b*x + c


y = quadratic(abc[:,0], abc[:,1], abc[:,2], x)

But this returns:

operands could not be broadcast together with shapes (4,) (100,)

When I manually enter the a, b and c values, I get a 100-element 1d array, so I would expect this to return a (4,100) array. What gives?

InsertCheesyLine · Accepted Answer · 2022-12-01 05:31:32Z

1

In numpy the operation x * y is performed element wise where one or both values can be expanded to make them compatible. This is called broadcasting. In the example above your arrays have different dimensions (100,0) and (4,3), hence the error.

When multiplying matrixes you should be using dot instead.

import numpy as np

x = np.linspace(-10,10,100)
abc = np.array([[1,2,1],
      [1,3,5],
      [12.5,-6.4,-1.25],
      [4,2,1]])

np.dot(x, abc[:,0])

answered Dec 1, 2022 at 5:31

InsertCheesyLine

1,4182 gold badges15 silver badges26 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Mercury · Accepted Answer · 2022-12-01 08:55:01Z

1

abc[:, i] is of shape (4). You need it to be of shape (4, 1) to broadcast against x of shape (100) to produce your desired (4, 100) output---so you will need to do abc[:, i, None] to add that extra dimension.

The following should work:

y = quadratic(abc[:, 0, None], abc[:, 1, None], abc[:, 2, None], x)

answered Dec 1, 2022 at 8:55

Mercury

4,1811 gold badge15 silver badges43 bronze badges

Collectives™ on Stack Overflow

using arrays of different sizes within a function

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related