Comparing two numpy arrays with different lengths line-wise-CodePudding

I am trying to compare all to all elements of two different lenght arrays. Lets say I have an array:

A = np.array([[15,25,22],[200,200,20]])

And second array is an array I want to compare with:

B = np.array([[150., 350.],
 [250., 450.],
 [150., 350.],
 [400., 600.],
 [400., 600.],
 [650., 850.],
 [550., 750.],
 [650., 850.]])

What I need is to compare all the elements on index 0 of all arrays in A (A[:, 0]) with all elements on index 0 in all arrays in array B (B[:, 0]).

If A is just a simple array such as A=[25,40,25], then it is simply done as:

smaller = np.array(A[0] > B[:, 0]).astype('int')

I thought I can transform it to 2d comparison such that

smaller = np.array(A[:, 0] > B[:, 0]).astype('int')

This is not working, error is clear, ValueError: operands could not be broadcast together with shapes (2,) (8,). I understand that this way I cannot compare it but I was not able to find the way how to do so.

My desired output would look like this:

[[False, False, False, False, False, False, False, False],
 [True, False, True, False, False, False, False, False]]

CodePudding user response：

This is what I can have

np.array([A[i,0] > B[:, 0] for i in range(A.shape[0])])

CodePudding user response：

Thanks to @MichaelSzczesny I got the solution I was looking for. Simply adding [:, None] to the comparison worked:

smaller = np.array(A[:, 0][:, None] > B[:, 0]).astype('int')

As explained in Use of None in Array indexing in Python, it adds an axis to the array. Thus from

[ 15 200]

We get

[[ 15]
 [200]]