Is there an easy way to subtract along the axis of 2-D array?
a=np.array([[610,550],[400,360]])
np.sum(a,axis=0) # Results in array([1010, 910])
np.subtract(a[0],a[1]) # Results in array([210, 190])
np.subtract(a,axis=0) # Errors out obviously
np.subtract along an axis
Viewed 270
1 Answers
np.subtract subtracts the corresponding elements of two arrays. Subtraction of two arrays along an axis is meaningless, since it happens to each element regardless of its position relative to its neighbors. Notice that in your example, np.sum only accepts one array, not two.
To find the differences between successive elements of a single array, use np.diff, which accepts an axis argument exactly as you would expect
np.diff(a, axis=0)
You can manually implement diff by calling subtract on an array sliced as 1: and :-1 along the desired axis. In your example:
a[1:] - a[:-1]
Related