I create an array in numpy as
a = np.ones([5 , 5])
I will then get an output as 5x5 array full of 1s. I would like to keep the outer elements as 1, and inner elements as 0. So I would like output to be:
[[ 1. 1. 1. 1. 1.]
[ 1. 0. 0. 0. 1.]
[ 1. 0. 0. 0. 1.]
[ 1. 0. 0. 0. 1.]
[ 1. 1. 1. 1. 1.]]
Is there any way with which we could do this in a single line? (I have read about inner() but I dont know how to get it working with this single array)
Yes, we can use slicing for this:
a[1:-1, 1:-1] = 0
Or for a generic multidimensional array:
a[(slice(1, -1),) * a.ndim] = 0
but usually it would be better to construct such matrix in another way. This produces:
>>> a = np.ones([5 , 5])
>>> a[1:-1, 1:-1] = 0
>>> a
array([[1., 1., 1., 1., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 1., 1., 1., 1.]])
and for instance for a 3d case (imagine some sort of cube):
>>> a = np.ones([5 , 5, 5])
>>> a[(slice(1, -1),) * a.ndim] = 0
>>> a
array([[[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.]],
[[1., 1., 1., 1., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 1., 1., 1., 1.]],
[[1., 1., 1., 1., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 1., 1., 1., 1.]],
[[1., 1., 1., 1., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 0., 0., 0., 1.],
[1., 1., 1., 1., 1.]],
[[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.]]])
a[np.s_[1:-1,] * a.ndim] = 0
np.s_[1:-1,] simply returns a (slice(1,-1),) tuple, so it will only cost more computational resources, and furthermore more dependency on numpy whereas slice is a vanilla Python class.slice function__getitem__ from the s_ attribute.return s_[1:-1]) it takes 2.986656811001012s for 10M runs, for return (slice(1, -1),), it takes 1.8174990950210486s.