In lecture we are finding gradient on a 2,2 shape tensor,

I’m curious why do we have to do reduce sum ?

`y = tf.reduced_sum(x)`

z is the actual function which is can be written as `f(x) = y^2`

and x is the tensor input, How do we come to a conclusion as to use reduced_sum of x and use that as input to the above function ?

I’m trying to understand this from a purely mathematical point of view.