Need some assistance with this. I never took multivariable calculus in college, so all of this is new to me.
Anyways, let's suppose that $$ x = \left\{ x_1, x_2, \cdots, x_n \right\}^T $$ and $$ y = \left\{ y_1, y_2,\cdots, y_n \right\}^T $$ and $$ f\left(x, y\right) = x\log \left(y\right)^T + \left(1 - x\right)\log(1 - y)^T $$
The task is to find the partial derivative of $$ f\left(x, y\right) $$ with respect to $$ y $$ assuming that $$ x $$ is a constant.
The partial derivative can be easily ascertained by simplifying:
$$ f_y\left(x, y\right) = \frac{x}{y^T} + \frac{x - 1}{\left(1 - y\right)^T} $$
My question is, is the above correct? If yes, how am I supposed to interpret it? I'm having a difficult time switching between treating both $$ x $$ and $$ y $$ like I would regular vectors and treating them as arbitrary variables.
Thank you for reading.
matlabnotation. So it is not surprising that you are confused, the interpretation requires some arbitrary assumptions (Is scalar+vector component-wise? Are functions applied component-wise?). – orion Nov 05 '14 at 09:13