Suppose $f_1(x,y)$ and $f_2(x,y)$ are two bounded real-valued measurable functions on $\mathbb{R}^2$.
Suppose $f_1(x,y) \geq f_2(x,y) \forall (x,y) \in \mathbb{R}^2$.
Can we write $f_1(x,y) = g(x) + f_2(x,y)$?, where $g(x)$ is some bounded real-valued measurable function on $\mathbb{R}$.
In other words, can a single variable function make up for the slack.?
If not how to define closeness and approximate?
Normally not. This requires that $f_2(x,y)-f_1(x,y)$ be a function of $x$ only. As an example, let $f_1(x,y)=0, f_2(x,y)=e^{-(x^2+y^2)}$. Because $f_2$ depends on $y$ you can't.