So, I have a function: $f(x) = a\cdot x + b\cdot x + \varepsilon$, where $\varepsilon \sim \mathcal{N}(0, a+2b)$
I want to derive this function to one of the parameters (for example $a$) when performing a sensitivity analysis, but I get stuck with the epsilon part. Somebody has an idea?