I am trying to find the variance of $u$.
I know that if:
$u = a \cdot X + b \cdot Y$
Then:
$var(u) = a^2 \cdot var(X) + b^2 \cdot var(Y) + 2\cdot a\cdot b \cdot cov(X,Y)$
The problem I'm trying to solve is to find $var(u)$ when:
$u = X(a-Y)$
Where $a$ is a constant and $X$ and $Y$ are random (normally distributed) variables.
SOLUTION
Thanks to all who helped. I have managed to solve this. For the benefit of others facing this problem in the future, here is my solution:
$u = X(a-Y)$
Expand:
$u = X \cdot a - X \cdot Y$
The Wikipedia page for the Delta Method gives the following formula:
$Var(h_r) = \sum_{i}\left(\frac{\partial h_r}{\partial B_i}\right)^2 Var(B_i) + \sum_{i}\sum_{j\neq i} \left(\frac{\partial h_r}{\partial B_i}\right)\left(\frac{\partial h_r}{\partial B_j}\right)Cov \left(B_i,B_j\right)$
So I need to calculate the partial derivatives with respect to each variable:
$\frac{\partial u}{\partial X}=a-Y$
$\frac{\partial u}{\partial Y}=-X$
Putting these into the formula gives:
$Var(u)=(a-Y)^2.Var(X) + (-X)^2.Var(Y)+2.(a-Y).(-X).Cov(X,Y)$
Simplifying slightly:
$Var(u)=(a-Y)^2.Var(X) + X^2.Var(Y)-2.X.(a-Y).Cov(X,Y)$
Thanks again for your pointers. Hopefully someone finds this useful one day.