I am interested in evaluating the integral
$$ \iint_{E} xy dA $$ where $E$ is the region bounded by $xy = 3$, $xy = 1$, $y = 3x$, and $y = x$.
This gives a region of integration that looks like this:
Using the substitution $x = u/v$ and $y = v$, we obtain a Jacobian of
$$\frac{\partial(x,y)}{\partial(u,v)} = 1 \cdot \frac{1}{v} - 0 \cdot \frac{-u}{v^2} = \frac{1}{v} > 0$$
which I think should turn our integral into:
$$ \int_1^3\int_{1}^3 u \cdot \frac{1}{v} du dv = 4 \ln(3). $$
However, the answer is supposed to be $2 \ln (3)$, so I seem to be missing a factor of $1/2$ somewhere.
I changed my bounds by noting that $v = y$, and $y$ goes from $1$ to $3$ in the graph above. Also, $x = u/v$ implies $u = xv = xy$, and $u = xy$ goes from $1$ to $3$ as well.
So where is my mistake exactly?


The problem is that "$y$ goes from $1$ to $3$ in the graph above" doesn't mean that you should integrate from $v=1$ to $v=3$ after the change of variables - you would then effectively be integrating over this region:
Indeed we can kind of smell that something should be off about that step, because there are many different regions where the range of values of $y$ is $[1, 3]$, but they don't all have the same area.
The problem is that for different values of $u$, the range of values $y$ can take in your region is different. For a fixed $u$, we have $u \le y^2 \le 3u$, so you should integrate from $v = \sqrt u$ to $v = \sqrt{3u}$:
\begin{equation*} \int_1^3 \int_{v=\sqrt u}^{v=\sqrt{3u}} \frac uv \,\mathrm dv \,\mathrm du \end{equation*}
which does give the correct answer (it is a pretty confusing coincidence that the answer happens to be half of what you got!). As mentioned in the comments, it's perhaps a bit easier to choose a change of variables which gives a nicer boundary, like $xy$ and $x/y$.