michielm wrote:
I get my answer in the following way:

is defined as
![U_{cap}=\frac{1}{2}\gamma \int \int \left[ \left( \frac{\partial z}{\partial x} \right) ^2 +\left(\frac{\partial z}{\partial y}\right)^2 - \theta^2 \right] dx dy U_{cap}=\frac{1}{2}\gamma \int \int \left[ \left( \frac{\partial z}{\partial x} \right) ^2 +\left(\frac{\partial z}{\partial y}\right)^2 - \theta^2 \right] dx dy](/CBB/latexrender/pictures/9c9f6e3aa9cdc9446fc99f0c2f8c1fc4.png)
Let's first calculate the partial derivates:

and

If I square both I get:

and

The part between square brackets then becomes:
![\left[\left(\frac{\partial z}{\partial x}\right)^2 +\left(\frac{\partial z}{\partial y}\right)^2 - \theta^2 \right]=\frac{\theta^2}{\pi}\int^{\infty}_{-\infty} \tilde{\eta} k e^{ikx} e^{-|k|y} dk \left[\left(\frac{\partial z}{\partial x}\right)^2 +\left(\frac{\partial z}{\partial y}\right)^2 - \theta^2 \right]=\frac{\theta^2}{\pi}\int^{\infty}_{-\infty} \tilde{\eta} k e^{ikx} e^{-|k|y} dk](/CBB/latexrender/pictures/a68ec6790998f93cc744fa0b1c798f82.png)
(the first and the last term in

cancel against

and

respectively)
Plugging this into

I end up with:
![U_{cap}=\frac{1}{2}\gamma \frac{\theta^2}{\pi} \int \int \left[\int^{\infty}_{-\infty} \tilde{\eta} |k| e^{ikx} e^{-|k|y} dk\right] dx dy U_{cap}=\frac{1}{2}\gamma \frac{\theta^2}{\pi} \int \int \left[\int^{\infty}_{-\infty} \tilde{\eta} |k| e^{ikx} e^{-|k|y} dk\right] dx dy](/CBB/latexrender/pictures/868b127dd7c1b40792bbc3f864566893.png)
Which then results in:

I hope this is more clear. Is there any mistake in here (or in my thinking)?
Why does

?
You only know it is

. Ditto for

. But they are inconsequential.
What I am worry about is the y-integral, because it doesn't necessarily converge. Indeed, it does not converge if we are working with

. Maybe you are working with

? But if that is the case, what about x? If you do not work on the whole of

then you need something extras in your Fourier inversion formula for

.