I'ven been reading Dimitris Koukoulopouls' analytic number theory book.
I'm stuck at p.63, the following part.
Let $f : \mathbb{R} \to \mathbb{C}$ be twice-continuously differentiable function of $\mathbb{R}$ and $f^{(j)} << \frac{1}{x^2}$ for $|x| \ge 1$ and $j \in \{0,1,2\}$, so that
$$\hat{f}(\xi)= \int_{\mathbb{R}} f(x) e^{-2\pi i \xi x} dx$$
and he defines $$S_f(x) := \sum_{n \ge 1} f(nx)$$
Next, he uses Poisson's summation formula to deduce that $$\sum_{n\in \mathbb{Z}} f(nx) = \frac{1}{x} \sum_{n \in \mathbb{Z}} \hat{f} (n/x)$$
I don't see why this is true.
Here is what I've tried to justify the above equation but failed:
Set $$g(t) = \sum_{n \in \mathbb{Z}} f(t+nw)$$
Then this $g$ is a periodic function with period $w$.
So its $m$-th Fourier coefficient is: $$c_m = \frac{1}{w} \int_{0}^{w} g(t) e^{-2\pi i m t} dt$$
Then
$\begin{align*}
c_m &= \frac{1}{w} \int_{0}^{w} \sum_{n \in \mathbb{Z}} f(t+nw) e^{-2\pi i m t} dt \\
& = \sum_{n \in \mathbb{Z}} \frac{1}{w} \int_{0}^{w} f(t+nw) e^{-2 \pi i mt} dt
\end{align*}$
And Substituting $t+nw= y$ in the equation gives $$c_m = \sum_{n \in \mathbb{Z}} \frac{1}{w}\int_{wn}^{w(n+1)} f(y) e^{-2 \pi i m (y-nw)} dy $$
I can't see how to proceed, could you give me any advice?
Try defining $g$ in terms of $f$ (no sum, just $f$).