On page 23 of Anton Deitmar's book on automorphic forms, he presents a proof that for a function
$$g\in C^\infty(\mathbb{R}/\mathbb{Z}),$$
the Fourier coefficients
$$c_k(g)=\int_0^1g(t)e^{-2\pi i k t}dt,$$
are of rapid decay, meaning that
$$\forall N\in\mathbb{N}, k^Nc_k(g), \text{is bounded as a function of } k.$$
Using integration by parts, the author manages to obtain the inequality (for $k\neq 0$)
$$|c_k(g)|\leq\frac{1}{(4\pi^2k^2)^n}\left|\int_0^1g^{(2n)}(t)e^{-2\pi i k t}dt\right|,$$
for some $n\in\mathbb{N}$. If I understand correctly the integral is bounded as a function of $k$ because both
$$\int_0^1g^{(2n)}(t)dt, \text{ and } \int_0^1e^{-2\pi i k t}dt,$$
are. Furthermore, multiplying by $k^n$, the coefficient becomes $((4\pi^2)^nk)^{-1}$, which is certainly bounded as a function of (non-zero) $k$, and thus is of rapid decay? (because the choice of $n$ is arbitrary)
Secondly, the author goes on to claim that the rapid decay of the $c_k(g)$ implies that the series
$$\sum_{k\in\mathbb{z}}|c_k(g)|,$$
converges. I'm having trouble seeing how this follows from rapid convergence. I played around with the series a bit, but to no avail.