# is there a nice way to find the fourier transform of…

I am looking for a nice way to calculate the FT of the following function

$f(x)=\biggl(\sum_{n=1}^{c}~a_n~e^{-\frac{i}{2}~x~b_n}\biggr)^d$,

where $d,c>0$, $a_n$ and $b_n$ are real coefficients, strictly monotonously rising in $n$ and $x$ is the free variable and $c$ might go to $\infty$.

I used mathematica to calculate it, but without specification of $d$ and when $c\to\infty$, there is no way that the programme will do it.

-
Are you sure $f$ is integrable? And is $d$ an integer? – Davide Giraudo Aug 5 '12 at 12:40
it is from a physics book. it seems the author does not worry to much about the dirichlet conditions. – Hamurabi Aug 5 '12 at 12:56
And how is the Fourier transform defined? – Davide Giraudo Aug 5 '12 at 13:10
What's the domain of $f(x)$? – Mhenni Benghorbal Aug 5 '12 at 13:19
What is the name of the physics book you got it from? – Fabian Aug 5 '12 at 13:21

If $d$ is a positive integer, then you can use the multinomial theorem to expand your expression then take the Fourier transform with the appropriate condition on $\sum b_k$.
If $d$ is not an integer, it looks rather difficult... – Fabian Aug 5 '12 at 14:46
If $d$ is a positive integer, expand $f$ as $f(x) = \sum_{n_1} \cdots \sum_{n_d} a_{n_1} \cdots a_{n_d} e^{\frac{i}{2} x (b_{n_1}+ \cdots + b_{n_d})}$, and then, formally, $\hat{f}(\omega) = \sum_{n_1} \cdots \sum_{n_d} a_{n_1} \cdots a_{n_d} \delta(\omega - \frac{b_{n_1}+ \cdots + b_{n_d}}{2})$. – copper.hat Aug 5 '12 at 15:03