Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I usually work in number theory so I am not familiar with Fourier transforms, I have read up on them and know the basics but it never seems to be in number theory language.

I am trying to find the transform of a primitive Dirichlet character $\chi(n) \bmod q$. I know this is a periodic function and $\chi(n)=\exp\left(\frac{Kv(n)}{\phi(p^\alpha)}\right)$ but I have no idea have to find its transform or the transform of $f(n)\chi(n)$

Yes you are right, say how would you calculate $\sum_(n\epsilon Z) f(n)\chi(n)$

share|cite|improve this question
It should be the discrete Fourier transform? For example the discrete Fourier transform of the Legendre symbol function is related to the quadratic gauss sum. – quanta Mar 17 '11 at 1:08
NIST Digital Library of Mathematical Functions has a section on Periodic Number-Theoretic Functions that might be helpful. – Eric Nitardy Mar 18 '11 at 2:55

I finally found the answer, I feel dumb because it is quite simple, but this answer cannot be found at many places on the web.

Consider the gaussian sum of a character modulo $q$ :

$$\tau(\chi) = \sum_{n=1}^{q-1} \chi(n) e^{\textstyle\frac{2 i \pi n}{q}}$$

I wrote $\sum_{n=1}^{q-1}$ but I could have written $\sum_{n \in G}$ the group of the inversible modulo $q$ (those $n$ with $gcd(n,q)=1$), because $\chi(n) = 0$ if $n \not\in G$. And this is important because it leads to the main trick :

take any $a \in G$ (which is inversible modulo $q$), thus the application $n \to a \,.n$ is a bijection from $G$ to itself, so that : $$\forall a \in G, \qquad \tau(\chi) = \sum_{n=1}^{q-1} \chi(a n) e^{\textstyle\frac{2 i \pi a n}{q}}$$

and simply because $\chi(a n) = \chi(a) \chi(n)$ : $$\tau(\chi) = \chi(a) \sum_{n=1}^{q-1} \chi(n) e^{\textstyle\frac{2 i \pi a n}{q}}$$

i.e. :

$$\sum_{n=1}^{q-1} \chi(n) e^{\textstyle\frac{2 i \pi a n}{q}} = \tau(\chi) \bar{\chi}(a)$$

for $a \in G$, but the uniqueness/inversibility of the Fourier transform implies that the other values of $\chi$'s Fourier transform must be $0$, so that this is true for every $a$.

finally, the discrete Fourier transform (of length $q$) of a Dirichlet character $\chi$ modulo $q$ is $\bar{\chi}\, \tau(\chi)$.

note that in the same way we have also that $\sum_{n=1}^{q-1} \chi(n) e^{-2i \pi n k / q} = \bar{\chi}(k) \overline{\tau(\bar{\chi})}$. then, writing a Fourier series representation for the distribution :

$$\delta_\chi(x) = \sum_{n=1}^\infty \chi(n) \delta(x-n) = \frac{1}{q} \sum_{k=1}^\infty \bar{\chi}(k)\left(\tau(\chi) e^{-2 i \pi k x / q} + \overline{\tau(\bar{\chi})} e^{2 i \pi k x / q}\right)$$ and with $L(s,\chi) = \int_0^\infty \delta_\chi(x) x^{-s} dx$ we can get the functional equation :

$$L(s,\chi) = \sum \chi(n) n^{-s} = L(1-s,\bar{\chi}) A(s)$$

with $A(s) = \frac{1}{q}\int_0^\infty \left(\tau(\chi) e^{-2 i \pi x / q} + \overline{\tau(\bar{\chi})} e^{2 i \pi x / q}-2\right) x^{-s}dx$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.