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

Every Fourier transform formula that I know of has a corresponding Fourier series analog, except the multiplication formula $$\int_{-\infty}^\infty f(x)\hat{g}(x) dx=\int_{-\infty}^\infty \hat{f}(y)g(y) dy.$$

So my question is whether a Fourier series analog for this formula exists.

share|cite|improve this question
up vote 9 down vote accepted

Your formula is another way of writing Parseval's theorem, namely that Fourier transform preserves inner products:

$$\int_{-\infty}^{\infty} f \overline{g} = \int_{-\infty}^{\infty} \hat{f}\overline{\hat{g}} .$$

(You should combine this with the identities $\hat{\hat{g}}(x) = g(-x)$ and $\overline{\hat{g}}(y) = \hat{\overline{g}}(-y).$)

There is also a Fourier series version of Parseval's theorem: namely if $f$ (with period 1) has Fourier series $a_n$ and $g$ (with period 1) has Fourier series $b_n$, then $$\int_0^1 f \overline{g} = \sum_{n = -\infty}^{\infty} a_n \overline{b}_n.$$

However, I don't see how to rewrite this in a way analogous to your formula, just because (unlike in the Fourier transform context, where the function $f$ and the Fourier transform $\hat{g}$ are both functions on the real line) a periodic function and its Fourier series are different beasts, and so it's not clear (to me) how one could mix them together in an integral or something similar.

Added: As p.s. observes in their answer, one could argue that the identity in question is weaker than Parseval, since to deduce Parseval from it requires an application of Fourier inversion (the formula $\hat{\hat{g}}(x) = g(-x)$ used above).

share|cite|improve this answer

Recall the adjoint of an operator $F$ on inner product spaces is defined by $\langle Fg,h \rangle=\langle g, F^*h \rangle$. The identity is just an expression of the fact that the Fourier transform is the complex conjugate of its adjoint: $\langle Fg,\bar{h} \rangle=\langle g, F^* \bar{h} \rangle=\langle \bar{F^*} h,\bar{g} \rangle=\langle Fh,\bar{g} \rangle$. (This is a strictly weaker statement than Parseval's equality.) In general, the conjugate adjoint of the Fourier operator from $H_1$ to $H_2$ is the Fourier operator from $H_2$ to $H_1$.

share|cite|improve this answer
Dear user981570, I have made note of your answer in my own. Regards, – Matt E Oct 31 '11 at 11:46

Given an orthogonal function system $\phi_{n}(x)$ ($n=1,2,\dots$) with norm

$$\left\vert \left\vert \phi _{n}\right\vert \right\vert =\sqrt{\left( \phi_{n}\cdot \overline{\phi }_{n}\right) }=\displaystyle\sqrt{\displaystyle\int_{a}^{b}\phi _{n}\left( x\right) \,\overline{\phi }_{n}\left( x\right) \;dx}$$

and two functions $f(x)$ e $g(x)$ represented by the Fourier series

$$f(x)\sim\displaystyle\sum_{n\ge 1} c_{n}\phi_{n}(x)$$

$$g(x)\sim\displaystyle\sum_{n\ge 1} d_{n}\phi_{n}(x)$$

with coefficients $c_n,d_n$

$$c_n=\dfrac{\left ( f\cdot \overline{\phi }_{n}\right)}{\left\vert \left\vert \phi _{n}\right\vert \right\vert ^{2}}=\dfrac{\displaystyle\int_{a}^{b}f\left( x\right)\overline{\phi }_{n}\left( x\right) \;dx}{\left\vert \left\vert \phi _{n}\right\vert \right\vert ^{2}}$$

$$d_n=\dfrac{\left ( f\cdot \overline{\phi }_{n}\right)}{\left\vert \left\vert \phi _{n}\right\vert \right\vert ^{2}}=\dfrac{\displaystyle\int_{a}^{b}g\left( x\right)\overline{\phi }_{n}\left( x\right) \;dx}{\left\vert \left\vert \phi _{n}\right\vert \right\vert ^{2}},$$

and if

$$\displaystyle\int_{a}^{b}\left\vert f(x)-\displaystyle\sum_{n=1}^{\infty}c_{n}\phi_{n}(x)\right\vert ^{2}dx=0$$


$$\displaystyle\int_{a}^{b}\left\vert g(x)-\displaystyle\sum_{n=1}^{\infty}d_{n}\phi_{n}(x)\right\vert ^{2}dx=0,$$

then the following equality holds

$$\displaystyle\int_{a}^{b}f(x)\overline{g}_{n}(x)\; dx=\displaystyle\sum_{n\ge 1} c_{n}\overline{d}_{n}||\phi_n||^2,$$

a particular case of which, for $g(x)=f(x)$, is

$$\displaystyle\int_{a}^{b}|f(x)|^2\; dx=\displaystyle\sum_{n\ge 1} |c_{n}|^2||\phi_n||^{2}.$$

share|cite|improve this answer

There is such a formula, but $f$ and $g$ live on different domains: Let $f: t\mapsto f(t)$ be periodic with period $1$ and $g: k\mapsto g(k)$ be a function defined on $\mathbb Z$. Both these functions have Fourier transforms defined by $$\hat f(k)=\int_0^1 f(t) e^{-2\pi i k t} dt, \ \ \ \hat g(t)=\sum_{k=-\infty}^\infty g(k)e^{-2\pi i k t}.$$ Put $h(t):= \overline{\hat g(t)}$; this is a function of period 1. Then $$h(t)= \sum_{k=-\infty}^\infty \overline{g(k)}e^{2\pi i k t},$$ so the Fourier coefficients of $h$ are given by $\hat h(k)=\overline{g(k)}$. Therefore, by Parseval's theorem for Fourier series one has $$\int_0^1 f(t)\hat g(t)dt = \int_0^1 f(t)\overline{h(t)}dt=\sum_{k=-\infty}^\infty \hat f(k)\overline{\hat h(k)}=\sum_{k=-\infty}^\infty \hat f(k) g(k).$$

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.