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

The full problem is this:

Let $f:[-\pi,\pi]\rightarrow\mathbb{C}$ be Riemann integrable. Let $\{c_n\}_{n\in\mathbb{Z}}\subset\mathbb{C}$. Prove that if $s_N=\sum_{n=-N}^Nc_ne^{inx}$ converges in $L^2$ to $f(x)$ then $c_n=\frac{1}{2\pi}\int_{-\pi}^\pi f(x)e^{-inx}dx$.

I have no idea where to start on this. Any hint at all would help immensely!

Thank you!

share|cite|improve this question

Write an error function as follows:

$$\begin{align}E(c) &= \int_{-\pi}^{\pi} dx \: \left |f(x) - \sum_{n=-N}^N c_n e^{i n x} \right |^2\\ &= \int_{-\pi}^{\pi} dx \: \left (f(x) - \sum_{n=-N}^N c_n e^{i n x} \right ) \left(f^*(x) - \sum_{n=-N}^N c_n^* e^{-i n x} \right )\\ &= \int_{-\pi}^{\pi} dx \: \left (|f(x)|^2 - f(x)\sum_{n=-N}^N c_n^* e^{-i n x} - f^*(x) \sum_{n=-N}^N c_n e^{i n x} + \sum_{n=-N}^N\: \sum_{n'=-N}^N c_n c_n^* e^{i (n-n') x} \right )\end{align}$$

where $c$ is the vector of coefficients $[c_n]_{n=-N}^N$. Note that, when the sum converges to $f$ in an $L^2$ sense, then

$$\lim_{N \rightarrow \infty} E(c) = 0$$

To find a critical coefficient $c_k$ that minimizes $E$, we solve

$$\frac{\partial E}{\partial c_k^*} = 0$$

Doing this, we obtain

$$\sum_{n=-N}^N c_n \int_{-\pi}^{\pi} dx \: e^{i (n-k) x} = \int_{-\pi}^{\pi} dx \: f(x) e^{-i k x}$$

Using the sifting property of the integral on the LHS, we deduce that

$$c_k \cdot 2 \pi = \int_{-\pi}^{\pi} dx \: f(x) e^{-i k x}$$

which is a Fourier coefficient.

share|cite|improve this answer

$L^2([-\pi,\pi])$ is a Hilbert space for the hermitian product $$\langle f,g \rangle = \frac{1}{2\pi}\int_{-\pi}^\pi \overline{f(x)}g(x)\,dx$$

Furthermore, the family of functions $e_n\colon x \mapsto e^{inx}$, $n \in \Bbb Z$ is orthonormal with respect to $\langle,\rangle$. (In fact, it is a Hilbert basis.) This yields

$$ c_n = \frac{1}{2\pi}\int_{-\pi}^\pi s_N(x)e^{-inx}\,dx = \langle e_n,s_N\rangle \xrightarrow[N\to\infty]{} \langle e_n,f\rangle $$ Indeed, $$ |\langle e_n,f\rangle - \langle e_n,s_N\rangle| = |\langle e_n,f-s_N\rangle| \leq \|e_n\|_{L^2}\|f-s_N\|_{L^2}. $$

share|cite|improve this answer
I don't see how you got the very first step. How are you able to say that $c_n$ is equal to that? Where, for instance, does the $\frac{1}{2\pi}$ come from? – crf Apr 5 '13 at 17:49
It comes from $\langle,\rangle$. – Siméon Apr 5 '13 at 18:06

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.