I sort of understand the principle of the Fourier series, but when I watch the wiki page I don't understand how to get from:

${a_0 \over 2} + \sum_{n=1}^N[a_n cos({2\pi n x \over P}) + b_n sin({2\pi n x \over P})]$


$\sum_{n=-N}^N c_n e^{i{2\pi n x \over P}}$

To be clear what I don't understand is the transition from using the coefficients $a_n$ and $b_n$ to $c_n$. I am familiar with the Euler's formula. I noticed that the sum goes from -N to N in the second equation. In the article they write:

$\begin{align} a_n & = c_n + c_{-n} \\[12pt] b_n & = i(c_n - c_{-n}) \\[12pt] c_n & = \begin{cases} \frac{1}{2}(a_n - i b_n) & \text{for } n \ne 0, \\[12pt] \frac{1}{2}a_0 & \text{for }n = 0. \end{cases} \end{align}$

So to get an intuition I tried to take a simple example in which N = 4 for example. I assume if I was taking n=-1 and n=1 for example, summing the terms using the above relationships, etc. I would get back to $a_1 cos(...) + b_1 sin(...)$. But I didn't succeed. I realise it probably takes quite some time to show how to get from one to another, but I would be grateful if someone could show me or at least put me on the right track.

Thank you.


We need consider only one $n > 0$ (the case $n = 0$ is easily verified). In the exponential form, we consider the two terms with indices $n$ and $-n$,

$$\begin{align} c_n e^{2\pi inx/P} + c_{-n} e^{-2\pi inx/P} &= c_n (\cos (2\pi nx/P) + i\sin (2\pi nx/P)) + c_{-n}(\cos (2\pi nx/P) - i \sin (2\pi nx/P))\\ &= (c_n + c_{-n})\cos (2\pi nx/P) + i(c_n - c_{-n})\sin (2\pi nx/P), \end{align}$$

so $a_n = c_n + c_{-n}$, and $b_n = i(c_n - c_{-n})$. In the trigonometric form, we consider

$$\begin{align} a_n \cos (2\pi nx/P) + b_n\sin (2\pi nx/P) &= \frac{a_n}{2}\left(e^{2\pi inx/P} + e^{-2\pi inx/P}\right) + \frac{b_n}{2i}\left(e^{2\pi inx/P} - e^{-2\pi inx/P}\right)\\ &= \frac{a_n - ib_n}{2}e^{2\pi inx/P} + \frac{a_n + ib_n}{2}e^{-2\pi inx/P} \end{align},$$

so we get $c_n = \frac{a_n-ib_n}{2}$ and $c_{-n} = \frac{a_n + ib_n}{2}$.

| cite | improve this answer | |
  • $\begingroup$ Fantastic, thank you so much for this quick and very precise (straight to the point) answer. I still have a small question: your wrote $c_{-n}e^{ix}$ as $c_{-n}(cos(x) - i \: sin(x)$. Why? Or is there a small typo on the left inside. Should it actually be $c_{-n}e^{-ix}$? Thank you. $\endgroup$ – Marc Ourens Oct 5 '13 at 13:00
  • $\begingroup$ That was a typo I didn't see. The coefficient $c_{-n}$ of course belongs to the exponential $e^{-2\pi inx/P}$, not $e^{2\pi inx/P}$. Sorry. $\endgroup$ – Daniel Fischer Oct 5 '13 at 13:02
  • $\begingroup$ That's all good. It at least shows I made an effort to make sense of what you wrote. Thank you again. $\endgroup$ – Marc Ourens Oct 5 '13 at 13:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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