Does anyone know how to represent the Discrete Fourier transform (DFT) coefficient, $X[k]$, with respect to the Continuous time-Fourier series (CT-FT) coefficient, $X_k$? I come to the conclusion as $X[k] = NX_k$. $\\\\$

As we know Fourier series is for a periodic signal. So DFT coefficient is just a sampled version of the Fourier series coefficient. Let $P$ be the period of CT signal $x(t)$, and $N$ be the sampled period of $x[n]$. We sampled $x(t)$ by $T=P/N$ as sample period. From $$x[n] = x(nP/N)\\=\sum_{k=-inf}^{inf}X_k e^{2\pi*k/P*(nP/N)}\\ =\sum_{k=-inf}^{inf}X_k e^{2\pi*kn/N}$$ Since for DFT, $$x[n] =1/N\sum_{k=-0}^{N-1}X[k] e^{2\pi*kn/N}$$, we can say that $X[k]=NX_k$ for $k=0,1,...,N-1$? $$\\\\$$Is my interpretation correct?


No, it's not that simple. In general, sampling will introduce aliasing. The correct derivation is as follows. Let $T$ be the period of the continuous-time signal $x(t)$, and let $T_s$ be the sampling period satisfying $T=NT_s$ with integer $N$. It follows that the sampled sequence $x_d[n]=x(nT_s)$ is also periodic with period $N$.

We have

$$x(t)=\sum_{k=-\infty}^{\infty}c_ke^{j\frac{2\pi k}{T} t}\tag{1}$$

For the sampled sequence we obtain

$$x_d[n]=x(nT_s)=\sum_{k=-\infty}^{\infty}c_ke^{j\frac{2\pi k}{T} nT_s}=\sum_{k=-\infty}^{\infty}c_ke^{j\frac{2\pi k}{N} n}\tag{2}$$

where we've used $T_s=T/N$. Note that the term $e^{j\frac{2\pi k}{N} n}$ is $N$-periodic in $k$, so we can rewrite $(2)$ as

$$x_d[n]=\sum_{k=0}^{N-1}\sum_{l=-\infty}^{\infty}c_{k+lN}e^{j\frac{2\pi k}{N} n}\tag{3}$$

Comparing Eq. $(3)$ to the IDFT

$$x_d[n]=\frac{1}{N}\sum_{k=0}^{N-1}X[k]e^{j\frac{2\pi kn}{N}}\tag{4}$$



Form Eq. $(5)$ it is obvious that the DFT coefficients $X[k]$ are a (scaled and) aliased version of the Fourier coefficients $c_k$ of the continuous-time function.

  • $\begingroup$ Omg! You are amazing. I totally know where I got wrong. I feel that my derivation is a little bit wrong so I decide to ask people on this platform. Thank you for your help! $\endgroup$ – Yok Jye Tang Jan 31 at 17:06

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.