if fft is a discrete fourier transform

Question

Julian Oviedo 2016-10-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/306634-if-fft-is-a-discrete-fourier-transform

评论： Star Strider 2016-10-10

if the fft (fast fourier transform) is a discrete fourier transform, so Why the spectrum is not repetitive? https://en.wikipedia.org/wiki/Discrete_Fourier_transform

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Massimo Zanetti 2016-10-10

No one expects the spectrum is "repetitive" (I assume that by "repetitive" you mean "periodic").

Take a very simple example. The signal you want to transform is a sinusoid, then its transform is just an impulse function.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2016-10-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/306634-if-fft-is-a-discrete-fourier-transform#answer_238383

The fast Fourier transform operates on finite sequences with specific sampling intervals (frequencies). Unlike the analytical Fourier transform that extends from -Inf to +Inf in the frequency domain, the discrete Fourier transform extends only as far as the Nyquist frequency, the highest uniquely resolvable frequency in a sampled signal (and the frequency above which ‘aliasing’ occurs if the analog-to-digital converter does not have a hardware anti-aliasing filter on its input to prevent higher frequencies from being sampled). See signal processing textbooks by Proakis and others (or any number of books devoted to the Fourier transform and its applications) for details.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 2

Julian Oviedo 2016-10-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/306634-if-fft-is-a-discrete-fourier-transform#answer_238384

Im studing with "Introduction To Digital Signal Processing And Filter Design (2005) - Shenoi - Wiley " as well I'm using matlab to make a better study.

************************, the discrete Fourier transform extends only as far as the Nyquist frequency, ****************************

So, that the reason why there is not periodic form on the fft.

But, If we take care that if a generate a 100 sample vector, it's not periodic so the spectum must be infinity.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Star Strider 2016-10-10

The two-sided discrete Fourier transform exists only in the region of ±Nyquist frequency, not infinity. In a sampled signal, you know only what is in the signal you have. What may exist in the signal if you sampled it longer is completely unknown to you. Nothing in or derived from a sampled signal is, or ever will be, infinite.

I don’t have Introduction to Digital Signal Processing and Filter Design in my library, but by the description, it should cover this.

请先登录，再进行评论。