"Frank " <allinone_2003@yahoo.com.hk> wrote in message <hvo3p5$ar2$1@fred.mathworks.com>...
> Hi Wayne,
>
> Thanks for your advice. In fact, I want to ask a conceptual question.
>
> DTFT:
> X(e^(j w/K)) = sum_{n=0}^{N1}x(n) e^{jwn/K}......................(1)
>
> IDTFT:
> x(p) = integrate_{ pi}^{pi} X(e^(j w/K)) e^{jwp}dw/2/pi
> = sum_{n=0}^{N1}x(n) sinc(pn/K);......................................(2)
>
> However, when I compute the FFT of (2), it is different from (1).
>
> Can you solve this problem?
>
> Thanks a lot.
>
> Frank
Frank, we you say you have the DTFT at X(e^(j w/k)), this is continuous function of w. All you're doing is dilating the frequency axis with w/K. Are you trying to sample the continuous frequency axis to arrive at the DFT?? If so, then you want X(e^j(2*pi*k/N)) k=0,.....N1 The arriving at the inverse DFT would be a summation over k.
If you are working in a compute, you are not going to be computing the DTFT. Again, that is a complexvalued function of a continuous variable.
Wayne
