Re: Zero Padding in radix 2 FFT
- From: "Andor" <andor.bariska@xxxxxxxxx>
- Date: 6 Jan 2006 07:33:06 -0800
stevenj@xxxxxxxxxxxx wrote:
....
> When used to compute the Fourier-series coefficients by sampling an
> arbitrary periodic function, the DFT is a trapezoidal-rule
> approximation for the integral.
Perhaps this is some misunderstanding. Consider some 1-periodic,
band-limited function f defined by
f(t) := a_0 + sum_{k=1}^K (a_k cos(2 pi k t) + b_k sin(2 pi k t) )
for arbitrary real coefficients a_k and b_k.
The Fourier coeffcients a_k and b_k can be perfectly (not
approximately) computed with the DFT of a vector that represents a
sufficiently sampled version of f(t). I don't think you will dispute
this.
Or are you saying that for periodic *non-band-limited* functions, the
DFT coefficients only approximate the Fourier series coefficients?
Which would be obvious, as non-band-limited functions can't be sampled
without throwing away information about the function.
Regards,
Andor
.
- References:
- Re: Zero Padding in radix 2 FFT
- From: stevenj
- Re: Zero Padding in radix 2 FFT
- From: robert bristow-johnson
- Re: Zero Padding in radix 2 FFT
- From: John Monro
- Re: Zero Padding in radix 2 FFT
- From: Jerry Avins
- Re: Zero Padding in radix 2 FFT
- From: John Monro
- Re: Zero Padding in radix 2 FFT
- From: Jerry Avins
- Re: Zero Padding in radix 2 FFT
- From: John Monro
- Re: Zero Padding in radix 2 FFT
- From: Jerry Avins
- Re: Zero Padding in radix 2 FFT
- From: John Monro
- Re: Zero Padding in radix 2 FFT
- From: Jerry Avins
- Re: Zero Padding in radix 2 FFT
- From: stevenj
- Re: Zero Padding in radix 2 FFT
- Prev by Date: Re: C values that are not known at compile time
- Next by Date: Re: Bandpass FIR filter varying over time
- Previous by thread: Re: Zero Padding in radix 2 FFT
- Next by thread: Re: Zero Padding in radix 2 FFT
- Index(es):
Relevant Pages
|