Re: Real-data DFT in Octave

[Torbjörn Rathsman]

Doug Stewart skrev 2012-10-06 16:05:

On Sat, Oct 6, 2012 at 10:02 AM, Torbjörn Rathsman <address@hidden> wrote:

Doug Stewart skrev 2012-10-06 15:46:

On Sat, Oct 6, 2012 at 9:35 AM, Sergei Steshenko <address@hidden> wrote:

----- Original Message -----
> From: Torbjörn Rathsman <address@hidden>
> To: address@hidden
> Cc:
> Sent: Saturday, October 6, 2012 3:18 PM
> Subject: Real-data DFT in Octave
>
> Is it possible to without much effort call the real to complex dft routins from
> octave. It is easier to manipulate the spectrum that way, since i am guarantied
> that the mirror part becomes correct. The purpose in this case is to compute
> filtered derivative of second order.
> _______________________________________________
> Help-octave mailing list
> address@hidden
> https://mailman.cae.wisc.edu/listinfo/help-octave

I am not sure what the problem is, but dealing with real input data I use 'conj' and 'fliplr' in the end when I need to produce mirrored spoectrum for inverse FFT.

Regards,
  Sergei.

_______________________________________________
Help-octave mailing list
address@hidden
https://mailman.cae.wisc.edu/listinfo/help-octave

fft works for me

a=randn(100,1);

b=fft(a);

c=ifft(b);

d=real(c);

e=a-d;

e(1:10)

this gives close to zero,which is good.

--

DAS

https://linuxcounter.net/user/206392.html

But you did not clobber b.  You need to mirror it the changes if you were to.

Yes if you change b then you must do it correctly 

--

DAS

https://linuxcounter.net/user/206392.html

Yes and if fftw:s r2c and c2r routine were used, I wouldn't need to bother, because it is inherent from the choise of algorithm.