[daip] sinc-function (fft-shift) gridding

[Eric Greisen]

Dirk Petry wrote:
> Hi,
> a quick question concerning the subroutines you kindly provided with the 
> email
> quoted below:
> The parameter
>    DELI          D        No. channels to shift by
> in XCSHFT is a real number but as far as I understand and my tests
> show, only _integer_ numbers work. At least this is the case in my
> implementation which uses FFT. I.e. one can not shift by a fraction of a 
> channel.
> This would also be supported by the shift theorem for discrete FT.
> 
> Can you confirm that the routine should not work for non-integers?

That is absolutely false.  It feeds in a phase shift which is applied to 
the data prior to any FFT and so can have any value whatsoever.

Eric Greisen