Comments about FFT



On Aug 24, 2009, at 8:41 PM, Raymond Toy wrote
>>
>> I might have switched the 'forward'/'backward' fft routines by  
>> accident..
> Ooh, that would be very bad if no one noticed!
>
> I will try to take a look at this soon.
>
> Ray
>

I would've checked it anyway. I don't think it's a big deal as long  
as it matches the version described in documentation. It's a common  
trouble with fft, people are surprised when their fft doesn't match  
the definition in Numerical Recipes, so it's no big deal, happens  
with all fft implementations..

My preference is that normalisation is done on both fft and ifft with  
1/sqrt(N) but all variations are common... I would make this  
configurable (where normalization is done, perhaps also the sign of  
exp() but that's just a naming thing)... Not hard to do...

Regards,
Ziga