Talk:Arithmetic complexity of the discrete Fourier transform

Merge

Should this page be merged to Discrete Fourier transform? Please comment below with merge or don't merge. D O N D E groovily Talk to me 04:29, 22 February 2012 (UTC)[reply]

There is also Fast Fourier transform#Computational issues with more information on the topic than here. Presently "Arithmetic complexity of the discrete Fourier transform" is not so useful, it has to be either rewritten or merged. Incnis Mrsi (talk) 06:58, 22 February 2012 (UTC)[reply]

Don't merge but delete: The page Discrete Fourier transform is explicitly devoted to the theory, not to the computation. The computation is considered in Fast Fourier transform. Thus, if a merge should be done, it should be into the latter. But the content of this page is already treated in Fast Fourier transform#Computational issues as "Moreover, explicit algorithms that achieve this count [linear lower bound on the number of multiplication] are known (Heideman & Burrus, 1986; Duhamel, 1990). Unfortunately, these algorithms require too many additions to be practical, at least on modern computers with hardware multipliers." Although short, this description is much more useful and accurate than this page, as it compare it with the other complexity results on FFT, while the present article presents only an awful formula which is of no help for anything and say nothing about the relationship of this particular result with the knowledge in this area. D.Lazard (talk) 13:22, 22 February 2012 (UTC)[reply]

In fact, "redirect" is better than "delete". I'll proceed myself. D.Lazard (talk) 14:34, 22 February 2012 (UTC)[reply]