Comments on, "A note on the computational complexity of the arithmetic Fourier transform" by N. Tepedelenlioglu

The commenter states that, because of the use of Farey sequences the result stated by N. Tepedelenlioglu (ibid., vol.37, no.7, p.1146-7, 1989) becomes a statement about D/sub N/, the number of elements in a Farey sequence of order N. Tepedelenlioglu's formula is then equivalent to a formula about Farey sequences given by M.R. Schroder (1986) C.M. Rader (1977). Thus, for example, while a standard FTT algorithm requires 1024 samples to yield 1024 spectral components, AFT would take more than 300000 samples to do the same job.<
>