Comments on "A neural net approach to discrete Hartley and Fourier transforms" by A.D. Culhane et al

By exploiting the structure and the dynamics of a neural net proposed recently for the computation of the discrete Fourier transform (DFT), it is shown to be possible to reduce the neural net given in the above-titled paper (see ibid., vol.36, no.5, p. 695-703, 1989) to a statical conductor array followed by a single row of Hopfield nets with local feedback only. Some modification is also described for better practical implementation. The modified circuit will compute the discrete Hartley transform with the same precision as claimed in the above-titled paper, while at the same time achieve structural modularity which is required to design chips for transforming a large number of samples.<>