Subvector techniques for DFT parallel processing

Runge´s “method of folding” provided a turn-of-the-century algorithm for hand computation of the Fourier transform. As such, it was an original “DFT” using discrete sampled data in subvector formats, to recover harmonics from diesel vibrations, from salient-pole alternator waveforms, etc.. We present preliminary findings for our subvector part-time research; these have promise by comparison with a standard FFT for highly parallel processing in that tier constraints are absent. In Runge´s (c. 1900) commentary papers he noted extensive complex symmetry in Fourier coefficient arrays for his subvectors. Findings show for limited examples, a modified Runge DFT requires about the same number of computer operations as an FFT. Each stage in a series of subvector stages is autonomous, leading to an attractive topology for massively parallel structures with arbitrary IP (independent processor) assignments