Accuracy of the discrete Fourier transform and the fast Fourier transform

被引:64
作者
Schatzman, JC
机构
[1] Department of Mathematics, University of Wyoming, Laramie, WY 82071
关键词
fast Fourier transform (FFT); discrete Fourier transform (DFP);
D O I
10.1137/S1064827593247023
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Fast Fourier transform (FFT)-based computations can be far more accurate than the slow transforms suggest. Discrete Fourier transforms computed through the FFT are far more accurate than slow transforms, and convolutions computed via FFT are far more accurate than the direct results. However, these results depend critically on the accuracy of the FFT software employed, which should generally be considered suspect. Popular recursions fur fast computation of the sine/cosine table (or twiddle factors) are inaccurate due to inherent instability. Some analyses of these recursions that have appeared heretofore in print, suggesting stability, are incorrect. Even in higher dimensions, the FFT is remarkably stable.
引用
收藏
页码:1150 / 1166
页数:17
相关论文
共 12 条
[1]   INDEX MAPPINGS FOR MULTIDIMENSIONAL FORMULATION OF DFT AND CONVOLUTION [J].
BURRUS, CS .
IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING, 1977, 25 (03) :239-242
[2]  
CALVETTI D, 1991, MATH COMPUT, V56, P755, DOI 10.1090/S0025-5718-1991-1068824-0
[3]  
CHU CY, 1987, 87882 CORN U DEP COM
[4]  
GENTLEMAN WM, AFIPS C P, V29, P563
[5]  
Gottlieb D., 1977, NUMERICAL ANAL SPECT
[6]   THE ACCURACY OF FLOATING-POINT SUMMATION [J].
HIGHAM, NJ .
SIAM JOURNAL ON SCIENTIFIC COMPUTING, 1993, 14 (04) :783-799
[7]   ACCUMULATION OF ROUND-OFF ERROR IN FAST FOURIER TRANSFORMS [J].
KANEKO, T ;
LIU, B .
JOURNAL OF THE ACM, 1970, 17 (04) :637-+
[8]  
OLIVER J, 1975, J I MATH APPL, V16, P247
[9]  
SCHATZMAN J, UNPUB MATH COMP
[10]  
SCHATZMAN J, 1996, IEEE T SIGNAL PROCES, V44