Accumulation of Round-Off Error in Fast Fourier Transforms

Authors:
Toyohisa Kaneko;Bede Liu
Affiliations:
IBM, Thomas J. Watson Research Center, Yorktown Heights, New York and Princeton University, Department of Electrical Engineering, Princeton, New Jersey;Princeton University, Department of Electrical Engineering, Princeton, New Jersey
Venue:
Journal of the ACM (JACM)
Year:
1970

Citing 1
Cited 10

Rounding Errors in Algebraic Processes

Rounding Errors in Algebraic Processes

Fault-Tolerant FFT Networks

IEEE Transactions on Computers
On Local Roundoff Errors in Floating-Point Arithmetic

Journal of the ACM (JACM)
Floating Point Fast Fourier Transform Computation Using Double Precision Floating Point Accumulators

ACM Transactions on Mathematical Software (TOMS)
Implementing Clenshaw-Curtis quadrature, II computing the cosine transformation

Communications of the ACM
Inverses of Multivariable Polynomial Matrices by Discrete Fourier Transforms

Multidimensional Systems and Signal Processing
Fixed-point fast Hartley transform error analysis

Signal Processing
Response of model simulated weather parameters to round-off-errors on different systems

Environmental Modelling & Software
Error Analysis and Verification of an IEEE 802.11 OFDM Modem using Theorem Proving

Electronic Notes in Theoretical Computer Science (ENTCS)
On the Precision Attainable with Various Floating-Point Number Systems

IEEE Transactions on Computers
The expansion in Gegenbauer polynomials: A simple method for the fast computation of the Gegenbauer coefficients

Journal of Computational Physics

Quantified Score

Hi-index	0.03

Visualization

Abstract

The fast Fourier transform (FFT) is an algorithm to compute the discrete Fourier coefficients with a substantial time saving over conventional methods. The finite word length used in the computer causes an error in computing the Fourier coefficients. This paper derives explicit expressions for the mean square error in the FFT when floating-point arithmetics are used. Upper and lower bounds for the total relative mean square error are given. The theoretical results are in good agreement with the actual error observed by taking the FFT of data sequences.