| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 4641463 | Journal of Computational and Applied Mathematics | 2008 | 20 Pages |
Abstract
This paper presents some new results on numerical stability for multivariate fast Fourier transform of nonequispaced data (NFFT). In contrast to fast Fourier transform (of equispaced data), the NFFT is an approximate algorithm. In a worst case study, we show that both approximation error and roundoff error have a strong influence on the numerical stability of NFFT. Numerical tests confirm the theoretical estimates of numerical stability.
Related Topics
Physical Sciences and Engineering
Mathematics
Applied Mathematics
Authors
Daniel Potts, Manfred Tasche,