Shifted Fourier Matrices and Their Tridiagonal Commutors

It is known that, for $ n \ge 3 $, the n × n Fourier matrix $ {F} = n^{-1/2} [e^{2 \pi {{\text{\textit{\r{\i}}}}} \mu \nu/n}] $ ($ 0 \le \mu < n $, $ 0 \le \nu < n $) commutes with a nonscalar tridiagonal matrix T and also with another matrix X that is "almost" tridiagonal. These matrices are important in selecting eigenvectors for the Fourier matrix itself. The purpose of this paper is to generalize those results to matrices $ F_n({\tau},{{\alpha}}) $, variants of the Fourier matrix depending on a base-choosing parameter $\tau$ and a shift parameter $\alpha$. These \emph{shifted Fourier matrices} are defined by $ F_n({\tau},{{\alpha}}) = %\fffnta{n}{\tau}{\alfa} = n^{-1/2} [e^{2\pi{{\text{\textit{\r{\i}}}}} \tau (\mu - m + a)(\nu - m + a)}] $, where $ {m} = (n - 1)/2 $ and $ a = \alpha/\tau $. We show that $ F_n({\tau},{{\alpha}}) $ commutes with a nonscalar tridiagonal matrix $ T_n({\tau},{{\alpha}}) $ for all values of the shift parameter $\alpha$, as long as the base $ q = e^{2\pi{{\text{\textit{\r{\i}}}}} \tau} $ determined by $ \tau $ is an $n$th root of unity, and also for all values of $\tau$ in the "centered" case corresponding to $ \alpha = 0 $. Furthermore, we show that, in certain more specialized cases, $ F_n({\tau},{{\alpha}}) $ also commutes with a matrix $ X_n({\tau},{{\alpha}}) $ that has $ \pm 1 $ in the upper-right and lower-left corners and is otherwise tridiagonal. In most cases, $ T_n({\tau},{{\alpha}}) $ and $ X_n({\tau},{{\alpha}}) $ are essentially the only matrices of their band-structure that commute with $ F_n({\tau},{{\alpha}})$.

[1]  W. W. Sawyer On the Matrix With Elements 1/(r+s-1) , 1974, Canadian Mathematical Bulletin.

[2]  G. Bongiovanni,et al.  One-dimensional and two-dimensional generalised discrete fourier transforms , 1976 .

[3]  Richard O. Rowlands The odd discrete Fourier transform , 1976, ICASSP.

[4]  M. Bellanger,et al.  Odd-time odd-frequency discrete Fourier transform for symmetric real-valued series , 1976, Proceedings of the IEEE.

[5]  F. Grünbaum Eigenvectors of a Toeplitz Matrix: Discrete Version of the Prolate Spheroidal Wave Functions , 1981 .

[6]  F. Grünbaum Toeplitz matrices commuting with tridiagonal matrices , 1981 .

[7]  B. Dickinson,et al.  Eigenvectors and functions of the discrete Fourier transform , 1982 .

[8]  F. Grünbaum,et al.  A remark on Hilbert's matrix , 1982 .

[9]  F. Grünbaum,et al.  The eigenvectors of the discrete Fourier transform: A version of the Hermite functions , 1982 .

[10]  David H. Bailey,et al.  The Fractional Fourier Transform and Applications , 1991, SIAM Rev..

[11]  Ronald L. Graham,et al.  Concrete Mathematics, a Foundation for Computer Science , 1991, The Mathematical Gazette.

[12]  Stephen A. Martucci,et al.  Symmetric convolution and the discrete sine and cosine transforms , 1993, IEEE Trans. Signal Process..

[13]  Gilbert Strang,et al.  The Discrete Cosine Transform , 1999, SIAM Rev..

[14]  Z. Zalevsky,et al.  The Fractional Fourier Transform: with Applications in Optics and Signal Processing , 2001 .