Displacement structure approach to Chebyshev-Vandermonde and related matrices (Q1892633)

Let \(x:= (x_ 1,\dots, x_ n)\in \mathbb{C}^ n\), and let \(T_ j\) and \(U_ j\) denote the \(j\)th Chebyshev polynomials of the first and second kinds, respectively. Then the Chebyshev-Vandermonde matrices \(V_ T(x)\) and \(V_ U(x)\) are the \(n\times n\) matrices with \((i,j)\)th entries \(T_{j- 1}(x_ i)\) and \(U_{j- 1}(x_ i)\), respectively. These matrices have been widely studied and fast \(O(n^ 2)\)-algorithms for inverting them are known [see, for example, \textit{I. Gohberg} and the second author, Numer. Math. 67, No. 1, 71-92 (1994; Zbl 0791.65013)]. Taking \(D:= \text{diag}(x_ 1^{- 1},\dots, x^{-1}_ n)\) and \(W\) as a suitable constant matrix, the authors show that the ``displacement operator'' \(R\to DR- RW\) maps each of \(V_ U(x)\) and \(V_ T(x)\text{diag}(1/2, 1,\dots, 1)\) onto a matrix of rank 1 from which the original matrix can be easily recomputed. They use this to rederive and generalize earlier results on the Chebyshev-Vandermonde matrices and their inverses. The authors acknowledge that some similar results were obtained by \textit{G. Heinig}, \textit{W. Hoppe} and \textit{K. Rost} [Wiss. Z. Techn. Univ. Karl- Marx-Stadt 31, No. 2, 196-202 (1989; Zbl 0709.41001)]. \{For a survey of general displacement operators see the first author and \textit{A. H. Sayed}, ``Displacement structure: Theory and applications'', SIAM Review (to appear)\}.