Preconditioners for ill‐posed Toeplitz matrices with differentiable generating functions

Both theoretical analysis and numerical experiments in the literature have shown that the Tyrtyshnikov circulant superoptimal preconditioner for Toeplitz systems can speed up the convergence of iterative methods without amplifying the noise of the data. Here we study a family of Tyrtyshnikov‐based preconditioners for discrete ill‐posed Toeplitz systems with differentiable generating functions. In particular, we show that the distribution of the eigenvalues of these preconditioners has good regularization features, since the smallest eigenvalues stay well separated from zero. Some numerical results confirm the regularization effectiveness of this family of preconditioners. Copyright © 2009 John Wiley & Sons, Ltd.     

Toeplitz matrices with Toeplitz inverses revisited

A necessary and sufficient condition derived by Huang and Cline for a nonsingular Toeplitz matrix to have a Toeplitz inverse is shown to hold under more general hypotheses than indicated by them.     

Toeplitz matrices commuting with tridiagonal matrices

we prove that if R is a nonscalar Toeplitz matrix R_{i, j}=r_?i?j? which commutes with a tridiagonal matrix with simple spectrum, then

$\text{r}{}_{\mathrm{k}}\text{r}{}_1\text{=}\text{u}{}_{\mathrm{k-1}}\text{r}{}_2\text{r}{}_1\text{cos p}\text{u}{}_{\mathrm{k-1}}\text{(cos p)}$

, k=4, 5,…, with U_k the Chebychev polynomial of the second kind, where p is determined from

$\text{cos p=}\text{1}\text{2}\text{r}{}^2{}_1\text{?r}{}_1\text{r}{}_3\text{r}{}^2{}_2\text{?r}{}_1\text{r}{}_3$

.     

Toeplitz approximate inverse preconditioner for banded Toeplitz matrices

In this paper we introduce a new preconditioner for banded Toeplitz matrices, whose inverse is itself a Toeplitz matrix. Given a banded Hermitian positive definite Toeplitz matrixT, we construct a Toepliz matrixM such that the spectrum ofMT is clustered around one; specifically, if the bandwidth ofT is , all but eigenvalues ofMT are exactly one. Thus the preconditioned conjugate gradient method converges in +1 steps which is about half the iterations as required by other preconditioners for Toepliz systems that have been suggested in the literature. This idea has a natural extension to non-banded and non-Hermitian Toeplitz matrices, and to block Toeplitz matrices with Toeplitz blocks which arise in many two dimensional applications in signal processing. Convergence results are given for each scheme, as well as numerical experiments illustrating the good convergence properties of the new preconditioner.Partly supported by a travel fund from the Deutsche Forschungsgemeinschaft.Research supported in part by Oak Ridge Associated Universities grant no. 009707.     

Noncirculant Toeplitz matrices all of whose powers are Toeplitz

Let a, b and c be fixed complex numbers. Let M _n (a, b, c) be the n × n Toeplitz matrix all of whose entries above the diagonal are a, all of whose entries below the diagonal are b, and all of whose entries on the diagonal are c. For 1 ⩽ k ⩽ n, each k × k principal minor of M _n (a, b, c) has the same value. We find explicit and recursive formulae for the principal minors and the characteristic polynomial of M _n (a, b, c). We also show that all complex polynomials in M _n (a, b, c) are Toeplitz matrices. In particular, the inverse of M _n (a, b, c) is a Toeplitz matrix when it exists.     

Nonsymmetric Toeplitz matrices that commute with tridiagonal matrices

Translated from Matematicheskie Zametki, Vol. 55, No. 5, pp. 69–79, May, 1994.     

Band matrices with Toeplitz inverses

It is shown that a square band matrix H=(h_ij) with h_ij=0 for j? i>r and i?j>s, where r+s is less than the order of the matrix, has a Toeplitz inverse if and only if it has a special structure characterized by two polynomials of degrees r and s, respectively.     

Toeplitz matrices with banded inverses

Given a Toeplitz matrix T with banded inverse [i.e., (T^?1)_ij=0 for j?i>p], we show that the elements of T can be expressed in terms of the roots of a polynomial. Then, using properties we have previously established, we generalize this result appropriately to allow singular T and show that the converse also holds. Finally, we give a sufficient condition for the decay of the elements of T as one moves away from the diagonal.     

Infinite extensions of Toeplitz matrices

A formula for the smallest of the ranks of infinite Toeplitz extensions of a finite rectangular Toeplitz matrix is derived. Bibliography: 3 titles._________Translated from Zapiski Nauchnykh Seminarov POMI, Vol. 296, 2003, pp. 5–14.     

QR factorization of Toeplitz matrices

Summary This paper presents a new algorithm for computing theQR factorization of anm×n Toeplitz matrix inO(mn) operations. The algorithm exploits the procedure for the rank-1 modification and the fact that both principal (m–1)×(n–1) submatrices of the Toeplitz matrix are identical. An efficient parallel implementation of the algorithm is possible.     

An inverse problem for Toeplitz matrices

A simplified solution to an inverse problem for Toeplitz matrices using central mass sequences is presented. Some connections with discrete transmission lines are mentioned.     

Approximate inverse-free preconditioners for Toeplitz matrices

In this paper, we propose approximate inverse-free preconditioners for solving Toeplitz systems. The preconditioners are constructed based on the famous Gohberg-Semencul formula. We show that if a Toeplitz matrix is generated by a positive bounded function and its entries enjoys the off-diagonal decay property, then the eigenvalues of the preconditioned matrix are clustered around one. Experimental results show that the proposed preconditioners are superior to other existing preconditioners in the literature.     

The conditioning of Toeplitz band matrices

In this paper, practical conditions to check the well-conditioning of a family of nonsingular Toeplitz band matrices are obtained. All the results are based on the location of the zeros of a polynomial associated with the given family of Toeplitz matrices.The same analysis is also used to derive uniform componentwise bounds for the entries of the inverse matrices in such family.     

The inverses of Toeplitz band matrices

The elements of the inverse of a Toeplitz band matrix are given in terms ofthe solution of a difference equation. The expression for these elements is a quotient of determinants whose orders depend the number of nonzero superdiagonals but not on the order of the matrix. Thus, the formulae are particularly simple for lower triangular and lower Hessenberg Toeplitz matrices. When the number of nonzero superdiagonals is small, sufficient conditions on the solution of the abovementioned difference equation can be given to ensure that the inverse matrix is positive. If the inverse is positive, the row sums can be expressed in terms of the solution of the difference equation.     

Toeplitz Preconditioners for Toeplitz Systems with Nonnegative Generating Functions

The preconditioned conjugate gradient method is employed tosolve Toeplitz systems T_nx = b where the generating functionsof the n-by-n Toeplitz matrices T_n are continuous nonnegativeperiodic functions defined in [–,]. The preconditionedC_n are band Toeplitz matrices with band-widths independent ofn. We prove that the spectra of C_n^-1T_n are uniformly boundedby constants independent of n. In particular, we show that thesolutions of T_nx = b can be obtained in O(nlogn) operations.