Computing a matrix symmetrizer exactly using modified multiple modulus residue arithmetic

A symmetric solution X satisfying the matrix equation XA = A^tX is called a symmetrizer of the matrix A. A general algorithm to compute a matrix symmetrizer is obtained. A new multiple-modulus residue arithmetic called floating-point modular arithmetic is described and implemented on the algorithm to compute an error-free matrix symmetrizer.     

Symmetrizing a Hessenberg matrix: Designs for VLSI parallel processor arrays

A symmetrizer of a nonsymmetric matrix A is the symmetric matrixX that satisfies the equationXA =A ^tX, wheret indicates the transpose. A symmetrizer is useful in converting a nonsymmetric eigenvalue problem into a symmetric one which is relatively easy to solve and finds applications in stability problems in control theory and in the study of general matrices. Three designs based on VLSI parallel processor arrays are presented to compute a symmetrizer of a lower Hessenberg matrix. Their scope is discussed. The first one is the Leiserson systolic design while the remaining two, viz., the double pipe design and the fitted diagonal design are the derived versions of the first design with improved performance.     

Applications of shortest path algorithms to matrix scalings

Summary A symmetric scaling of a nonnegative, square matrixA is a matrixXAX ^–1, whereX is a nonsingular, nonnegative diagonal matrix. By associating a family of weighted directed graphs with the matrixA we are able to adapt the shortest path algorithms to compute an optimal scaling ofA, where we call a symmetric scalingA ofA optimal if it minimizes the maximum of the ratio of non-zero elements.Dedicated to Professor F.L. Bauer on the occasion of his 60th birthdayThe first author was partially supported by the Deutsche Forschungsgemeinschaft under grant GO 270/3, the second author by the U.S. National Science Foundation under grand MCS-8026132     

The symmetric procrustes problem

The following symmetric Procrustes problem arises in the determination of the strain matrix of an elastic structure: find the symmetric matrixX which minimises the Frobenius (or Euclidean) norm ofAX — B, whereA andB are given rectangular matrices. We use the singular value decomposition to analyse the problem and to derive a stable method for its solution. A perturbation result is derived and used to assess the stability of methods based on solving normal equations. Some comparisons with the standard, unconstrained least squares problem are given.     

Symmetrizers of matrices

A symmetrizer of a given pair of matrices, A and B, is defined as a matrix X for which the product AXB is symmetric. Right and left symmetrizers of a given matrix A are defined accordingly. The main results of the paper are general representations of all three types of symmetrizers. The problem considered arose in connection with certain questions pertaining to admissible linear estimation in a Gauss-Markoff model.     

A Jacobi eigenreduction algorithm for definite matrix pairs

Summary We propose a Jacobi eigenreduction algorithm for symmetric definite matrix pairsA, J of small to medium-size real symmetric matrices withJ ²=I,J diagonal (neitherJ norA itself need be definite). Our Jacobi reduction works only on one matrix and usesJ-orthogonal elementary congruences which include both trigonometric and hyperbolic rotations and preserve the symmetry throughout the process. For the rotation parameters only the pivotal elements of the current matrix are needed which facilitates parallelization. We prove the global convergence of the method; the quadratic convergence was observed in all experiments. We apply our method in two situations: (i) eigenreducing a single real symmetric matrix and (ii) eigenreducing an overdamped quadratic matrix pencil. In both cases our method is preceded by a symmetric indefinite decomposition and performed in its one-sided variant on the thus obtained factors. Our method outdoes the standard methods like standard Jacobi orqr/ql in accuracy in spite of the use of hyperbolic transformations which are not orthogonal (a theoretical justification of this behaviour is made elsewhere). The accuracy advantage of our method can be particularly drastic if the eigenvalues are of different order. In addition, in working with quadratic pencils our method is shown to either converge or to detect non-overdampedness.     

Singular-value-like decomposition for complex matrix triples

The classical singular value decomposition for a matrix A∈C^m×n is a canonical form for A that also displays the eigenvalues of the Hermitian matrices AA^∗ and A^∗A. In this paper, we develop a corresponding decomposition for A that provides the Jordan canonical forms for the complex symmetric matrices and . More generally, we consider the matrix triple , where are invertible and either complex symmetric or complex skew-symmetric, and we provide a canonical form under transformations of the form , where X,Y are nonsingular.     

Quadratically constrained least squares and quadratic problems

Summary We consider the following problem: Compute a vectorx such that Ax–b₂=min, subject to the constraint x₂=. A new approach to this problem based on Gauss quadrature is given. The method is especially well suited when the dimensions ofA are large and the matrix is sparse.It is also possible to extend this technique to a constrained quadratic form: For a symmetric matrixA we consider the minimization ofx ^T A x–2b ^T x subject to the constraint x₂=.Some numerical examples are given.This work was in part supported by the National Science Foundation under Grant DCR-8412314 and by the National Institute of Standards and Technology under Grant 60NANB9D0908.     

On symmetrizers for Gauss methods

Summary In this paper the maximum attainable order of a special class of symmetrizers for Gauss methods is studied. In particular, it is shown that a symmetrizer of this type for thes-stage Gauss method can attain order 2s-1 only for 1 s 3, and that these symmetrizers areL-stable. A classification of the maximum attainable order of symmetrizers for some higher stages is presented. AnL-stable symmetrizer is also shown to exist for each of the methods studied.     

The simultaneous computation of a few of the algebraically largest and smallest eigenvalues of a large,sparse, symmetric matrix

A heuristic argument and supporting numerical results are given to demonstrate that a block Lanczos procedure can be used to compute simultaneously a few of the algebraically largest and smallest eigenvalues and a corresponding eigenspace of a large, sparse, symmetric matrixA. This block procedure can be used, for example, to compute appropriate parameters for iterative schemes used in solving the equationAx=b. Moreover, if there exists an efficient method for repeatedly solving the equation (A–I)X=B, this procedure can be used to determine the interior eigenvalues (and corresponding eigenvectors) ofA closest to .     

Computing the inertia from sign patterns

A symmetric matrix A is said to be sign-nonsingular if every symmetric matrix with the same sign pattern as A is nonsingular. Hall, Li and Wang showed that the inertia of a sign-nonsingular symmetric matrix is determined uniquely by its sign pattern. The purpose of this paper is to present an efficient algorithm for computing the inertia of such symmetric matrices. The algorithm runs in time for a symmetric matrix of order n with m nonzero entries. In addition, it is shown to be NP-complete to decide whether the inertia of a given symmetric matrix is not determined by its sign pattern.     

Localization of dominant eigenpairs and planted communities by means of Frobenius inner products

We propose a new localization result for the leading eigenvalue and eigenvector of a symmetric matrix A. The result exploits the Frobenius inner product between A and a given rank-one landmark matrix X. Different choices for X may be used, depending on the problem under investigation. In particular, we show that the choice where X is the all-ones matrix allows to estimate the signature of the leading eigenvector of A, generalizing previous results on Perron-Frobenius properties of matrices with some negative entries. As another application we consider the problem of community detection in graphs and networks. The problem is solved by means of modularity-based spectral techniques, following the ideas pioneered by Miroslav Fiedler in mid-’70s.     

On matrix equations and

With the help of the concept of Kronecker map, an explicit solution for the matrix equation X−AXF=C is established. This solution is neatly expressed by a symmetric operator matrix, a controllability matrix and an observability matrix. In addition, the matrix equation is also studied. An explicit solution for this matrix equation is also proposed by means of the real representation of a complex matrix. This solution is neatly expressed by a symmetric operator matrix, two controllability matrices and two observability matrices.     

Open selections on smooth fans

Given a dendroid X, an open selection is an open map such that s(A)∈A for every A∈C(X). We show that a smooth fan X admits an open selection if and only if X is locally connected.     

On the linearity of regression

Summary A stochastic process X={X _t :tT| is called spherically generated if for each random vector , there exist a random vector Y=(Y₁,..., Y _m) with a spherical (radially symmetric) distribution and a matrix A such that X is distributed as AY. X is said to have the linear regression property if (X ₀¦X ₁,..., X _n) is a linear function of X ₁,..., X _n whenever the X _j's are elements of the linear span of X. It is shown that providing the linear span of X has dimension larger than two, then X has the linear regression property if and only if it is spherically generated. The class of symmetric stable processes which are spherically generated is shown to coincide with the class of socalled sub-Gaussian processes, characterizing those stable processes having the linear regression property.This research was supported by a grant from the University of Wisconsin-Milwaukee     

The expected values of zonal polynomials of elliptically contoured distributions

Based on the work of Constantine, Waal, Shah, Khatri, and Muirhead, we have made further studies on the expected values of zonal polynomials and elementary symmetric functions of a Wishart matrix in this paper. Under certain conditions, we found the expected values of zonal polynomials and elementary symmetric functions of matrixXX, where the matrixX has matrix elliptically contoured distribution and obtain two interesting results. Three examples are given.     

Difference sets and inverting the difference operator

For a setA of non-negative numbers, letD(A) (the difference set ofA) be the set of nonnegative differences of elements ofA, and letD ^k be thek-fold iteration ofD. We show that for everyk, almost every set of non-negative integers containing 0 arises asD ^k (A) for someA. We also give sufficient conditions for a setA to be the unique setX such that 0X andD ^k (X)=D ^k (A). We show that for eachm there is a setA such thatD(X)=D(A) has exactly 2 ^m solutionsX with 0X.This work was supported by grants DMS 92-02833 and DMS 91-23478 from the National Science Foundation. The first author acknowledges the support of the Hungarian National Science Foundation under grants, OTKA 4269, and OTKA 016389, and the National Security Agency (grant No. MDA904-95-H-1045).Lee A. Rubel died March 25, 1995. He is very much missed by his coauthors.     

A hybrid Arnoldi-Faber iterative method for nonsymmetric systems of linear equations

Summary We present here a new hybrid method for the iterative solution of large sparse nonsymmetric systems of linear equations, say of the formAx=b, whereA ^{N, N} , withA nonsingular, andb ^N are given. This hybrid method begins with a limited number of steps of the Arnoldi method to obtain some information on the location of the spectrum ofA, and then switches to a Richardson iterative method based on Faber polynomials. For a polygonal domain, the Faber polynomials can be constructed recursively from the parameters in the Schwarz-Christoffel mapping function. In four specific numerical examples of non-normal matrices, we show that this hybrid algorithm converges quite well and is approximately as fast or faster than the hybrid GMRES or restarted versions of the GMRES algorithm. It is, however, sensitive (as other hybrid methods also are) to the amount of information on the spectrum ofA acquired during the first (Arnoldi) phase of this procedure.     

On a theorem of Stein-Rosenberg type in interval analysis

Summary In classical numerical analysis the asymptotic convergence factor (R ₁-factor) of an iterative processx ^m+1=Ax^m+b coincides with the spectral radius of then×n iteration matrixA. Thus the famous Theorem of Stein and Rosenberg can at least be partly reformulated in terms of asymptotic convergence factor. Forn×n interval matricesA with irreducible upper bound and nonnegative lower bound we compare the asymptotic convergence factor ( _T ) of the total step method in interval analysis with the factor _S of the corresponding single step method. We derive a result similar to that of the Theorem of Stein and Rosenberg. Furthermore we show that _S can be less than the spectral radius of the real single step matrix corresponding to the total step matrix |A| where |A| is the absolute value ofA. This answers an old question in interval analysis.