Cones, graphs and optimal scalings of matrices

Characterizations are given of the optimal scalings of a complex square matrix within its diagonal similarity class and its restricted diagonal equivalence class with respect to the maximum element norm. The characterizations are in terms of a finite number of products, principally circuit and diagonal products. The proofs proceed by reducing the optimal scaling problems from the multiplicative matrix level in succession to an additive matrix level, a graph theoretic level, and a geometric level involving duality theorems for cones. At the geometric level, the diagonal similarity and the restricted diagonal equivalence problems are unified.     

Characterizations of EP matrices and weighted-EP matrices

A square complex matrix A is said to be EP if A and its conjugate transpose A^∗ have the same range. In this paper, we first collect a group of known characterizations of EP matrix, and give some new characterizations of EP matrices. Then, we define weighted-EP matrix, and present a wealth of characterizations for weighted-EP matrix through various rank formulas for matrices and their generalized inverses.     

Computing optimal scalings by parametric network algorithms

Asymmetric scaling of a square matrixA 0 is a matrix of the formXAX ^–1 whereX is a nonnegative, nonsingular, diagonal matrix having the same dimension ofA. Anasymmetric scaling of a rectangular matrixB 0 is a matrix of the formXBY ^–1 whereX andY are nonnegative, nonsingular, diagonal matrices having appropriate dimensions. We consider two objectives in selecting a symmetric scaling of a given matrix. The first is to select a scalingA of a given matrixA such that the maximal absolute value of the elements ofA is lesser or equal that of any other corresponding scaling ofA. The second is to select a scalingB of a given matrixB such that the maximal absolute value of ratios of nonzero elements ofB is lesser or equal that of any other corresponding scaling ofB. We also consider the problem of finding an optimal asymmetric scaling under the maximal ratio criterion (the maximal element criterion is, of course, trivial in this case). We show that these problems can be converted to parametric network problems which can be solved by corresponding algorithms.This research was supported by NSF Grant ECS-83-10213.     

Characterizations of totally balanced matrices

A (0,1)-matrix is totally balanced if it does not contain as a submatrix the incidence matrix of any cycle of length at least 3. Several alternative characterizations of these matrices are presented. These characterizations follow from properties of strongly chordal graphs, studied by Farber, and maximal totally balanced matrices, studied by Anstee. Using these characterizations, efficient recognition algorithms for totally balanced matrices are presented. In addition, a new completion algorithm for building a maximal totally balanced matrix from an arbitrary totally balanced matrix follows from these results.     

Characterizations of products of symmetric matrices

Characterizations are obtained for matrices C of the form C = AΣ, where A, Σ are n×n matrices over the real field such that A is symmetric and C is nonnegative definite. Among others, a proof of recent generalization of Cochran's theorem is given.     

Characterizations of sign patterns of inverse-positive matrices

A result of Johnson, Leighton, and Robinson characterizing sign patterns of real matrices with nonzero entries whose inverses are (entrywise) positive is generalized. The restriction to matrices with nonzero entries is removed, and additional five equivalent conditions are established. One of them, using a graph-theoretical approach, expedites a simple criterion for recognition of such sign patterns.     

Characterizations of the beta distribution on symmetric matrices

This paper extends to the beta-Wishart distribution on symmetric matrices, two characterizations of the beta distributions on , due to Seshadri and Wesolowski and based on some properties of constancy regression.     

Characterizations of the optimal stable allocation mechanism

The stable allocation problem is the generalization of (0,1)-matching problems to the allocation of real numbers (hours or quantities) between two separate sets of agents. The same unique-optimal matching (for one set of agents) is characterized by each of three properties: “efficiency”, “monotonicity”, and “strategy-proofness”.     

Characterizations of rectangular totally and strictly totally positive matrices

An n×m real matrix A is said to be totally positive (strictly totally positive) if every minor is nonnegative (positive). In this paper, we study characterizations of these classes of matrices by minors, by their full rank factorization and by their thin QR factorization.     

Characterizations of an approximate minimum in optimal control

By using the classical variational methods based on geodesic coverings of a domain and on Hilbert's independent integral, further characterizations of an approximate solution in problems of control are described. The starting point is the Ekeland-type characterization, the variational principle. As consequences, sufficient conditions for optimality are obtained in a form similar to the Weierstrass conditions from the calculus of variations.The author is grateful to the referee for the valuable counterexamples to the first version of the paper.     

Characterizations of optimal solution sets of convex infinite programs

In this paper, several Lagrange multiplier characterizations of the solution set of a convex infinite programming problem are given. Characterizations of solution sets of cone-constrained convex programs are derived as well. The procedure is then adopted to a semi-convex problem with convex constraints. For this problem, we present firstly a necessary and sufficient condition for optimality and secondly a characterization of its optimal solution set, based on a Lagrange multiplier associated with a given solution and on directional derivatives of the objective function.      

A note on optimal block-scaling of matrices

Summary After pointing out that two recent results on optimal blockscaling are equivalent, a new short and simple proof of both results is given.Dedicated to Prof. Dr. F.L. Bauer on the occasion of his 60th birthday     

Covariance and correlation matrices of an optimal stochastic system

The orthogonal projection lemma is applied to derive differential equations for the correlation and covariance matrices of filter estimators, whose solution is essentially simplified in the stationary state. The matrix similarity transformation and pseudocommutativity are used to construct an exact algebraic expression for the estimate covariance and correlation matrices.Translated from Vychislitel'naya i Prikladnaya Matematika, No. 72, pp. 88–94, 1990.     

Classes of matrices associated with the optimal assignment problem

Let E=[e_ij] be a matrix with integral elements, and let x be an indeterminate defined over the rational field Q. We investigate matrices of the form X=[x^e_ij] (i = 1,…, m; j = 1,…, n; m ⩽ n). We may multiply the lines (rows or columns) of the matrix X by suitable integral powers of x in various ways and thereby transform X into a matrix Y=[x^f_ij] such that the f_ij are nonnegative integers and each line of Y contains at least one element x⁰ = 1. We call Y a normalized form of X, and we denote by S(X) the class of all normalized forms associated with a given matrix X. The classes S(X) have a fascinating combinatorial structure, and the present paper is a natural outgrowth and extension of an earlier study. We introduce new concepts such as an elementary transformation called an interchange. We prove, for example, that two matrices in the same class are transformable into one another by interchanges. Our analysis of the class S(X) also yields new insights into the structure of the optimal assignments of the matrix E by way of the diagonal products of the matrix X.     

Upper bounds for nearly optimal diagonal scaling of matrices

Optimal diagonal scaling of an n×n matrix A consists in finding a diagonal matrix D that minimizes a condition number of AD. Often a nearly optimal scaling of A is achieved by taking a diagonal matrix D₁ such that all diagonal elements of D₁A^TAD₁ are equal to one. It is shown in this paper that the condition number of AD₁ can be at least (n/2)^1/2 times the minimal one. Some questions for a further research are posed.     

Essentially optimal computation of the inverse of generic polynomial matrices

We present an inversion algorithm for nonsingular n×n matrices whose entries are degree d polynomials over a field. The algorithm is deterministic and, when n is a power of two, requires O^∼(n³d) field operations for a generic input; the soft-O notation O^∼ indicates some missing log(nd) factors. Up to such logarithmic factors, this asymptotic complexity is of the same order as the number of distinct field elements necessary to represent the inverse matrix.     

Constrained minimax approximation and optimal preconditioned for Toeplitz matrices

A good preconditioner is extremely important in order for the conjugate gradients method to converge quickly. In the case of Toeplitz matrices, a number of recent studies were made to relate approximation of functions to good preconditioners. In this paper, we present a new result relating the quality of the Toeplitz preconditionerC for the Toeplitz matrixT to the Chebyshev norm (f– g)/f, wheref and g are the generating functions forT andC, respectively. In particular, the construction of band-Toeplitz preconditioners becomes a linear minimax approximation problem. The case whenf has zeros (but is nonnegative) is especially interesting and the corresponding approximation problem becomes constrained. We show how the Remez algorithm can be modified to handle the constraints. Numerical experiments confirming the theoretical results are presented.     

An optimal bound for the spectral variation of two matrices

We establish a bound for the spectral variation of two complex n × n matrices A,B in terms of ∥A∥, ∥B∥, and ∥A ? B∥. Here ∥ ∥ denotes the spectral norm. It is always better than a bound previously given by Bhatia and Friedland, and it is optimal. We describe the set of pairs A,B for which the bound is attained.