Simultaneous block diagonalization of two real symmetric matrices

The simultaneous diagonalization of two real symmetric (r.s.) matrices has long been of interest. This subject is generalized here to the following problem (this question was raised by Dr. Olga Taussky-Todd, my thesis advisor at the California Institute of Technology): What is the first simultaneous block diagonal structure of a nonsingular pair of r.s. matrices ? For example, given a nonsingular pair of r.s. matrices S and T, which simultaneous block diagonalizations X′SX = diag(A₁, , A_k), X′TX = diag(B₁,, B_k) with dim A_i = dim B_i and X nonsingular are possible for 1 ? k ? n; and how well defined is a simultaneous block diagonalization for which k, the number of blocks, is maximal? Here a pair of r.s. matrices S and T is called nonsingular if S is nonsingular.If the number of blocks k is maximal, then one can speak of the finest simultaneous block diagonalization of S and T, since then the sizes of the blocks A_i are uniquely determined (up to permutations) by any set of generators of the pencil P(S, T) = {aS + bT|a, tb ε R} via the real Jordan normal form of S^?1T. The proof uses the canonical pair form theorem for nonsingular pairs of r.s. matrices. The maximal number k and the block sizes dim A_i are also determined by the factorization over C of ? (λ, μ) = det(λS + μT) for λ, μ ε R.     

On some algebraic problems in connection with general elgenvalue algorithms

Two real matrices A,B are S-congruent if there is a nonsingular upper triangular matrix R such that A = R^TBR. This congruence relation is studied in the set of all nonsingular symmetric and that of all skew-symmetric matrices. Invariants and systems of representation are give. The results are applied to the question of decomposability of a matrix in a product of an isometry and an upper triangular matrix, a problem crucial in eigenvalue algorithms.     

Transposing a matrix by similarity operations

A matrix T is said to co-transpose a square matrix A if T^?1AT=A′ and T^?1A′T=A. For every n?3 there exists a real n×n matrix which cannot be co-transposed by any matrix. However, it is shown that the following classes of real matrices can be co-transposed by a symmetric matrix of order two: 2×2 matrices, normal matrices, and matrices whose square is symmetric.     

Linear transformations on symmetric matrices that preserve commutativity

It is shown that if a nonsingular linear transformation T on the space of n-square real symmetric matrices preserves the commutativity, where n ?3, then T(A) = λQAQ^t + Q(A)I_n for all symmetric matricesA, for some scalar λ, orthogonal matrix Q, and linear functional Q.     

G-matrices,J-orthogonal matrices,and their sign patterns

A real matrix A is a G-matrix if A is nonsingular and there exist nonsingular diagonal matrices D₁ and D₂ such that A^?T = D₁AD₂, where A^?T denotes the transpose of the inverse of A. Denote by J = diag(±1) a diagonal (signature) matrix, each of whose diagonal entries is +1 or ?1. A nonsingular real matrix Q is called J-orthogonal if Q^TJQ = J. Many connections are established between these matrices. In particular, a matrix A is a G-matrix if and only if A is diagonally (with positive diagonals) equivalent to a column permutation of a J-orthogonal matrix. An investigation into the sign patterns of the J-orthogonal matrices is initiated. It is observed that the sign patterns of the G-matrices are exactly the column permutations of the sign patterns of the J-orthogonal matrices. Some interesting constructions of certain J-orthogonal matrices are exhibited. It is shown that every symmetric staircase sign pattern matrix allows a J-orthogonal matrix. Sign potentially J-orthogonal conditions are also considered. Some examples and open questions are provided.     

Pairs of alternating forms and products of two skew-symmetric matrices

Let k be a field, and let S,T,S₁,T₁ be skew-symmetric matrices over k with S,S₁ both nonsingular (if k has characteristic 2, a skew-symmetric matrix is a symmetric one with zero diagonal). It is shown that there exists a nonsingular matrix P over k with P'SP = S₁, P'TP = T₁ (where P' denotes the transpose of P) if and only if S^-1T and S^-1₁T₁ are similar. It is also shown that a 2m×2m matrix over k can be factored as ST, with S,T skew-symmetric and S nonsingular, if and only if A is similar to a matrix direct sum B⊕B where B is an m×m matrix over k. This is equivalent to saying that all elementary divisors of A occur with even multiplicity. An extension of this result giving necessary and sufficient conditions for a square matrix to be so expressible, without assuming that either S or T is nonsingular, is included.     

On the reduction of pairs of hermitian or symmetric matrices to diagonal form by congruence

Let A₁, A₂ be given n-by-n Hermitian or symmetric matrices, and consider the simultaneous transformations A_i→SA_iS^* if A_i is Hermitian or A_i→SA_iS^T if A_i is symmetric. We give necessary and sufficient conditions for the existence of a unitary S which reduces both A₁ and A₂ to diagonal form in this way. When at least one of A₁ or A₂ is nonsingular, we give necessary and sufficient conditions for a reduction of this sort with a nonsingular S. These results are a generalization of the classical theorem from mechanics that a positive definite matrix and a Hermitian matrix can always be diagonalized simultaneously by a nonsingular congruence.     

Congruence of Hermitian matrices by Hermitian matrices

Two Hermitian matrices A,B∈M_n(C) are said to be Hermitian-congruent if there exists a nonsingular Hermitian matrix C∈M_n(C) such that B=CAC. In this paper, we give necessary and sufficient conditions for two nonsingular simultaneously unitarily diagonalizable Hermitian matrices A and B to be Hermitian-congruent. Moreover, when A and B are Hermitian-congruent, we describe the possible inertias of the Hermitian matrices C that carry the congruence. We also give necessary and sufficient conditions for any 2-by-2 nonsingular Hermitian matrices to be Hermitian-congruent. In both of the studied cases, we show that if A and B are real and Hermitian-congruent, then they are congruent by a real symmetric matrix. Finally we note that if A and B are 2-by-2 nonsingular real symmetric matrices having the same sign pattern, then there is always a real symmetric matrix C satisfying B=CAC. Moreover, if both matrices are positive, then C can be picked with arbitrary inertia.     

Author index to volume 20

This paper considers the conjecture that given a real nonsingular matrix A, there exist a real diagonal matrix Λ with $\text{¦λ}{}_{\mathrm{ii}}\text{λ = 1}$ and a permutation matrix P such that (ΛPA) is positive stable. The conjecture is shown to be true for matrices of order 3 or less and may not be true for higher order matrices. A counterexample is presented in terms of a matrix of order 65. In showing this, an interesting matrix M_l of order 2^l = 64, which satisfies the matrix equation 2^l-1(M_l + M^T_l), has been used. The stability analysis is done by first decomposing the nonsingular matrix into its polar form. Some interesting results are presented in the study of eigenvalues of a product of orthogonal matrices. A simple function is derived in terms of these orthogonal matrices, which traces a hysteresis loop.     

Some angularity and inertia theorems related to normal matrices

Motivated by the definition of the inertia, introduced by Ostrowski and Schneider, a notion of angularity of a matrix is defined. The angularity characterizes the distribution of arguments of eigenvalues of a matrix. It is proved that if B and C are nonsingular matrices, then B¹AB and C¹AC have the same angularity provided they are diagonal. Some well-known inertia theorems (e.g. Sylvester's law) have been deduced as corollaries of this result. The case when C is permitted to be singular is discussed next. Finally, we prove that (a) any linear transformation T, on the set of n by n complex matrices, mapping Hermitian matrices into themselves and preserving the inertia of each Hermitian matrix is of the form T(A)=C¹AC or T(A)=C^1LA′C where C is some nonsingular matrix, and (b) any linear transformation T mapping normal matrices into normal matrices and preserving the angularity of each normal matrix is also of one of the above forms, but with C=kU where k≠0 and U is unitary.     

On a characterization of tridiagonal matrices by M. Fiedler

In this note two new proofs are given of the following characterization theorem of M. Fiedler: Let C_n, n?2, be the class of all symmetric, real matrices A of order n with the property that rank (A + D) ? n - 1 for any diagonal real matrix D. Then for any A ε C_n there exists a permutation matrix P such that PAP^T is tridiagonal and irreducible.     

A trace inequality for M-matrices and the symmetrizability of a real matrix by a positive diagonal matrix

The question of whether a real matrix is symmetrizable via multiplication by a diagonal matrix with positive diagonal entries is reduced to the corresponding question for M-matrices and related to Hadamard products. In the process, for a nonsingular M-matrix A, it is shown that tr(A^-1A^T) ? n, with equality if and only if A is symmetric, and that the minimum eigenvalue of A^-1 ° A is ? 1 with equality in the irreducible case if and only if A is positive diagonally symmetrizable.     

Retarded equations on the sphere induced by linear equations

Using a Poincaré compactification, the linear homogeneous system of delay equations {x = Ax(t ? 1) (A is an n × n real matrix) induces a delay system π(A) on the sphere Sⁿ. The points at infinity belong to an invariant submanifold S^{n ? 1} of Sⁿ. For an open and dense set of 2 × 2 matrices A with distinct eigenvalues, the system π(A) has only hyperbolic critical points (including the critical points at infinity). For an open and dense set of 2 × 2matrices $\text{A}$ with complex eigenvalues, the nonwandering set at infinity is the union of an odd number of hyperbolic periodic orbits; if $\text{(}\text{det}\text{A}\text{)}{}^{\mathrm{<mtext>1</mtext><mtext>2</mtext>}}\text{<}\text{3π}\text{2}$, the restriction of $\text{π(}\text{A}\text{)}$ to S¹ is Morse-Smale. For n = 1 there exist periodic orbits of period 4 provided that $\text{?A >}\text{π}\text{2}$ and Hopf bifurcation of a center occurs for ?A near $\text{(}\text{π}\text{2}\text{) + 2kπ, k ? Z}$.     

On the semiaxes of touching quadrics

In this paper the semiaxes of quadrics which touch each other are investigated. If two quadrics Q, Q′ are given, it turns out that we can rotate Q′ such that it touches Q in two opposite points if and only if the squared semiaxes of Q, Q′ do not separate each other. This is equivalent to the statement that for two symmetric matrices A,B there is an orthogonal matrix S with det(A — S ^TBS) = 0 if and only if the eigenvalues of A and B do not separate each other.     

The Lyapunov order for real matrices

The real Lyapunov order in the set of real n×n matrices is a relation defined as follows: A?B if, for every real symmetric matrix S, SB+B^tS is positive semidefinite whenever SA+A^tS is positive semidefinite. We describe the main properties of the Lyapunov order in terms of linear systems theory, Nevenlinna-Pick interpolation and convexity.     

A note on generating <Emphasis Type="Italic">P</Emphasis>-matrices

We prove that for any \({A,B\in\mathbb{R}^{n\times n}}\) such that each matrix S satisfying min(A, B) ≤ S ≤ max(A, B) is nonsingular, all four matrices A ^?1 B, AB ^?1, B ^?1 A and BA ^?1 are P-matrices. A practical method for generating P-matrices is drawn from this result.     

Pertubation bounds for the definite generalized eigenvalue problem

Let A and B be Hermitian matrices, and let c(A, B) = inf{|x^H(A + iB)x|:6 = 1}. The eigenvalue problem Ax = λBx is called definite if c(A, B)>0. It is shown that a definite problem has a complete system of eigenvectors and that its eigenvalues are real. Under pertubations of A and B, the eigenvalues behave like the eigenvalues of a Hermitian matrix in the sense that there is a 1-1 pairing of the eigenvalues with the perturbed eigenvalues and a uniform bound for their differences (in this case in the chordal metric). Pertubation bounds are also developed for eigenvectors and eigenspaces.     

On sesquilinear forms over fields with involution

Let k be a field of characteristic ≠2 with an involution σ. A matrix A is split if there is a change of variables Q such that (Q^σ)^TAQ consists of two complementary diagonal blocks. We classify all matrices that do not split. As a consequence we obtain a new proof for the following result. Given a square matrix A there is a matrix S such that (S^σ)^TAS=A^T and S^σS=I.     

Linear transformations on symmetric matrices that preserve the permanent

It is shown that if a linear transformation T on the space of n-square symmetric matrices over any subfield of the real field preserves the permanent, where n ? 3, then T(A)= ± PAP^t for all symmetric matrices A and a fixed generalized permutation matrix P with per P= ± 1.     

A theorem in combinatorial matrix theory

Let Z be a matrix of order n, and suppose that the elements of Z consist of only two elements x and y, which are elements of a field F. We call Z an (x,y)-matrix over F. In this paper we study the matrix equation ZEZ^T = D+λJ, where Z is a nonsingular (x,y)-matrix over F, Z^T is the transpose of Z, D and E are nonsingular diagonal matrices, J is the matrix of 1's and λ is an element of F. Our main theorem shows that the column sums of Z are severely restricted. This result generalizes a number of earlier investigations that deal with symmetric block designs and related configurations. The problems that emerge are of interest from both a combinatorial and a matrix theoretic point of view.