Search

The Cayley transform of a matrix A , C (A) = (I + A) − 1 (I − A) , has been investigated in the contexts of skew-symmetry, eigenvalue mapping and as a factorization technique for several matrix positivity classes. Here, we extend the discussion to Cayley transforms of almost skew-Hermitian matrices, positive definite matrices, semipositive matrices, as well as subclasses of the M -matrices, H -matrices and their inverses. [ABSTRACT FROM AUTHOR]

We classify maximal digraphs whose Hermitian spectral radius is at most 2. [ABSTRACT FROM AUTHOR]

Let λ j and a j j , 1 ≤ j ≤ n , be the eigenvalues and the diagonal entries of a Hermitian matrix A , both enumerated in the increasing order. We prove some inequalities that are stronger than the Schur majorisation inequalities ∑ j = 1 r λ j ≤ ∑ j = 1 r a j j , 1 ≤ r ≤ n. [ABSTRACT FROM AUTHOR]

In this paper, we determine the underlying graphs of the mixed graphs with smallest Hermitian eigenvalue greater than − 3. Furthermore, we characterize all mixed graphs on n ≥ 11 vertices with smallest Hermitian eigenvalue greater than − 3. The mixed graphs on n ≤ 10 vertices with smallest Hermitian eigenvalue greater than − 3 could be also obtained easily with the help of computer because their underlying graphs are determined. Roughly speaking, we completely determine the mixed graphs with smallest Hermitian eigenvalue greater than − 3. [ABSTRACT FROM AUTHOR]

Yoopyo Hong proved in 1989 that each Hermitian matrix A is orthogonally *congruent to a matrix of the form ε 1 A 1 ⊕ ⋯ ε r A r ⊕ B 1 ⊕ ⋯ ⊕ B s , in which A 1 , ... , A r , B 1 , ... , B s are uniquely determined by the orthogonal *congruence class of A, and ε 1 , ... , ε r ∈ { 1 , − 1 }. We prove that ε 1 , ... , ε p are uniquely determined by the orthogonal *congruence class of A as well. As an application, we present a canonical form of a pair (A , B) consisting of a Hermitian matrix A and a nonsingular symmetric matrix B with respect to transformations (S ⁎ A S , S T B S) with a nonsingular S. [ABSTRACT FROM AUTHOR]

In the block version of the classical two-sided Jacobi method for the Hermitian eigenvalue problem, the off-diagonal elements of iterated matrix A (k) converge to zero. However, this fact alone does not necessarily guarantee that A (k) converges to a fixed diagonal matrix. The same is true for the matrix of accumulated unitary transformations Q (k). We prove that under certain assumptions A (k) indeed converges to a fixed diagonal matrix, whose diagonal elements are the eigenvalues of the input matrix A. Next it is shown that for a simple eigenvalue the corresponding column of Q (k) converges to the corresponding eigenvector. For a multiple eigenvalue or a cluster of eigenvalues, we prove that the orthogonal projectors constructed from the corresponding columns of Q (k) converge to the orthogonal projector onto the eigenspace corresponding to those eigenvalues. Moreover, the appropriate convergence bounds are obtained for all discussed cases. Convergence results are also valid for the parallel block-Jacobi method with dynamic ordering. The developed theory is illustrated by numerical example. [ABSTRACT FROM AUTHOR]

The main goal of this paper is to prove the following determinantal inequality: det ⁡ (A k + | B s 2 A s 2 | 2 t s ) ≤ det ⁡ (A k + A t B t) ≤ det ⁡ (A k + | A s 2 B s 2 | 2 t s ) for any positive semi-definite matrices A and B , and for all 0 ≤ t ≤ s ≤ k. It generalizes several known determinantal inequalities, and one main consequence of it confirms Lin's conjecture which states that for positive semi-definite matrices A and B , det ⁡ (A 2 + A t B t) ≤ det ⁡ (A 2 + | A B | t) for 0 ≤ t ≤ 2. We conclude with another related determinantal inequality. [ABSTRACT FROM AUTHOR]

In 2017, M. Lin formulated two conjectures concerning determinantal inequalities for positive semi-definite matrices A and B , and which can be stated as follows det ⁡ (A 2 + | A B | p) ≥ det ⁡ (A 2 + | B A | p) for p ≥ 0 and det ⁡ (A 2 + | A B | p) ≥ det ⁡ (A 2 + A p B p) for 0 ≤ p ≤ 2. The main goal of this paper is to confirm the first conjecture in a slightly more general setting namely in the case when A and B are Hermitian, and also to prove the second conjecture when 0 ≤ p ≤ 4 3. Various related inequalities are then presented and we conclude with an open log-majorization question. [ABSTRACT FROM AUTHOR]

The singular values of a matrix A are defined as the square roots of the eigenvalues of A ⁎ A , and the energy of A denoted by E (A) is the sum of its singular values. The energy of a graph G , E (G) , is defined as the sum of absolute values of the eigenvalues of its adjacency matrix. In this paper, we prove that if A is a Hermitian matrix with the block form A = ( B D D ⁎ C ) , then E (A) ≥ 2 E (D). Also, we show that if G is a graph and H is a spanning subgraph of G such that E (H) is an edge cut of G , then E (H) ≤ E (G) , i.e., adding any number of edges to each part of a bipartite graph does not decrease its energy. Let G be a connected graph of order n and size m with the adjacency matrix A. It is well-known that if G is a bipartite graph, then E (G) ≥ 4 m + n (n − 2) | det ⁡ (A) | 2 n . Here, we improve this result by showing that the inequality holds for all connected graphs of order at least 7. Furthermore, we improve a lower bound for E (G) given in Oboudi (2019) [14]. [ABSTRACT FROM AUTHOR]

Matrices in R n × n with determinant equal to ∑ i ≤ n μ (i) , where μ represents the Möbius function, have been studied for decades. The motivation to study such matrices is the close connection that they have to the prime number theorem, Dirichlet convolution, and related concepts. We introduce two parameterized families of bordered Hermitian matrices that possess similar properties. Each family is comprised of matrices M s ∈ C (n − 1) × (n − 1) that satisfy det ⁡ M 0 = ∑ i ≤ n | μ (i) | , det ⁡ M 1 = (∑ i ≤ n μ (i)) 2 , and we show det ⁡ M s is a quadratic polynomial in s. We apply the Cauchy interlacing theorem to show that, for each matrix in one of the families, the product of all of the subdominant eigenvalues is bounded above by 6 / π 2 + O (n − 1 / 2). [ABSTRACT FROM AUTHOR]

Abstract We investigate the change in the multiplicities of the eigenvalues of a Hermitian matrix with a simple graph G , when edges are inserted into G or removed from G. We focus upon cases in which the multiplicity of the eigenvalue does not change due to inserting or removing edges incident to a vertex. Furthermore, we show how the change in the multiplicities of the eigenvalues occur, when two disjoint graphs are connected with one edge, based upon the status of the vertices that are connected. Lastly, we give the possible classifications of cut-edges in a graph and characterize the occurrence of each possible status. [ABSTRACT FROM AUTHOR]

In 1893 Jacques Hadamard introduced the famous inequality concerning the determinant of the Gram matrix. The generalizations of this inequality are the main subject of considerations presented in this paper. Calculations executed for the gramian of order three revealed that the description of gramian hides a great potential enabling to strengthen essentially the Hadamard inequality. The research, undertaken in this direction, resulted in discovering the more subtle versions of the Hadamard inequality, also by involving the “language” of the positive semi-definite Hermitian matrices. [ABSTRACT FROM AUTHOR]

For a Hermitian matrix A ∈ C p × p , given vectors x , y ∈ C p and for suitable functions f , the bilinear form y ⁎ f ( A ) x is estimated by extending the extrapolation method proposed by C. Brezinski in 1999. Families of one term and two term estimates e f , ν , ν ∈ C and e ˆ f , n , k , n , k ∈ Z , respectively, are derived by extrapolation of the moments of the matrix A . For the positive definite case, bounds for the optimal value of ν , which leads to an efficient one term estimate in only one matrix vector product, are derived. For f ( A ) = A − 1 , a formula approximating this optimal value of ν is specified. Numerical results for several matrix functions and comparisons are provided to demonstrate the effectiveness of the extrapolation method. [ABSTRACT FROM AUTHOR]

Fundamental theorem of geometry of hermitian matrices characterizes all bijective maps on the space of all hermitian matrices, which preserve adjacency in both directions. In this and subsequent paper we characterize maps on the set of all invertible hermitian matrices over a finite field, which preserve adjacency in one direction. In this first paper it is shown that maps that preserve adjacency on the set of all invertible hermitian matrices over a finite field are necessarily bijective, so the corresponding graph on invertible hermitian matrices, where edges are defined by the adjacency relation, is a core. Besides matrix theory, the proof relies on results from several other mathematical areas, including spectral and chromatic graph theory, and finite geometry. [ABSTRACT FROM AUTHOR]

Maps that preserve adjacency on the set of all invertible hermitian matrices over a finite field are characterized. It is shown that such maps form a group that is generated by the maps A ↦ P A P ⁎ , A ↦ A σ , and A ↦ A − 1 , where P is an invertible matrix, P ⁎ is its conjugate transpose, and σ is an automorphism of the underlying field. Bijectivity of maps is not an assumption but a conclusion. Moreover, adjacency is assumed to be preserved in one direction only. [ABSTRACT FROM AUTHOR]

The enhanced principal rank characteristic sequence (epr-sequence) of a symmetric n × n matrix is a sequence ℓ 1 ℓ 2 ⋯ ℓ n where ℓ k is A , S , or N according as all, some, or none of its principal minors of order k are nonzero. Such sequences give more information than the (0,1) pr-sequences previously studied (where basically the k th entry is 0 or 1 according as none or at least one of its principal minors of order k is nonzero). Various techniques including the Schur complement are introduced to establish that certain subsequences such as NAN are forbidden in epr-sequences over fields of characteristic not two. Using probabilistic methods over fields of characteristic zero, it is shown that any sequence of A s and S s ending in A is attainable, and any sequence of A s and S s followed by one or more N s is attainable; additional families of attainable epr-sequences are constructed explicitly by other methods. For real symmetric matrices of orders 2, 3, 4, and 5, all attainable epr-sequences are listed with justifications. [ABSTRACT FROM AUTHOR]

Given an n × n matrix, its principal rank characteristic sequence is a sequence of length n + 1 of 0s and 1s where, for k = 0 , 1 , … , n , a 1 in the k th position indicates the existence of a principal submatrix of rank k and a 0 indicates the absence of such a submatrix. The principal rank characteristic sequences for symmetric matrices over various fields are investigated, with all such attainable sequences determined for all n over any field with characteristic 2. A complete list of attainable sequences for real symmetric matrices of order 7 is reported. [ABSTRACT FROM AUTHOR]

We consider the problem of determining those undirected n -vertex graphs with a corresponding Hermitian matrix that admits only two distinct eigenvalues, with multiplicities k and n--k. After giving some general algebraic characterizations of these dual multiplicity graphs, we then prove two major graph theoretic necessary conditions on such graphs. Construction techniques are then developed, and these lead to a characterization of dual multiplicity graphs for which the lesser multiplicity is two. [ABSTRACT FROM AUTHOR]

Abstract: Let be a positive semidefinite block matrix with square matrices M and N of the same order and denote . The main results are the following eigenvalue majorization inequalities: for any complex number z of modulus 1, If, in addition, K is Hermitian, then for any real number , while if K is skew-Hermitian, then for any real number , where O is the zero matrix of compatible size. These majorization inequalities generalize some results due to Furuichi and Lin, Turkmen, Paksoy and Zhang, Lin and Wolkowicz. [Copyright &y& Elsevier]

Abstract: We consider the conditions under which the Cayley transform of the Kronecker product of two Hermitian matrices can be again presented as a Kronecker product of two matrices and, if so, if it is a product of the Cayley transforms of the two Hermitian matrices. We also study the related question: given two matrices, which matrix under the Cayley transform yields the Kronecker product of their Cayley transforms. [Copyright &y& Elsevier]

Abstract: We show some majorization inequalities and apply them to derive norm, eigenvalue, singular value, and trace inequalities of matrices. We also present a generalization and a different proof of a recent result on majorization concerning partitioned Hermitian positive semidefinite matrices. [Copyright &y& Elsevier]

Abstract: Estimating upper bounds of the spectrum of large Hermitian matrices has long been a problem with both theoretical and practical significance. Algorithms that can compute tight upper bounds with minimum computational cost will have applications in a variety of areas. We present a practical algorithm that exploits k-step Lanczos iteration with a safeguard step. The k is generally very small, say 5–8, regardless of the large dimension of the matrices. This makes the Lanczos iteration economical. The safeguard step can be realized with marginal cost by utilizing the theoretical bounds developed in this paper. The bounds establish the theoretical validity of a previous bound estimator that has been successfully used in various applications. Moreover, we improve the bound estimator which can now provide tighter upper bounds with negligible additional cost. [Copyright &y& Elsevier]

Abstract: We consider products of unitary operators with at most two points in their spectra, 1 and . We prove that the scalar operator is a product of k such operators if for . Also we prove that for , only a countable number of scalar operators can be decomposed in a product of four operators from the mentioned class. As a corollary we show that every unitary operator on an infinite-dimensional space is a product of finitely many such operators. [Copyright &y& Elsevier]

Abstract: Polygon transformations based on taking the apices of similar triangles constructed on the sides of an initial polygon are analyzed as well as the limit polygons obtained by iteratively applying such transformations. In contrast to other approaches, this is done with respect to two construction parameters representing a base angle and an apex perpendicular subdivision ratio. Furthermore, a combined transformation leading to circulant Hermitian matrices is proposed, which eliminates the rotational effect of the basic transformation. A finite set of characteristic parameter subdomains is derived for which the sequence converges to specific eigenpolygons. Otherwise, limit polygons turn out to be linear combinations of up to three eigenpolygons. This leads to a full classification of circulant Hermitian similar triangles based polygon transformations and their limit polygons. As a byproduct classical results as Napoleon’s theorem and the Petr–Douglas–Neumann theorem can be easily deduced. [Copyright &y& Elsevier]

Abstract: The inertia of a Hermitian matrix is defined to be a triplet composed of the numbers of the positive, negative and zero eigenvalues of the matrix counted with multiplicities, respectively. In this paper, we show some basic formulas for inertias of block Hermitian matrices. From these formulas, we derive various equalities and inequalities for inertias of sums, parallel sums, products of Hermitian matrices, submatrices in block Hermitian matrices, differences of outer inverses of Hermitian matrices. As applications, we derive the extremal inertias of the linear matrix expression with respect to a variable Hermitian matrix X. In addition, we give some results on the extremal inertias of Hermitian solutions to the matrix equation , as well as the extremal inertias of a partial block Hermitian matrix. [Copyright &y& Elsevier]

Abstract: Denote by a graph which is the vertex set and is an adjacency relation on a subset of . In this paper, the good distance graph is defined. Let and be two good distance graphs, and be a map. The following theorem is proved: is a graph isomorphism is a bounded distance preserving surjective map in both directions is a distance preserving surjective map in both directions (where is a positive integer), etc. Let be a division ring with an involution such that both and is not a field of characteristic 2 with , where and is the center of . Let be the set of Hermitian matrices over . It is proved that is a good distance graph, where for all . [Copyright &y& Elsevier]

Abstract: Necessary and sufficient conditions are derived for a 2-by-2 partitioned matrix to have {1}-, {1,2}-, {1,3}-, {1,4}-inverses and the Moore–Penrose inverse with Banachiewicz–Schur forms. As applications, the Banachiewicz–Schur forms of {1}-, {1,2}-, {1,3}-, {1,4}-inverses and the Moore–Penrose inverse of a 2-by-2 partitioned Hermitian matrix are also given. [Copyright &y& Elsevier]

Abstract: In this paper, by using normal maps originated by Lewis [A.S. Lewis, Group invariance and convex matrix analysis, SIAM J. Matrix Anal. Appl. 17 (1996) 927–949], we present a method for producing majorization inequalities on eigenvalues and singular values of matrices. Some classical and new inequalities are derived. [Copyright &y& Elsevier]

Abstract: We study theoretical properties of two inexact Hermitian/skew-Hermitian splitting (IHSS) iteration methods for the large sparse non-Hermitian positive definite system of linear equations. In the inner iteration processes, we employ the conjugate gradient (CG) method to solve the linear systems associated with the Hermitian part, and the Lanczos or conjugate gradient for normal equations (CGNE) method to solve the linear systems associated with the skew-Hermitian part, respectively, resulting in IHSS(CG, Lanczos) and IHSS(CG, CGNE) iteration methods, correspondingly. Theoretical analyses show that both IHSS(CG, Lanczos) and IHSS(CG, CGNE) converge unconditionally to the exact solution of the non-Hermitian positive definite linear system. Moreover, their contraction factors and asymptotic convergence rates are dominantly dependent on the spectrum of the Hermitian part, but are less dependent on the spectrum of the skew-Hermitian part, and are independent of the eigenvectors of the matrices involved. Optimal choices of the inner iteration steps in the IHSS(CG, Lanczos) and IHSS(CG, CGNE) iterations are discussed in detail by considering both global convergence speed and overall computation workload, and computational efficiencies of both inexact iterations are analyzed and compared deliberately. [Copyright &y& Elsevier]

Abstract: Two Hermitian matrices are said to be Hermitian-congruent if there exists a nonsingular Hermitian matrix such that . 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 . Moreover, if both matrices are positive, then C can be picked with arbitrary inertia. [Copyright &y& Elsevier]

Abstract: We propose a new QR-like algorithm, symmetric squared QR (SSQR) method, that can be readily parallelized using commonly available parallel computational primitives such as matrix–matrix multiplication and QR decomposition. The algorithm converges quadratically and the quadratic convergence is achieved through a squaring technique without utilizing any kind of shifts. We provide a rigorous convergence analysis of SSQR and derive structures for several of the important quantities generated by the algorithm. We also discuss various practical implementation issues such as stopping criteria and deflation techniques. We demonstrate the convergence behavior of SSQR using several numerical examples. [Copyright &y& Elsevier]

Abstract: Let K be a field of characteristic not two or three with an automorphism − of order two. Let . We characterize additive mappings T from one F-vector space of Hermitian matrices over K to another that preserve rank less than or equal to one for the following cases: (i) the image of T contains a matrix of rank at least three, (ii) every nonzero endomorphism of F is surjective. [Copyright &y& Elsevier]

Groß [Linear Algebra Appl. 326 (2001) 215] developed characterizations of the minus and star partial orders between the squares of Hermitian nonnegative definite matrices referring to the concept of the space preordering. In the present paper, his results are generalized by deleting the nonnegative definiteness assumption and supplemented by alternative characterizations. [Copyright &y& Elsevier]

Baksalary and Pukelsheim [Linear Algebra Appl. 151 (1991) 135] considered the problem of how an order between two Hermitian nonnegative definite matrices A and B is related to the corresponding order between the squares A2 and B2, in the sense of the star partial ordering, the minus partial ordering, and the Lo¨wner partial ordering. In the present paper, possibilities of generalizing and strengthening their results are studied from two points of view: by widening the class of matrices considered and by replacing the squares by arbitrary powers. [Copyright &y& Elsevier]

Certain classes of matrices are indicated for which the star, left-star, right-star, and minus partial orderings, or some of them, are equivalent. Characterizations of the left-star and right-star orderings, similar to those devised by Hartwig and Styan [Linear Algebra Appl. 82 (1986) 145] for the star and minus orderings, are established along with other auxiliary results, which are of independent interest as well. Some inheritance-type properties of matrices are also given. The class of EP matrices appears to be essential in several points of our considerations. [Copyright &y& Elsevier]

We describe the set of all (α,n), for which the scalar complex matrix αIn is a sum of k idempotent Hermitian matrices, and get the minimal number of summands for each αIn. [Copyright &y& Elsevier]

If −∞<α<β<∞ let mid(α,x,β)=α if x<α, x if α&les;x&les;β, β if x>β. Let An=Bn+Pn where Bn and Pn are n×n Hermitian matrices. We show that if &par;Pn&par;F2=o(n) then, for any [α,β], (A) ∑i=1n|F(mid(α,λi(An),β))−F(mid(α,λi(Bn),β))|=o(n) if F∈C[α,β]. (Eigenvalues numbered in nondecreasing order.) We consider the special case where {Pn} are real Hankel matrices. We also show that if rank(Pn)=o(n) then (A) holds for every [α,β] and F∈C[α,β]. Combining these results yields a result concerning Cn=Bn+En+Rn, where &par;En&par;2F=o(n) and rank(Rn)=o(n). We also consider the case where the conditions on {En} are stated in terms of Schatten p-norms. Finally, we show that if {Tn} are Hermitian Toeplitz matrices generated by f∈C[−π,π] with minimum mf and maximum Mf, (2(i−1)−n)π/n&les;ξin&les;(2i−n)π/n, 1&les;i&les;n, and τn is a permutation of {1,2,…,n} such that f(ξτn(1),n)&les;f(ξτn(2),n)&les;⋯&les;f(ξτn(n),n), then ∑i=1n|F(λi(Tn))−F(f(ξτn(i),n))|=o(n) if F∈C[mf,Mf]. [Copyright &y& Elsevier]

For Hermitian matrices, whose graph is a given tree, the relationships among vertex degrees, multiple eigenvalues and the relative position of the underlying eigenvalue in the ordered spectrum are discussed in detail. In the process, certain aspects of special vertices, whose removal results in an increase in multiplicity are investigated. [Copyright &y& Elsevier]