Tiziano Politi

An upper bound for the condition number of a matrix in spectral norm

In this note we generalize an upper bound given in Guggenheimer et al. (College Math. J. 26(1) (1995) 2) for the condition number of a matrix as a function of the determinant, the Frobenius norm and of k singular values. If no singular value is known it is possible to derive an upper bound for the condition number applying lower and upper bounds for the product of a subset of singular values.

On the Low-Rank Approximation of Data on the Unit Sphere

In various applications, data in multidimensional space are normalized to unit length. This paper considers the problem of best fitting given points on the m-dimensional unit sphere Sm-1 by k-dimensional great circles with k much less than m. The task is cast as an algebraically constrained low-rank matrix approximation problem. Using the fidelity of the low-rank approximation to the original data as the cost function, this paper offers an analytic expression of the projected gradient which, on one hand, furnishes the first order optimality condition and, on the other hand, can be used as a numerical means for solving this problem.

A Formula for the Exponential of a Real Skew-Symmetric Matrix of Order 4

In this short paper the formula of the exponential matrix eA when A is a kew-symmetric real matrix of order 4 is derived. The formula is a generalization of the well known Rodrigues formula for skew-symmetric matrices of order 3.

A Continuous Technique for the Weighted Low-Rank Approximation Problem

This paper concerns with the problem of approximating a target matrix with a matrix of lower rank with respect to a weighted norm. Weighted norms can arise in several situations: when some of the entries of the matrix are not observed or need not to be treated equally. A gradient flow approach for solving weighted low rank approximation problems is provided. This approach allows the treatment of both real and complex matrices and exploits some important features of the approximation matrix that optimization techniques do not use. Finally, some numerical examples are provided.

One step semi-explicit methods based on the Cayley transform for solving isospectral flows

Abstract This note deals with the numerical solution of the matrix differential system Y′ = [B(t,Y), Y], Y(0) = Y0, t ⩾ 0, where Y0 is a real constant symmetric matrix, B maps symmetric into skew-symmetric matrices, and [B(t,Y),Y] is the Lie bracket commutator of B(t,Y) and Y, i.e. [B(t,Y),Y] = B(t,Y)Y − YB(t,Y). The unique solution of (1) is isospectral, that is the matrix Y(t) preserves the eigenvalues of Y0 and is symmetric for all t (see [1, 5]). Isospectral methods exploit the Flaschka formulation of (1) in which Y(t) is written as Y(t) = U(t)Y0UT(t), for t ⩾ 0, where U(t) is the orthogonal solution of the differential system U′ = B(t, UY0UT)U, U(0) = I, t ⩾ 0, (see [5]). Here a numerical procedure based on the Cayley transform is proposed and compared with known isospectral methods.

A discrete approach for the inverse singular value problem in some quadratic group

In this paper the solution of an inverse singular value problem is considered. First the decomposition of a real square matrix A = UΣV is introduced, where U and V are real square matrices orthogonal with respect to a particular inner product defined through a real diagonal matrix G of order n having all the elements equal to ±1, and Σ is a real diagonal matrix with nonnegative elements, called G-singular values. When G is the identity matrix this decomposition is the usual SVD and Σ is the diagonal matrix of singular values. Given a set {σ1,..., σn} of n real positive numbers we consider the problem to find a real matrix A having them as G-singular values. Neglecting theoretical aspects of the problem, we discuss only an algorithmic issue, trying to apply a Newton type algorithm already considered for the usual inverse singular value problem.

Newton-type methods for solving nonlinear equations on quadratic matrix groups

In this paper we consider numerical methods for solving nonlinear equations on matrix Lie groups. Recently Owren and Welfert (Technical Report Numerics, No 3/1996, Norwegian University of Science and Technology, Trondheim, Norway, 1996) have proposed a method where the original nonlinear equation F(Y) = 0 is transformed into a nonlinear equation on the Lie algebra of the group, thus Newton-type methods may be applied which require the evaluation of exponentials of matrices. Here the previous transformation will be performed by the Cayley approximant of the exponential map. This approach has the advantage that no exponentials of matrices are needed. The numerical tests reported in the last section seem to show that our approach is less expensive and provides a larger convergence region than the method of Owren and Welfert.

Computation of functions of Hamiltonian and skew-symmetric matrices

In this paper we consider numerical methods for computing functions of matrices being Hamiltonian and skew-symmetric. Analytic functions of this kind of matrices (i.e., exponential and rational functions) appear in the numerical solutions of ortho-symplectic matrix differential systems when geometric integrators are involved. The main idea underlying the presented techniques is to exploit the special block structure of a Hamiltonian and skew-symmetric matrix to gain a cheaper computation of the functions. First, we will consider an approach based on the numerical solution of structured linear systems and then another one based on the Schur decomposition of the matrix. Splitting techniques are also considered in order to reduce the computational cost. Several numerical tests and comparison examples are shown.

Variable step-size techniques in continuous Runge-Kutta methods for isospectral dynamical systems

Abstract In this paper we consider numerical methods for the dynamical system L′ = [B(L),L], L(0) = L0, where L0 is a n × n symmetric matrix, [if[B(L),L] is the commutator of B(L) and L, and B(L) is a skew-symmetric matrix for each symmetric matrix L. The differential system is isospectral, i.e., L(t) preserves the eigenvalues of L0, for t⩾0. The matrix B(L) characterizes the flow, and for special B(·), the solution matrix L(t) tends, as t increases, to a diagonal matrix with the same eigenvalues of L0. In [11] a modification of the MGLRK methods, introduced in [2], has been proposed. These procedures are based on a numerical approximation of the Flaschka formulation of (∗) by Runge-Kutta (RK) methods. Our numerical schemes (denoted by EdGLRKs consist in solving the system (∗) by a continuous explicit Runge-Kutta method (CERK) and then performing a single step of a Gauss-Legendre RK method, for the Flaschka formulation of (∗), in order to convert the approximation of L(t) to an isospectral solution. The problems of choosing a constant time step or a variable time step strategy are both of great importance in the application of these methods. In this paper, we introduce a definition of stability for the isospectral numerical methods. This definition involves a potential function associated to the isospectral flow. For the class EdGLRKs we propose a variable step-size strategy, based on this potential function, and an optimal constant time step h in the stability interval. The variable time step strategy will be compared with a known variable step-size strategy for RK methods applied to these dynamical systems. Numerical tests will be given and a comparison with the QR algorithm will be shown.

Applying fixed point homotopy to nonlinear DAEs deriving from switching circuits

The time-domain analysis of switching circuits is a time consuming process as the change of switch state produces sharp discontinuities in switch variables. In this paper, a method for fast time-domain analysis of switching circuits is described. The proposed method is based on piecewise temporal transient analysis windows joined by DC analysis at the switching instants. The DC analysis is carried out by means of fixed point homotopy to join operating points between consecutive time windows. The proposed method guarantees accurate results reducing the number of iterations needed to simulate the circuit.