The (M, N)-symmetric Procrustes problem
Introduction
Let denote the set of n-dimensional real vectors, and denote the set of real matrices, real symmetric matrices, real antisymmetric matrices and orthogonal matrices, respectively. The notation stands for the Moore-Penrose inverse and the Frobenius norm of a matrix A, respectively. For and , define as Hadamard product of matrices A and B. Definition 1 Given , we say that is -symmetric ifWe denote by the set of all -symmetric matrix.
In this paper, we consider the following problems: Problem I Given , find such that Problem II Given , find such that
In electricity, control theory and processing of digital signals, we often need to study the Procrustes problem of the well-known equation with the unknown matrix A. The problem with A being symmetric, bisymmetric, centrosymmetric and positive semidefinite symmetric were studied (see [1], [2], [3], [4], [5], [6], [7], [8]). Taking for A the set of orthogonal matrices yields the orthogonal Procrustes problem (see [9]), the set of symmetric matrices yiedls the symmetric Procrustes problem (see [10]). By analogy with the problem mentioned above, we will refer to Problem I as the -symmetric Procrustes problem. Usually, by applying the structure properties of unknown matrix A and appropriate matrix decompositions (the singular-value decomposition, the generalized singular-value decomposition, etc.), the solution of the Procrustes problem were given. But this approach can not be used to solve Problem I. In this paper, we initiate an efficient method: Firstly, by the structure property of A and a set of orthonormal basis of a subspace, we find out a solution of Problem I. Secondly, using the solution and the Project Theorem, we transfer Problem I to the problem of finding the -symmetric solution of a consistent matrix equation . Finally, we find out the -symmetric solution of the consistent matrix equation.
Problem II, that is, the optimal approximation problem of a matrix with the given matrix restriction, is proposed in the processes of test or recovery of linear systems due to incomplete data or revising given data. A preliminary estimate of the unknown matrix A can be obtained by the experimental observation values and the information of statical distribution. The optimal estimate of A is a matrix satisfying the given matrix restriction for A and being the best approximation of . Various aspects of the optimal approximation problem associated with were considered, see [1], [2] and reference therein.
The paper is organized as follows. In Section 2, we will discuss the structure properties of matrices in , and provide the general expression of the solutions of Problem I. In Section 3, we will prove the existence and uniqueness of the solution of Problem II and derive the expression of this unique solution. In Section 4, we will give the algorithm to compute the approximate solution and the numerical experiment.
Section snippets
The general form of solutions of Problem I
For convenience, here we give the singular-value decomposition (SVD) of a matrix X, and the generalized singular-value decomposition (GSVD) of a matrix pair . Proofs and properties concerning the SVD and GSVD can be found in [11], [12], [13].
Given a matrix of rank r, its SVD iswhere , , .
Given two matrices , the GSVD of the matrix pair iswhere W is a nonsingular matrix, ,
The expression of solution of Problem II
The solution set of Problem I is a closed convex set. Therefore, there exists a unique solution for Problem II. Lemma 8 Given , , , there exists a unique matrix such thatand can be expressed aswhere . Proof For , we haveIn Eq. (21), is a continuously differentiable function of
Numerical computation of the solution of Problem II
Now we give the procedure to compute the optimal approximate solution of Problem II and a experiment example. Algorithm Input and ; Make the GSVD of the matrix pair according to (2); Calculate according to (6), calculate ; Make the SVD of the matrix X according to (1), make the GSVD of the matrix pair according to (9); Partition the matrix according to (12), partition according to (23); According to (24) calculate .
Example
Given
Acknowledgements
Research supported by National Natural Science Foundation of China (10571047), by Specialized Research Fund for the Doctoral Program of Higher Education (20060532014) and by the College Fund of College of Applied Sciences, Beijing University of Technology.
References (15)
- et al.
The least-squares solution of inconsistent matrix equation over symmetric and antisymmetric matrices
Appl. Math. Lett.
(2003) Least-squares solution of over positive semidefinite symmetric P
Lin. Alg. Appl.
(1996)- et al.
Least-square solutions for inverse problems of centrosymmetric matrices
Comput. Math. Appl.
(2005) Two kinds of inverse eigenvalue problems for real symmetric matrices
Math. Numer. Sinica.
(1988)Positive semidefinite matrices: Characterization via conical hulls and least-squares solution of a matrix equation
SIAM J. Control Optim.
(1998)- et al.
Least-squares solutions of inverse problems for anti-symmetric matrices
J. Eng. Math.
(1993) - et al.
Least-squares of inverse problems for bisymmetric matrices
Math. Numer. Sin.
(2000)
Cited by (4)
Procrustes: A python library to find transformations that maximize the similarity between matrices
2022, Computer Physics CommunicationsCitation Excerpt :Motivated by the scientific significance of Procrustes problem and the absence of a dedicated package, we designed the Procrustes library: a flexible, easy-to-use, and easy-to-extend Python library for finding the optimal transformation that makes two matrices as close as possible to each other. As summarized in Table 1, the Procrustes library includes algorithms for orthogonal [9], rotational [9–11], symmetric [12–14], and permutation [17,16] Procrustes problems and their two-sided counterparts [19,20,7]. Notably, the softassign algorithm [42,23] is applied for solving the two-sided permutation Procrustes problem, which gives higher accuracy than other approaches, with commensurate computational costs.
Procrustes problems and parseval quasi-dual frames
2014, Acta Applicandae MathematicaeThe hermitian R -conjugate generalized procrustes problem
2013, Abstract and Applied AnalysisLeast squares (P, Q)-orthogonal symmetric solutions of the matrix equation and its optimal Approximation
2010, Electronic Journal of Linear Algebra