Abstract
In this article, we study a boundary value problem of a class of generalized linear discretetime systems whose coefficients are square constant matrices. By using matrix pencil theory, we obtain formulas for the solutions and we give necessary and sufficient conditions for existence and uniqueness of solutions. Moreover, we provide some numerical examples. These kinds of systems are inherent in many physical and engineering phenomena.
Keywords:
linear difference equations; boundary value problem; matrix pencil; discrete time system; matrix difference equations1 Introduction
Linear matrix difference equations (LMDEs) are systems in which the variables take their values at instantaneous time points. Discrete time systems differ from continuous time ones in that their signals are in the form of sampled data. With the development of the digital computer, the discrete time system theory plays an important role in control theory. In real systems, the discrete time system often appears when it is the result of sampling the continuoustime system or when only discrete data are available for use. LMDEs are inherent in many physical, engineering, mechanical, and financial/actuarial models. In this article, our purpose is to study the solutions of generalized linear discretetime boundary value problems into the mainstream of matrix pencil theory. A boundary value problem consists of finding solutions which satisfies an ordinary matrix difference equation and appropriate boundary conditions at two or more points. Thus, we consider
with known boundary values of type
where (i.e., the algebra of square matrices with elements in the field ). For the sake of simplicity, we set and .
Systems of type (1) are more general, including the special case when F = I_{n}, where I_{n }is the identity matrix of .
The matrix pencil theory has extensively been used for the study of linear difference equations with time invariant coefficients, see for instance [15]. A matrix pencil is a family of matrices sF  G, parametrized by a complex number s. When G is square and F = I_{n}, where I_{n }is the identity matrix, the zeros of the function det (sF  G) are the eigenvalues of G. Consequently, the problem of finding the nontrivial solutions of the equation
is called the generalized eigenvalue problem. Although the generalized eigenvalue problem looks like a simple generalization of the usual eigenvalue problem, it exhibits some important differences. In the first place, it is possible for det (sF  G) to be identically zero, independent of s. Second, it is possible for F to be singular, in which case the problem has infinite eigenvalues. To see this, write the generalized eigenvalue problem in the reciprocal form
If F is singular with a null vector X, then , so that X is an eigenvector of the reciprocal problem corresponding to eigenvalue s^{1 }= 0; i.e., s = ∞. It might be thought that infinite eigenvalues are special, unhappy cases to be ignored in our perturbation problem but that is a misconception (see also [69]).
2 Mathematical background and notation
This brief section introduces some preliminary concepts and definitions from matrix pencil theory, which are being used throughout the article. Linear systems of type (1) are closely related to matrix pencil theory, since the algebraic, geometric, and dynamic properties stem from the structure by the associated pencil sF  G.
Definition 2.1. Given F,G ∈ M_{nm }and an indeterminate s ∈ F, the matrix pencil sF  G is called regular when m = n and det (sF  G) ≠ 0. In any other case, the pencil will be called singular.
Definition 2.2. The pencil sF  G is said to be strictly equivalent to the pencil if and only if there exist nonsingular and such as
In this article, we consider the case that pencil is regular.
The class of sF  G is characterized by a uniquely defined element, known as a complex Weierstrass canonical form, sF_{w } Q_{w}, see [5], specified by the complete set of invariants of the pencil sF  G.
This is the set of elementary divisors (e.d.) obtained by factorizing the invariant polynomials into powers of homogeneous polynomials irreducible over field F. In the case where sF  G is a regular, we have e.d. of the following type:
• e.d. of the type s^{p }are called zero finite elementary divisors (z. f.e.d.)
• e.d. of the type (s  a)^{π}, a ≠ 0 are called nonzero finite elementary divisors (nz. f.e.d.)
• e.d. of the type are called infinite elementary divisors (i.e.d.).
Let B_{1}, B_{2}, ..., B_{n }be elements of . The direct sum of them denoted by B_{1 }⊕ B_{2 }⊕ ··· ⊕ B_{n }is the block diag {B_{1}, B_{2}, ..., B_{n}}.
Then, the complex Weierstrass form sF_{w } Q_{w }of the regular pencil sF  G is defined by sF_{w } Q_{w }:= sI_{p } J_{p }⊕ sH_{q } I_{q}, where the first normal Jordantype element is uniquely defined by the set of f.e.d.
of sF  G and has the form
and also the q blocks of the second uniquely defined block sH_{q } I_{q }correspond to the i.e.d.
of sF  G and has the form
Thus, H_{q }is a nilpotent element of with index , then
We denote with O the zero matrix. are defined as
3 Main resultsSolution space form of a consistent boundary value problem
In this section, the main results for a consistent boundary value problem of types (1) and (2) are analytically presented. Moreover, it should be stressed that these results offer the necessary mathematical framework for interesting applications.
Definition 3.1. The boundary value problem (1) and (2) is said to be consistent if it possesses at least one solution.
Consider the problem (1) with known boundary conditions (2). From the regularity of sF  G, there exist nonsingular matrices P and Q such that (see also Section 2),
and
where are defined by (10), (11), (12) and moreover
Note that and , where p + q = n.
Lemma 3.1. System (1) is divided into two subsystems:
and the subsystem
Proof. Consider the transformation
Substituting the previous expression into (1) we obtain
whereby, multiplying by P, we arrive at
Moreover, we can write Z_{k }as , where and . Taking into account the above expressions, we arrive easily at (16) and (17).
Proposition 3.2. The subsystem (16) has general solution
Proposition 3.3. The subsystem (17) has the unique solution
Proof. Let q_{* }be the index of the nilpotent matrix H_{q}, i.e. (), we obtain the following equations
and
The conclusion, i.e., , is obtained by repetitive substitution of each equation in the next one, and using the fact that .
The boundary value problem
A necessary and sufficient condition for the boundary value problem to be consistent is given by the following result
Theorem 3.1. The boundary value problem (1), (2) is consistent, if and only if
Where . The matrix Q_{p }has column vectors the p linear independent eigenvectors of the finite generalized eigenvalues of sFG (see [1] for an algorithm of the computation of Q_{p}).
Proof. Let Q = [Q_{p}Q_{q}], where and ; Combining propositions (3.2) and (3.3), we obtain
or
The solution exists if and only if
or
It is obvious that, if there is a solution of the boundary value problem, it needs not to be unique. The necessary and sufficient conditions, for uniqueness, when the problem is consistent, are given by the following theorem.
Theorem 3.2. Assume the boundary value problem (1), (2). Then when it is consistent, it has a unique solutions if and only if
Then the formula of the unique solution is
where C is the solution of the equation
Proof. Let the boundary value problem (1), (2) be consistent, then from Theorem 3.1 and (22) the solution is
with
and
It is clear that for given A, B, D the problem (1), (2) has a unique solution if and only if the system (24) has a unique solution. Since , the solution is unique for system (24) if and only if the matrix is left invertible. This fact is equivalent to:
Then the formula of the unique solution is
where C is the solution of the equation
Other type of boundary conditions
Assume that the matrix difference equation (1) has a different type of boundary conditions. Let the boundary conditions be
where . Then we can state the following theorem.
Theorem 3.3. The boundary value problem (1), (25) is consistent, if and only if
Moreover when it is consistent, it has a unique solution if and only
and the linear system
gives a unique solution for the constant column C.
Proof. From (22) and (25) the solution exists if and only if
or
It is obvious that a consistent solution of the boundary value problem (1), (25), is unique if and only if the system (28) gives a unique solution for C. Since , the solution is unique if and only if the matrices are left invertible or .
4 Numerical example
Consider the boundary value problem (1), (2), where
and A, B the identity and zero matrices, respectively. The invariants of sF  G are s  1, s  2, s  3 (finite elementary divisors) and (infinite elementary divisor of degree 3). Then
and the columns of Q_{p }are the eigenvectors of the generalized eigenvalues 1, 2, 3, respectively. Then
where ()^{T }is the transpose tensor.
4.1 Example 1
Let
Then
and by calculating C from (24) we get
and the unique solution of the system by substituting in (22) is
4.2 Example 2
Let
Then
and the problem is not consistent.
5 Conclusions
The aim of this article was to give necessary and sufficient conditions for existence and uniqueness of solutions for generalized linear discretetime boundary value problems of a class of linear rectangular matrix difference equations whose coefficients are square constant matrices. By taking into consideration that the relevant pencil is regular, we use the Weierstrass canonical form to decompose the difference system into two subsystems. Afterwards, we provide analytical formulas when we have a consistent problem. Moreover, as a further extension of this article, we can discuss the case where the pencil is singular. Thus, the Kronecker canonical form is required. For all these, there is some research in progress.
Competing interests
The author declares that they have no competing interests.
Acknowledgements
The author would like to express his sincere gratitude to Professor G. I. Kalogeropoulos for his fruitful discussion that improved the article. The author would also like to thank the anonymous referees for their comments.
References

Kalogeropoulos, G: Matrix Pencils and Linear System Theory. City University Press, London (1985)

Rough, WJ: Linear System Theory. Prentice Hall, Upper Saddle River, New Jersey (1996)

Dai, L: Singular Control Systems. Lecture Notes in Control and information Sciences, SpringerVerlag, Heidelberg (1988)

Cambell, SL: Singular Systems of Differential Equations. Cole Publishing Company (1980)

Gantmacher, FR: The Theory of Matrices, Volume I and II. Chelsea (1959)

Datta, BN: Numerical Linear Algebra and Applications. Cole Publishing Company (1995)

Dassios, IK: Solutions of higherorder homogeneous linear matrix differential equations for consistent and nonconsistent initial conditions: regular case. ISRN Math Anal 2011. 14, Article ID 183795 (2011)

Kalogeropoulos, GI, Psarrakos, P: A note on the controllability of Higher Order Systems. Appl Math Lett. 17(12), 1375–1380 (2004). Publisher Full Text

Kalogeropoulos, GI, Psarrakos, P, Karcanias, N: On the computation of the Jordan canonical form of regular matrix polynomials. Linear Algebra Appl. 385, 117–130 (2004)