Next Article in Journal
Stability-Optimized Graph Convolutional Network: A Novel Propagation Rule with Constraints Derived from ODEs
Previous Article in Journal
The Global Dynamics of the Painlevé–Gambier Equations XVIII, XXI, and XXII
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Tripled Fixed Points, Obtained by Ran-Reunrings Theorem for Monotone Maps in Partially Ordered Metric Spaces

1
Department of Algebra and Geometry, Faculty of Mathematics and Informatics, Konstantin Preslavsky University of Shumen, 115 Universitetska Str., 9700 Shumen, Bulgaria
2
Department of Mathematical Analysis, Faculty of Mathematics and Informatics, University of Plovdiv Paisii Hilendarski, 24 Tsar Assen Str., 4000 Plovdiv, Bulgaria
*
Author to whom correspondence should be addressed.
Mathematics 2025, 13(5), 760; https://doi.org/10.3390/math13050760
Submission received: 28 January 2025 / Revised: 17 February 2025 / Accepted: 21 February 2025 / Published: 25 February 2025

Abstract

:
Using the deep result of Ran & Reunrings, we generalize existing results for tripled fixed points. In contrast to the previously known results for tripled fixed points of maps with or without the mixed monotone property in partially ordered complete metric spaces, we demonstrate that it is possible to obtain results for the existence and uniqueness of such points for arbitrary maps with a type of monotonicity, with the partial ordering in the Cartesian product arising from the maps itself. We prove theorems that ensure the existence and uniqueness for tripled fixed points for maps with different types of monotone properties. We obtain sufficient conditions for the existence and uniqueness of systems of three nonlinear matrix equations. The obtained results are illustrated by solving systems of matrix equations.

1. Introduction

It is well known that algebraic discrete-type Riccati equations play a central role in modern control theory and signal processing. These equations arise in many important applications, such as in optimal control theory, dynamic programming, stochastic filtering, statistics, and other fields of pure and applied mathematics [1,2,3]. Nonlinear matrix equations are commonly used in many fields of scientific and engineering computing. Research on the existence and properties of the solution to the matrix equations, as well as the corresponding numerical methods, has important theoretical significance and practical value. In recent years, researchers have shown great interest in the matrix equation
X i = 1 m A i X p i A i = Q ,
where 0 < p i 1 ( i = 1 , 2 , , m ) , with their primary focus being to explore the conditions for the existence of positive definite solutions, perturbation analysis, and developing iterative methods to solve these equations. Notably, there have been several noteworthy studies on the special case for m = 1 , 2 and p i = 1 [4,5].
In a series of articles [6,7], authors have studied the system of matrix equation X ± A X q A = Q and have obtained deep necessary conditions for the existence of a solution. More complicated matrix equations are studied in [8]. The studies [9,10] are also related to the presented generalizations in this study. Additionally, in [11], a nonlinear system of matrix equations is investigated through an approach different from ours.
Fixed point results have been recently used across various fields, including those assumed not to be a closely related to them and their applications: fixed points of principal bundles over algebraic curves [12]; fixed points in Higgs bundles over a compact and connected Riemann surface [13,14] and the Hitchin integrable system [13]; and fixed points with applications in physics [15].
Let ( X , d ) be a metric space, and let T : X X be a self-map. The contraction mapping theorem [16] and the abstract monotone iterative technique [17] are widely known and apply in many contexts. The first result on fixed points in partially ordered spaces is found in [18], but [17] laid the groundwork for numerous studies in this area. Recently [19,20,21,22,23], a trend has emerged to relax the contraction condition by requiring it only in a partially ordered metric space ( X , d ) , meaning that instead of requiring d ( T x , T y ) α d ( x , y ) for all x , y X , the condition is needed only when x y , where ≼ is a partial order on ( X , d ) .
A recent application of fixed points has arisen in solving matrix equations that include the inverted matrix X 1 [24]. The authors transform the matrix equation X + A X 1 A B X 1 B = Q , where A and B are arbitrary N × N square matrices and Q is a positive definite N × N matrix, into a system of two matrix equations:
X = Q A X 1 A + B Y 1 B Y = Q A Y 1 A + B X 1 B ,
and apply the notion of coupled fixed points in metric spaces with a partial ordering from [25].
The system of Equation (1) can be viewed as a map F : X ( N ) × X ( N ) X ( N ) , where X ( N ) represents the space of N × N matrices, defined by F ( X , Y ) = Q A X 1 A + B Y 1 B . We seek the coupled fixed point ( X , Y ) [26], where X = F ( X , Y ) and Y = F ( Y , X ) .
It is known that if certain classical assumptions on the map F : X × X X and the underlying normed space ( X , · ) hold, then X = Y , as in [27]. In [27], two maps F , G : X × X X are introduced, allowing the system of equations X = F ( X , Y ) and Y = G ( X , Y ) to have solutions where X 0 Y 0 . If G ( X , Y ) = F ( Y , X ) , we obtain the well-known result on coupled fixed points [25].
Following the trend noted above [23,28,29] and the observations in [27], this paper expands the concept of mixed monotone maps to provide a unified framework for a broader class of problems, where existing results become particular cases of the new findings. We establish criteria for guaranteeing the uniqueness and existence of solutions for the matrix equation system:
X = Q 1 + α 1 A 1 X k A 1 + α 2 B 1 Y p B 1 + α 3 C 1 Y q C 1 Y = Q 2 + α 1 A 2 X k A 2 + α 2 B 2 Y p B 2 + α 3 C 2 Y q C 2 Z = Q 3 + α 1 A 3 X k A 3 + α 2 B 3 Y p B 3 + α 3 C 3 Y q C 3
for α i { 1 , + 1 } for i = 1 , 2 , 3 and k , p , q ( 0 , 1 ] .
Our work generalizes the results of [17] on the existence and uniqueness of fixed points in partially ordered metric spaces for monotone mappings by removing the assumption of continuity. We adapt the notion of tripled fixed points so that the solution ( X , Y , Z ) , X = Y = Z of Equation (2) represents a solution of
X = Q + α 1 A X k A + α 2 B X p B + α 3 C X q C .
The idea of tripled fixed points was introduced in different aspects in [30,31,32]. Later on, a more sophisticated notion of n-tuples of fixed points was proposed in [33].
Starting with the introduction of coupled fixed points, as well as tripled and n-tuple ones, there has been one drawback: all results lead to solutions ( x 1 , x 2 , x k ) for k = 2 , 3 , n , so that x i = x j for i , j = 1 , 2 , n . This reduction in the classes of possible systems of equations for which solutions can be found is overcome by using the concept of generalized coupled fixed points [27] and triples of fixed points [34,35].
It seems that most of the illustrative examples for solving of systems of algebraic equations in partially ordered space naturally satisfy the contractive type conditions for all elements x , y X instead of only for the partially ordered ones. Following [17,24,36], it seems that a natural application of the fixed point theorem for monotone maps [17,18] exists in the investigation of matrix equations.

2. Materials and Methods

2.1. Tripled Fixed Point

In what follows, let ( X , d ) be a metric space, N be the set of natural numbers, and R be the set of real numbers. Whenever we consider a normed space ( X , · ) , we will assume that the metric in X is the one endowed by the norm, i.e., d ( x , y ) = x y .
Starting with the publications [18,26], it is assumed that a partially ordered set is a well-known notion. Just for completeness of the presentation, we will recall its definition, following the presentation from [37].
Definition 1
([37]). Let X be a set and p X × X be a relation. The relation p is said to be:
(d1.i) 
reflexive, provided that ( x , x ) p for every x X
(d1.ii) 
antisymmetric, provided that if ( x , y ) p and ( y , x ) p then x = y
(d1.iii) 
transitive, provided that ( x , y ) p and ( y , z ) p then ( x , z ) p
Definition 2
([37]). Let X be a set. A partial order on X is a relation p X × X that is reflexive, antisymmetric, and transitive. The ordered pair ( X , p ) is called a partially ordered set.
For a partially ordered set ( X , p ) , there are various partial orders with their traditional notations. For example, the relation δ N × N that checks if m N is a divisor of n N is denoted by m | n . Therefore, we write m | n to denote that ( m , n ) δ . This is the well-known divisibility relation on N . Whenever practical, for generic partially ordered sets, the partial order relation ≼ is used. Therefore, we will denote ( x , y ) p by x y , which is easier to perceive.
We will denote by ( X , d , ) a metric space with a partial ordering.
Definition 3
([30]). The ordered triple ( ξ , η , ζ ) X × X × X is said to be a tripled fixed points for the map F : X × X × X X if there hold the equalities ξ = F ( ξ , η , ζ ) , y = F ( η , ξ , η ) , and ζ = F ( ζ , η , ξ ) .
The notion of mixed monotone property for a map of two variables is generalized for a map of three variables F : X × X × X X in [30].
Definition 4
([30]). Let ( X , ) be a partially ordered set and F : X × X × X X . We say that F satisfies the mixed monotone property if F ( x , y , z ) is monotonely nondecreasing on its first x and third z variables and monotonely non increasing on its second variable y, i.e., for any x , y , z X , the following holds:
f o r x 1 , x 2 X , x 1 x 2 t h e r e   h o l d s F ( x 1 , y , z ) F ( x 2 , y , z ) ,
f o r y 1 , y 2 X , y 1 y 2 t h e r e   h o l d s F ( x , y 1 , z ) F ( x , y 2 , z )
and
f o r z 1 , z 2 X , z 1 z 2 t h e r e   h o l d s F ( x , y , z 1 ) F ( x , y , z 2 ) .
A sufficient condition for the existence and uniqueness of tripled fixed points is presented in [30], which generalizes the results from [26].
Let ( X , ρ ) be a metric space. The Cartesian product space X × X × X is equipped with the metric ρ 1 ( ( x , y , z ) , ( u , v , w ) ) = ρ ( x , u ) + ρ ( y , v ) + ρ ( z , w ) .
Let ( X , ) be a partially ordered set. Following [30], we introduce a partial order in the Cartesian product space X × X × X as follows: for any ( x , y , z ) , ( u , v , w ) X × X × X we say that ( u , v , w ) ( x , y , z ) if x u , y v , and z w .
Theorem 1
([30]). Let ( X , ρ , ) be a partially ordered complete metric space and F : X × X × X X be a continuous map with the mixed monotone property in X. Let there be constants α , β , γ [ 0 , 1 ) , such that α + β + γ < 1 and there holds the inequality
ρ ( F ( x , y , z ) , F ( u , v , w ) ) α ρ ( x , u ) + β ρ ( y , v ) + γ ρ ( z , w )
for any x u , y v , z w .
If there exist x 0 , y 0 , z 0 X so that x 0 F ( x 0 , y 0 , z 0 ) , y 0 F ( y 0 , x 0 , y 0 ) , and z 0 F ( z 0 , y 0 , x 0 ) , then there are ξ , η , ζ X , which are a tripled fixed point for F, i.e., ξ = F ( ξ , η , ζ ) , η = F ( η , ξ , η ) , and ζ = F ( ζ , η , ξ ) .
If for any ordered triples ( x , y , z ) , ( x 1 , y 1 , z 1 ) X × X × X there is an ordered tripled ( u , v , w ) X × X × X , which is comparable with both ( x , y , z ) and ( x 1 , y 1 , z 1 ) , then the tripled fixed point ( x , y , z ) is unique.
Due the technique of the proof, the tripled fixed point introduced in [30] is defined as ξ = F ( ξ , η , ζ ) , y = F ( η , ξ , η ) , and ζ = F ( ζ , η , ξ ) , not as ξ = F ( ξ , η , ζ ) , η = F ( η , ζ , ξ ) , and ζ = F ( ζ , η , ξ )
We will follow the approach introduced in [34,38].
Definition 5
([34,38]). Let X be a set and F : X × X × X X , G : X × X × X X , and H : X X . The ordered triple ( F , G , H ) is called an ordered triple of maps.
Definition 6
([34,38]). Let X be a set and ( F , G , H ) be an ordered triple of maps. An element ( x , y , z ) X × X × X is called a triple fixed point if
ξ = F ( ξ , η , ζ ) η = G ( ξ , η , ζ ) ζ = H ( ξ , η , ζ ) .
If G ( x , y , z ) = F ( y , x , y ) and H ( x , y , z ) = F ( z , y , x ) we obtain the tripled fixed points investigated in [30], i.e.,
ξ = F ( ξ , η , ζ ) η = F ( η , ξ , η ) ζ = F ( ζ , η , ξ ) .

2.2. Fixed Points for Monotone Maps in Partially Ordered Complete Metric Spaces

Definition 7
([17,37]). Let ( X , ) be a partially ordered set. We say that a map T : X X is monotone if it is either order preserving, i.e., T x T y for all x Y , or order reversing, i.e., f ( x ) y for all x y .
The next result is a slight generalization of the main result from [17].
Theorem 2
([17,36]). Let ( X , d , ) be a partially ordered complete metric space and f : X X be a monotone contractive map.
Let one of the following conditions hold:
(th2.i) 
f is continuous
(th2.ii) 
for any convergent sequence lim n z n = z
  • if z n z n + 1 then z n z
  • if z n z n + 1 then z n z .
If there exists z 0 X such that either z 0 f z 0 or z 0 f z 0 , then there is a fixed point ξ X of f, which is a limit of the sequence { z n } n = 1 , defined by z n = f z n 1 = f n z 0 .
The error estimations are listed below:
(th2.I) 
A priori error estimate d ( z n , ξ ) k n 1 k d ( z 0 , z 1 )
(th2.II) 
A posteriori error estimate d ( z n + 1 , ξ ) k 1 k d ( z n , z n + 1 )
(th2.III) 
The rate of convergence d ( z n + 1 , ξ ) k 1 k d ( z n , z n + 1 ) .
The fixed point ξ is unique if each pair of elements x , y X additionally has a lower bound and an upper bound.
Proposition 1
([36]). Let ( X , d , ) be a partially ordered metric space and f : X X be a monotone contractive map. Then inequality (4) is also satisfied for any x , y X that are comparable.
Proposition 2
([36]). Let ( X , d , ) be a partially ordered metric space and f : X X be a monotone contractive map. Let ξ be a fixed point for f and x 0 X be comparable with ξ. Then each element of the iterated sequence x n = f x n 1 = f n x 0 , for n N is comparable with ξ.
Proposition 3
([39]). Let ( X , ) be a partially ordered set. Every pair of elements x , y X has either a lower or an upper bound if and only if there is z X , comparable with both of them.
A different approach is used in [33]. Instead of considering a partial order, the notion of an F-invariant M set is proposed. We will recall the basic notions from [33] for maps F : X 3 X instead of F : X n X . An element ( x 1 , x 2 , x 3 ) X 3 is said to be a fixed point of order 3 if
x 1 = F ( x 1 , x 2 , x 3 ) x 2 = F ( x 2 , x 3 , x 1 ) x 3 = F ( x 3 , x 1 , x 2 ) .
Let F : X 3 X . A non-empty subset M X 6 is said to be an F–invariant one if for any x 1 , x 2 , x 3 , y 1 , y 2 , y 3 X there holds
( x 1 , x 2 , x 3 , y 1 , y 2 , y 3 ) M ( x 2 , x 3 , y 1 , y 2 , y 3 , x 1 ) M ( x 3 , y 1 , y 2 , y 3 , x 1 , x 2 ) M ( y 1 , y 2 , y 3 , x 1 , x 2 , x 3 ) M ( y 2 , y 3 , x 1 , x 2 , x 3 , y 1 ) M ( y 3 , x 1 , x 2 , x 3 , y 1 , y 2 ) M
and if ( x 1 , x 2 , x 3 , y 1 , y 2 , y 3 ) M , then
( F ( x 1 , x 2 , x 3 ) , F ( x 2 , x 3 , x 1 ) , F ( x 3 , x 1 , x 2 ) , F ( y 1 , y 2 , y 3 ) , F ( y 2 , y 3 , y 1 ) , F ( y 3 , y 1 , y 2 ) ) M
Theorem 3.
Let ( X , d ) be a complete metric space, F : X 3 X , be a continuous mapping, and M X 6 be a non empty F–invariant set. There is ( x 1 , x 2 , x 3 ) X 3 so that
( F ( x 1 , x 2 , x 3 ) , F ( x 2 , x 3 , x 1 ) , F ( x 3 , x 1 , x 2 ) , x 1 , x 2 , x 3 ) M .
For any ( x , y , z , u , v , w ) M there holds the inequality
d ( F ( x , y , z ) , F ( u , v , w ) ) k 3 ( d ( x , u ) + d ( y , v ) + d ( z , w ) ) .
Then, F has a fixed point of order 3. If, in addition, for any ( x 1 , x 2 , x 3 ) , ( y 1 , y 2 , y 3 ) X 3 there is ( z 1 , z 2 , z 3 ) X 3 so that ( x 1 , x 2 , x 3 , z 1 , z 2 , z 3 ) , ( y 1 , y 2 , y 3 , z 1 , z 2 , z 3 ) M , then the fixed point is unique and its components are equal.

2.3. Matrix Equations

Throughout this section, we will denote by H ( N ) the set of all N × N Hermitian matrices. From [40], we will denote by A the conjugated A matrix. A is called Hermitian if A = A , skew-Hermitian if A = A , unitary if A A = I , where I is identity matrix, and the matrix is normal if A A = A A . A matrix A is said to be positive semidefinite if x , A x 0 for all nonzero vectors x. The notation A 0 will be used to express the fact that A is a positive matrix. If x , A x > 0 for all nonzero x, we will say A is a positive definite. We will then write A > 0 . A positive matrix is strictly positive if and only if it is invertible. If A and B are Hermitian matrices, then we say A B ( A > B ) if the matrix A B is positive semidefinite (positive definite), i.e., A B 0 ( A B > 0 ) . For any matrix A the matrix A A is always positive, and its unique positive definite square root is denoted by | A | . The eigenvalues of | A | counted with multiplicities are called the singular values of A . We will always enumerate these in decreasing order, and use for them the notation σ 1 ( A ) σ 2 ( A ) σ n ( A ) . If rank A = k , then σ k ( A ) > 0 , but σ k + 1 ( A ) = = σ n ( A ) = 0 . Every matrix A is unitarily equivalent (or unitarily similar) to an upper triangular matrix T, i.e., A = Q T Q , where Q is unitary.
We will endow H ( N ) with a partial ordering ≼, i.e., A = ( a i , j ) B = ( b i , j ) if a i , j b i , j for all i , j = 1 , 2 , N . If A , B H ( N ) , we will write A B if B A . If A B and A B , we will denote it by A B . By A = 0 we will denote the N × N matrix, that satisfies a i j = 0 for all i , j = 1 , N . If A H ( N ) by A 0 ( A 0 ) we will denote that A is a positive semi-definite (positive definite). For A , B H ( N ) by A B ( A B ) we will assume that A B is a positive semi-definite (positive definite) matrix. If there hold the inequalities A X B we will use the notation X [ A , B ] . We will denote by A the conjugated transpose, and by r ( A ) we will denote the spectral radius of A . We will denote by . the spectral norm, i.e., A = A + ( A A ) , where A + ( A A ) is the largest eigenvalue of A A . The N × N identity matrix will be written as I. We will endow the space H ( N ) with the norm · tr . Let it be recalled, just for completeness, that A tr = j = 1 N σ j ( A ) , where σ j ( A ) , j = 1 , 2 , , N are the singular values of A .
The next lemmas will be useful later.
Lemma 1
([17]). Let A 0 and B 0 be N × N matrices, then 0 t r ( A B ) A t r ( B ) .
Lemma 2
([40]). If 0 < θ 1 , and P and Q are positive definite matrices of the same order with P , Q b I > 0 , then for every unitarily invariant norm | | | · | | | there hold the inequalities | | | P θ Q θ | | | θ b θ 1 | | | P Q | | | and | | | P θ Q θ | | | θ b ( θ + 1 ) | | | P Q | | | .
Lemma 3
([40]). Let A H ( N ) satisfying I A I , then A < 1 .
Theorem 4
(e.g., [41]). Let S be a non-empty, compact, convex subset of a normed vector space. Every continuous function f : S S mapping S into itself has a fixed point.

3. Results

3.1. Tripled Fixed Points for Maps with the Mixed Monotone Property in Partially Ordered Metric Spaces

Proposition 4.
Let ( X , d , ) be a metric space with a partial ordering, F k : X × X × X X , k = 1 , 2 , 3 be such that ( F 1 , F 2 , F 3 ) has the mixed monotone property. If ( X × X × X , ) is endowed with the partial ordering w = ( x , y , z ) ( p , q , s ) = u , provided that x p , y q , and z s , then T z = ( F 1 ( w ) , F 2 ( w ) , F 3 ( w ) ) is monotone increasing map.
Proof. 
From the mixed monotone property of ( F 1 , F 2 , F 3 ) for any w = ( x , y , z ) ( p , q , s ) = u , i.e., x p , y q , and z s , it follows
F k ( x , y , z ) F ( p , y , z ) F k ( p , q , z ) F k ( p , q , s )
for k = 1 , 2 , 3 . Thus
T w = T ( x , y , z ) = ( F 1 ( x , y , z ) , F 2 ( x , y , z ) , F 3 ( x , y , z ) )   ( F 1 ( p , q , s ) , F 2 ( p , q , s ) , F 3 ( p , q , s ) ) = T ( p , q , s ) = T u .
Consequently, T is a monotone increasing with respect to the ordering ≼ in the Cartesian product space ( X × X × X , ) . □
Till the end of this subsection we will endow the Cartesian product set X × X × X with the partial order w = ( x , y , z ) ( p , q , s ) = u , whenever x p , y q , provided that ( X , ) be a partially ordered set.
Theorem 5.
Let ( X , ρ , ) be a partially ordered complete metric space, and ( F 1 , F 2 , F 3 ) be an ordered triple of continuous maps with the mixed monotone property. Let there exist α [ 0 , 1 ) , so that the inequality
k = 1 3 ρ ( F k ( x , y , z ) , F k ( u , v , w ) ) α ( ρ ( x , u ) + ρ ( y , v ) + ρ ( z , w ) )
holds for all x u , y v and z w .
Let one of the following hold:
(th5.i) 
F k , k = 1 , 2 , 3 are continuous maps
(th5.ii) 
for any convergent sequence lim n ( x n , y n , z n ) = ( x , y , z ) , ( x n , y n , z n ) X × X × X
  • if ( x n , y n , z n ) ( x n + 1 , y n + 1 , z n + 1 ) then ( x n , y n , z n ) ( x , y , z )
  • if ( x n , y n , z n ) ( x n + 1 , y n + 1 , z n + 1 ) then ( x n , y n , z n ) ( x , y , z ) .
If there are x 0 , y 0 , z 0 X so that one of the following holds
  • x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , and x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
  • x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , and x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then, a tripled fixed point ( ξ 1 , ξ 2 , ξ 3 ) X × X × X exists, which is a limit of the sequence x 1 ( n ) , x 2 ( n ) , x 3 ( n ) n = 0 , defined by x k ( n ) = F k x 1 ( n ) , x 2 ( n ) , x 3 ( n ) for k = 1 , 2 , 3 and n N .
The following error estimates hold:
(th5.I) 
a priori error estimate
max k = 1 , 2 , 3 d x k ( n ) , ξ k k n 1 k k = 1 3 d x k ( 0 ) , x k ( 1 )
(th5.II) 
a posteriori error estimate
max k = 1 , 2 , 3 d x k ( n + 1 ) , ξ k k 1 k k = 1 3 d x k ( n ) , x k ( n + 1 )
(th5.III) 
the rate of convergence
max k = 1 , 2 , 3 d x k ( n + 1 ) , ξ k k 1 k k = 1 3 d x k ( n ) , ξ k .
If each pair of components ( x , y , z ) , ( u , v , w ) ( X × X × X , ) , where ( x 1 , x 2 , x 3 ) ( y 1 , y 2 , y 3 ) if x 1 y 1 , x 2 y 2 , and x 3 y 3 , has either a lower bound or an upper bound, then ( ξ 1 , ξ 2 , ξ 3 ) is a unique coupled fixed point.
Remark 1.
It should be pointed out that a particular case of Theorem 5 is proven in [38], where a variational technique is used, and only the uniqueness and existence result is obtained without the convergence of the iterated sequence or error estimates. It is also assumed that it holds that F k , k = 1 , 2 , 3 are continuous maps and x 0 F 1 ( x 0 , y 0 , z 0 ) , and y 0 F 2 ( x 0 , y 0 , z 0 ) , and z 0 F 3 ( x 0 , y 0 , z 0 ) .
Proof. 
Let the partial ordering w = ( u , v , z ) ( p , q , s ) = u be assigned to ( X × X × X , ) , given that p x , q v , and s z . Let the function T be defined by
T ( x , y , z ) = ( F 1 ( x , y , z ) , F 2 ( x , y , z ) , F 3 ( x , y , z ) ) .
From Proposition 4, if X × X × X is endowed with the partial ordering ( x , y , z ) ( p , q , s ) , then T is an increasing map.
We will endow the partially ordered Cartesian product ( X × X × X , ) with the metric ρ ( ( x 1 , x 2 , x 3 ) , ( y 1 , y 2 , y 3 ) ) = k = 1 3 d ( x k , y k ) , which satisfies the property
lim n ρ x 1 ( n ) , x 2 ( n ) , x 3 ( n ) , ( x 1 , x 2 , x 3 ) = 0
if and only if lim n d x k ( n ) , x k = 0 for all k = 1 , 2 , 3 .
We will rewrite (5) for the map T and the metric ρ ( · , · ) . Let us denote
x = ( x 1 , x 2 , x 3 ) , u = ( u 1 , u 2 , u 3 ) X × X × X
and x u . Then
ρ ( T x , T u ) = ρ ( ( F 1 ( x ) , F 2 ( x ) , F 3 ( x ) ) , ( F 1 ( u ) , F 2 ( u ) , F 3 ( u ) ) ) = k = 1 3 d ( F k ( x 1 , x 2 , x 3 ) , F k ( u 1 , u 2 , u 3 ) ) α k = 1 3 d ( x k , u k ) = k ρ ( ( x 1 , x 2 x 3 ) , ( u 1 , u 2 , u 3 ) ) = k ρ ( x , u ) .
Consequently, T : X × X × X X × X × X satisfies (4), i.e., T is a monotone contractive map.
If F k , k = 1 , 2 , 3 are continuous in ( X , d ) , then T is continuous in ( X × X × X , ρ ) .
Assumption (th5.ii) in Theorem 5 is the assumption for the map f in Theorem 2 to be a contractive one.
If
x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , a n d x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) T x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) .
If
x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , a n d x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) T x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) .
As a result, it can be concluded that there is a tripled fixed point for the ordered triple of maps ( F 1 , F 2 , F 3 ) , or that there is a fixed point ( ξ 1 , ξ 2 , ξ 3 ) X × X × X of T, because all the conditions in Theorem 2 are fulfilled.
The error estimates follow directly by the definition of the metric function ρ ( · , · ) in X × X × X . □
Theorem 6.
Adding to the hypotheses of Theorem 5 the following conditions:
(th6.i) 
F 2 ( ξ , η , ζ ) = F 1 ( η , ζ , ξ ) and F 3 ( ξ , η , ζ ) = F 1 ( ζ , ξ , η )
(th6.ii) 
if each pair of components ( x , y , z ) , ( u , v , w ) ( X × X × X , ) , where ( x 1 , x 2 , x 3 ) ( y 1 , y 2 , y 3 ) if x 1 y 1 , x 2 y 2 , and x 3 y 3 , has either a lower bound or an upper bound.
Then the components of the tripled fixed point ( ξ 1 , ξ 2 , ξ 3 ) are equal, i.e., ξ 1 = ξ 2 = ξ 3 .
Proof. 
The proof follows directly from [33]. Indeed, let define the F invariant set M by ( x , y , z , u , v , w ) M , if and only if either ( x , y , z ) ( u , v , w ) or ( x , y , z ) ( u , v , w ) , matching all the conditions of Theorem 3. □

3.2. Tripled Fixed Points for Maps with the Total Monotone Property in Partially Ordered Metric Spaces

Definition 8.
Let ( X , ) be a partially ordered set and ( F 1 , F 2 , F 3 ) be an ordered triple of maps. We say that ( F 1 , F 2 , F 3 ) has the total monotone property if for any x , x 1 , x 2 , y , y 1 , y 2 , z , z 1 , z 2 X the following holds:
F i ( x 1 , y , z ) F i ( x 2 , y , z ) , h o l d s   f o r   e v e r y i = 1 , 2 , 3 , p r o v i d e d   t h a t x 1 x 2
F i ( x , y 1 , z ) F i ( x , y 2 , z ) , h o l d s   f o r   e v e r y i = 1 , 2 , 3 , p r o v i d e d   t h a t y 1 y 2
and
F i ( x , y , z 1 ) F i ( x , y , z 2 ) , h o l d s   f o r   e v e r y i = 1 , 2 , 3 , p r o v i d e d   t h a t z 1 z 2 .
Till the end of this subsection, we will endow the Cartesian product set X × X × X with the partial order w = ( x , y , z ) ( p , q , s ) = u , whenever x p , y q , provided that ( X , ) be a partially ordered set.
Theorem 7.
Let ( X , ρ , ) be a partially ordered complete metric space, and ( F 1 , F 2 , F 3 ) be an ordered triple of continuous maps with the total monotone property. Let there exist α [ 0 , 1 ) , so that the inequality
k = 1 3 ρ ( F k ( x , y , z ) , F k ( u , v , w ) ) α ( ρ ( x , u ) + ρ ( y , v ) + ρ ( z , w ) )
holds for all x u , y v and z w .
Let one of the following hold:
(th7.i) 
F k , k = 1 , 2 , 3 are continuous maps
(th7.ii) 
for any convergent sequence lim n ( x n , y n , z n ) = ( x , y , z ) , ( x n , y n , z n ) X × X × X
  • if ( x n , y n , z n ) ( x n + 1 , y n + 1 , z n + 1 ) then ( x n , y n , z n ) ( x , y , z )
  • if ( x n , y n , z n ) ( x n + 1 , y n + 1 , z n + 1 ) then ( x n , y n , z n ) ( x , y , z ) .
If there are x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) X so that one of the following holds
  • x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , and x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
  • x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , and x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then a tripled fixed point ( ξ 1 , ξ 2 , ξ 2 ) X × X × X exists.
The following error estimates hold:
(th7.I) 
a priori error estimate
max k = 1 , 2 , 3 d x k ( n ) , ξ k k n 1 k k = 1 3 d x k ( 0 ) , x k ( 1 )
(th7.II) 
a posteriori error estimate
max k = 1 , 2 , 3 d x k ( n + 1 ) , ξ k k 1 k k = 1 3 d x k ( n ) , x k ( n + 1 )
(th7.III) 
the rate of convergence
max k = 1 , 2 , 3 d x k ( n + 1 ) , ξ k k 1 k k = 1 3 d x k ( n ) , ξ k .
If each pair of components ( x , y , z ) , ( u , v , w ) ( X × X × X , ) also has a lower bound or an upper bound, then ( ξ 1 , ξ 2 , ξ 3 ) is a unique coupled fixed point.
Proof. 
Let the partial ordering w = ( u , v , z ) ( p , q , s ) = u be assigned to ( X × X , ) , given that p x , q v , and s z . Let the function T be defined by
T ( x , y , z ) = ( F 1 ( x , y , z ) , F 2 ( x , y , z ) , F 3 ( x , y , z ) ) .
From Proposition 4, if X × X × X is endowed with the partial ordering ( x , y , z ) ( p , q , s ) , then T is an increasing map.
We will endow the partially ordered Cartesian product ( X × X × X , ) with the metric ρ ( ( x 1 , x 2 , x 3 ) , ( y 1 , y 2 , y 3 ) ) = k = 1 3 d ( x k , y k ) , that satisfies the property
lim n ρ ( ( x 1 ( n ) , x 2 ( n ) , x 3 ( n ) ) , ( x 1 , x 2 , x 3 ) ) = 0
if and only if lim n d ( x k ( n ) , x k ) = 0 for all k = 1 , 2 , 3 .
We will rewrite (7) for the map T and the metric ρ ( · , · ) . Let us denote x = ( x 1 , x 2 , x 3 ) , u = ( u 1 , u 2 , u 3 ) X × X × X and x u . Then
ρ ( T x , T u ) = ρ ( ( F 1 ( x ) , F 2 ( x ) , F 3 ( x ) ) , ( F 1 ( u ) , F 2 ( u ) , F 3 ( u ) ) ) = k = 1 3 d ( F k ( x k , x 2 , x 3 ) , F k ( u 1 , u 2 , u 3 ) ) α k = 1 3 d ( x k , u k ) = α ρ ( ( x 1 , x 2 x 3 ) , ( u 1 , u 2 , u 3 ) ) = α ρ ( x , u ) .
Consequently, T : X × X × X X × X × X satisfies (4), i.e., T is a monotone contractive map.
If F k , k = 1 , 2 , 3 are continuous in ( X , d ) then T is continuous in ( X × X × X , ρ ) .
Assumption (th7.ii) in Theorem 5 is the assumption for the map f in Theorem 2 to be a contractive one.
If
x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , a n d x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) T x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) .
If
x 1 ( 0 ) F 1 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , x 2 ( 0 ) F 2 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) , a n d x 3 ( 0 ) F 3 x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 )
then x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) T x 1 ( 0 ) , x 2 ( 0 ) , x 3 ( 0 ) .
As a result, it can be concluded that there is a tripled fixed point for the ordered triple of maps ( F 1 , F 2 , F 3 ) , or that there is a fixed point ( ξ 1 , ξ 2 , ξ 3 ) X × X × X of T, because all the conditions in Theorem 2 are fulfilled.
The error estimates follow directly by the definition of the metric function ρ ( · , · ) in X × X × X . □
Theorem 8.
Adding to the hypotheses of Theorem 7 the following conditions:
(th8.i) 
F 2 ( ξ , η , ζ ) = F 1 ( η , ζ , ξ ) and F 3 ( ξ , η , ζ ) = F 1 ( ζ , ξ , η )
(th8.iI) 
any two triples ( x , y , z ) , ( u , v , w ) X × X × X there is a triple ( p , q , r ) X × X × X ) , comparable with both ( x , y , z ) , ( u , v , w ) .
The components of the tripled fixed point ( ξ 1 , ξ 2 , ξ 3 ) are equal, i.e., ξ 1 = ξ 2 = ξ 3 .
Proof. 
The proof follows directly from [33]. Indeed, let us define the F invariant set M by ( x , y , z , u , v , w ) M , if and only if either ( x , y , z ) ( u , v , w ) or ( x , y , z ) ( u , v , w ) , matching all the conditions of Theorem 3. □

4. Applications of Theorems 5 and 7 in Solving of Systems of Matrix Equations

4.1. Application of Theorem 5

We will use the notation Ω a I = { X H ( N ) : X a I } . Let the system of matrix Equation (9)
X i = Q i + k = 1 3 ( 1 ) k A i k X k θ k A i k , i , k = 1 , 2 , 3
be considered, where θ k ( 0 , 1 ] , A i k X ( N ) , Q i H ( N ) and Q i , A i k A i k , P ( N ) for i , k = 1 , 2 , 3 .
Theorem 9.
Let there be 0 < a 1 , a 3 < a 2 , so that
(th9.i) 
Q i a 1 θ 1 A i 1 A i 1 a 1 θ 3 A i 3 A i 3 a 1 I , for i = 1 , 2 , 3
(th9.ii) 
a 2 θ 2 a 3 θ 3 A 11 A 11 a 1 θ 1 a 3 θ 3 A 12 A 12 + a 1 θ 1 a 2 θ 2 A 13 A 13 a 1 θ 1 a 2 θ 2 a 3 θ 3 ( Q 1 a 1 I ) , a 2 θ 2 a 3 θ 3 A 21 A 21 a 1 θ 1 a 3 θ 3 A 22 A 22 + a 1 θ 1 a 2 θ 2 A 23 A 23 a 1 θ 1 a 2 θ 2 a 3 θ 3 ( Q 2 a 2 I ) , a 2 θ 2 a 3 θ 3 A 31 A 31 a 1 θ 1 a 3 θ 3 A 32 A 32 + a 1 θ 1 a 2 θ 2 A 33 A 33 a 1 θ 1 a 2 θ 2 a 3 θ 3 ( Q 3 a 3 I )
(th9.iii) 
δ = max θ 1 i = 1 3 A i 1 A i 1 a 1 θ 1 + 1 , θ 2 i = 1 3 A i 2 A i 2 a 1 θ 2 + 1 , θ 3 i = 1 3 A i 3 A i 3 a 1 θ 3 + 1 < 1
are satisfied. Then the system (9) has a unique solution X ˜ i , Ω a I for i = 1 , 2 , 3 .
Moreover the sequences X i ( n ) n = 0 , i = 1 , 2 , 3 , defined as follows
X i ( 0 ) = a i I , X i ( n + 1 ) = Q i A i 1 X 1 ( n ) θ 1 A i 1 + A i 2 X 2 ( n ) θ 2 A i 2 A i 3 X 3 ( n ) θ 3 A i 3 , f o r i = 1 , 2 , 3
converge to the solution X ˜ i , i = 1 , 2 , 3 of considered system of matrix equations, in the space ( H ( N ) , · tr , ) , respectively.
The following error estimations are listed below:
(th9.I) 
a priori error estimate
max i = 1 , 2 , 3 X i ( n ) X ˜ i tr δ n 1 δ M X 1 ( 1 ) , ( X 1 ( 0 ) , X 2 ( 1 ) , X 2 ( 0 ) , X 3 ( 1 ) , X 3 ( 0 )
(th9.II) 
a posteriori error estimate
max i = 1 , 2 , 3 X i ( n + 1 ) X ˜ i tr δ 1 δ M X 1 ( n + 1 ) , X 1 ( n ) , X 2 ( n + 1 ) , X 2 ( n ) , X 3 ( n + 1 ) , X 3 ( n )
(th9.III) 
the convergence rate
max i = 1 , 2 , 3 X i ( n + 1 ) X ˜ i tr δ 1 δ M X 1 ( n ) , X ˜ 1 , X 2 ( n ) , X ˜ 2 , X 3 ( n ) , X ˜ 3 ,
where we use the notation M ( U 1 , V 1 , U 2 , V 2 , U 3 , V 3 ) = i = 1 3 U i V i tr .
If, in addition, for i = 1 , 2 , 3 Q i , Q i a i θ 1 A i 1 A i 1 a i θ 3 A i 3 A i 3 a i I P ( N ) and F i are continuous, then there is n 0 N so that ( X ˜ 1 , X ˜ 2 , X i ˜ ) [ a 1 I , n 0 Q 1 ] × [ a 1 I , n 0 Q 2 ] × [ a 1 I , n 0 Q 3 ] .
Proof. 
Let us define the maps F i : H ( N ) × H ( N ) × H ( N ) H ( N ) by
F i ( X 1 , X 2 , X 3 ) = Q i + k = 1 3 ( 1 ) k A i k X k θ k A i k .
Let X i Ω a 1 I , i.e., X i a 1 I , i = 1 , 2 , 3 . Then, from Q i , A i k A i k , P ( N ) for i , k = 1 , 2 , 3 and a we obtain
F i ( X 1 , X 2 , X 3 ) = Q i + k = 1 3 ( 1 ) k A i k X k θ k A i k   Q i A i 1 X 1 θ 1 A i 1 A i 3 X 3 θ 3 A i 3   Q i a 1 θ 1 A i 1 A i 1 a 1 θ 3 A i 3 A i 3 a 1 I .
Thus F i : Ω a 1 I × Ω a 1 I × Ω a 1 I Ω a 1 I .
We will show that the ordered triple of maps ( F 1 , F 2 , F 3 ) satisfies the mixed monotone property. Let X , U Ω a 1 I be such that X U then X θ i U θ i . Therefore, by A i k A i k P ( N ) we get
F i ( X , Y , Z ) F i ( U , Y , Z ) = A i 1 ( X θ 1 U θ 1 ) A i 1 = A i 1 ( U θ 1 X θ 1 ) A i 1 0 ,
F i ( Y , X , Z ) F i ( Y , U , Z ) = A i 2 ( X θ 2 U θ 2 ) A i 2 = A i 2 ( U θ 2 X θ 2 ) A i 2 0
and
F i ( Y , Z , X ) F i ( Y , Z , U ) = A i 3 ( X θ 3 U θ 3 ) A i 3 = A i 3 ( U θ 3 X θ 3 ) A i 3 0 .
Thus, the ordered pair of maps satisfies the mixed monotone property.
If we set X i ( 0 ) = a i I , for i = 1 , 2 , 3 using (th9.ii), we can easily show that X 1 ( 0 ) F 1 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) , X 2 ( 0 ) F 2 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) and X 3 ( 0 ) F 3 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) .
The inequalities
F 1 ( a 1 I , a 2 I , a 3 I ) = Q 1 a 1 θ 1 A 11 A 11 + a 2 θ 2 A 12 A 12 a 3 θ 3 A 13 A 13 a 1 I ,
F 2 ( a 1 I , a 2 I , a 3 I ) = Q 2 a 1 θ 1 A 21 A 21 + a 2 θ 2 A 22 A 22 a 3 θ 3 A 23 A 23 a 2 I ,
F 3 ( a 1 I , a 2 I , a 3 I ) = Q 3 a 1 θ 1 A 31 A 31 + a 2 θ 2 A 32 A 32 a 3 θ 3 A 33 A 33 a 3 I
are equivalent (th9.ii). Thus there holds x F i ( x , y , z ) , y F 2 ( x , y , z ) and z F 3 ( x , y , z ) for some ( x , y , z ) .
There exists a greatest lower bound and a least upper bound for each X , Y H ( N ) .
The maps F i , i = 1 , 2 , 3 are continuous.
Let us put
T ( x , y , z ) = ( F 1 ( x , y , z ) , F 2 ( x , y , z ) , F 3 ( x , y , z ) ) : Ω a 1 I × Ω a 1 I × Ω a 1 I Ω a 1 I × Ω a 1 I × Ω a 1 I
and ( Ω a 1 I × Ω a 1 I × Ω a 1 I , | | | · | | | , ) , where ( x , y , z ) ( u , v , w ) if x u , y v , and z w and | | | ( x , y , z ) | | | = x tr + y tr + z tr . We can write down the chain of inequalities
S = | | | T ( X 1 , X 2 , X 3 ) T ( U 1 , U 2 , U 3 ) | | |   = | | | ( F 1 ( X ) F 1 ( U ) , F 2 ( X ) F ( U ) , F 3 ( X ) F ( U ) ) | | |   = i = 1 3 F i ( X 1 , X 2 , X 3 ) F i ( U 1 , U 2 , U 3 ) tr   = i = 1 3 j = 1 3 A i j U j θ j X j θ j A i j   = i = 1 3 tr j = 1 3 A i j A i j U j θ j X j θ j   j = 1 3 i = 1 3 A i j A i j . tr ( U j θ j X j θ j )   j = 1 3 i = 1 3 A i j A i j . θ j a 1 θ j + 1 tr ( X j U j )   max j = 1 , 2 , 3 i = 1 3 A i j A i j j = 1 3 θ j a 1 θ j + 1 tr ( X j U j )   = max j = 1 , 2 , 3 θ j i = 1 3 A i j A i j a 1 θ j + 1 j = 1 3 tr ( X j U j )   = δ j = 1 3 X j U j tr   = δ | | | ( X 1 , X 2 , X 3 ) ( U 1 , U 2 , U 3 ) | | | ,
where δ = max j = 1 , 2 , 3 θ j i = 1 3 A i j A i j a 1 θ j + 1 < 1 .
Consequently, the ordered pair of maps ( F 1 , F 2 , F 3 ) satisfies the assumption in Theorem 5, and thus a unique ordered pair ( X ˜ 1 , X ˜ 2 , X ˜ 3 ) Ω a 1 I × Ω a 1 I × Ω a 1 I exists, which is a solution to system (9).
We have proven the existence of unique solutions X ˜ 1 , X ˜ 2 , X ˜ 3 a 1 I . We will give a shorter interval where the solutions are situated.
Let there hold the inequalities
Q i , Q i a 1 θ 1 A i 1 A i 1 a 1 θ 3 A i 3 A i 3 a 1 I P ( N ) f o r i = 1 , 2 , 3
For any Q i P ( N ) there is n 0 N so that
n 0 θ 1 N A 11 Q 1 θ 1 A 11 + a 1 θ 2 A 12 A 12 n 0 θ 3 N A 13 Q 1 θ 3 A 13 ( n 0 1 ) Q 1 ,
n 0 θ 1 N A 21 Q 1 θ 1 A 21 + a 1 θ 2 A 22 A 22 n 0 θ 3 N A 23 Q 1 θ 3 A 23 ( n 0 1 ) Q 2 ,
and
n 0 θ 1 N A 31 Q 1 θ 1 A 31 + a 1 θ 2 A 32 A 32 n 0 θ 3 N A 33 Q 3 θ 3 A 33 ( n 0 1 ) Q 3 ,
The last three inequalities are equivalent to
S 1 = F 1 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) = Q 1 A 11 ( n 0 Q 1 ) θ 1 A 11 + A 12 ( a 1 I ) θ 2 A 12 A 13 ( n 0 Q 3 ) θ 3 A 13 = Q 1 n 0 θ 1 N A 11 Q 1 θ 1 A 11 + a 1 θ 2 A 12 A 12 n 0 θ 3 N A 13 Q 3 θ 3 A 13 n 0 Q 1 ,
S 2 = F 2 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) = Q 2 A 21 ( n 0 Q 1 ) θ 1 A 21 + A 22 ( a 1 I ) θ 2 A 22 A 23 ( n 0 Q 3 ) θ 3 A 23 = Q 2 n 0 θ 1 N A 21 Q 1 θ 1 A 21 + a 1 θ 2 A 22 A 22 n 0 θ 3 N A 23 Q 3 θ 3 A 23 n 0 Q 2 ,
and
S 3 = F 3 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) = Q 3 A 31 ( n 0 Q 1 ) θ 1 A 31 + A 32 ( a 1 I ) θ 2 A 32 A 33 ( n 0 Q 3 ) θ 3 A 33 = Q 3 n 0 θ 1 N A 31 Q 1 θ 1 A 31 + a 1 θ 2 A 32 A 32 n 0 θ 3 N A 33 Q 3 θ 3 A 33 n 0 Q 3 .
From (12) it follows
S 4 = F 1 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 1 A 11 ( a 1 I ) θ 1 A 11 + A 12 ( n 0 Q 2 ) θ 2 A 12 A 13 ( a 1 I ) θ 3 A 13 Q 1 a 1 θ 1 A 11 A 11 a 1 θ 3 A 13 A 13 a 1 I ,
and
S 5 = F 2 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 2 A 21 ( a 1 I ) θ 1 A 21 + A 22 ( n 0 Q 2 ) θ 2 A 22 A 23 ( a 1 I ) θ 3 A 23 Q 2 a 1 θ 1 A 21 A 21 a 1 θ 3 A 23 A 23 a 1 I ,
and
S 6 = F 3 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 3 A 31 ( a 1 I ) θ 1 A 31 + A 32 ( n 0 Q 2 ) θ 2 A 32 A 33 ( a 1 I ) θ 3 A 33 Q 3 a 1 θ 1 A 31 A 31 a 1 θ 3 A 33 A 33 a 1 I .
Let X [ a 1 I , n 0 Q 1 ] , Y [ a 1 I , n 0 Q 2 ] , and Z [ a 1 I , n 0 Q 3 ] , i.e.,
a 1 I X n 0 Q 1 a n d a 1 I Y n 0 Q 2 a n d a 1 I Z n 0 Q 3 .
Using the mixed monotone property of ( F 1 , F 2 , F 3 ) and (13)–(18), we obtain
a 1 I F 1 ( a 1 I , n 0 Q 2 , a 1 I ) F 1 ( X , Y , Z ) F 1 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) n 0 Q 1 ,
a 1 I F 2 ( a 1 I , n 0 Q 2 , a 1 I ) F 2 ( X , Y , Z ) F 2 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) n 0 Q 2 ,
and
a 1 I F 3 ( a 1 I , n 0 Q 2 , a 1 I ) F 3 ( X , Y , Z ) F 3 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) n 0 Q 3 .
Consequently, from (19)–(21), it follows that
T : [ a 1 I , n 0 Q 1 ] × [ a 1 I , n 0 Q 2 ] × [ a 1 I , n 0 Q 3 ] [ a 1 I , n 0 Q 1 ] × [ a 1 I , n 0 Q 2 ] × [ a 1 I , n 0 Q 3 ] ,
where T ( X , Y , Z ) = ( F 1 ( X , Y , Z ) , F 2 ( X , Y , Z ) , F 3 ( X , Y , Z ) ) . As far as T is a continuous map and [ a 1 I , n 0 Q 1 ] × [ a 1 I , n 0 Q 2 ] × [ a 1 I , n 0 Q 3 ] is compact and convex according to Theorem 4, it follows that T has at least one fixed point in [ a 1 I , n 0 Q 1 ] × [ a 1 I , n 0 Q 2 ] × [ a 1 I , n 0 Q 3 ] . □
We will use the notation Ω a 1 I = { X H ( N ) : X a 1 I } . Let the matrix Equation (22)
X = Q A X θ A + B X θ B C X θ C
be considered, where, θ ( 0 , 1 ] , A , B , C X ( N ) , Q H ( N ) and Q , A A , B B , C C P ( N ) .
Theorem 10.
Let there be 0 < a < b , so that
(th10.i) 
Q a θ A A a θ C C a I
(th10.ii) 
b θ A A a θ B B + b θ C C a θ b θ ( Q a I )
and
b θ A A a θ B B + b θ C C a θ b θ ( Q b I ) ,
(th10.iii) 
δ = max θ A A a θ + 1 , θ B B a θ + 1 , θ C C a θ + 1 < 1 / 3 .
Then the system
X 0 = a I , Y 0 = b I , Z 0 = a I , X n + 1 = Q A X n θ A + B Y n θ B C Z n θ C Y n + 1 = Q A Y n θ A + B Z n θ B C X n θ C Z n + 1 = Q A Z n θ A + B X n θ B C Y n θ C
has a unique solution ( X , Y , Z ) , such that X = Y = Z , which is the unique solution of Equation (22).

4.2. Application of Theorem 7

We will use the notation Ω a 1 I = { X H ( N ) : X a 1 I } . Let the system of matrix Equation (23)
X i = Q i k = 1 3 A i k X k θ k A i k , i = 1 , 2 , 3
be considered, where θ k ( 0 , 1 ] , A i k X ( N ) , Q i H ( N ) and Q i , A i k A i k , P ( N ) for i , k = 1 , 2 , 3 .
Theorem 11.
Let there be 0 < a 1 , a 3 < a 2 , so that
(th11.i) 
Q i k = 1 3 a 1 θ k A i k A i k a 1 I , for i = 1 , 2 , 3
(th11.ii) 
a 2 θ 2 a 3 θ 3 A i 1 A i 1 + a 1 θ 1 a 3 θ 3 A i 2 A i 2 + a 1 θ 1 a 2 θ 2 A i 3 A i 3 a 1 θ 1 a 2 θ 2 a 3 θ 3 ( Q i a 1 I ) , for i = 1 , 2 , 3
(th11.iii) 
δ = max θ 1 i = 1 3 A i 1 A i 1 a 1 θ 1 + 1 , θ 2 i = 1 3 A i 2 A i 2 a 1 θ 2 + 1 , θ 3 i = 1 3 A i 3 A i 3 a 1 θ 3 + 1 < 1
are satisfied.
Then the system (23) has a unique solution X ˜ i , Ω a I for i = 1 , 2 , 3 .
Moreover, the sequences X i ( n ) n = 0 , i = 1 , 2 , 3 , defined as follows
X i ( 0 ) = a i I , X i ( n + 1 ) = Q i A i 1 X 1 ( n ) θ 1 A i 1 A i 2 X 2 ( n ) θ 2 A i 2 A i 3 X 3 ( n ) θ 3 A i 3 , f o r i = 1 , 2 , 3
converge to the solution X ˜ i , i = 1 , 2 , 3 of considered system of matrix equations, in the space ( H ( N ) , · tr , ) , respectively.
The error estimations are listed below:
(th11.I) 
a priori error estimate
max i = 1 , 2 , 3 X i ( n ) X ˜ i tr δ n 1 δ M X 1 ( 1 ) , ( X 1 ( 0 ) , X 2 ( 1 ) , X 2 ( 0 ) , X 3 ( 1 ) , X 3 ( 0 )
(th11.II) 
a posteriori error estimate
max i = 1 , 2 , 3 X i ( n + 1 ) X ˜ i tr δ 1 δ M X 1 ( n + 1 ) , X 1 ( n ) , X 2 ( n + 1 ) , X 2 ( n ) , X 3 ( n + 1 ) , X 3 ( n )
(th11.III) 
the convergence rate
max i = 1 , 2 , 3 X i ( n + 1 ) X ˜ i tr δ 1 δ M X 1 ( n ) , X ˜ 1 , X 2 ( n ) , X ˜ 2 , X 3 ( n ) , X ˜ 3 ,
where we use the notation M ( U 1 , V 1 , U 2 , V 2 , U 3 , V 3 ) = i = 1 3 U i V i tr .
Proof. 
Let us define the maps F i : H ( N ) × H ( N ) × H ( N ) H ( N ) by
F i ( X 1 , X 2 , X 3 ) = Q i k = 1 3 A i k X k θ k A i k .
Let X i Ω a 1 I , i.e., X i a 1 I , i = 1 , 2 , 3 . Then, from Q i , A i k A i k , P ( N ) for i , k = 1 , 2 , 3 and (th9.i) we obtain
F i ( X 1 , X 2 , X 3 ) = Q i k = 1 k A i k X k θ k A i k a 1 I .
Thus, F i : Ω a 1 I × Ω a 1 I × Ω a 1 I Ω a 1 I .
We will show that the ordered triple of maps ( F 1 , F 2 , F 3 ) satisfies the total monotone property. Let X , U Ω a 1 I be such that X U then X θ i U θ i . Therefore, by A i k A i k P ( N ) we obtain
F i ( X , Y , Z ) F i ( U , Y , Z ) = A i 1 ( X θ 1 U θ 1 ) A i 1 = A i 1 ( U θ 1 X θ 1 ) A i 1 0 ,
F i ( Y , X , Z ) F i ( Y , U , Z ) = A i 2 ( X θ 2 U θ 2 ) A i 2 = A i 2 ( U θ 2 X θ 2 ) A i 2 0 ,
and
F i ( Y , Z , X ) F i ( Y , Z , U ) = A i 3 ( X θ 3 U θ 3 ) A i 3 = A i 3 ( U θ 2 X θ 2 ) A i 3 0 .
Thus, the ordered pair of maps satisfies the total monotone property.
If we set X i ( 0 ) = a i I for i = 1 , 2 , 3 using (th11.ii), we can easily show that
X 1 ( 0 ) F 1 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) ,
X 2 ( 0 ) F 2 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) ,
and
X 3 ( 0 ) F 3 X 1 ( 0 ) , X 2 ( 0 ) , X 3 ( 0 ) .
The inequalities
F i ( a 1 I , a 2 I , a 3 I ) = Q i a 1 θ 1 A i 1 A i 1 a 2 θ 2 A i 2 A i 2 a 3 θ 3 A i 3 A i 3 a i I
for every i = 1 , 2 , 3 are equivalent (th11.ii). Thus, there holds x F i ( x , y , z ) , y F 2 ( x , y , z ) and z F 3 ( x , y , z ) for some ( x , y , z ) .
There exists a greatest lower bound and a least upper bound for each X , Y H ( N ) .
The maps F i , i = 1 , 2 , 3 are continuous.
Let us put
T ( x , y , z ) = ( F 1 ( x , y , z ) , F 2 ( x , y , z ) , F 3 ( x , y , z ) ) : Ω a 1 I × Ω a 1 I × Ω a 1 I Ω a 1 I × Ω a 1 I × Ω a 1 I
and ( Ω a 1 I × Ω a 1 I × Ω a 1 I , | | | · | | | , ) , where ( x , y , z ) ( u , v , w ) if x u , y v , and z w and | | | ( x , y , z ) | | | = x tr + y tr + z tr . We can write down the chain of inequalities
S = | | | T ( X 1 , X 2 , X 3 ) T ( U 1 , U 2 , U 3 ) | | |   = | | | ( F 1 ( X ) F 1 ( U ) , F 2 ( X ) F ( U ) , F 3 ( X ) F ( U ) ) | | |   = i = 1 3 F i ( X 1 , X 2 , X 3 ) F i ( U 1 , U 2 , U 3 ) tr   = i = 1 3 j = 1 3 A i j U j θ j X j θ j A i j   = i = 1 3 tr j = 1 3 A i j A i j U j θ j X j θ j   j = 1 3 i = 1 3 A i j A i j . tr ( U j θ j X j θ j )   j = 1 3 i = 1 3 A i j A i j . θ j a 1 θ j + 1 tr ( X j U j )   max j = 1 , 2 , 3 i = 1 3 A i j A i j j = 1 3 θ j a 1 θ j + 1 tr ( X j U j )   = max j = 1 , 2 , 3 θ j i = 1 3 A i j A i j a 1 θ j + 1 j = 1 3 tr ( X j U j )   = δ j = 1 3 X j U j tr   = δ | | | ( X 1 , X 2 , X 3 ) ( U 1 , U 2 , U 3 ) | | | ,
where δ = max j = 1 , 2 , 3 θ j i = 1 3 A i j A i j a 1 θ j + 1 < 1 .
Consequently, the ordered pair of maps ( F 1 , F 2 , F 3 ) satisfies the assumption in Theorem 7, and thus a unique ordered pair ( X ˜ 1 , X ˜ 1 , X ˜ 3 ) Ω a I × Ω a I × Ω a I exists, which is a solution to system (23). □
Following the ideas from [24] we will try to solve the matrix equation
X = Q A X θ A B X θ B C X θ C .
Let us put A 11 = A 21 = A 31 = A , A 12 = A 22 = A 32 = B , A 13 = A 23 = A 33 = C , θ 1 = θ 2 = θ 3 = θ , and Q 1 = Q 2 = Q 3 = Q in (23). Let us denote F ( X , Y , Z ) = Q A X θ A + B Y θ B C Z θ C . Then, it holds that F 1 ( X , Y , Z ) = F 1 ( X , Y , Z ) , F 2 ( X , Y , Z ) = F ( Y , Z , X ) , and F 3 ( X , Y , Z ) = F ( Z , X , Y ) , and consequently, we can apply Theorems 7 and 11.
Theorem 12.
Let there be 0 < a , so that
(th12.i) 
Q a θ A A a θ B B a θ C C a I ,
(th12.ii) 
δ = max θ A A a θ + 1 , θ B B a θ + 1 , θ C C a θ + 1 < 1 / 3
are satisfied.
Then, Equation (25) has a unique solution X ˜ , which is the solutions of the system of matrix equations
X 0 = Y 0 = Z 0 = a I , X n + 1 = Q A X n θ A B Y n θ B C Z n θ C Y n + 1 = Q A Y n θ A B Z n θ B C X n θ C Z n + 1 = Q A Z n θ A B X n θ B C Y n θ C ,
where X ˜ = Y ˜ = Z ˜ Ω a I . The error estimations from Theorem 11 hold.

5. Illustrative Examples

Let us put
A 11 = 0.021 0.01 0.1 0.1 0.03 0.1 1 0.01 0.2 , A 21 = 0.01875 0.00438 0.125 0.125 0.05 0.00625 0.0025 0.0125 0.0625 ,
A 31 = 0.03667 0.00333 0.00333 0.01667 0.66 0.06 0.03333 0.06333 0.79333 , A 12 = 0.08333 0.01667 0.01667 0.025 0.08333 0.025 0.03333 0.03333 0.08333 ,
A 22 = 0.18 0.54 0.54 0.72 1.8 0.72 0.9 0.9 1.8 , A 32 = 0.015 0.0005 0.0005 0 0.014 0.001 0.001 0.0015 0.017 ,
A 13 = 0.2 0.07 0.11 0.07 0.06 0.13 0.11 0.13 0.14 , A 23 = 0.0066 0.0066 0.016 0.0024 0.0072 0.01 0.064 0.1426 0.00268 ,
A 33 = 0.02 0.025 0.006 0 0.07 0.06 0.005 0.005 0.16 , Q 1 = 3.5 0.5 0.5 0.5 3.5 1 0.5 1 4 ,
Q 2 = 7.5 0.5 2.5 0.5 7.5 4.5 2.5 4.5 34 , Q 3 = 10.5 7.5 1.5 7.5 12.45 1.5 1.5 1.5 12 ,
and a 1 = 2 , a 2 = 45 , a 3 = 2.5 , θ 1 = 1 5 , θ 2 = 1 13 , and θ 3 = 1 7 .
We give four numerical examples to illustrate that the matrix sequences X k , Y k and Z k , generated by the relevant systems, converge to the positive definite solutions ( X ˜ , Y ˜ , Z ˜ ) . We use the stopping criterion
max X n + 1 X ˜ tr , Y n + 1 Y ˜ tr , Z n + 1 Z ˜ tr 10 10 .

5.1. Application of Theorem 9

Example 1.
Let us consider the system of matrix equations
X = Q 1 A 11 X θ 1 A 11 + A 12 Y θ 2 A 12 A 13 Z θ 3 A 13 Y = Q 2 A 21 X θ 1 A 21 + A 22 Y θ 2 A 22 A 23 Z θ 3 A 23 Z = Q 3 A 31 X θ 1 A 31 + A 32 Y θ 2 A 32 A 33 Z θ 3 A 33 .
To solve (27) we will apply Theorem 9. It is easy to calculate A i k A i k 0 , Q i , A i k A i k , P ( N ) .
Q 1 a 1 θ 1 A 11 A 11 a 1 θ 3 A 13 A 13 a 1 I 0.56873 0.48442 0.28972 0.48442 1.47604 0.99233 0.28972 0.99233 1.90375 0 ,
Q 2 a 1 θ 1 A 21 A 21 a 1 θ 3 A 23 A 23 a 1 I 5.48233 0.49104 2.49687 0.49104 5.48133 4.49832 2.49687 4.49832 31.98263 0 ,
Q 3 a 1 θ 1 A 31 A 31 a 1 θ 3 A 33 A 33 a 1 I 8.49724 7.51083 1.52266 7.51083 10.06226 1.48751 1.52266 1.48751 9.42247 0 ,
S 7 = Q 1 a 1 θ 1 A 11 A 11 + a 1 θ 2 A 12 A 12 a 1 θ 3 A 13 A 13 a 1 I   4.57683 0.48590 0.29110 0.48590 5.48298 0.99313 0.29110 0.99313 5.91099 0 ,
S 8 = Q 2 a 1 θ 1 A 21 A 21 + a 1 θ 2 A 22 A 22 a 1 θ 3 A 23 A 23 a 2 I   36.50218 2.13523 4.01993 2.13523 34.27858 6.45655 4.01993 6.45655 7.99542 0 ,
and
S 9 = Q 3 a 1 θ 1 A 31 A 31 + a 1 θ 2 A 32 A 32 a 1 θ 3 A 33 A 33 a 3 I   7.99742 7.51085 1.52263 7.51085 9.56257 1.48762 1.52263 1.48762 8.92352 0 .
It holds that
λ + ( A 12 A 12 + A 22 A 22 + A 32 A 32 ) < 8.16296 ,
and
λ + ( A 13 A 13 + A 23 A 23 + A 33 A 33 ) < 0.10286 .
Consequently, we obtain
θ 1 A 11 A 11 + A 21 A 21 + A 31 A 31 a k + 1 < 0.0921 ,
θ 2 A 12 A 12 + A 22 A 22 + A 32 A 32 a l + 1 < 0.1042 ,
and
θ 3 A 13 A 13 + A 23 A 23 + A 33 A 33 a r + 1 < 0.0207 .
Thus, the assumptions of Theorem 9 are satisfied. It remains to find the interval, where the coupled fixed point is situated. From
F 1 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 1 a 1 θ 1 A 11 A 11 + n 0 θ 2 N A 12 Q 2 θ 2 A 12 a 1 θ 3 A 13 A 13   2.57112 0.48487 0.28983 0.48487 3.47833 0.99249 0.28983 0.99249 3.90577 a 1 I ,
F 1 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) = Q 1 n 0 θ 1 N A 11 Q 1 θ 1 A 11 + a 1 θ 2 A 12 A 12 n 0 θ 3 N A 13 Q 3 θ 3 A 13   3.46108 0.50102 0.48938 0.50102 3.50547 1.00068 0.48938 1.00068 4.00030 n 0 Q 1 ,
F 2 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 2 a 1 θ 1 A 21 A 21 + n 0 θ 2 N A 22 Q 2 θ 2 A 22 a 1 θ 3 A 23 A 23   7.83296 1.07306 3.01249 1.07306 8.66008 5.16682 3.01249 5.16682 35.01603 a 1 I ,
F 2 ( n 0 Q 1 , a 1 , n 0 Q 3 ) = Q 2 n 0 θ 1 N A 21 Q 1 θ 1 A 21 + a 1 θ 2 A 22 A 22 n 0 θ 3 N A 23 Q 3 θ 3 A 23   8.78918 2.58797 4.43509 2.58797 11.61436 6.98805 4.43509 6.98805 37.83901 n 0 Q 2 ,
F 3 ( a 1 I , n 0 Q 2 , a 1 I ) = Q 3 a θ 1 A 31 A 31 + n 0 θ 2 N A 32 Q 2 θ 2 A 32 a θ 3 A 33 A 33   10.49730 7.51083 1.52266 7.51083 12.06232 1.48751 1.52266 1.48751 11.42254 a 1 I ,
and
F 3 ( n 0 Q 1 , a 1 I , n 0 Q 3 ) = Q 3 n 0 θ 1 N A 31 Q 1 θ 1 A 31 + a 1 θ 2 A 32 A 32 n 0 θ 3 N A 33 Q 3 θ 3 A 33   10.50007 7.50055 1.50091 7.50055 12.43414 1.49972 1.50091 1.49972 11.97628 n 0 Q 3 ,
it follows that
X ˜ 2.57112 0.48487 0.28983 0.48487 3.47833 0.99249 0.28983 0.99249 3.90577 , 3.46108 0.50102 0.48938 0.50102 3.50547 1.00068 0.48938 1.00068 4.00030 ,
Y ˜ 7.83296 1.07306 3.01249 1.07306 8.66008 5.16682 3.01249 5.16682 35.01603 , 8.78918 2.58797 4.43509 2.58797 11.61436 6.98805 4.43509 6.98805 37.83901 ,
and
Z ˜ 10.49730 7.51083 1.52266 7.51083 12.06232 1.48751 1.52266 1.48751 11.42254 , 10.50007 7.50055 1.50091 7.50055 12.43414 1.49972 1.50091 1.49972 11.97628 .
We calculate the first iteration by putting X 0 = a 1 I , Y 0 = a 2 I and Z 0 = a 3 I .
From δ = max k i = 1 3 A i 1 A i 1 a k + 1 , l i = 1 3 A i 2 A i 2 a l + 1 , r i = 1 3 A i 3 A i 3 a r + 1 = 0.10418 < 1 the approximate solutions with ε = 10 10 , using the a posteriori error estimate, will be
X 7 2.65564 0.48345 0.30662 0.48345 3.4719 0.98924 0.30662 0.98924 3.89897 ,
Y 7 8.52857 2.21300 4.04991 2.21300 10.95537 6.49705 4.04991 6.49705 37.11390 ,
Z 7 10.49742 7.50890 1.52009 7.50890 12.09889 1.50965 1.52009 1.50965 11.46865 ,
and the a posteriori error is 0.177145 10 11 .

5.2. Application of Theorem 10

Example 2.
Let us consider the matrix equation
X = Q A X θ A + B X θ B C X θ C
with Q = Q 3 , A = A 31 , B = A 32 , C = A 33 , a = a 1 , b = a 2 and θ = θ 1 .
We will apply Theorem 10 using similar equations, and we obtain
Q 3 a 1 θ 1 A 31 A 31 a 1 θ 1 A 33 A 33 a 1 I 0 ,
Q 3 a 1 θ 1 A 31 A 31 + a 2 θ 1 A 32 A 32 a 1 θ 1 A 33 A 33 a 1 I 0 ,
Q 3 a 1 θ 1 A 31 A 31 + a 2 θ 1 A 32 A 32 a 1 θ 1 A 33 A 33 a 2 I 0 ,
and
δ = max i = 1 , 2 , 3 θ 1 A 3 i A 3 i a 1 θ 1 + 1 0.06935 < 1 3 .
The approximate solutions with ε = 10 10 , using the a posteriori error estimate. We calculate the first iteration by using X 0 = a 1 I , Y 0 = a 2 I , and Z 0 = a 1 I . Since X, Y, and Z are solutions in the system of matrix equations
X = Q A X θ A + B Y θ B C Z θ C Y = Q A Y θ A + B Z θ B C X θ C Z = Q A Z θ A + B X θ B C Y θ C ,
we obtain
X 7 10.49758 7.50948 1.51646 7.50948 12.15627 1.49521 1.51646 1.49521 11.56359 ,
Y 7 10.49814 7.51053 1.51603 7.51053 12.16282 1.49843 1.51603 1.49843 11.59221 ,
Z 7 10.49597 7.51482 1.54115 7.51482 11.75557 1.46112 1.54115 1.46112 10.96544 .
The solution of the matrix equation X = Q A X θ A + B X θ B C X θ C is
X ˜ 10.49852 7.51085 1.51504 7.51085 12.16908 1.50299 1.51504 1.50299 11.62825
and the a posteriori error is 0.73220 10 12 .

5.3. Application of Theorem 11

Example 3.
Let us consider the system of matrix equations
X = Q 1 A 11 X θ 1 A 11 A 12 Y θ 2 A 12 A 13 Z θ 3 A 13 Y = Q 2 A 21 X θ 1 A 21 A 22 Y θ 2 B 22 A 23 Z θ 3 A 23 Z = Q 3 A 31 X θ 1 A 31 A 32 Y θ 2 A 32 A 33 Z θ 3 A 33 .
We will apply Theorem 11. It is easy to check the conditions of the theorem. The approximate solutions with ε = 10 10 , can be found using the a posteriori error estimate. For X 0 = a 1 I , Y 0 = a 2 I and Z 0 = a 3 I , after ten iterations, we obtain
X 10 2.64049 0.48059 0.30594 0.48059 3.45778 0.98833 0.30594 0.98833 3.88657 a 1 I ,
Y 10 6.37028 1.38652 0.89325 1.38652 3.64197 2.3888 0.89325 2.3888 30.80573 a 1 I ,
Z 10 10.49703 7.50889 1.52012 7.50889 12.09826 1.50972 1.52012 1.50972 11.46785 a 1 I ,
and the a posteriori error is 0.442151 10 11 .

5.4. Application of Theorem 12

Example 4.
Let us consider the matrix equation
X = Q A X θ A B X θ B C X θ C
with Q = Q 1 , A = A 31 , B = A 12 , C = A 33 and θ = θ 1 .
We will apply Theorem 12 using similar calculations and obtain
Q 1 a 1 θ 1 A 31 A 31 a 1 θ 1 A 12 A 12 a 1 θ 1 A 33 A 33 a 1 I 0 ,
and
δ = max θ 1 A 31 A 31 a 1 θ 1 + 1 , θ 1 A 12 A 12 a 1 θ 1 + 1 , θ 1 A 33 A 33 a 1 θ 1 + 1 0.06935 < 1 3 .
The approximate solutions with ε = 10 10 are found using the a posteriori error estimate. We calculate the first iteration by using X 0 = a 1 I , Y 0 = a 1 I , and Z 0 = a 1 I . Since X, Y, and Z are solutions in the system of matrix equations
X = Q A 31 X θ A 31 A 12 Y θ A 12 A 33 Z θ A 33 Y = Q A 31 Y θ A 31 A 12 Z θ A 12 A 33 X θ A 33 Z = Q A 31 Z θ A 31 A 12 X θ A 12 A 33 Y θ A 33 ,
we obtain
X 7 3.49096 0.50839 0.52003 0.50839 3.13866 1.01389 0.52003 1.01389 3.46439 ,
Y 7 3.49096 0.50839 0.52003 0.50839 3.13866 1.01389 0.52003 1.01389 3.46439 ,
Z 7 3.49096 0.50839 0.52003 0.50839 3.13866 1.01389 0.52003 1.01389 3.46439 .
The solution of the matrix Equation (31) is
X ˜ 3.49162 0.50911 0.51900 0.50911 3.14736 1.02078 0.51902 1.02078 3.51266
and the a posteriori error is 0.425568 10 11 .

6. Discussion

We would like to comment on the connection between the presented results and those known so far. Similar results for the existence and uniqueness of fixed point triples were obtained in [35], without requiring the mixed monotone property for the considered triple of maps, but of course the contraction condition (5) needs to be satisfied for all ordered triples. We will mention that the application of the results [35] and the illustrative example [30] are related to linear maps, for which if the inequality is satisfied under the requirements only for comparable triples of elements, then it is satisfied for all ordered triples. In contrast to the above, when applying results from the theory of fixed points in solving matrix equations or systems of matrix equations, the assumption that the contraction holds only for elements that are comparable is of essential importance [17]. This illustrates the deep observation and the proposed technique in [17] for the study of nonlinear matrix equations [17] and systems of nonlinear matrix equations [24].
In [42], comprehensive observation in solving matrix inequalities systems is offered. Is it possible to solve the system of matrix inequalities involved in the current study using the method from [42].
A very natural generalization can involve obtaining sufficient conditions for the existence and uniqueness of n-tupled fixed points introduced in [33], but in the context of the presented investigation, i.e., for n ordered maps ( F 1 , , F n ) , and applying it in solving matrix equations.
An interesting application in obtaining exact solutions for nonlinear systems of equations [27], where the approximate methods do not find the exact solution as shown in [27], is to change the system of equations to a best proximity points problem. Is it possible to apply the ideas from [27] in the context of matrix equations, maps with the mixed monotone property, and to obtain the exact solutions?

7. Conclusions

In the present paper, with the help of the deep results in [17,36,43,44], we have constructed a generalization of the coupled fixed points introduced in [25,26]. We have demonstrated that there is a close connection between the type of monotonicity of the maps under study and the definition of partial ordering in the proposed idea in [26] of studying coupled fixed points for maps with the mixed monotone property in partially ordered metric spaces. This allows us to solve various classes of systems of matrix equations. The presented idea of considering an ordered pair of maps instead of one map generalizes the technique proposed by [24] for solving only symmetric systems of matrix equations.

Author Contributions

Conceptualization, A.A., C.D., A.I., H.K. and B.Z.; methodology, A.A., C.D., A.I., H.K. and B.Z.; investigation, A.A., C.D., A.I., H.K. and B.Z.; writing—original draft preparation, A.A., C.D., A.I., H.K. and B.Z.; writing—review and editing, A.A., C.D., A.I., H.K. and B.Z. The listed authors have contributed equally to this research and are listed in alphabetical order. All authors have read and agreed to the published version of the manuscript.

Funding

This research was financed and supported by the European Union-NextGenerationEU, through the National Recovery and Resilience Plan of the Republic of Bulgaria, project DUECOS BG-RRP-2.004-0001-C01. The first author is partially funded by Scientific Research Grant RD-08-80/03.02.2025 of Shumen University and the second author by Scientific Research Grant RD-08-53/22.01.2025 of Shumen University.

Data Availability Statement

Data sharing is not applicable.

Acknowledgments

The authors appreciate the efforts made by the anonymous reviewers to improve the quality and presentation of their work and thank the journal for its kind invitation to submit the manuscript for consideration.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Anderson, W.N.; Morley, T.D.; Trapp, G.E. Positive Solutions to X = ABX−1B*. Linear Algebra Appl. 1990, 134, 53–62. [Google Scholar] [CrossRef]
  2. Engwerda, J.C. On the Existence of a Positive Definite Solution of the Matrix Equation X + ATX−1A = I. Linear Algebra Appl. 1993, 194, 91–108. [Google Scholar] [CrossRef]
  3. Zhan, X. Computing the Extremal Positive Definite Solution of a Matrix Equation. SIAM J. Sci. Comput. 1996, 247, 337–345. [Google Scholar] [CrossRef]
  4. Guo, C.; Lancaster, P. Iterative Solution of Two Matrix Equations. Math. Comput. 1999, 68, 1589–1603. [Google Scholar] [CrossRef]
  5. Ferrante, A.; Levy, B.C. Hermitian Solution of the Equation X = Q + NX−1N*. Linear Algebra Appl. 1996, 247, 359–373. [Google Scholar] [CrossRef]
  6. Hasanov, V.; El-Sayed, S.M. On the Positive Definite Solutions of Nonlinear Matrix Equation X + A*XδA = Q. Linear Algebra Appl. 2006, 412, 154–160. [Google Scholar] [CrossRef]
  7. Hasanov, V. Positive Definite Solutions of the Matrix Equations X ± A*XqA = Q. Linear Algebra Appl. 2005, 404, 166–182. [Google Scholar] [CrossRef]
  8. Hasanov, V.; Hakkaev, S. Newton’s Method for a Nonlinear Matrix Equation. C. R. Acad. Bulgare Sci. 2015, 68, 973–982. [Google Scholar]
  9. Nikolova, K.; Hasanov, V. On Inexact Newton Methods for Solving Two Nonlinear Matrix Equations. Ann. Acad. Rom. Sci. Ser. Math. Appl. 2024, 16, 311–327. [Google Scholar] [CrossRef]
  10. Hasanov, V. Positive Definite Solutions of a Linearly Perturbed Matrix Equation. Ann. Acad. Rom. Sci. Ser. Math. Appl. 2021, 13, 5–19. [Google Scholar] [CrossRef]
  11. Din, F.U.; Alshaikey, S.; Ishtiaq, U.; Din, M.; Sessa, S. Single and Multi-Valued Ordered-Theoretic Perov Fixed-Point Results for θ-Contraction with Application to Nonlinear System of Matrix Equations. Mathematics 2024, 12, 1302. [Google Scholar] [CrossRef]
  12. Antón-Sancho, Á. Triality and Automorphisms of Principal Bundles Moduli Spaces. Adv. Geom. 2024, 24, 421–435. [Google Scholar] [CrossRef]
  13. Antón-Sancho, Á. Spin(8, C)-Higgs Bundles and the Hitchin Integrable System. Mathematics 2024, 12, 3436. [Google Scholar] [CrossRef]
  14. Antón-Sancho, Á. Fixed Points of Involutions of G-Higgs Bundle Moduli Spaces over a Compact Riemann Surface with Classical Complex Structure Group. Front. Math. 2024, 19, 1025–1039. [Google Scholar] [CrossRef]
  15. Antón-Sancho, Á. Principal SO(2n, C)-Bundle Fixed Points Over a Compact Riemann Surface. J. Math. Phys. Anal. Geom. 2024, 20, 139–152. [Google Scholar] [CrossRef]
  16. Banach, B. Sur les opérations dan les ensembles abstraits et leurs applications aux integrales. Fundam. Math. 1922, 3, 133–181. [Google Scholar] [CrossRef]
  17. Ran, A.; Reurings, M. A Fixed Point Theorem in Partially Ordered Sets and Some Applications to Matrix Equations. Proc. Am. Math. Soc. 2004, 132, 1435–1443. [Google Scholar] [CrossRef]
  18. Turinici, M. Abstract Comparison Principles and Multivariable Gronwall-Bellman Inequalities. J. Math. Anal. Appl. 1986, 117, 100–127. [Google Scholar] [CrossRef]
  19. Dehaish, B.A.B.; Alharbi, R.K. Common Fixed Points Approximation of Two Generalized Alpha Nonexpansive Mappings in Partially Ordered Uniformly Convex Banach Space. Math. Sci. 2023, 17, 379–385. [Google Scholar] [CrossRef]
  20. Çevik, C.; Özeken, Ç.C. Coupled Fixed Point Results for New Classes of Functions on Ordered Vector Metric Space. Acta Math. Hungar. 2024, 172, 1–18. [Google Scholar] [CrossRef]
  21. Majid, N.A.; Jumaili, A.A.; Ng, Z.C.; Lee, S.K. Some Applications of Fixed Point Results for Monotone Multivalued and Integral Type Contractive Mappings. Fixed Point Theory Algorithms Sci. Eng. 2023, 2023, 11. [Google Scholar] [CrossRef]
  22. Nikam, V.; Shukla, A.K.; Gopal, D. Existence of a System of Fractional Order Differential Equations via Generalized Contraction Mapping in Partially Ordered Banach Space. Int. J. Dyn. Cont. 2024, 12, 125–135. [Google Scholar] [CrossRef]
  23. Paramasivam, E.; Sampath, S. Tripled Fixed Point Theorems in Partially Ordered ϵ-Chainable Metric Spaces for Uniformly Locally Contractive Mappings. AIP Conf. Proc. 2023, 2852, 020010. [Google Scholar] [CrossRef]
  24. Berzig, M.; Duan, X.; Samet, B. Positive Definite Solution of the Matrix Equation X = QA*X−1A + B*X−1B via Bhaskar-Lakshmikantham Fixed Point Theorem. Math. Sci. 2012, 6, 27. [Google Scholar] [CrossRef]
  25. Bhaskar, T.G.; Lakshmikantham, V. Fixed Point Theorems in Partially Ordered Metric Spaces and Applications. Nonlinear Anal. Theory Methods Appl. 2006, 65, 1379–1393. [Google Scholar] [CrossRef]
  26. Guo, D.; Lakshmikantham, V. Coupled Fixed Points of Nonlinear Operators with Applications. Nonlinear Anal. Theory Methods Appl. 1987, 11, 623–632. [Google Scholar] [CrossRef]
  27. Zlatanov, B. Coupled Best Proximity Points for Cyclic Contractive Maps and Their Applications. Fixed Point Theory 2021, 22, 431–452. [Google Scholar] [CrossRef]
  28. Erfanifar, R.; Hajarian, M. Efficient Iterative Schemes Based on Newton’s Method and Fixed-Point Iteration for Solving Nonlinear Matrix Equation Xp = Q ± A(X−1 + B)−1AT. Eng. Comput. 2023, 40, 2862–2890. [Google Scholar] [CrossRef]
  29. Sayevand, K.; Erfanifar, H.; Esmaili, H. The Maximal Positive Definite Solution of the Nonlinear Matrix Equation X + A*X−1A + B*X−1B = I. Math. Sci. 2023, 17, 337–350. [Google Scholar] [CrossRef]
  30. Berinde, V.; Borcut, M. Tripled Fixed Point Theorems for Contractive Type Mappings in Partially Ordered Metric Spaces. Nonlinear Anal. Theory Methods Appl. 2011, 74, 4889–4897. [Google Scholar] [CrossRef]
  31. Aydi, H.; Karapinar, E.; Postolache, M. Tripled Coincidence Point Theorems for Weak ϕ-Contractions in Partially Ordered Metric Spaces. Fixed Point Theory Appl. 2012, 2012, 44. [Google Scholar] [CrossRef]
  32. Borcut, M.; Pscurar, M.; Berinde, V. Tripled Fixed Point Theorems for Mixed Monotone Kannan Type Contractive Mappings. Appl. Math. Comput. 2014, 2014, 120203. [Google Scholar] [CrossRef]
  33. Samet, B.; Vetro, C. Coupled Fixed Point, f-Invariant Set and Fixed Point of n-Order. Ann. Funct. Anal. 2010, 1, 46–56. [Google Scholar] [CrossRef]
  34. Dzhabarova, Y.; Zlatanov, B. A Note on the Market Equilibrium in Oligopoly with Three Industrial Players. AIP Conf. Proc. 2022, 2449, 070013. [Google Scholar] [CrossRef]
  35. Ilchev, A.; Ivanova, V.; Kulina, H.; Yaneva, P.; Zlatanov, B. Investigation of Equilibrium in Oligopoly Markets with the Help of Tripled Fixed Points in Banach Spaces. Econometrics 2024, 12, 18. [Google Scholar] [CrossRef]
  36. Ali, A.; Dinkova, C.; Ilchev, A.; Zlatanov, B. Bhaskar-Lakshmikantham Fixed Point Theorem vs. Ran-Reunrings One and Some Possible Generalizations and Applications in Matrix Equations. AIMS Math. 2024, 9, 21890–21917. [Google Scholar] [CrossRef]
  37. Simovici, D.A.; Djeraba, C. Mathematical Tools for Data Mining: Set Theory, Partial Orders, Combinatorics; Springer: London, UK, 2008. [Google Scholar] [CrossRef]
  38. Ajeti, L.; Ilchev, A. A Variational Principle and Triple Fixed Points. AIP Conf. Proc. 2025. in print. [Google Scholar]
  39. Nieto, J.J.; Rodríguez-López, R. Contractive Mapping Theorems in Partially Ordered Sets and Applications to Ordinary Differential Equations. Order 2005, 22, 223–239. [Google Scholar] [CrossRef]
  40. Bhatia, R. Matrix Analysis. Graduate Texts in Mathematics; Springer: New York, NY, USA, 1997. [Google Scholar] [CrossRef]
  41. Zeidler, E. Nonlinear Functional Analysis and Its Applications, I—Fixed-Point Theorems; Springer: New York, NY, USA, 1986. [Google Scholar]
  42. Hasanov, V. On the Matrix Equation X + A*X−1AB*X−1B = I. Linear Multilinear Algebra 2018, 66, 1783–1798. [Google Scholar] [CrossRef]
  43. Petruşel, A. Fixed Points vs. Coupled Fixed Points. J. Fixed Point Theory Appl. 2018, 20, 150. [Google Scholar] [CrossRef]
  44. Petruşel, A.; Petruşel, G.; Xiao, Y.-B.; Yao, J.-C. Fixed Point Theorems for Generalized Contractions with Applications to Coupled Fixed Point Theory. J. Nonlinear Convex Anal. 2018, 19, 71–81. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ali, A.; Dinkova, C.; Ilchev, A.; Kulina, H.; Zlatanov, B. Tripled Fixed Points, Obtained by Ran-Reunrings Theorem for Monotone Maps in Partially Ordered Metric Spaces. Mathematics 2025, 13, 760. https://doi.org/10.3390/math13050760

AMA Style

Ali A, Dinkova C, Ilchev A, Kulina H, Zlatanov B. Tripled Fixed Points, Obtained by Ran-Reunrings Theorem for Monotone Maps in Partially Ordered Metric Spaces. Mathematics. 2025; 13(5):760. https://doi.org/10.3390/math13050760

Chicago/Turabian Style

Ali, Aynur, Cvetelina Dinkova, Atanas Ilchev, Hristina Kulina, and Boyan Zlatanov. 2025. "Tripled Fixed Points, Obtained by Ran-Reunrings Theorem for Monotone Maps in Partially Ordered Metric Spaces" Mathematics 13, no. 5: 760. https://doi.org/10.3390/math13050760

APA Style

Ali, A., Dinkova, C., Ilchev, A., Kulina, H., & Zlatanov, B. (2025). Tripled Fixed Points, Obtained by Ran-Reunrings Theorem for Monotone Maps in Partially Ordered Metric Spaces. Mathematics, 13(5), 760. https://doi.org/10.3390/math13050760

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop