Next Article in Journal
A Microsimulation Model for Sustainability and Detailed Adequacy Analysis of the Retirement Pension System
Next Article in Special Issue
Neural Networks as Positive Linear Operators
Previous Article in Journal
Sustainability, Accuracy, Fairness, and Explainability (SAFE) Machine Learning in Quantitative Trading
Previous Article in Special Issue
A Subgradient Extragradient Framework Incorporating a Relaxation and Dual Inertial Technique for Variational Inequalities
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Triple Mann Iteration Method for Variational Inclusions, Equilibria, and Common Fixed Points of Finitely Many Quasi-Nonexpansive Mappings on Hadamard Manifolds

1
Department of Mathematics, Shanghai Normal University, Shanghai 200234, China
2
Research Center for Interneural Computing, China Medical University Hospital, China Medical University, Taichung 404327, Taiwan
3
Academy of Romanian Scientists, 50044 Bucharest, Romania
*
Author to whom correspondence should be addressed.
Mathematics 2025, 13(3), 444; https://doi.org/10.3390/math13030444
Submission received: 14 January 2025 / Revised: 23 January 2025 / Accepted: 25 January 2025 / Published: 28 January 2025

Abstract

:
In this paper, we introduce a triple Mann iteration method for approximating an element in the set of common solutions of a system of quasivariational inclusion issues, which is an equilibrium problem and a common fixed point problem (CFPP) of finitely many quasi-nonexpansive operators on a Hadamard manifold. Through some suitable assumptions, we prove that the sequence constructed in the suggested algorithm is convergent to an element in the set of common solutions. Finally, making use of the main result, we deal with the minimizing problem with a CFPP constraint and saddle point problem with a CFPP constraint on a Hadamard manifold, respectively.

1. Introduction

It is well known that the theory of variational inequalities, introduced by Hartman and Stampacchia [1] in the early 1960s, has played an important role in the study of physics, engineering, and economics, and it encompasses as special cases complementarity problems, optimization problems, the problem of finding zeros of operators, etc. Within the period of the past 20 years, a great deal of effort has gone into the existence and algorithms for variational inequality problems, related optimization problems, and related fixed-point problems; see, e.g., [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18] and the references therein.
Let M be a Hadamard manifold and C be a nonempty subset of M . In 2012, Calao et al. [19] first studied the following equilibrium problem of finding y * C such that
Φ ( y * , x ) 0
for all x C , in which the function Φ is of equilibrium. It was proven in [19] that there holds the existence of solutions to the issue (1) for Φ . Moreover, the outcomes were employed to settle mixed variational inequalities, fixed point issues, and Nash equilibrium issues in M . Using Picard’s iterative method, the authors introduced an iterative algorithm for finding a solution of issue (1). Subsequently, it was pointed out in [20] that there hold certain gaps involving the existence argument for mixed variational inequality problems and the domain for the resolvent implicating the equilibrium issue of [19].
Recently, the authors [21,22] investigated the problem of finding an element x * Fix ( T ) ( A + B ) 1 0 , with T being the nonexpansive operator, B being the maximal monotone vector field, and A being the continuous monotone vector field in a Hadamard manifold M . They proposed some Halpern-type and Mann-type iterative methods. Under some mild conditions, they proved that the iterative sequences fabricated in the suggested methods are strongly convergent to an element in the set of common solutions of the fixed-point problem of T and the variational inclusion problem for A and B.
Assume C M , where the bounded C is of both closedness and geodesic convexity in M ; let T r Φ : M C be the resolvent of equilibrium function Φ for r > 0 , and let exp x 1 be the inverse of the exponential map exp x : T x M M at x M . Very recently, Chang et al. [23] considered the problem of finding an element
x * Ω : = Fix ( S ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) ,
in which Fix ( S ) is the fixed point set of the quasi-nonexpansive operator S : C C , EP ( Φ ) is the equilibrium point set of Φ , and B : C 2 T M is the maximal monotone vector field; for i = 1 , , N , each continuous A i : C T M is a monotone vector field, and m = 1 M ( A m + B ) 1 0 is the common singularity set of a quasivariational inclusion system. They proposed the following splitting iterative algorithm, that is, for any initial x 0 C , the sequence { x n } is defined as follows:
u n i = J λ B exp x n ( λ A i x n ) , i = 1 , 2 , , M , y n = S u n i n with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) , x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) n 0 ,
where { α n } ( 0 , 1 ) , α n 1 , n = 0 α n = , and J λ B exp I ( λ A i ) : C M is the mapping defined by B , A i and λ > 0 for i = 1 , 2 , , M . Under some mild assumptions, it was proven that the { x n } constructed in (3) is strongly convergent to an element of Ω .
Inspired and aroused by the works above, this article considers the issue of seeking an element
y * Ω : = ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) ,
in Hadamard manifold M , where i = 1 N Fix ( S i ) is the set of common fixed points of finitely many quasi-nonexpansive mappings { S i } i = 1 N , and Φ , B , A i , i = 1 , 2 , , M are the same as above. We propose a triple Mann iteration method and show that the sequence fabricated in suggested algorithm is convergent to an element in the set of common solutions for issue (4). Finally, making use of the main result, we deal with the minimizing issue with a CFPP constraint and saddle-point issue with a CFPP constraint on Hadamard manifolds, respectively. Our main result improves, extends, and develops Chang et al. [23], Theorem 3.1, in some aspects.

2. Basic Notions and Tools

To address the problem (4), one first releases certain preliminaries on geometric theory of manifolds, including some concepts and basic tools. It is worth mentioning that these can be found in many introductory books on Riemannian and differential geometry (see, e.g., [24]).
Assume M is a differentiable manifold of finite dimension. Then, the tangent space of M at x M is denoted as T x M . The tangent bundle of M is formulated as T M = x M T x M . Naturally, it is a manifold. The inner product R x ( · , · ) in T x M is termed as a Riemannian metric in T x M . The tensor field R ( · , · ) is termed as a Riemannian metric on M if and only if, for each x M , the tensor R x ( · , · ) is a Riemannian metric on T x M . The associated norm with R x ( · , · ) on T x M is written as · x . The differentiable manifold M equipped with Riemannian metric R ( · , · ) is termed as Riemannian manifold. Let the curve γ : [ a , b ] M joining u to v (i.e., γ ( a ) = u and γ ( b ) = v ) be piecewise smooth. Then, the length of γ is defined as l ( γ ) = a b γ ( t ) d t . Moreover, the Riemannian distance d ( v , y ) , inducing the original topology on M , is formulated as minimizing this length over the set of all such curves joining v to y.
The Riemannian manifold M is of completeness if and only if, for each y M , any geodesic emanating from y is definable with respect to (w.r.t.) t in R : = ( , ) . The geodesic joining v to y in M is termed as the minimal geodesic if and only if the length of it equals to d ( v , y ) . Riemannian manifold M endowed by Riemannian distance d is the metric space ( M , d ) . From the Hopf–Rinow result [24], it is easily known that, in case M is of completeness, each pair of elements in M is joined with the minimal geodesic. Meanwhile, the metric space ( M , d ) is of completeness, and each closed bounded subset is of compactness.
Suppose that Riemannian manifold M is of completeness. The map exp y : T y M M at y is the exponential one formulated as exp y v = γ v ( 1 , y ) for all v T y M , in which γ ( · ) = γ v ( · , y ) denotes the geodesic initiating from y with velocity v, that is, γ v ( 0 , y ) = y and γ v ( 0 , y ) = v . Then, exp y t v = γ v ( t , y ) for any t R . It is clear to check exp y 0 = γ v ( 0 , y ) = y , in which 0 denotes the zero tangent vector. It is noteworthy that map exp y is the differentiable exponential one on T y M w.r.t. y M .
A complete simply connected Riemannian manifold of nonpositive sectional curvature is called a Hadamard Manifold. In the rest of this paper, we always assume that M is a Hadamard Manifold of finite dimension without causing confusion.
Proposition 1 
(see [24]). For any two points x , y M , there exists a unique normalized geodesic γ : [ 0 , 1 ] M joining x = γ ( 0 ) to y = γ ( 1 ) , which is actually a minimal geodesic denoted by
γ ( t ) = exp x t exp x 1 y t [ 0 , 1 ] .
Moreover, for each { x n } in M satisfying x n x 0 in M , there hold the following:
exp x n 1 y exp x 0 1 y and exp y 1 x n exp y 1 x 0 y M .
It is easy to check that the following lemma is valid.
Lemma 1. 
(i) If γ : [ 0 , 1 ] M is a geodesic joining x to y, then we have
d ( γ ( t 1 ) , γ ( t 2 ) ) = | t 1 t 2 | d ( x , y ) t 1 , t 2 [ 0 , 1 ] .
(From now on, d ( x , y ) indicates the Riemannian distance).
(ii) For each y , z , u , v , w M and t [ 0 , 1 ] , there hold the inequalities below:
d ( exp y t exp y 1 v , z ) ( 1 t ) d ( y , z ) + t d ( v , z ) ;
d 2 ( exp y t exp y 1 v , z ) ( 1 t ) d 2 ( y , z ) + t d 2 ( v , z ) t ( 1 t ) d 2 ( y , v ) ;
d ( exp y t exp y 1 v , exp u t exp u 1 w ) ( 1 t ) d ( y , u ) + t d ( v , w ) .
The set C in M is of geodesic convexity if and only if, for each y , v C , the geodesic joining y to v belongs to C. The geodesic convex hull of set D M is the smallest geodesic convex set containing D in M , and it is written as co ( D ) .
In what follows, unless otherwise specified, we always assume that C M , where C is of both closedness and geodesic convexity in Hadamard manifold M , and Fix ( S ) denotes the set of fixed points of an operator S.
The function h : C R { + } = ( , ] is termed as being of geodesic convexity if and only if, for each geodesic γ ( t ) ( t [ 0 , 1 ] ) joining y , v C , the h γ is of convexity, that is,
h ( γ ( t ) ) t h ( γ ( 0 ) ) + ( 1 t ) h ( γ ( 1 ) ) = t h ( y ) + ( 1 t ) h ( v ) .
Assume the metric space X is complete, and let D X . The sequence { x n } in X is of Fejér monotonicity w.r.t. D if and only if, for each v D and each non-negative n, one has d ( x n , v ) d ( x n + 1 , v ) .
Lemma 2 
(see [25,26]). Assume the metric space X is complete, and let D X . In the case that { y n } X is of Fejér monotonicity w.r.t. D, { y n } is of boundedness. In addition, in the case that { y n } has a cluster point y lying in D, { y n } is convergent to y.
An operator S from C to itself is termed as being the following:
(a) Contractive if and only if ( 0 , 1 ) such that
d ( S y , S v ) d ( y , v ) y , v C
(in particular, in the case of = 1 , S is termed as being nonexpansive);
(b) Quasi-nonexpansive if and only if Fix ( S ) is not equal to ∅, and
d ( S y , v ) d ( y , v ) y C , v Fix ( S ) ;
(c) Firmly nonexpansive [27] if and only if, for each y , v in C, the map ψ : [ 0 , 1 ] [ 0 , ] formulated by
ψ ( t ) : = d ( exp y t exp y 1 S y , exp v t exp v 1 S v ) t [ 0 , 1 ]
is nonincreasing.
Proposition 2 
(see [27]). Let S : C C be a mapping. Then, the following statements are equivalent:
(i) S is firmly nonexpansive;
(ii) For all x , y C and t [ 0 , 1 ] ,
d ( S x , S y ) d ( exp x t exp x 1 S x , exp y t exp y 1 S y ) ;
(iii) for all x , y C ,
R ( exp S x 1 S y , exp S x 1 x ) + R ( exp S y 1 S x , exp S y 1 y ) 0 .
Lemma 3 
(see [23]). Let S : C C be firmly nonexpansive, and assume Fix ( S ) is not equal to . Then, for each y C and v Fix ( S ) , there holds the inequality below:
d 2 ( S y , v ) d 2 ( y , v ) d 2 ( S y , y ) .
It is remarkable that, when Fix ( S ) , by Lemma 3, one obtains the relations below:
The firm nonexpansivity of S the nonexpansivity of S the quasinonexpansivity of S.
However, the converse relations do not hold.
An operator S from C to itself is termed as being demiclosed at zero if and only if, for each { y n } C satisfying y n y * C and d ( y n , S y n ) 0 , one has y * Fix ( S ) .
In what follows, one uses Ω ( M ) to denote the collection of all single-valued vector fields A : M T M such that A y T y M y M , and D ( A ) the domain of A defined by
D ( A ) = { y M : A y T y M } .
Also, one uses X ( M ) to indicate the collection of all set-valued vector fields B : M 2 T M such that B y T y M y M , and D ( B ) is the domain of B formulated by D ( B ) = { y M : B y } .
Next, one furnishes some vital notions (see [28]) that will be helpful to demonstrate the major outcome:
(i) A single-valued vector field A Ω ( M ) is termed as being monotone if and only if
R ( A y , exp y 1 v ) R ( A v , exp v 1 y ) y , v M .
(ii) A set-valued vector field B X ( M ) is termed as being the following:
(a) Monotone if and only if, for each y , z D ( B )
R ( v , exp y 1 z ) R ( w , exp z 1 y ) v B y , w B z ;
(b) Maximal monotone if and only if it is monotone and, for the given y D ( B ) and v T y M , the relation
R ( v , exp y 1 z ) R ( w , exp z 1 y ) z D ( B ) , w B z
implies v B y .
(iii) For given λ > 0 , the resolvent of B for λ > 0 is a set-valued mapping J λ B : M 2 T M formulated by
J λ B ( y ) : = { x M : y exp x λ B x } y M .
Lemma 4 
(see [21]). Given a single-valued vector field A Ω ( M ) of monotonicity and a set-valued vector field B X ( M ) of maximal monotonicity, one then has that for each y M , there holds the equivalence of the assertions below:
(i) y ( A + B ) 1 0 ;
(ii) y = J λ B ( exp y ( λ A y ) ) λ > 0 .
In addition, we recall the resolvent of a bifunction on Hadamard manifold M introduced in Calao et al. [19], as outlined below:
Assume C M , where C is of both closedness and geodesic convexity in M . Given a bifunction Φ : C × C R , the resolvent of Φ is set-valued mapping T r Φ : M 2 C such that
T r Φ ( y ) = { x C : Φ ( x , z ) 1 r exp x 1 y , exp x 1 z 0 z C } y M .
Lemma 5 
(see [19,20]). Let Φ : C × C R be a bifunction satisfying the following conditions:
(A1) Φ ( x , x ) = 0 x C ;
(A2) Φ ( x , y ) + Φ ( y , x ) 0 x , y C , i.e., Φ is monotone;
(A3) x Φ ( x , y ) is upper semicontinuous for each y C ;
(A4) y Φ ( x , y ) is geodesic convex and lower semicontinuous for each x C ;
(A5)(compact set) D in M such that
y C D z C D s . t . Φ ( y , z ) < 0 .
One then has that for each r > 0 , the assertions listed below hold:
(i) The resolvent T r Φ is nonempty and single-valued;
(ii) The resolvent T r Φ is firmly nonexpansive;
(iii) Fix ( T r Φ ) = EP ( Φ ) , i.e., the fixed point set of T r Φ is the equilibrium point set of Φ;
(iv) EP ( Φ ) is of both closedness and geodesic convexity.

3. Major Outcomes

Let M be a Hadamard manifold of finite dimension and the nonempty C M be of both closedness and geodesic convexity in M . Assume throughout that the following conditions hold:
  • B : C 2 T M is a multi-valued vector field of maximal monotonicity, A i : C T M is a single-valued vector field of both continuity and monotonicity for i = 1 , 2 , , M , and J λ B ( exp I ( λ A i ) ) : C M is the mapping defined by B , A i and λ > 0 for i = 1 , 2 , , M ;
  • Φ : C × C ( , ) is the bifunction fulfilling the hypotheses (A1)–(A5) of Lemma 5 and T r Φ : M 2 C is the resolvent of Φ for r > 0 and D ( T r Φ ) is of both closedness and geodesic convexity;
  • { S i } i = 1 N is a finite family of L-Lipschitzian and demiclosed at zero quasi-nonexpansive self-mappings on C, where L 1 ;
  • Ω : = ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) ;
  • For given λ > 0 and ρ ( 0 , 1 ) , the inequality holds
    d ( exp x ( λ A i x ) , exp y ( λ A i y ) ) ( 1 ρ ) d ( x , y ) x , y C , i { 1 , 2 , , M } ;
  • { α n } ( 0 , 1 ) and { β n } , { γ n } ( 0 , 1 ] such that the following are true:
    (i)
    lim n α n = 1 , and n = 1 ( 1 α n ) = ;
    (ii)
    0 < lim inf n β n , and 0 < lim inf n γ n .
In terms of Xu and Kim [18], we write S n : = S n mod N for integer n 1 with the mod function taking values in the set { 1 , 2 , , N } , that is, if n = j N + q for some integers j 0 and 0 q < N , then S n = S N if q = 0 and S n = S q if 0 < q < N .
Theorem 1. 
Let M , C , { A i } i = 1 M , B , { J λ B exp I ( λ A i ) } i = 1 M , Φ , T r Φ , and { S i } i = 1 N be chosen as in the above conditions. Given a starting x 0 C arbitrarily. Let { x n } , { u n i } i = 1 M and { y n } be the sequences defined as follows:
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) , i = 1 , 2 , , M , y n = exp x n β n exp x n 1 ( S n u n i n ) with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) , x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) n 0 .
Then, we have the following: (i) The sequence { x n } converges to x * Ω a solution of problem (4);
(ii) In addition, if β n = γ n = 1 n 0 , and S i = S , i = 1 , 2 , , N , then the sequence { x n } defined by
u n i = J λ B exp x n ( λ A i x n ) , i = 1 , 2 , , M , y n = S u n i n with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) , x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) n 0 ,
converges to a solution of problem (2);
(iii) In addition, if β n = γ n = 1 n 0 , and A j = B = 0 (the null operator), j = 1 , 2 , , M , then the sequence { x n } defined by
x n + 1 = exp x n α n exp x n 1 ( T r Φ S n x n ) n 0 ,
converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) ;
(iv) In addition, if for Φ = 0 , β n = γ n = 1 n 0 , and S i = I i { 1 , 2 , , N } , { x n } is the sequence constructed as
x n + 1 = exp x n α n exp x n 1 ( J λ B exp x n ( λ A i n x n ) ) n 0 ,
then { x n } converges to an element x * i = 1 M ( A i + B ) 1 0 ;
(v) In addition, if for Φ = 0 , β n = γ n = 1 n 0 , and A j = B = 0 j { 1 , 2 , , M } , { x n } is the sequence constructed as
x n + 1 = exp x n α n exp x n 1 ( S n x n ) n 0 ,
then { x n } converges to x * i = 1 N Fix ( S i ) a common fixed point of { S i } i = 1 N ;
(vi) In addition, if β n = γ n = 1 n 0 , and A i = A , i = 1 , 2 , , M , then the sequence { x n } defined by
x n + 1 = exp x n α n exp x n 1 ( T r Φ ( S n ( J λ B exp x n ( λ A x n ) ) ) ) n 0 ,
converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( A + B ) 1 0 ;
(vii) In addition, if β n = γ n = 1 n 0 , and A i = 0 , i = 1 , 2 , , M , then the sequence { x n } defined by
x n + 1 = exp x n α n exp x n 1 ( T r Φ ( S n ( J λ B x n ) ) ) n 0 ,
converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) B 1 0 .
Proof. 
First of all, using the conditions in Theorem 1 and Lemmas 4 and 5, one knows the following:
(i) The resolvent T γ Φ of function Φ is the single-valued mapping of firm nonexpansivity such that Fix ( T γ Φ ) = EP ( Φ ) γ ( 0 , ) ;
(ii) If z Ω , then one has that λ > 0 and γ > 0 ,
z ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) = ( i = 1 N Fix ( S i ) ) Fix ( T γ Φ ) ( i = 1 M Fix ( J λ B exp I ( λ A i ) ) ) .
We now divide the rest of the proof into several steps.
Step 1. We claim that the sequences { x n } , { y n } , { u n i } , i = 1 , 2 , , M , { S n u n i n } and { T r Φ y n } ( r > 0 ) all are bounded in C.
Indeed, by condition (17), for each i = 1 , 2 , , M and λ > 0 , J λ B exp I ( λ A i ) is a mapping of quasi-nonexpansivity. Take a fixed p Ω arbitrarily. Then, for i = 1 , 2 , , M , by (17) and Lemma 1 (ii), we have
d ( u n i , p ) = d ( exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) , exp p γ n exp p 1 ( J λ B exp I ( λ A i ) p ) ) ( 1 γ n ) d ( x n , p ) + γ n d ( J λ B exp x n ( λ A i x n ) , J λ B exp p ( λ A i p ) ) ( 1 γ n ) d ( x n , p ) + γ n ( 1 ρ ) d ( x n , p ) = ( 1 γ n ρ ) d ( x n , p ) d ( x n , p ) .
Since S i is quasi-nonexpansive for i = 1 , 2 , , N , from Lemma 1 (ii), (18), and (20) we obtain
d ( y n , p ) = d ( exp x n β n exp x n 1 ( S n u n i n ) , exp p β n exp p 1 ( S n p ) ) ( 1 β n ) d ( x n , p ) + β n d ( S n u n i n , S n p ) ( 1 β n ) d ( x n , p ) + β n d ( u n i n , p ) ( 1 β n ) d ( x n , p ) + β n d ( x n , p ) = d ( x n , p ) .
Thus, from Lemma 1 (ii), Lemma 5, (18), and (21), we have
d 2 ( x n + 1 , p ) = d 2 ( exp x n α n exp x n 1 ( T r Φ y n ) , exp p α n exp p 1 ( T r Φ p ) ) ( 1 α n ) d 2 ( x n , p ) + α n d 2 ( T r Φ y n , T r Φ p ) α n ( 1 α n ) d 2 ( x n , T r Φ y n ) ( 1 α n ) d 2 ( x n , p ) + α n d 2 ( y n , p ) α n ( 1 α n ) d 2 ( x n , T r Φ y n ) ( 1 α n ) d 2 ( x n , p ) + α n d 2 ( x n , p ) α n ( 1 α n ) d 2 ( x n , T r Φ y n ) = d 2 ( x n , p ) α n ( 1 α n ) d 2 ( x n , T r Φ y n ) .
This implies that, for all n 0 and p Ω ,
d ( x n + 1 , p ) d ( x n , p ) ,
that is, { x n } is of Fejér monotonicity w.r.t. Ω . Via Lemma 2, { x n } is of boundedness and so are the sequences { u n j } , j = 1 , 2 , , M , { y n } , { T r Φ y n } . As a result, the limit lim n d ( x n , p ) exists for each p Ω .
Step 2. We claim that for i = 1 , 2 , , M ,
d ( x n , u n i ) 0 , d ( x n , y n ) 0 and d ( x n + 1 , x n ) 0 as n .
Indeed, from (20), (21), and (23) we obtain that
d ( x n , u n i ) d ( x n , p ) + d ( u n i , p ) 2 d ( x n , p ) for i = 1 , 2 , , M , d ( x n , y n ) d ( x n , p ) + d ( y n , p ) 2 d ( x n , p ) , d ( x n + 1 , x n ) d ( x n , p ) + d ( x n + 1 , p ) 2 d ( x n , p ) .
Taking p Ω and setting ξ = sup n 0 d ( x n , p ) , by Lemma 1 (ii), we deduce that, for each i = 1 , 2 , , M ,
d ( x n , u n i ) 2 d ( x n , p ) = 2 d ( exp x n 1 α n 1 exp x n 1 1 ( T r Φ y n 1 ) , exp p α n 1 exp p 1 ( T r Φ p ) ) 2 { ( 1 α n 1 ) d ( x n 1 , p ) + α n 1 d ( T r Φ y n 1 , T r Φ p ) } 2 { ( 1 α n 1 ) d ( x n 1 , p ) + α n 1 d ( y n 1 , p ) } 2 { ( 1 α n 1 ) ξ + α n 1 d ( x n 1 , p ) } .
By induction, one deduces that, for 1 m n 1 ,
d ( x n , u n i ) 2 ξ j = m n 1 { ( 1 α j ) i = j + 1 n 1 α i } + 2 ξ j = m n 1 α j , i = 1 , 2 , , M .
Taking the limit in (26) as n , we obtain
lim n d ( x n , u n i ) 2 ξ j = m { ( 1 α j ) i = j + 1 α i } + 2 ξ j = m α j .
Since { α n } ( 0 , 1 ) , α n 1 ( n ) and n = 0 ( 1 α n ) = , we obtain that
lim m j = m { ( 1 α j ) i = j + 1 α i } = 0 , lim m j = m α j = 0 .
Consequently,
lim n d ( x n , u n i ) = 0 , i = 1 , 2 , , M .
In a similar way, from (25), we can also show that
d ( x n , y n ) 0 and d ( x n , x n + 1 ) 0 as n .
Thus, the assertions in (24) are valid.
Step 3. We claim that, for i = 1 , 2 , , M ,
d ( x n , J λ B exp I ( λ A i ) x n ) 0 , d ( x n , S n x n ) 0 and d ( x n , T r Φ x n ) 0 as n .
Indeed, by (18) and Lemma 1 (ii), we obatain
d ( x n , J λ B exp I ( λ A i ) x n ) d ( x n , u n i ) + d ( u n i , J λ B exp I ( λ A i ) x n ) = d ( x n , u n i ) + d ( exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) , J λ B exp I ( λ A i ) x n ) d ( x n , u n i ) + ( 1 γ n ) d ( x n , J λ B exp I ( λ A i ) x n ) ,
which hence yields
d ( x n , J λ B exp I ( λ A i ) x n ) 1 γ n d ( x n , u n i ) .
Noticing 0 < lim inf n γ n , from (24) we have
lim n d ( x n , J λ B exp I ( λ A i ) x n ) = 0 , i = 1 , 2 , , M .
Since each S n is L-Lipschitzian, using Lemma 1 (ii), one has
d ( S n x n , x n ) d ( S n x n , y n ) + d ( y n , x n ) = d ( y n , x n ) + d ( exp x n β n exp x n 1 ( S n u n i n ) , S n x n ) d ( x n , y n ) + ( 1 β n ) d ( x n , S n x n ) + β n d ( S n u n i n , S n x n ) d ( x n , y n ) + ( 1 β n ) d ( x n , S n x n ) + β n L d ( u n i n , x n ) ,
which hence leads to
d ( x n , S n x n ) 1 β n [ d ( x n , y n ) + β n L d ( u n i n , x n ) ] 1 β n [ d ( x n , y n ) + L d ( u n i n , x n ) ] .
Noticing 0 < lim inf n β n , from (24), we have
lim n d ( x n , S n x n ) = 0 .
Also, since T r Φ is firmly nonexpansive, by Lemma 1 (ii), we obatain
d ( x n , T r Φ x n ) d ( x n , x n + 1 ) + d ( x n + 1 , T r Φ x n ) = d ( x n , x n + 1 ) + d ( exp x n α n exp x n 1 ( T r Φ y n ) , T r Φ x n ) d ( x n , x n + 1 ) + ( 1 α n ) d ( x n , T r Φ x n ) + α n d ( T r Φ y n , T r Φ x n ) d ( x n , x n + 1 ) + ( 1 α n ) d ( x n , T r Φ x n ) + α n d ( y n , x n ) ,
which immediately implies that
d ( x n , T r Φ x n ) 1 α n [ d ( x n , x n + 1 ) + α n d ( y n , x n ) ] 1 α n [ d ( x n , x n + 1 ) + d ( y n , x n ) ] .
Noticing lim n α n = 1 , from (24), we obtain
lim n d ( x n , T r Φ x n ) = 0 .
Step 4. We claim that { x n } converges to some point x * Ω .
Indeed, since C is closed and geodesic convex in finite dimensional M , and { x n } is of boundedness in C, one has that { x n k } { x n } such that { x n k } is convergent to an element x * in C. We first claim x * i = 1 N Fix ( S i ) . Without loss of generality, we may assume l = n k mod N for all k. Since combining d ( x n , x n + 1 ) 0 (due to (24)) and x n k x * guarantees x n k + j x * for all j 1 , we deduce from (28) that d ( x n k + j , T l + j x n k + j ) = d ( x n k + j , T n k + j x n k + j ) 0 . Then, the demiclosedness assumption of S i , i = 1 , 2 , , N implies that x * Fix ( T l + j ) for all j. This ensures that x * i = 1 N Fix ( S i ) . Again, note that J λ B exp I ( λ A i ) , i = 1 , 2 , , M and T r Φ are all nonexpansive mappings, which are also demiclosed at zero. So, it follows from (27) and (29) that x * ( i = 1 M Fix ( J λ B exp I ( λ A i ) ) ) Fix ( T r Φ ) . Since Fix ( J λ B exp I ( λ A i ) ) = ( A i + B ) 1 0 for i = 1 , 2 , , M and Fix ( T r Φ ) = EP ( Φ ) , we have x * ( i = 1 M Fix ( J λ B exp I ( λ A i ) ) ) Fix ( T r Φ ) = ( i = 1 M ( A i + B ) 1 0 ) EP ( Φ ) . Consequently, x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) = Ω . Since { x n } is Fejér monotone with respect to Ω , in terms of Lemma 2, we obtain lim n x n = x * . The conclusion (i) of Theorem 1 is proved.
Step 5. We claim that there hold the conclusions (ii), (iii), (iv), (v), (vi), and (vii) in Theorem 1:
(1) Indeed, (a) if γ n = 1 n 0 , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A i x n ) ;
(b) If β n = 1 n 0 and S i = S , i = 1 , 2 , , N , then we obtain
y n = exp x n β n exp x n 1 ( S n u n i n ) = S u n i n with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) .
Therefore, (18) reduces to the following iterative algorithm:
u n i = J λ B exp x n ( λ A i x n ) , i = 1 , 2 , , M , y n = S u n i n with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) , x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) n 0 ,
which converges to x * Fix ( S ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) as a solution of problem (2). The conclusion (ii) of Theorem 1 is proved.
(2) Indeed, (a) if γ n = 1 n 0 and A i = B = 0 , i = 1 , 2 , , M , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A i x n ) = x n ;
(b) If β n = 1 n 0 , we obtain
y n = exp x n β n exp x n 1 ( S n u n i n ) = S n u n i n = S n x n .
Therefore, (18) reduces to the following iterative algorithm:
x n + 1 = exp x n α n exp x n 1 ( T r Φ S n x n ) n 0 ,
which converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) . The conclusion (iii) is proved.
(3) Indeed, (a) if β n = γ n = 1 n 0 and S i = I , i = 1 , 2 , , N , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A i x n ) ,
and hence
y n = exp x n β n exp x n 1 ( S n u n i n ) = u n i n = J λ B exp x n ( λ A i n x n ) ;
(b) If Φ = 0 , then we have T r Φ = I r > 0 , and hence
x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) = exp x n α n exp x n 1 y n .
Therefore, (18) reduces to the following iterative algorithm:
x n + 1 = exp x n α n exp x n 1 ( J λ B exp x n ( λ A i n x n ) ) n 0 ,
which converges to an element x * i = 1 M ( A i + B ) 1 0 . The conclusion (iv) is proved.
(4) Indeed, if β n = γ n = 1 n 0 and A i = B = 0 , i = 1 , 2 , , M , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A i x n ) = x n ,
and hence
y n = exp x n β n exp x n 1 ( S n u n i n ) = S n u n i n = S n x n ;
(b) If Φ = 0 , then we have T r Φ = I r > 0 , and hence
x n + 1 = exp x n α n exp x n 1 ( T r Φ y n ) = exp x n α n exp x n 1 y n .
Therefore, (18) reduces to the following iterative algorithm:
x n + 1 = exp x n α n exp x n 1 ( S n x n ) n 0 ,
which converges to x * i = 1 N Fix ( S i ) a common fixed point of { S i } i = 1 N . The conclusion (v) is proved.
(5) Indeed, (a) if γ n = 1 n 0 , and A i = A , i = 1 , 2 , , M , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A x n ) ;
(b) If β n = 1 n 0 , then we obtain
y n = exp x n β n exp x n 1 ( S n u n i n ) = S n u n i n = S n ( J λ B exp x n ( λ A x n ) ) .
Therefore, (18) reduces to the following iterative algorithm:
x n + 1 = exp x n α n exp x n 1 ( T r Φ ( S n ( J λ B exp x n ( λ A x n ) ) ) ) n 0 ,
which converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( A + B ) 1 0 . The conclusion (vi) is proved.
(6) Indeed, (a) if γ n = 1 n 0 , and A i = 0 , i = 1 , 2 , , M , then we obtain that, for i = 1 , 2 , , M and λ > 0 ,
u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) = J λ B exp x n ( λ A i x n ) = J λ B x n ;
(b) If β n = 1 n 0 , then we obtain
y n = exp x n β n exp x n 1 ( S n u n i n ) = S n u n i n = S n ( J λ B x n ) .
Therefore, (18) reduces to the following iterative algorithm:
x n + 1 = exp x n α n exp x n 1 ( T r Φ ( S n ( J λ B x n ) ) ) n 0 ,
which converges to an element x * ( i = 1 N Fix ( S i ) ) EP ( Φ ) B 1 0 . The conclusion (vii) is proved. □
Remark 1. 
Compared with Theorem 3.1 in Chang et al. [23], our Theorem 1 improves, extends, and develops it in the following aspects:
(i) The problem of finding an element of Fix ( S ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) in [23] is extended to develop our problem of finding an element of ( i = 1 N Fix ( S i ) ) EP ( Φ ) ( i = 1 M ( A i + B ) 1 0 ) , where { S i } i = 1 N is a finite family of L-Lipschitzian ( L 1 ) demiclosed at zero and which features quasi-nonexpansive self-mappings on C.
(ii) The boundedness assumption of the nonempty closed and geodesic convex subset C M in [23], Theorem 3.1, is dropped by our Theorem 1, and there is only the assumption of the nonempty closed and geodesic convex subset C M in our Theorem 1.
(iii) Because M is diffeomorphic to an Euclidean space R m , M has the same topology and differential structure as R m . Moreover, Hadamard manifolds and Euclidean spaces have some similar geometrical properties. Therefore, the convergence statement of the sequence { x n } in our Theorem 1 is more general.
(iv) The splitting iterative algorithm of Theorem 3.1 in [23] is developed into the triple Mann-type iteration method of this paper, i.e., the iteration steps u n i = J λ B exp x n ( λ A i x n ) , i = 1 , 2 , , M , and y n = S u n i n with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) in [23], Theorem 3.1, are extended to develop the ones u n i = exp x n γ n exp x n 1 ( J λ B exp I ( λ A i ) x n ) , i = 1 , 2 , , M , and y n = exp x n β n exp x n 1 ( S n u n i n ) with i n { 1 , 2 , , M } s . t . d ( u n i n , x n ) = max 1 i M d ( u n i , x n ) in our Theorem 1, respectively.

4. Applicable Examples

Assume throughout that M is a Hadamard manifold of finite dimension, and the nonempty C M is of both closedness and geodesic convexity in M .

4.1. The Minimizing Problem with CFPP Constraint

Suppose that f , g : C R { + } are two proper functions of both lower semicontinuity and geodesic convexity. One first considers the minimizing problem of seeking an element y * C such that
f ( y * ) + g ( y * ) = min y C { f ( y ) + g ( y ) } ,
where Ω 1 denotes the set of solutions of the minimization issue (30), i.e.,
Ω 1 : = { x C : f ( x ) + g ( x ) = min y C { f ( y ) + g ( y ) } } .
This matter possesses a specific scenario for linear inverse issues, and some authors have furnished its applications in compressed sensing, image reconstruction, data retrieve and machine learning; refer to [29,30,31,32].
Next, one writes
Φ ( y , v ) : = Θ ( y , v ) + g ( v ) g ( y ) y , v C ,
in which Θ ( y , v ) = f ( v ) f ( y ) . It is easily known that function Φ ( y , v ) : C × C R { + } satisfies the conditions below:
(A1) Φ ( x , x ) = 0 x C ;
(A2) Φ ( x , y ) + Φ ( y , x ) 0 x , y C , i.e., Φ is monotone;
(A3) x Φ ( x , y ) is upper semicontinuous for each y C ;
(A4) y Φ ( x , y ) is geodesic convex and lower semicontinuous for each x C ;
Assume that { S i } i = 1 N is a finite family of L-Lipschitzian ( L 1 ) demiclosed at zero and with quasi-nonexpansive self-mappings on C. In Theorem 1, we put β n = γ n = 1 n 0 , and A i = B = 0 , i = 1 , 2 . , M . Then, using Theorem 1 (iii), we derive the following result.
Theorem 2. 
Assume C , M , g , f , Φ , Θ and { S i } i = 1 N are chosen as in the above conditions, and T r Φ : M 2 C ( r > 0 ) is the resolvent of Φ formulated in (16), where D ( T r Φ ) is of both closedness and geodesic convexity. Suppose there hold the relations below:
(i)(compact set) D M such that
y C D v C D s . t . Φ ( y , v ) < 0 ;
(ii) { α n } ( 0 , 1 ) , α n 1 , n = 0 ( 1 α n ) = , and ( i = 1 N Fix ( S i ) ) Ω 1 .
Then the sequence { x n } defined by
x n + 1 = exp x n α n exp x n 1 ( T r Φ ( S n x n ) ) n 0 ,
converges to an element x * ( i = 1 N Fix ( S i ) ) Ω 1 .

4.2. The Saddle Point Problem (SPP) with CFPP Constraint

Given Hadamard manifolds M 1 , M 2 , we consider geodesic convex subsets C 1 , C 2 of M 1 , M 2 , respectively. The mapping H : C 1 × C 2 R is termed as a saddle function iff the following hold:
(i) v H ( y , v ) is of geodesic convexity on C 2 w.r.t. y C 1 ;
(ii) y H ( y , v ) is of geodesic concavity, that is, y H ( y , v ) is of geodesic convexity on C 1 w.r.t. v C 2 .
An element z * = ( y * , v * ) is termed as a saddle point of H if and only if
H ( y , v * ) H ( y * , v * ) H ( y * , v ) z = ( y , v ) C 1 × C 2 ,
where Ω 2 denotes the saddle point set of problem (33).
Let V H : C 1 × C 2 2 T M 1 × 2 T M 2 be an associated multi-valued vector field with saddle function H formulated below
V H ( y , v ) = ( H ( · , v ) ) ( y ) × ( H ( y , · ) ) ( v ) ( y , v ) C 1 × C 2 .
Let M = M 1 × M 2 denote a product space. Then, it is a Hadamard manifold, and the tangent space of it at z = ( x , y ) is T z M = T y M 1 × T v M 2 (for more information, refer to Page 239 of [24]). The relevant metric is specified below
R z u , u = R y x , x + R v t , t u = ( x , t ) , u = ( x , t ) T z M .
The product of two geodesics in M 1 and M 2 denotes a geodesic in the product manifold M . So, for arbitrary two elements z = ( y , v ) and z = ( y , v ) in M , one has
exp z 1 z = exp ( y , v ) 1 ( y , v ) = ( exp y 1 y , exp v 1 v ) .
The vector field V : M 1 × M 2 2 T M 1 × 2 T M 2 is termed as being monotone if and only if, for each z = ( y , v ) , z = ( y , v ) , u = ( x , t ) V ( z ) and u = ( x , t ) V ( z ) , one has
x , exp y 1 y + t , exp v 1 v x , exp y 1 y + t , exp v 1 v .
Lemma 6 
(see [33]). Suppose that H is a saddle function on C = C 1 × C 2 , and V H is the multi-valued vector field formulated in (34). Then, V H is of maximal monotonicity.
It can be readily seen that an element z * = ( y * , v * ) C = C 1 × C 2 is a saddle point of H if and only if z * is a singularity of V H . Next, for i = 1 , 2 , , N , let the mapping S i : C C be L-Lipschitzian ( L 1 ), demiclosed at zero, and quasi-nonexpansive.
In Theorem 1, we put β n = γ n = 1 n 0 , B = V H , A i = 0 , i = 1 , 2 , , M and Φ = 0 . By Theorem 1 (vi), we immediately obtain the assertion below.
Theorem 3. 
Suppose that H : C = C 1 × C 2 ( , ) is a saddle function, and V H : C 1 × C 2 2 T M 1 × 2 T M 2 is the corresponding vector field of maximal monotonicity. Assume { S i } i = 1 N is a finite family of L-Lipschitzian ( L 1 ), demiclosed at zero, and with quasi-nonexpansive self-mappings on C such that ( i = 1 N Fix ( S i ) ) Ω 2 . Choose the initial point x 0 C , and define { x n } as follows:
x n + 1 : = exp x n α n exp x n 1 S n ( J λ V H x n ) n ,
for λ > 0 , with { α n } ( 0 , 1 ) satisfying α n 1 and n = 0 ( 1 α n ) = . Then, { x n } converges to an element x * ( i = 1 N Fix ( S i ) ) Ω 2 .

5. Conclusions

In this study, we have devised and analyzed a triple Mann iteration method for approximating a common solution of a system of quasi-variational inclusion problems, an equilibrium problem and a common fixed point problem (CFPP) of finitely many quasi-nonexpansive mappings on Hadamard manifolds. With the help of some suitable assumptions, we have shown that the iterative sequence constructed by the proposed algorithm converges to a common solution. In addition, we have furnished the applications of the major outcomes in some specific matters such as the minimization problem with CFPP constraint and saddle point problem with CFPP constraint on Hadamard manifolds. Our major outcomes are the improvement and development of those in Chang et al. [23]. Finally, it is worth mentioning that a section of our subsequent investigation is focused on presenting the new convergence outcome for the modification of our devised method with countably many quasi-nonexpansive mappings on Hadamard manifolds.

Author Contributions

Methodology, L.-C.C. and Y.-Y.H. Validation, S.-Y.L.; Writing—review & editing, J.-C.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the 2020 Shanghai Leading Talents Program of the Shanghai Municipal Human Resources and Social Security Bureau (20LJ2006100), the Innovation Program of Shanghai Municipal Education Commission (15ZZ068), and the Program for Outstanding Academic Leaders in Shanghai City (15XD1503100).

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Hartman, P.; Stampacchia, G. On some nonlinear elliptic differential functional equations. Acta Math. 1966, 115, 153–188. [Google Scholar] [CrossRef]
  2. Eskandani, G.Z.; Raeisi, M.; Lotfikar, R.H. New algorithms for solving pseudo-monotone variational inequalities in Banach spaces. J. Math. Stat. 2024, 53, 981–1000. [Google Scholar] [CrossRef]
  3. Chang, S.S.; Wen, C.F.; Yao, J.C. Forward-backward splitting method for solving a system of quasi-variational inclusions. Bull. Malays. Math. Sci. Soc. 2019, 42, 2169–2189. [Google Scholar] [CrossRef]
  4. Raeisi, M.; Chalack, M.; Eskandani, G.Z. Gradient projection-type algorithms for solving ϕ-strongly pseudomonotone equilibrium problems in Banach Spaces. Optimization 2022, 71, 2749–2767. [Google Scholar] [CrossRef]
  5. Ceng, L.C.; Wen, C.F. Systems of variational inequalities with hierarchical variational inequality constraints for asymptotically nonexpansive and pseudocontractive mappings. Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 2019, 113, 2431–2447. [Google Scholar] [CrossRef]
  6. Eskandani, G.Z.; Kim, J.K.; Moharami, R.; Lim, W.H. An extragradient method for equilibrium problems and fixed point problems of an asymptotically nonexpansive mapping in CAT(0) spaces. J. Nonlinear Convex Anal. 2022, 23, 1509–1523. [Google Scholar]
  7. Khammahawong, K.; Kumam, P.; Chaipunya, P. Splitting algorithms of common solutions between equilibrium and inclusion problems on Hadamard manifolds. Linear Nonlinear Anal. 2020, 6, 227–243. [Google Scholar]
  8. Zhao, T.Y.; Wang, D.Q.; Ceng, L.C.; He, L.; Wang, C.Y.; Fan, H.L. Quasi-inertial Tseng’s extragradient algorithms for pseudomonotone variational inequalities and fixed point problems of quasi-nonexpansive operators. Numer. Funct. Anal. Optim. 2020, 42, 69–90. [Google Scholar] [CrossRef]
  9. Ceng, L.C.; Postolache, M.; Yao, Y. Iterative algorithms for a system of variational inclusions in Banach spaces. Symmetry 2019, 11, 811. [Google Scholar] [CrossRef]
  10. Chang, S.S.; Lee, H.W.J.; Chan, C.K. A new method for solving equilibrium problem fixed point problem and variational inequality problem with application to optimization. Nonlinear Anal. 2009, 70, 3307–3319. [Google Scholar] [CrossRef]
  11. Mohammadi, M.; Eskandani, G.Z. A strong convergence theorem for quasi-equilibrium problems in Banach spaces. Acta Math. Sci. Ser. B (Engl. Ed.) 2022, 42, 221–232. [Google Scholar] [CrossRef]
  12. Eskandani, G.Z.; Raeisi, M. An iterative explicit algorithm for solving equilibrium problems in Banach spaces. Bull. Malays. Math. Sci. Soc. 2021, 44, 4299–4321. [Google Scholar] [CrossRef]
  13. Ceng, L.C.; Postolache, M.; Wen, C.F.; Yao, Y. Variational inequalities approaches to minimization problems with constraints of generalized mixed equilibria and variational inclusions. Mathematics 2019, 7, 270. [Google Scholar] [CrossRef]
  14. Ceng, L.C.; Köbis, E.; Zhao, X. On general implicit hybrid iteration method for triple hierarchical variational inequalities with hierarchical variational inequality constraints. Optimization 2020, 69, 1961–1986. [Google Scholar] [CrossRef]
  15. Mohammadi, M.; Eskandani, G.Z. Approximation solutions of quasi-equilibrium problems in Banach spaces. Miskolc Math. Notes 2020, 21, 261–272. [Google Scholar] [CrossRef]
  16. Li, C.; López, G.; Martín-Márquez, V. Iterative algorithms for nonexpansive mappings on Hadamard manifolds. Taiwan. J. Math. 2010, 14, 541–559. [Google Scholar]
  17. Moharami, R.; Eskandani, G.Z. An extragradient algorithm for solving equilibrium problem and zero point problem in Hadamard spaces. Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. RACSAM 2020, 114, 152. [Google Scholar] [CrossRef]
  18. Xu, H.K.; Kim, T.H. Convergence of hybrid steepest-descent methods for variational inequalities. J. Optim. Theory Appl. 2003, 119, 185–201. [Google Scholar] [CrossRef]
  19. Colao, V.; López, G.; Marino, G.; Martín-Márquez, V. Equilibrium problems in Hadamard manifolds. J. Math. Anal. Appl. 2012, 388, 61–71. [Google Scholar] [CrossRef]
  20. Wang, X.; López, G.; Li, C.; Yao, J.C. Equilibrium problems on Riemannian manifolds with applications. J. Math. Anal. Appl. 2019, 473, 866–891. [Google Scholar] [CrossRef]
  21. Ansari, Q.H.; Babu, F.; Li, X. Variational inclusion problems in Hadamard manifolds. J. Nonlinear Convex Anal. 2018, 19, 219–237. [Google Scholar]
  22. Al-Homidan, S.; Ansari, Q.H.; Babu, F. Halpern- and Mann-type algorithms for fixed points and inclusion problems on Hadamard manifolds. Numer. Funct. Anal. Optim. 2019, 40, 621–653. [Google Scholar] [CrossRef]
  23. Chang, S.S.; Yao, J.C.; Yang, L.; Wen, C.F.; Wu, D.P. Convergence analysis for variational inclusion problems equilibrium problems and fixed point in Hadamard Manifolds. Numer. Funct. Anal. Optim. 2021, 42, 567–582. [Google Scholar] [CrossRef]
  24. Sakai, T. Riemannian Geometry (Translations of Mathematical Monographs); American Mathematical Society: Providence, RI, USA, 1996; Volume 149. [Google Scholar]
  25. Ceng, L.C.; Li, X.; Qin, X. Parallel proximal point methods for systems of vector optimization problems on Hadamard manifolds without convexity. Optimization 2020, 69, 357–383. [Google Scholar] [CrossRef]
  26. Ferreira, O.P.; Oliveira, P.R. Proximal point algorithm on Riemannian manifolds. Optimization 2002, 51, 257–270. [Google Scholar] [CrossRef]
  27. Li, C.; López, G.; Martín-Márquez, V.; Wang, J.H. Resolvents of set-valued monotone vector fields in Hadamard manifolds. Set-Valued Anal. 2011, 19, 361–383. [Google Scholar] [CrossRef]
  28. Neto, J.X.d.; Ferreira, O.P.; Pérez, L.R.L. Monotone point-to-set vector fields. Balk. J. Geom. Appl. 2000, 5, 69–79. [Google Scholar]
  29. Beck, A.; Teboulle, M. A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM J. Imaging Sci. 2009, 2, 183–202. [Google Scholar] [CrossRef]
  30. Chang, S.S.; Wen, C.F.; Yao, J.C. Common zero point for a finite family of inclusion problems of accretive mappings in Banach spaces. Optimization 2018, 67, 1183–1196. [Google Scholar] [CrossRef]
  31. Cruz, J.Y.B. On proximal subgradient splitting method for minimizing the sum of two nonsmooth convex functions. Set-Valued Var. Anal. 2017, 25, 245–263. [Google Scholar] [CrossRef]
  32. Zhao, X.; Ng, K.F.; Li, C.; Yao, J.C. Linear regularity and linear convergence of projection-based methods for solving convex feasibility problems. Appl. Math. Optim. 2018, 78, 613–641. [Google Scholar] [CrossRef]
  33. Li, C.; López, G.; Martín-Márquez, V. Monotone vector fields and the proximal point algorithm on Hadamard manifolds. J. Lond. Math. Soc. 2009, 79, 663–683. [Google Scholar] [CrossRef]
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

Ceng, L.-C.; Huang, Y.-Y.; Li, S.-Y.; Yao, J.-C. Triple Mann Iteration Method for Variational Inclusions, Equilibria, and Common Fixed Points of Finitely Many Quasi-Nonexpansive Mappings on Hadamard Manifolds. Mathematics 2025, 13, 444. https://doi.org/10.3390/math13030444

AMA Style

Ceng L-C, Huang Y-Y, Li S-Y, Yao J-C. Triple Mann Iteration Method for Variational Inclusions, Equilibria, and Common Fixed Points of Finitely Many Quasi-Nonexpansive Mappings on Hadamard Manifolds. Mathematics. 2025; 13(3):444. https://doi.org/10.3390/math13030444

Chicago/Turabian Style

Ceng, Lu-Chuan, Yun-Yi Huang, Si-Ying Li, and Jen-Chih Yao. 2025. "Triple Mann Iteration Method for Variational Inclusions, Equilibria, and Common Fixed Points of Finitely Many Quasi-Nonexpansive Mappings on Hadamard Manifolds" Mathematics 13, no. 3: 444. https://doi.org/10.3390/math13030444

APA Style

Ceng, L.-C., Huang, Y.-Y., Li, S.-Y., & Yao, J.-C. (2025). Triple Mann Iteration Method for Variational Inclusions, Equilibria, and Common Fixed Points of Finitely Many Quasi-Nonexpansive Mappings on Hadamard Manifolds. Mathematics, 13(3), 444. https://doi.org/10.3390/math13030444

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