Case 1 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 11 | 0.015498 |
Dixit et al. (1.4) | 75 | 0.699281 |
Lorenz and Pock (1.5) | 15 | 0.019471 |
The goal of this study is to develop a novel iterative algorithm for approximating the solutions of the monotone inclusion problem and fixed point problem of a finite family of demimetric mappings in the context of a real Hilbert space. The proposed algorithm is based on the inertial extrapolation step strategy and combines forward-backward and Tseng's methods. We introduce a demimetric operator with respect to M-norm, where M is a linear, self-adjoint, positive and bounded operator. The algorithm also includes a new step for solving the fixed point problem of demimetric operators with respect to the M-norm. We study the strong convergence behavior of our algorithm. Furthermore, we demonstrate the numerical efficiency of our algorithm with the help of an example. The result given in this paper extends and generalizes various existing results in the literature.
Citation: Anjali, Seema Mehra, Renu Chugh, Salma Haque, Nabil Mlaiki. Iterative algorithm for solving monotone inclusion and fixed point problem of a finite family of demimetric mappings[J]. AIMS Mathematics, 2023, 8(8): 19334-19352. doi: 10.3934/math.2023986
[1] | Yasir Arfat, Supak Phiangsungnoen, Poom Kumam, Muhammad Aqeel Ahmad Khan, Jamshad Ahmad . Some variant of Tseng splitting method with accelerated Visco-Cesaro means for monotone inclusion problems. AIMS Mathematics, 2023, 8(10): 24590-24608. doi: 10.3934/math.20231254 |
[2] | Jamilu Abubakar, Poom Kumam, Jitsupa Deepho . Multistep hybrid viscosity method for split monotone variational inclusion and fixed point problems in Hilbert spaces. AIMS Mathematics, 2020, 5(6): 5969-5992. doi: 10.3934/math.2020382 |
[3] | Doaa Filali, Mohammad Dilshad, Mohammad Akram . Generalized variational inclusion: graph convergence and dynamical system approach. AIMS Mathematics, 2024, 9(9): 24525-24545. doi: 10.3934/math.20241194 |
[4] | Yali Zhao, Qixin Dong, Xiaoqing Huang . A self-adaptive viscosity-type inertial algorithm for common solutions of generalized split variational inclusion and paramonotone equilibrium problem. AIMS Mathematics, 2025, 10(2): 4504-4523. doi: 10.3934/math.2025208 |
[5] | Kasamsuk Ungchittrakool, Natthaphon Artsawang . A generalized viscosity forward-backward splitting scheme with inertial terms for solving monotone inclusion problems and its applications. AIMS Mathematics, 2024, 9(9): 23632-23650. doi: 10.3934/math.20241149 |
[6] | Meiying Wang, Luoyi Shi, Cuijuan Guo . An inertial iterative method for solving split equality problem in Banach spaces. AIMS Mathematics, 2022, 7(10): 17628-17646. doi: 10.3934/math.2022971 |
[7] | Jun Yang, Prasit Cholamjiak, Pongsakorn Sunthrayuth . Modified Tseng's splitting algorithms for the sum of two monotone operators in Banach spaces. AIMS Mathematics, 2021, 6(5): 4873-4900. doi: 10.3934/math.2021286 |
[8] | Sani Salisu, Poom Kumam, Songpon Sriwongsa . One step proximal point schemes for monotone vector field inclusion problems. AIMS Mathematics, 2022, 7(5): 7385-7402. doi: 10.3934/math.2022412 |
[9] | Mohammad Dilshad, Fahad Maqbul Alamrani, Ahmed Alamer, Esmail Alshaban, Maryam G. Alshehri . Viscosity-type inertial iterative methods for variational inclusion and fixed point problems. AIMS Mathematics, 2024, 9(7): 18553-18573. doi: 10.3934/math.2024903 |
[10] | Mohammad Dilshad, Mohammad Akram, Md. Nasiruzzaman, Doaa Filali, Ahmed A. Khidir . Adaptive inertial Yosida approximation iterative algorithms for split variational inclusion and fixed point problems. AIMS Mathematics, 2023, 8(6): 12922-12942. doi: 10.3934/math.2023651 |
The goal of this study is to develop a novel iterative algorithm for approximating the solutions of the monotone inclusion problem and fixed point problem of a finite family of demimetric mappings in the context of a real Hilbert space. The proposed algorithm is based on the inertial extrapolation step strategy and combines forward-backward and Tseng's methods. We introduce a demimetric operator with respect to M-norm, where M is a linear, self-adjoint, positive and bounded operator. The algorithm also includes a new step for solving the fixed point problem of demimetric operators with respect to the M-norm. We study the strong convergence behavior of our algorithm. Furthermore, we demonstrate the numerical efficiency of our algorithm with the help of an example. The result given in this paper extends and generalizes various existing results in the literature.
Monotone inclusion problem (MIP) is the problem to identify a zero of the sum of two operators and it is described as
Findingζ∈Hsuchthat0∈(A+B)ζ, | (1.1) |
where A:H→H is an M-cocoercive operator, M is a linear, bounded operator on a real Hilbert space H and B:H→2H is a maximal monotone operator. Monotone inclusion problem incorporates various problems such as convex optimization, machine learning, statistical regression, signal and image processing, see [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17]. Lions and Mercier [18] defined the forward-backward splitting algorithm which is one of the most effective method for solving the problem (1.1) and it is defined by
ζn+1=(I+λnA)−1(I−λnB)ζnforalln∈N, | (1.2) |
where A is monotone, B is 1/L-cocoercive operator and λn∈(0,2/L) is a step size parameter. But the algorithm defined in (1.2) converges weakly to a solution of the monotone inclusion problem. Later, Tseng [15] improved this forward-backward splitting algorithm and also proved weak convergence of it. After that, Gibali and Thong [19] proposed a modified version of Tseng's splitting algorithm and proved strong convergence of the proposed algorithm. To speed up the convergence rate of the algorithms, Moudafi and Oliny [20] introduced the following algorithm which includes the inertial parameter.
{φn=ζn+ϵn(ζn−ζn−1),ζn+1=(I+λnA)−1(I−λnB)ζnforalln∈N, | (1.3) |
where ϵn∈[0,1) is inertial parameter. They proved the weak convergence of above algorithm (1.3) assuming the conditions ∑∞n=1ϵn‖ζn−ζn−1‖2<∞ and λn<2/L, where L is Lipschitz constant of operator B. Later on, the following preconditioning algorithm was defined by Lorenz and Pock [10] for solving the monotone inclusion problem and proved weak convergence of it.
{φn=ζn+ϵn(ζn−ζn−1),ζn+1=(I+λnM−1A)−1(I−λnM−1B)φn. | (1.4) |
It is clear that the algorithm (1.4) reduces to forward-backward algorithm (1.2) if we take ϵn=0 and M=I. Dixit et al.[4], in 2021, introduced accelerated preconditioning forward-backward normal S-iteration and proved weak convergence under few assumptions in a real Hilbert space H.
{φn=ζn+ϵn(ζn−ζn−1),ζn+1=JA,Bλ,M((1−γn)φn+γnJA,Bλ,M(φn))foralln∈N, | (1.5) |
where JA,Bλ,M=(I+λM−1A)−1(I−λM−1B),γn∈(0,1),ϵn∈[0,1) and λ∈[0,1). Next, Altiparmak and Karahan [21] proposed a new preconditioning forward-backward splitting algorithm and proved strong convergence of it.
{φn=ζn+ϵn(ζn−ζn−1),υn=JA,Bλ,M((1−βn)φn+βnJA,Bλ,M(φn)),ζn+1=(1−γn)JA,Bλ,M(υn)+γnf(υn)foralln∈N, | (1.6) |
where ϵn⊂[0,θ] with θ∈[0,1) and βn,γn∈(0,1) and f:H→H is a k- contraction mapping with respect to M-norm. Recently, in 2021, the same authors [22] proved strong convergence of a modified preconditioning algorithm for solving the monotone inclusion problem in the following manner:
{φn=ζn+ϵn(ζn−ζn−1),υn=(1−αn)φn+αnJA,Bλ,M(φn),κn=JA,Bλ,M((1−βn)υn+βnJA,Bλ,M(υn)),ζn+1=(1−γn)JA,Bλ,M(κn)+γnf(κn)foralln∈N, | (1.7) |
where JA,Bλ,M=(I+λM−1B)−1(I−λM−1A),αn,βn,γn∈(0,1),ϵn⊂[0,θ] with θ∈[0,1) and h:H→H is k-contraction with respect to M-norm.
On the other hand, the theory of fixed points has been an appealing topic of research. Several researchers have worked in this direction, see [23,24,25,26,27].
We have considered the Monotone inclusion problem and fixed point problem of a finite family of demimetric mappings in the context of Hilbert space in this paper as we defined the demimetric operator in the notion of inner product with respect to M-norm, where M is the linear, self-adjoint, positive and bounded operator. In the setting of a Hilbert space, that is, in the complete inner product space, the existence of a solution to the monotone inclusion problem is guaranteed under certain conditions. The authors defined contraction, nonexpansive, quasi-nonexpansive operators with respect to M-norm in these algorithms and proved weak and strong convergence of those algorithms under suitable assumptions. Our main contributions to this research are as follows:
● We introduce an operator namely, a demimetric operator with respect to M-norm;
● We define a new algorithm which is the combination of forward-backward and Tseng method for solving the monotone inclusion problem together with a new step for solving the fixed point problem of the finite family of demimetric operators;
● We also utilize the inertial extrapolation step strategy due to Polyak [29] to enhance the convergence rate of the proposed algorithm.
In this section, we list some definitions and lemmas which contribute significantly to our main result. Throughout the study, suppose H is a real Hilbert space, Q is a nonempty, closed and convex subset of H, M is a linear, self-adjoint, positive and bounded operator on H and Fix(U) denotes the set of all fixed points of the mapping U.
Definition 2.1. [30] Assume that A:H→2H is a set-valued operator. It is called monotone if
⟨ζ−φ,υ−κ⟩≥0forallζ,φ∈H,υ∈Aζandκ∈Aφ. |
The operator A is called maximal monotone if the graph of the operator A is not properly contained in the graph of any other monotone operator.
Definition 2.2. [31] Assume that Q is nonempty subset of H and ζ∈H. If for any υ∈H, there exists a unique point φ∈Q such that ‖φ−ζ‖≤‖υ−ζ‖forallυ∈H then φ is called metric projection of ζ onto Q. It is symbolized by φ=PQζ. If PQζ exists and it can be uniquely obtained for all ζ∈H, then PQ:H→Q is called metric projection operator. The operator PQ is nonexpansive and it satisfies the following inequality
⟨ζ−PQζ,ψ−PQζ⟩≤0forallψ∈Q. |
Definition 2.3. [32] Assume that M is a bounded and linear operator on H. The operator M is said to be self-adjoint if M∗=M where M∗ denotes the adjoint of M. If ⟨M(ζ),ζ)⟩>0 for all ζ(≠0)∈H, then M is called positive definite operator. The M-inner product is given by ⟨ζ,ψ⟩M=⟨ζ,M(ψ)⟩ for all ζ,ψ∈H and the corresponding M-norm is also given by ‖ζ‖2M=⟨ζ,M(ζ)⟩ for all ζ∈H by using the self-adjoint, linear and bounded operator M.
Definition 2.4. [4] Assume that Q is a nonempty subset of H, M is a positive definite operator on H and U:Q→H is an operator. Then U is called
(i) M-cocoercive operator if
‖Uζ−Uψ‖2M−1≤⟨ζ−ψ,Uζ−Uψ⟩,forallζ,ψ∈H, |
(ii) nonexpansive operator with respect to M -norm if
‖Uζ−Uψ‖M≤‖ζ−ψ‖Mforallζ,ψ∈H, |
(iii) quasi-nonexpansive operator with respect to M -norm if
‖Uζ−Uψ‖M≤‖ζ−ψ‖Mforallζ∈H,ψ∈Fix(U), |
(iv) h-contraction with respect to M-norm if there exists h∈[0,1) such that
‖Uζ−Uψ‖M≤h‖ζ−ψ‖Mforallζ,ψ∈H. |
Lemma 2.1. [30] For ζ,ψ,ω,ι∈H and α,β,γ∈[0,1], where α+β+γ=1, we have
(i) ‖ζ+ψ‖2≤‖ζ‖2+2⟨ψ,ζ+ψ⟩,
(ii) ‖βζ+(1−β)ψ‖2=β‖ζ‖2+(1−β)‖ψ‖2−β(1−β)‖ζ−ψ‖2,
(iii) ‖αζ+βψ+γω‖2=α‖ζ‖2+β‖ψ‖2+γ‖ω‖2−αβ‖ζ−ψ‖2−αγ‖ζ−ω‖2−βγ‖ψ−ω‖2,
(iv) 2⟨ζ−ψ,ω−ι⟩=‖ζ−ι‖2+‖ψ−ω‖2−‖ζ−ω‖2−‖ψ−ι‖2,
(v) ‖ζ±ψ‖2=‖ζ‖2±2⟨ζ,ψ⟩+‖ψ‖2.
Lemma 2.2. [33] Let U:Q→H be the nonexpansive operator with Fix(U)≠ϕ. Then the mapping I−U is demiclosed at origin, that is, for any sequence {ζn}∈H such that ζn→ζ∈H and ‖ζn−Uζn‖→0 as n→∞, we have ζ∈Fix(U).
Lemma 2.3. [34] Assume that S:Q→H is ξ-demimetric operator with ξ∈(−∞,1). Then, Fix(S) is closed and convex.
Lemma 2.4. [4] Suppose that A is M-cocoercive operator on H, B:H→2H is maximal monotone operator. Then, JA,Bλ,M=(I+λM−1B)−1(I−λM−1A) is nonexpansive with respect to M-norm for λ∈(0,1].
Lemma 2.5. [4] Suppose that A is M-cocoercive operator on H, B:H→2H is a maximal monotone operator and λ is a non-negative real number. Then, ζ∈H is a solution of monotone inclusion problem (1.1) if and only if (I+λM−1B)−1(I−λM−1A)(ζ)=ζ.
Lemma 2.6. [35] Suppose that {tn}⊂[0,∞) is a sequence of real numbers. Let
sn+1≤(1−pn)sn+pntnforalln∈N, |
where {pn}⊂[0,1] and {tn}⊂(−∞,∞) satisfying the following assumptions:
(1) ∑∞n=1pn=∞,
(2) lim supn→∞tn≤0.
Then limn→∞sn=0.
Lemma 2.7. [36] Assume that {ζn}⊆[0,∞) such that there exists a subsequence {ζnj} of {ζn} such that ζnj<ζnj+1. Then, there exists a nondecreasing sequence {nk} of natural numbers satisfying limk→∞nk=∞, ζnk≤ζnk+1 and ζk≤ζnk+1 for all k∈N. Also, nk is the greatest number n in the set {1,2,...,k} satisfying ζn<ζn+1.
In this part, we will prove the strong convergence of the following algorithm for finding a common solution of monotone inclusion and fixed point problem of a finite family of demimetric operators in the setting of a real Hilbert space H. Assume that A is M-cocoercive operator on H, B:H→2H is a maximally monotone operator. Suppose Si is a finite family of ξ-demimetric operators with ξ∈(−∞,1) such that I−Si is demiclosed at origin for all i=0,1,2,...N−1, h:H→H is a contraction mapping with respect to M-norm with constant k∈(0,1].
Algorithm 3.1. Let ζ0,ζ1∈H. Compute {φn},{υn},{κn}and{ζn} using
{φn=ζn+ϵn(ζn−ζn−1),υn=(1−αn)φn+αnJA,Bλ,M(φn),κn=JA,Bλ,M((1−βn)υn+βnJA,Bλ,M(υn)),ζn+1=γnh(ζn)+(1−γn−δn)JA,Bλ,M(κn)+δnSnκn, | (3.1) |
where Sn=1N∑N−1i=0(1−qn)I+qnSi
Definition 3.1. A mapping U:Q→H, where Q is closed, convex and nonempty subset of H is called ξ-demimetric with respect to M-norm, where ξ∈(−∞,1) if Fix(U)≠ϕ such that
⟨ζ−ζ∗,(I−U)ζ⟩M≥12(1−ξ)‖(I−U)ζ‖2M,forallζ∈Q,ζ∗∈Fix(U). |
Example 3.1. Let H=R3,Q=H and M(ζ)=(5ζ1,4ζ2,5ζ3) for all ζ=(ζ1,ζ2,ζ3)∈H. The mapping U:H→H defined by U(ζ)=−2ζ for all ζ=(ζ1,ζ2,ζ3) is 13-demimetric mapping with respect to M-norm.
Now, we prove a result for ξ-demimetric operator with respect to M-norm which is motivated by Lemma (2.2) of [37].
Lemma 3.1. Suppose S:Q→H is ξ-demimetric operator with respect to M-norm, where ξ∈(−∞,1) and Fix(S) is nonempty. Let P=(1−γ)I+γS, where γ∈(−∞,∞) with γ∈(0,1−ξ], then P:Q→H is a quasi-nonexpansive operator.
Proof. It is direct that Fix(S)=Fix(P). Now, since S:Q→H is demimetric operator, we have for any ζ∈Q and y∈Fix(P),
⟨ζ−y,ζ−Pζ⟩M=⟨ζ−y,γ(ζ−Sζ)⟩M=γ⟨ζ−y,ζ−Sζ⟩M≥γ2(1−ξ)‖(I−S)ζ‖2M=1−ξ2γ‖ζ−Pζ‖2M≥γ2γ‖ζ−Pζ‖2M=12‖ζ−Pζ‖2M. |
This implies that P is 0-demimetric mapping. Also, for ζ∈Q, y∈Fix(P) and using Lemma (2.1), we deduce
12‖ζ−Pζ‖2M≤⟨ζ−y,ζ−Pζ⟩M⟺‖ζ−Pζ‖2M≤2⟨ζ−y,ζ−Pζ⟩M⟺‖ζ−Pζ‖2M≤‖ζ−Pζ‖2M+‖ζ−y‖2M−‖Pζ−y‖2M⟺‖Pζ−y‖2M≤‖ζ−y‖2M⟺‖Pζ−y‖M≤‖ζ−y‖M. |
Hence, P is a quasi-nonexpansive operator.
Lemma 3.2. The mapping Sn defined by Sn=1N∑N−1i=0(1−qn)I+qnSi is quasi-nonexpansive. \proof Let ζ∗∈Ω.
Consider
‖Snζ−ζ∗‖M=‖1NN−1∑i=0((1−qn)I+qnSi)ζ−ζ∗‖M≤1NN−1∑i=0‖((1−qn)I+qnSi)ζ−ζ∗‖M≤1NN−1∑i=0‖ζ−ζ∗‖M=‖ζ−ζ∗‖M. |
Hence, Sn is quasi-nonexpansive.
Lemma 3.3. Assume that {ζn} is bounded sequence of real numbers and ζ∗∈Ω=(A+B)−1(0)∩N−1⋂i=0Fix(Si). If limn→∞‖JA,Bλ,Mζn−ζn‖M=0 and limn→∞‖κn−Snκn‖M=0. Then lim supn→∞⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤0.
Proof. Since the sequence {ζn} is bounded, so there exists a subsequence {ζnk} of {ζn} such that ζnk⇀ζ
Consider
lim supn→∞⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M=lim supk→∞⟨h(ζ∗)−ζ∗,ζnk+1−ζ∗⟩M=⟨h(ζ∗)−ζ∗,ζ−ζ∗⟩M. | (3.2) |
Using the condition ‖JA,Bλ,Mζn−ζn‖M→0 as n→∞ and by using Lemma (2.5), we deduce that ζ∈(A+B)−1(0). Also, since limn→∞‖ζn−κn‖=0 and ζnk⇀ζ, so we obtain κnk⇀ζ. By using the condition limn→∞‖κn−Snκn‖=0 and using Lemma (2.2), we obtain that ζ∈Fix(Si). Thus ζ∈Ω.
From Eq (3.2) and by the property of metric projection, we have
lim supn→∞⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤⟨h(ζ∗)−ζ∗,ζ−ζ∗⟩≤0. |
Hence the lemma is proved.
Theorem 3.1. Suppose that the solution set Ω=(A+B)−1(0)∩N−1⋂i=0Fix(Si) is non-empty and the sequence {ζn} is generated by algorithm (3.1), where {ϵn}⊂[0,θ] with θ∈[0,1),{αn},{βn},{γn},{δn}∈(0,1) such that the following conditions hold:
(i) 0<a≤αn≤b<1forsomea,b∈R,
(ii) 0<c≤βn≤d<1forsomec,d∈R,
(iii) limn→∞γn=0,∑∞n=1γn=∞,
(iv) ∑∞n=1ϵn‖ζn−ζn−1‖<∞,
(v) for any n∈N,0<a∗<lim infn→∞δn≤lim supn→∞δn<b∗<1−γn, where a∗,b∗∈R+.
Then the sequence {ζn} converges strongly to a point ζ∗∈Ω=PΩh(ζ∗).
Proof. First we prove that the sequence {ζn} is bounded. Let ζ∗∈Ω.
Consider
‖φn−ζ∗‖M=‖ζn+ϵn(ζn−ζn−1)−ζ∗‖M≤‖ζn−ζ∗‖M+‖ϵn(ζn−ζn−1)‖M=‖ζn−ζ∗‖M+ϵn‖ζn−ζn−1‖M. | (3.3) |
Since, JA,Bλ,M is nonexpansive with respect to M-norm, so using nonexpansiveness of JA,Bλ,M, we have
‖υn−ζ∗‖M=‖(1−αn)φn+αnJA,Bλ,M(φn)−ζ∗‖M≤(1−αn)‖φn−ζ∗‖M+αn‖JA,Bλ,M(φn)−ζ∗‖M≤(1−αn)‖φn−ζ∗‖M+αn‖φn−ζ∗‖M=‖φn−ζ∗‖M, | (3.4) |
and
‖κn−ζ∗‖M=‖JA,Bλ,M((1−βn)υn+βnJA,Bλ,M(υn))−ζ∗‖M≤‖(1−βn)υn+βnJA,Bλ,M(υn)−ζ∗‖M≤(1−βn)‖υn−ζ∗‖M+βn‖JA,Bλ,M(υn)−ζ∗‖M≤(1−βn)‖υn−ζ∗‖M+βn‖υn−ζ∗‖M=‖υn−ζ∗‖M. | (3.5) |
From Eqs (3.1)–(3.5) and using the fact that h is k-contraction and Sn is quasi-nonexpansive with respect to M-norm, we have
‖ζn+1−ζ∗‖M=‖γnh(ζn)+(1−γn−δn)JA,Bλ,M(κn)+δnSnκn−ζ∗‖M≤γn‖h(ζn)−ζ∗‖M+(1−γn−δn)‖JA,Bλ,M(κn)−ζ∗‖M+δn‖Snκn−ζ∗‖M≤γn‖h(ζn)−ζ∗‖M+(1−γn−δn)‖κn−ζ∗‖M+δn‖κn−ζ∗‖M=γn‖h(ζn)−ζ∗‖M+(1−γn)‖κn−ζ∗‖M≤γn‖h(ζn)−h(ζ∗)‖M+γn‖h(ζ∗)−ζ∗‖M+(1−γn)‖υn−ζ∗‖M≤γn‖h(ζn)−h(ζ∗)‖M+γn‖h(ζ∗)−ζ∗‖M+(1−γn)‖φn−ζ∗‖M≤γn‖h(ζn)−h(ζ∗)‖M+γn‖h(ζ∗)−ζ∗‖M+(1−γn)[‖ζn−ζ∗‖M+ϵn‖ζn−ζn−1‖M]≤γnk‖ζn−ζ∗‖M+γn‖h(ζ∗)−ζ∗‖M+(1−γn)‖ζn−ζ∗‖M+ϵn(1−γn)‖ζn−ζn−1‖M=(1−γn+γnk)‖ζn−ζ∗‖M+γn‖h(ζ∗)−ζ∗‖M+ϵn(1−γn)‖ζn−ζn−1‖M≤[1−γn(1−k)]‖ζn−ζ∗‖M+γn‖h(ζ∗)−ζ∗‖M+ϵn‖ζn−ζn−1‖M=[1−γn(1−k)]‖ζn−ζ∗‖M+γn‖h(ζ∗)−ζ∗‖M+ϵnγnγn‖ζn−ζn−1‖M. | (3.6) |
From assumptions (iii) and (iv), we have limn→∞ϵnγn‖ζn−ζn−1‖M=0. So, there exists a positive integer N1>0 such that ϵnγn‖ζn−ζn−1‖≤N1. By using Eq (3.6), we obtain
‖ζn+1−ζ∗‖M≤[1−γn(1−k)]‖ζn−ζ∗‖M+γn[N1+‖h(ζ∗)−ζ∗‖M]≤[1−γn(1−k)]‖ζn−ζ∗‖M+γn(1−k)[N1+‖h(ζ∗)−ζ∗‖M1−k]≤max.{‖ζn−ζ∗‖M,N1+‖h(ζ∗)−ζ∗‖M1−k}. |
Continuing like this,
‖ζn+1−ζ∗‖M≤max.{‖ζ1−ζ∗‖M,N1+‖h(ζ∗)−ζ∗‖M1−k}. |
Thus the sequence {ζn} is bounded and therefore, the sequences {φn},{υn},{κn} are also bounded.
Now, we show that ζn→ζ∗. Using Lemma (2.1), we obtain
‖φn−ζ∗‖2M=‖ζn+ϵn(ζn−ζn−1)−ζ∗‖2M≤‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M, | (3.7) |
‖υn−ζ∗‖2M=‖(1−αn)φn+αnJA,Bλ,M(φn)−ζ∗‖2M=(1−αn)‖φn−ζ∗‖2M+αn‖JA,Bλ,M(φn)−ζ∗‖2M−αn(1−αn)‖φn−JA,Bλ,M(φn)‖2M≤(1−αn)‖φn−ζ∗‖2M+αn‖φn−ζ∗‖2M−αn(1−αn)‖φn−JA,Bλ,M(φn)‖2M=‖φn−ζ∗‖2M−(1−αn)αn‖φn−JA,Bλ,M(φn)‖2M≤‖φn−ζ∗‖2M, | (3.8) |
and
‖κn−ζ∗‖2M=‖JA,Bλ,M((1−βn)υn+βnJA,Bλ,M(υn))−ζ∗‖2M≤‖(1−βn)υn+βnJA,Bλ,M(υn)−ζ∗‖2M=‖(1−βn)(υn−ζ∗)+βn(JA,Bλ,M(υn)−ζ∗)‖2M=(1−βn)‖υn−ζ∗‖2M+βn‖JA,Bλ,M(υn)−ζ∗‖2M−(1−βn)βn‖υn−JA,Bλ,M(υn)‖2M≤(1−βn)‖υn−ζ∗‖2M+βn‖υn−ζ∗‖2−(1−βn)βn‖υn−JA,Bλ,M(υn)‖2M≤‖υn−ζ∗‖2M. | (3.9) |
Using Eqs (3.7)–(3.9), Lemma (2.1) and using the fact that JA,Bλ,M is nonexpansive and Sn is quasi-nonexpansive, we obtain
‖ζn+1−ζ∗‖2M=‖γnh(ζn)+(1−γn−δn)JA,Bλ,M(κn)+δnSnκn−ζ∗‖2M=‖γn(h(ζn)−ζ∗)+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M=‖γn(h(ζn)−h(ζ∗))+γn(h(ζ∗)−ζ∗)+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M≤‖γn(h(ζn)−h(ζ∗))+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M+2⟨γn(h(ζ∗)−ζ∗),ζn+1−ζ∗⟩M=‖γn(h(ζn)−h(ζ∗))+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤γn‖h(ζn)−h(ζ∗)‖2M+(1−γn−δn)‖JA,Bλ,M(κn)−ζ∗‖2M+δn‖Snκn−ζ∗‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤γn‖h(ζn)−h(ζ∗)‖2M+(1−γn−δn)‖κn−ζ∗‖2M+δn‖κn−ζ∗‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M=(1−γn)‖κn−ζ∗‖2M+γn‖h(ζn)−h(ζ∗)‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤(1−γn)[‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M]+γn‖h(ζn)−h(ζ∗)‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤(1−γn)‖ζn−ζ∗‖2M+2ϵn(1−γn)‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n(1−γn)‖ζn−ζn−1‖2M+γnk‖ζn−ζ∗‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤[1−γn(1−k)]‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M+2γn⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M, | (3.10) |
which implies that
‖ζn+1−ζ∗‖2M≤[1−γn(1−k)]‖ζn−ζ∗‖2M+γn(1−k)[21−k⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩+2ϵnγn(1−k)‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2nγn(1−k)‖ζn−ζn−1‖2M]. | (3.11) |
Equation (3.11) is equivalent to
sn+1≤(1−pn)sn+pntn, | (3.12) |
where, sn=‖ζn−ζ∗‖2M, pn=γn(1−k) and tn=21−k⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩+2ϵnγn(1−k)‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2nγn(1−k)‖ζn−ζn−1‖2M
Now, we consider two possible cases on the sequence {‖ζn−ζ∗‖M}.
Case 1. Suppose that there exists some positive integer n0 such that the sequence {‖ζn−ζ∗‖M} is nonincreasing sequence for any n≥n0. Also, the sequence {‖ζn−ζ∗‖M} is bounded below by zero. So, it is convergent.
By using Lemma (2.1), Eqs (3.7) and (3.9), we have
‖ζn+1−ζ∗‖2M=‖γn(h(ζn)−ζ∗)+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn−δn)‖JA,Bλ,M(κn)−ζ∗‖2M+δn‖Snκn−ζ∗‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn−δn)‖κn−ζ∗‖2M+δn‖κn−ζ∗‖2M=γn‖h(ζn)−ζ∗‖2M+(1−γn)‖κn−ζ∗‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn)‖υn−ζ∗‖2M, | (3.13) |
and
‖υn−ζ∗‖2M=‖(1−αn)(φn−ζ∗)+αn(JA,Bλ,M(φn)−ζ∗)‖2M=(1−αn)‖φn−ζ∗‖2M+αn‖JA,Bλ,M(φn)−ζ∗‖2M−αn(1−αn)‖JA,Bλ,M(φn)−φn‖2M=‖φn−ζ∗‖2M−αn(1−αn)‖JA,Bλ,M(φn)−φn‖2M≤‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M−αn(1−αn)‖JA,Bλ,M(φn)−φn‖2M. | (3.14) |
From Eqs (3.13) and (3.14), we obtain
‖ζn+1−ζ∗‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn)[‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M−αn(1−αn)‖JA,Bλ,M(φn)−φn‖2M], |
which implies that
αn(1−αn)‖JA,Bλ,M(φn)−φn‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn)‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M−‖ζn+1−ζ∗‖2M≤γn‖h(ζn)−ζ∗‖2M+‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M−‖ζn+1−ζ∗‖2M. | (3.15) |
Since, ∑∞n=1ϵn‖ζn−ζn−1‖<∞, so we have
limn→∞ϵn‖ζn−ζn−1‖M=0. | (3.16) |
Using the condition limn→∞γn=0, Eq (3.16) and taking limit n→∞ in Eq (3.15), we get
limn→∞‖JA,Bλ,M(φn)−φn‖2M=0. | (3.17) |
Now, consider
‖υn−φn‖M=‖(1−αn)φn+αnJA,Bλ,M(φn)−φn‖M=‖αn(JA,Bλ,M(φn)−φn)‖M=αn‖JA,Bλ,M(φn)−φn‖M. | (3.18) |
Taking limit n→∞ in Eq (3.18) and using Eq (3.17), we obtain
limn→∞‖υn−φn‖M=0. | (3.19) |
By using triangle inequality, we have
‖JA,Bλ,M(φn)−υn‖M≤‖JA,Bλ,M(φn)−φn‖M+‖φn−υn‖M. | (3.20) |
Taking limit n→∞ in Eq (3.20) and using Eqs (3.17) and (3.19), we deduce
limn→∞‖JA,Bλ,M(φn)−υn‖M=0. | (3.21) |
Also,
‖JA,Bλ,M(υn)−υn‖M=‖JA,Bλ,M(υn)−JA,Bλ,M(φn)+JA,Bλ,M(φn)−υn‖M≤‖JA,Bλ,M(υn)−JA,Bλ,M(φn)‖M+‖JA,Bλ,M(φn)−υn‖M≤‖υn−φn‖M+‖JA,Bλ,M(φn)−υn‖M, | (3.22) |
Taking limit n→∞ in Eq (3.22) and using Eqs (3.19) and (3.21) in Eq (3.22), we have
limn→∞‖JA,Bλ,M(υn)−υn‖M=0, | (3.23) |
and
‖κn−φn‖M=‖κn−JA,Bλ,M(φn)+JA,Bλ,M(φn)−φn‖M≤‖κn−JA,Bλ,M(φn)‖M+‖JA,Bλ,M(φn)−φn‖M=‖JA,Bλ,M((1−βn)υn+βnJA,Bλ,M(υn))−JA,Bλ,M(φn)‖M+‖JA,Bλ,M(φn)−φn‖M≤‖(1−βn)υn+βnJA,Bλ,M(υn)−φn‖M+‖JA,Bλ,M(φn)−φn‖M=‖(υn−φn)+βn(JA,Bλ,M(υn)−υn)‖M+‖JA,Bλ,M(φn)−φn‖M≤‖υn−φn‖M+βn‖JA,Bλ,M(υn)−υn‖M+‖JA,Bλ,M(φn)−φn‖M. | (3.24) |
Taking limit n→∞ in Eq (3.24) and using Eqs (3.17), (3.19) and (3.23), we obtain
limn→∞‖κn−φn‖M=0. | (3.25) |
Again using Eq (3.1) and Lemma (2.1), we deduce that
‖ζn+1−ζ∗‖2M=‖γn(h(ζn)−ζ∗)+(1−γn−δn)(JA,Bλ,M(κn)−ζ∗)+δn(Snκn−ζ∗)‖2M≤γn‖h(ζn)−ζ∗‖2M+(1−γn−δn)‖JA,Bλ,M(κn)−ζ∗‖2M+δn‖Snκn−ζ∗‖2M−(1−γn−δn)δn‖JA,Bλ,M(κn)−Snκn‖2M. |
Using nonexpansiveness of JA,Bλ,M, quasi-nonexpansiveness of Sn, Eqs (3.7)–(3.9), we have
(1−γn−δn)δn‖JA,Bλ,M(κn)−Snκn‖2M≤γn‖h(ζn)−ζ∗‖2M+‖ζn−ζ∗‖2M+2ϵn‖ζn−ζ∗‖M‖ζn−ζn−1‖M+ϵ2n‖ζn−ζn−1‖2M−‖ζn+1−ζ∗‖2M. | (3.26) |
Taking limit n→∞ in Eq (3.26) and using Eq (3.16), condition limn→∞γn=0, we obtain
limn→∞‖JA,Bλ,M(κn)−Snκn‖M=0. | (3.27) |
By using triangle inequality, we have
‖JA,Bλ,M(κn)−κn‖M=‖JA,Bλ,M(κn)−JA,Bλ,M(φn)+JA,Bλ,M(φn)−φn+φn−κn‖M≤‖JA,Bλ,M(κn)−JA,Bλ,M(φn)‖M+‖JA,Bλ,M(φn)−φn‖+‖φn−κn‖M≤‖κn−φn‖M+‖JA,Bλ,M(φn)−φn‖M+‖φn−κn‖M. | (3.28) |
Taking limit n→∞ in Eq (3.28) and using Eqs (3.17) and (3.25), we get
limn→∞‖JA,Bλ,M(κn)−κn‖M=0. | (3.29) |
Also,
‖Snκn−κn‖M=‖Snκn−JA,Bλ,M(κn)+JA,Bλ,M(κn)−κn‖M≤‖Snκn−JA,Bλ,M(κn)‖M+‖JA,Bλ,M(κn)−κn‖M. | (3.30) |
Taking limit n→∞ in Eq (3.30) and using Eqs (3.27) and (3.29), we obtain
limn→∞‖Snκn−κn‖M=0. | (3.31) |
From Eqs (3.19) and (3.25) and using triangle inequality, we obtain
limn→∞‖κn−υn‖M=0. | (3.32) |
From Eq (3.1),
‖φn−ζn‖M=‖ζn+ϵn(ζn−ζn−1)−ζn‖M=‖ϵn(ζn−ζn−1)‖M. | (3.33) |
Taking limit n→∞ in Eq (3.33) and using Eq (3.16), we get
limn→∞‖φn−ζn‖M=0. | (3.34) |
Similarly, we can prove
limn→∞‖κn−ζn‖M=0. | (3.35) |
Now, from Eqs (3.31) and (3.35), we can prove
‖Snκn−ζn‖M=‖Snκn−κn+κn−ζn‖M≤‖Snκn−κn‖M+‖κn−ζn‖M, |
which implies
limn→∞‖Snκn−ζn‖M=0. | (3.36) |
Again consider
‖ζn+1−ζn‖M=‖ζn+1−Snκn+Snκn−ζn‖M≤‖ζn+1−Snκn‖M+‖Snκn−ζn‖M=‖γnh(ζn)+(1−γn−δn)JA,Bλ,M(κn)+δnSnκn−Snκn‖M+‖Snκn−ζn‖M=‖(JA,Bλ,M(κn)−Snκn)+γn(h(ζn)−JA,Bλ,M(κn))+δn(Snκn−JA,Bλ,M(κn))‖M+‖Snκn−ζn‖M. | (3.37) |
Taking limit n→∞ in Eq (3.37) and using the condition limn→∞γn=0, Eqs (3.27) and (3.36), we obtain
limn→∞‖ζn+1−ζn‖M=0. | (3.38) |
Also, using Eqs (3.34) and (3.38), we have
limn→∞‖ζn+1−φn‖M=0. | (3.39) |
From Eqs (3.17) and (3.34), we can prove
‖JA,Bλ,M(ζn)−ζn‖M=‖JA,Bλ,M(ζn)−JA,Bλ,M(φn)−(ζn−φn)+(JA,Bλ,M(φn)−φn)‖M≤2‖ζn−φn‖M+‖JA,Bλ,M(φn)−φn‖M. |
Taking limit n→∞ in above equation, we have
limn→∞‖JA,Bλ,M(ζn)−ζn‖M=0. | (3.40) |
Since, the sequence {ζn} is a bounded sequence of real numbers, so using Lemma (3.3), we obtain
lim supn→∞⟨h(ζ∗)−ζ∗,ζn+1−ζ∗⟩M≤0. | (3.41) |
Thus, by making use of Eq (3.16), we have lim supn→∞tn≤0. Hence from Lemma (2.6), ζn→ζ∗ as n→∞.
Case 2. There exists a subsequence {‖ζnj−ζ∗‖2M} of {‖ζn−ζ∗‖2M} such that ‖ζnj−ζ∗‖2M≤‖ζnj+1−ζ∗‖2M for any j∈N. By using Lemma (2.7), we see that there exists a nondecreasing sequence {ζnk} of N such that limk→∞nk=∞ and the following inequalities hold for all k∈N.
‖ζnk−ζ∗‖2M≤‖ζnk+1−ζ∗‖2M, | (3.42) |
and
‖ζk−ζ∗‖2M≤‖ζnk+1−ζ∗‖2M. | (3.43) |
Similar to Eq (3.15), we have
αnk(1−αnk)‖JA,Bλ,M(φnk)−φnk‖2M≤γnk‖h(ζnk)−ζ∗‖2M+‖ζnk−ζ∗‖2M+2ϵnk‖ζnk−ζ∗‖M‖ζnk−ζnk−1‖M+ϵ2nk‖ζnk−ζnk−1‖2M−‖ζnk+1−ζ∗‖2M. | (3.44) |
Using Eqs (3.16), (3.42) and condition limk→∞γnk=0 in Eq (3.44), we obtain
limk→∞‖JA,Bλ,M(φnk)−φnk‖M=0. | (3.45) |
Similarly as in Case 1, we have
limk→∞‖υnk−φnk‖M=0,limk→∞‖φnk−κnk‖M=0,limk→∞‖ζnk+1−φnk‖M=0,limk→∞‖ζnk+1−ζnk‖M=0,limk→∞‖JA,Bλ,M(ζnk)−ζnk‖M=0. |
Using Eq (3.12), we have
snk+1≤(1−pnk)snk+pnktnk. | (3.46) |
Using Eq (3.42) in Eq (3.46), we obtain
ξnksnk≤ξnktnk. | (3.47) |
As pnk>0, so Eq (3.47) implies snk≤tnk, that is, we have
‖ζnk+1−ζ∗‖2M≤21−k⟨h(ζ∗)−ζ∗,ζnk+1−ζ∗⟩+2ϵnkγnk(1−k)‖ζnk−ζ∗‖M‖ζnk−ζnk−1‖M+ϵ2nkγnk(1−k)‖ζnk−ζnk−1‖2M. | (3.48) |
Similar to Case 1, we obtain lim supn→∞tnk≤0. So, we have by using Lemma (2.6),
limk→∞‖ζnk+1−ζ∗‖2M=0. | (3.49) |
From Eq (3.43), we get ζk→ζ∗ as k→∞. This completes the proof.
In this section, we provide a numerical example to show the numerical efficiency of the proposed algorithm. We performed all numerical experiments on a Dell computer equipped with a 3.20 GHz Intel (R) Core (TM) i5-3470 CPU and 8 GB of memory. The MATLAB R 2022 a platform was used as the implementation environment.
Let H=R3. Define the operators h,A,B,M,Si:H→H as h(ζ)=ζ2,A(ζ)=(5ζ1,4ζ2,5ζ3),B=(4ζ1,4ζ2,4ζ3),M=A,Siζ=−3ζ for i=0,1. for all ζ=(ζ1,ζ2,ζ3)∈H. Clearly, M is self-adjoint, bounded and positive operator, A is M-cocoercive operator, B is maximally monotone operator, h is 12-contraction operator with respect to M-norm and Si is a finite family of 12-demimetric operators. To find the numerical values of ζn, we choose αn=12n,λ=0.5,qn=0.1,βn=12n,γn=1n,δn=n2(n+1),ϵn=0.8. We compare our algorithm (3.1) with algorithms (1.4) and (1.5) with different choices of ζ0 and ζ1. We consider three cases:
Case 1. ζ0=[20,20,20] and ζ1=[10,10,10].
Case 2. ζ0=[10,10,10] and ζ1=[5,5,5].
Case 3. ζ0=[−1,−1,−1] and ζ1=[−2,−2,−2].
The values of iteration number (n) for various choices of ζ0 and ζ1 are given in the Tables 1–3. Comparison of Algorithm (3.1) by taking En=‖ζn−ζn−1‖<10−4 with the algorithms Dixit et al. (1.4) and Lorenz and Pock (1.5) for Cases 1–3 are shown in Figures 1–3 respectively.
Case 1 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 11 | 0.015498 |
Dixit et al. (1.4) | 75 | 0.699281 |
Lorenz and Pock (1.5) | 15 | 0.019471 |
Case 2 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 10 | 0.001373 |
Dixit et al. (1.4) | 75 | 0.003053 |
Lorenz and Pock (1.5) | 14 | 0.001431 |
Case 3 | Iteration number(n) | cpu time (in seconds) |
Algorithm(3.1) | 9 | 0.001283 |
Dixit et al. (1.4) | 75 | 0.002931 |
Lorenz and Pock (1.5) | 13 | 0.001368 |
We introduced an algorithm for finding the common solution of monotone inclusion and fixed point problem of a finite family of demimetric mappings in a real Hilbert space and showed that our algorithm has strong convergence under some conditions. Moreover, we proved the algorithm has a better rate of convergence by giving a numerical example.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
The author Anjali delightedly acknowledge the University Grants Commission (UGC), New Delhi and the authors S. Haque and N. Mlaiki would like to thank the Prince Sultan University for paying the publication fees for this work through TAS LAB.
The authors declare no conflicts of interest.
[1] |
A. B. Abubakar, K. Muangchoo, A. H. Ibrahim, A. B. Muhammad, L. O. Jolaoso, K. O. Aremu, A new three-term Hestenes-Stiefel type method for nonlinear monotone operator equations and image restoration, IEEE Access, 9 (2021), 18262–18277. https://doi.org/10.1109/ACCESS.2021.3053141 doi: 10.1109/ACCESS.2021.3053141
![]() |
[2] |
P. L. Combettes, V. R. Wajs, Signal recovery by proximal forward-backward splitting, Multiscale Model. Sim., 4 (2005), 1168–1200. https://doi.org/10.1137/050626090 doi: 10.1137/050626090
![]() |
[3] |
V. Dadashi, M. Postolache, Forward–backward splitting algorithm for fixed point problems and zeros of the sum of monotone operators, Arab. J. Math., 9 (2020), 89–99. https://doi.org/10.1007/s40065-018-0236-2 doi: 10.1007/s40065-018-0236-2
![]() |
[4] |
A. Dixit, D. R. Sahu, P. Gautam, T. Som, J. C. Yao, An accelerated forward-backward splitting algorithm for solving inclusion problems with applications to regression and link prediction problems, J. Nonlinear Var. Anal., 5 (2021), 79–101. https://doi.org/10.23952/jnva.5.2021.1.06 doi: 10.23952/jnva.5.2021.1.06
![]() |
[5] |
A. Hanjing, S. Suantai, A fast image restoration algorithm based on a fixed point and optimization method, Mathematics, 8 (2020), 378. https://doi.org/10.3390/math8030378 doi: 10.3390/math8030378
![]() |
[6] | K. Janngam, S. Suantai, An accelerated forward-backward algorithm with applications to image restoration problems, Thai J. Math., 19 (2021), 325–339. |
[7] | D. Kitkuan, K. Muangchoo, Inertial relaxed CQ algorithm with an application to signal processing, Thai J. Math., 18 (2020), 1091–1103. |
[8] |
D. Kitkuan, P. Kumam, J. Martínez-Moreno, Generalized Halpern-type forward–backward splitting methods for convex minimization problems with application to image restoration problems, Optimization, 69 (2020), 1557–1581. 10.1080/02331934.2019.1646742 doi: 10.1080/02331934.2019.1646742
![]() |
[9] |
G. López, V. Martín-Márquez, F. Wang, H. K. Xu, Forward-backward splitting methods for accretive operators in Banach spaces, Abstract Appl. Anal., 2012 (2012), 109236. https://doi.org/10.1155/2012/109236 doi: 10.1155/2012/109236
![]() |
[10] |
D. A. Lorenz, T. Pock, An inertial forward-backward algorithm for monotone inclusions, J. Math. Imaging Vis., 51 (2015), 311–325. https://doi.org/10.1007/s10851-014-0523-2 doi: 10.1007/s10851-014-0523-2
![]() |
[11] |
D. R. Sahu, A. Pitea, M. Verma, A new iteration technique for nonlinear operators as concerns convex programming and feasibility problems, Numer. Algor., 83 (2020), 421–449. https://doi.org/10.1007/s11075-019-00688-9 doi: 10.1007/s11075-019-00688-9
![]() |
[12] |
K. Sitthithakerngkiet, J. Deepho, P. Kumam, A hybrid viscosity algorithm via modify the hybrid steepest descent method for solving the split variational inclusion in image reconstruction and fixed point problems, Appl. Math. Comput., 250 (2015), 986–1001. https://doi.org/10.1016/j.amc.2014.10.130 doi: 10.1016/j.amc.2014.10.130
![]() |
[13] | S. Sra, S. Nowozin, S. J. Wright, Optimization for machine learning, 2012. |
[14] |
P. Sunthrayuth, P. Cholamjiak, Iterative methods for solving quasi-variational inclusion and fixed point problem in q-uniformly smooth Banach spaces, Numer. Algor., 78 (2018), 1019–1044. https://doi.org/10.1007/s11075-017-0411-0 doi: 10.1007/s11075-017-0411-0
![]() |
[15] |
P. Tseng, A modified forward-backward splitting method for maximal monotone mappings, SIAM J. Control Optim., 38 (2000), 431–446. https://doi.org/10.1137/S0363012998338806 doi: 10.1137/S0363012998338806
![]() |
[16] | G. I. Usurelu, M. Postolache, Algorithm for generalized hybrid operators with numerical analysis and applications, J. Nonlinear Var. Anal., 6 (2022), 255–277. |
[17] |
Y. Dong, X. Zhu, An inertial splitting method for monotone inclusions of three operators, Int. J. Math. Stat. Oper. Res., 2 (2022), 43–60. https://DOI:10.47509/IJMSOR.2022.v02i01.04 doi: 10.47509/IJMSOR.2022.v02i01.04
![]() |
[18] |
P. L. Lions, B. Mercier, Splitting algorithms for the sum of two nonlinear operators, SIAM J. Numer. Anal., 16 (1979), 964–979. https://doi.org/10.1137/0716071 doi: 10.1137/0716071
![]() |
[19] |
A. Gibali, D. V.Thong, Tseng type method for solving inclusion problems and its applications, Calcolo, 55 (2018), 49. https://doi.org/10.1007/s10092-018-0292-1 doi: 10.1007/s10092-018-0292-1
![]() |
[20] |
A. Moudafi, M. Oliny, Convergence of a splitting inertial proximal method for monotone operators, J. Comput. Appl. Math., 155 (2003), 447–454. https://doi.org/10.1016/S0377-0427(02)00906-8 doi: 10.1016/S0377-0427(02)00906-8
![]() |
[21] |
E. Altiparmak, I. Karahan, A new preconditioning algorithm for finding a zero of the sum of two monotone operators and its application to image restoration problems, Int. J. Comput. Math., 99 (2022), 2482–2498. https://doi.org/10.1080/00207160.2022.2068146 doi: 10.1080/00207160.2022.2068146
![]() |
[22] | E. Altiparmak, I. Karahan, A Modified Preconditioning Algorithm for solving monotone inclusion problem and application to image restoration problem, U. Politeh. Buch. Ser. A, 84 (2022), 81–92. |
[23] |
W. Shatanawi, T. A. Shatnawi, New fixed point results in controlled metric type spaces based on new contractive conditions, AIMS Math., 8 (2023), 9314–9330. https://doi.org/10.3934/math.2023468 doi: 10.3934/math.2023468
![]() |
[24] |
A. Z. Rezazgui, A. A. Tallafha, W. Shatanawi, Common fixed point results via Aν-α-contractions with a pair and two pairs of self-mappings in the frame of an extended quasi b-metric space, AIMS Math., 8 (2023), 7225–7241. https://doi.org/10.3934/math.2023363 doi: 10.3934/math.2023363
![]() |
[25] |
M. Joshi, A. Tomar, T. Abdeljawad, On fixed points, their geometry and application to satellite web coupling problem in S-metric spaces, AIMS Math., 8 (2023), 4407–4441. https://doi.org/10.3934/math.2023220 doi: 10.3934/math.2023220
![]() |
[26] |
P. Debnath, Z. D. Mitrović, S. Y. Cho, Common fixed points of Kannan, Chatterjea and Reich type pairs of self-maps in a complete metric space, São Paulo J. Math. Sci., 15 (2021), 383–391. https://doi.org/10.1007/s40863-020-00196-y doi: 10.1007/s40863-020-00196-y
![]() |
[27] |
P. Debnath, Banach, Kannan, Chatterjea and Reich‐type contractive inequalities for multivalued mappings and their common fixed points, Math. Method. Appl. Sci., 45 (2022), 1587–1596. https://doi.org/10.1002/mma.7875 doi: 10.1002/mma.7875
![]() |
[28] |
P. Debnath, New common fixed point theorems for Gornicki-type mappings and enriched contractions, São Paulo J. Math. Sci., 16 (2022), 1401–1408. https://doi.org/10.1007/s40863-022-00283-2 doi: 10.1007/s40863-022-00283-2
![]() |
[29] |
B. T. Polyak, Some methods of speeding up the convergence of iteration methods, USSR Comp. Math. Math. Phys., 4 (1964), 1–17. https://doi.org/10.1016/0041-5553(64)90137-5 doi: 10.1016/0041-5553(64)90137-5
![]() |
[30] | H. H. Bauschke, P. L. Combettes, Convex analysis and monotone operator theory in Hilbert spaces, New York: Springer, 2011. |
[31] | A. Cegielski, Iterative methods for fixed point problems in Hilbert spaces, Springer, 2012. |
[32] | B. V. Limaye, Functional Analysis, New Age International, 1996. |
[33] | K. Goebel, W. A. Kirk, Topics in metric fixed point theory, Cambridge University Press, 1990. https://doi.org/10.1017/CBO9780511526152 |
[34] |
S. Takahashi, W. Takahashi, The split common fixed point problem and the shrinking projection method in Banach spaces, Optimization, 65 (2016), 281–287. https://doi.org/10.1080/02331934.2015.1020943 doi: 10.1080/02331934.2015.1020943
![]() |
[35] |
H. K. Xu, Iterative algorithms for nonlinear operators, J. London Math. Soc., 66 (2002), 240–256. https://doi.org/10.1112/S0024610702003332 doi: 10.1112/S0024610702003332
![]() |
[36] |
P. E. Maingé, A hybrid extra gradient-viscosity method for monotone operators and fixed point problems, SIAM J. Control Optim., 47 (2008), 1499–1515. https://doi.org/10.1137/060675319 doi: 10.1137/060675319
![]() |
[37] |
W. Takahashi, C. F. Wen, J. C. Yao, The shrinking projection method for a finite family of demimetric mappings with variational inequality problems in a Hilbert space, Fixed Point Theor., 19 (2018), 407-419. 10.24193/fpt-ro.2018.1.32 doi: 10.24193/fpt-ro.2018.1.32
![]() |
1. | Seema Mehra, Renu Chugh, Dania Santina, Nabil Mlaiki, An iterative approach for addressing monotone inclusion and fixed point problems with generalized demimetric mappings, 2024, 12, 26668181, 100953, 10.1016/j.padiff.2024.100953 | |
2. | Renu Chugh, Charu Batra, Fixed point theorems of enriched Ciric's type and enriched Hardy-Rogers contractions, 2023, 0, 2155-3289, 0, 10.3934/naco.2023022 |
Case 1 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 11 | 0.015498 |
Dixit et al. (1.4) | 75 | 0.699281 |
Lorenz and Pock (1.5) | 15 | 0.019471 |
Case 2 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 10 | 0.001373 |
Dixit et al. (1.4) | 75 | 0.003053 |
Lorenz and Pock (1.5) | 14 | 0.001431 |
Case 3 | Iteration number(n) | cpu time (in seconds) |
Algorithm(3.1) | 9 | 0.001283 |
Dixit et al. (1.4) | 75 | 0.002931 |
Lorenz and Pock (1.5) | 13 | 0.001368 |
Case 1 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 11 | 0.015498 |
Dixit et al. (1.4) | 75 | 0.699281 |
Lorenz and Pock (1.5) | 15 | 0.019471 |
Case 2 | Iteration number(n) | cpu time (in seconds) |
Algorithm (3.1) | 10 | 0.001373 |
Dixit et al. (1.4) | 75 | 0.003053 |
Lorenz and Pock (1.5) | 14 | 0.001431 |
Case 3 | Iteration number(n) | cpu time (in seconds) |
Algorithm(3.1) | 9 | 0.001283 |
Dixit et al. (1.4) | 75 | 0.002931 |
Lorenz and Pock (1.5) | 13 | 0.001368 |