In this research, we studied modified inertial composite subgradient extragradient implicit rules for finding solutions of a system of generalized equilibrium problems with a common fixed-point problem and pseudomonotone variational inequality constraints. The suggested methods consisted of an inertial iterative algorithm, a hybrid deepest-descent technique, and a subgradient extragradient method. We proved that the constructed algorithms converge to a solution of the considered problem, which also solved some hierarchical variational inequality.
Citation: Lu-Chuan Ceng, Shih-Hsin Chen, Yeong-Cheng Liou, Tzu-Chien Yin. Modified inertial subgradient extragradient algorithms for generalized equilibria systems with constraints of variational inequalities and fixed points[J]. AIMS Mathematics, 2024, 9(6): 13819-13842. doi: 10.3934/math.2024672
[1] | Lu-Chuan Ceng, Li-Jun Zhu, Tzu-Chien Yin . Modified subgradient extragradient algorithms for systems of generalized equilibria with constraints. AIMS Mathematics, 2023, 8(2): 2961-2994. doi: 10.3934/math.2023154 |
[2] | Francis Akutsah, Akindele Adebayo Mebawondu, Austine Efut Ofem, Reny George, Hossam A. Nabwey, Ojen Kumar Narain . Modified mildly inertial subgradient extragradient method for solving pseudomonotone equilibrium problems and nonexpansive fixed point problems. AIMS Mathematics, 2024, 9(7): 17276-17290. doi: 10.3934/math.2024839 |
[3] | Fei Ma, Jun Yang, Min Yin . A strong convergence theorem for solving pseudo-monotone variational inequalities and fixed point problems using subgradient extragradient method in Banach spaces. AIMS Mathematics, 2022, 7(4): 5015-5028. doi: 10.3934/math.2022279 |
[4] | Rose Maluleka, Godwin Chidi Ugwunnadi, Maggie Aphane . Inertial subgradient extragradient with projection method for solving variational inequality and fixed point problems. AIMS Mathematics, 2023, 8(12): 30102-30119. doi: 10.3934/math.20231539 |
[5] | Yuanheng Wang, Chenjing Wu, Yekini Shehu, Bin Huang . Self adaptive alternated inertial algorithm for solving variational inequality and fixed point problems. AIMS Mathematics, 2024, 9(4): 9705-9720. doi: 10.3934/math.2024475 |
[6] | Habib ur Rehman, Poom Kumam, Kanokwan Sitthithakerngkiet . Viscosity-type method for solving pseudomonotone equilibrium problems in a real Hilbert space with applications. AIMS Mathematics, 2021, 6(2): 1538-1560. doi: 10.3934/math.2021093 |
[7] | Habib ur Rehman, Wiyada Kumam, Poom Kumam, Meshal Shutaywi . A new weak convergence non-monotonic self-adaptive iterative scheme for solving equilibrium problems. AIMS Mathematics, 2021, 6(6): 5612-5638. doi: 10.3934/math.2021332 |
[8] | Ziqi Zhu, Kaiye Zheng, Shenghua Wang . A new double inertial subgradient extragradient method for solving a non-monotone variational inequality problem in Hilbert space. AIMS Mathematics, 2024, 9(8): 20956-20975. doi: 10.3934/math.20241020 |
[9] | Yasir Arfat, Muhammad Aqeel Ahmad Khan, Poom Kumam, Wiyada Kumam, Kanokwan Sitthithakerngkiet . Iterative solutions via some variants of extragradient approximants in Hilbert spaces. AIMS Mathematics, 2022, 7(8): 13910-13926. doi: 10.3934/math.2022768 |
[10] | Wenlong Sun, Gang Lu, Yuanfeng Jin, Zufeng Peng . Strong convergence theorems for split variational inequality problems in Hilbert spaces. AIMS Mathematics, 2023, 8(11): 27291-27308. doi: 10.3934/math.20231396 |
In this research, we studied modified inertial composite subgradient extragradient implicit rules for finding solutions of a system of generalized equilibrium problems with a common fixed-point problem and pseudomonotone variational inequality constraints. The suggested methods consisted of an inertial iterative algorithm, a hybrid deepest-descent technique, and a subgradient extragradient method. We proved that the constructed algorithms converge to a solution of the considered problem, which also solved some hierarchical variational inequality.
Throughout, assume H is a real Hilbert space with the inner product ⟨⋅,⋅⟩ and the norm ‖⋅‖. Assume ∅≠C⊂H is a closed and convex set. Let S:C→H be an operator and Θ:C×C→R be a bifunction. Use Fix(S) to mean the fixed-point set of S.
Recall that the equilibrium problem (EP) is to search an equilibrium point in EP(Θ), where
EP(Θ)={x∈C:Θ(x,y)≥0,∀y∈C}. |
Under the theory framework of equilibrium problems, there exists a unified way for exploring a broad number of problems originating in structural analysis, transportation, physics, optimization, finance, and economics [1,8,10,17,20,24,25,26,35]. In order to find an element in EP(Θ), one needs to make the hypotheses below:
(H1) Θ(v,v)=0,∀v∈C;
(H2) Θ(w,v)+Θ(v,w)≤0,∀v,w∈C;
(H3) limλ→0+Θ((1−λ)v+λu,w)≤Θ(v,w),∀u,v,w∈C;
(H4) For every v∈C, Θ(v,⋅) is convex and lower semicontinuous (l.s.c.).
In order to solve the equilibrium problems, in 1994, Blum and Oettli [1] obtained the following valuable lemma:
Lemma 1.1. [1] Assume that Θ:C×C→R fulfills the hypotheses (H1)–(H4). If ∀x∈H and ℓ>0, let TΘℓ:H→C be an operator formulated below:
TΘℓ(x):={y∈C:Θ(y,z)+1ℓ⟨z−y,y−x⟩≥0,∀z∈C}. |
Then, (i) TΘℓ is single-valued and satisfies ‖TΘℓv−TΘℓw‖2≤⟨TΘℓv−TΘℓw,v−w⟩,∀v,w∈H; and (ii) Fix(TΘℓ)=EP(Θ), and EP(Θ) is convex and closed.
In particular, in case of Θ(x,y)=⟨Ax,y−x⟩,∀x,y∈C, and the EP reduces to the classical variational ineqiality problem (VIP) of seeking x∈C such that
⟨Ax,y−x⟩≥0,∀y∈C. |
The solution set of the VIP is denoted by VI(C,A).
An effective approach for settling EP and VIP is the Korpelevich's algorithm [15]. The Korpelevich extragradient technique has been adapted and applied extensively; see e.g., the modified extragradient method [11,29,34], subgradient extragradient method [3,13,16,28,31,32], relaxed extragradient method [7], Tseng-type method [22,23,33], inertial extragradient method [14,27], and so on.
In 2010, Ceng and Yao [6] investigated the system generalized equilibrium problems (SGEP) of finding (x,y)∈C×C satisfying
{Θ1(x,u)+⟨B1y,u−x⟩+1α1⟨x−y,u−x⟩≥0,∀u∈C,Θ2(y,v)+⟨B2x,v−y⟩+1α2⟨y−x,v−y⟩≥0,∀v∈C, | (1.1) |
where B1,B2:H→H are two nonlinear operators, Θ1,Θ2:C×C→R are two bifunctions, and α1,α2>0 are two constants.
If Θ1=Θ2=0, then the SGEP comes down to the generallized variational inequalities considered in [5]: Find (x,y)∈C×C satisfying
{⟨α1B1y+x−y,u−x⟩≥0,∀u∈C,⟨α2B2x+y−x,v−y⟩≥0,∀v∈C, |
with constants α1,α2>0.
To solve problem (1.1), the authors in [6] used a fixed point technique. In fact, the SGEP (1.1) can be transformed into the fixed-point problem.
Lemma 1.2. [6] Suppose that the bifunctions Θ1,Θ2:C×C→R satisfy the hypotheses (H1)–(H4) and B1,B2:H→H are ρ-ism and σ-ism, respectively. Then, (u∗,v∗)∈C×C is a solution of SGEP (1.1) if and only if u†∈Fix(G), where G:=TΘ1α1(I−α1B1)TΘ2α2(I−α2B2) and v∗=TΘ2α2(I−α2B2)u∗ in which α1∈(0,2ρ) and α2∈(0,2σ).
On the other hand, in 2018, Cai, Shehu, and Iyiola [2] proposed the modified viscosity implicit rule for solving EP and a fixed-point problem: for x1∈C, let {xk} be the sequence constructed below:
{uk=σkxk+(1−σk)yk,vk=PC(uk−α2B2uk),yk=PC(vk−α1B1vk),xk+1=PC[ρkf(xk)+(I−ρkαF)Skyk],∀k≥1. |
Under suitable conditions, Cai, Shehu, and Iyiola [2] proved xk→u∗∈Fix(S)∩Fix(G), which solves the hierarchical variational inequality (HVI):
⟨(αF−f)u,v−u⟩≥0,∀v∈Fix(S)∩Fix(G). |
Moreover, Ceng and Shang [4] suggested an algorithm for solving the common fixed-point problem (CFPP) of finite nonexpansive mappings {Sr}Nr=1, an asymptotically nonexpansive mapping S and VIP.
Algorithm 1.1. [4] Let x1,x0∈H be arbitrary. Let γ>0,ℓ∈(0,1),ν∈(0,1), and xk be known. Calculate xk+1 via the following iterative steps:
Step 1. Set pk=Skxk+εk(Skxk−Skxk−1) and calculate yk=PC(pk−ζkApk), where ζk is the largest ζ∈{γ,γℓ,γℓ2,...} fulfilling ζ‖Apk−Ayk‖≤ν‖pk−yk‖.
Step 2. Compute zk=PCk(pk−ζkAyk) where Ck:={y∈H:⟨pk−ζkApk−yk,yk−y⟩≥0}.
Step 3. Compute xk+1=ρkf(xk)+σkxk+((1−σk)I−ρkαF)Skzk.
Let k:=k+1 and return to Step 1.
Motivated and inspired by the work in the literature, the main purpose of this article was to design two modified inertial composite subgradient extragradient implicit rules for solving the SGEP with the VIP and CFPP constraints. The suggested algorithms consisted of the subgradient extragradient rule, inertial iteration approach, and hybrid deepest-descent technique. We proved that the proposed algorithms converge to a solution of the SGEP with the VIP and CFPP constraints, which also solved some HVI.
Let C be a nonempty, convex, and closed subset of a real Hilbert space H. For all v,w∈C, an operator T:C→H is called
● asymptotically nonexpansive if ∃{ϖm}∞m=1⊂[0,+∞) satisfying ϖm→0(m→∞) and
‖Tmv−Tmw‖≤ϖm‖v−w‖+‖v−w‖,∀m≥1. |
In particular, in the case of ϖm=0,∀m≥1, and T is known as being nonexpansive.
● α-Lipschitzian if ∃α>0 such that ‖Tv−Tw‖≤α‖v−w‖;
● monotone if ⟨Tv−Tw,v−w⟩≥0;
● strongly monotone if there is ρ>0 such that ⟨Tv−Tw,v−w⟩≥ρ‖v−w‖2;
● pseudomonotone if ⟨Tv,w−v⟩≥0⇒⟨Tw,w−v⟩≥0;
● σ inverse strongly monotone (σ-ism) if there is σ>0 such that ⟨Tv−Tw,v−w⟩≥σ‖Tv−Tw‖2;
● sequentially weakly continuous if ∀{vl}⊂C, vl⇀v⇒Tvl⇀Tv.
Recall that the metric (or nearest point) projection from H onto C is the mapping PC:H→C which assigns to each point x∈C the unique point PC(x)∈C satisfying the property
‖x−PC(x)‖=infy∈C‖x−y‖. |
The following results are well-known ([12]):
(a) ‖PC(y)−PC(z)‖2≤⟨y−z,PC(y)−PC(z)⟩,∀y,z∈H;
(b) z=PC(y)⇔⟨x−z,y−z⟩≤0,∀y∈H,x∈C;
(c) ‖y−z‖2≥‖z−PC(y)‖2+‖y−PC(y)‖2,∀y∈H,z∈C;
(d) ‖y−z‖2=‖y‖2−2⟨y−z,z⟩−‖z‖2,∀y,z∈H;
(e) ‖ty+(1−t)x‖2=t‖y‖2+(1−t)‖x‖2−t(1−t)‖y−x‖2, ∀x,y∈H,t∈[0,1].
Lemma 2.1. [6] Suppose B:H→H is an η-ism. Then,
‖(I−αB)y−(I−αB)z‖2≤‖y−z‖2−α(2η−α)‖By−Bz‖2,∀y,z∈H,∀α≥0. |
When 0≤α≤2η, we have that I−αB is nonexpansive.
Lemma 2.2. [6] Let B1,B2:H→H be ρ-ism and σ-ism, respectively. Suppose that the bifunctions Θ1,Θ2:C×C→R satisfy the hypotheses (H1)–(H4). Then, G:=TΘ1α1(I−α1B1)TΘ2α2(I−α2B2) is nonexpansive when 0<α1≤2ρ and 0<α2≤2σ.
In particular, if Θ1=Θ2=0, using Lemma 1.1, we deduce that TΘ1α1=TΘ2α2=PC.
Corollary 2.1. [5] Let B1:H→H be ρ-ism and B2:H→H σ-ism. Define an operator G:H→C by G:=PC(I−α1B1)PC(I−α2B2). Then G is nonexpansive when 0<α1≤2ρ and 0<α2≤2σ.
Lemma 2.3. [9] If the operator A:C→H is continuous pseudomonotone, then v∈VI(C,A) if and only if ⟨Aw,w−v⟩≥0,∀w∈C.
Lemma 2.4. [30] Suppose {al}⊂[0,∞) s.t. al+1≤(1−ωl)al+ωlνl,∀l≥1, where {ωl} and {νl} satisfy: (i) {ωl}⊂[0,1]; (ii) ∑∞l=1ωl=∞, and (iii) lim supl→∞νl≤0 or ∑∞l=1|ωlνl|<∞. Then, we have liml→∞al=0.
Lemma 2.5. [18] Let X be a Banach space with a weakly continuous duality mapping. Let C be a nonempty closed convex subset of X and T:C→C an asymptotically nonexpansive mapping such that Fix(T)≠∅. Then I−T is demiclosed at zero.
Lemma 2.6. [19] Suppose that the real number sequence {Γm} is not decreasing at infinity: ∃{Γmk}⊂{Γm} s.t. Γmk<Γmk+1,∀k≥1. Let {ϕ(m)}m≥m0 be an integer sequence defined by
ϕ(m)=max{k≤m:Γk<Γk+1}. |
Then,
(i) ϕ(m0)≤ϕ(m0+1)≤⋯ and ϕ(m)→∞ as m→∞;
(ii) For all m≥m0, Γϕ(m)≤Γϕ(m)+1 and Γm≤Γϕ(m)+1.
Lemma 2.7. [30] Let λ∈(0,1], S:C→C be a nonexpansive operator, and F:C→H be a κ-Lipschitzian and η-strongly monotone operator. Set Sλv:=(I−λαF)Sv,∀v∈C. If 0<α<2ηκ2, then ‖Sλv−Sλw‖≤(1−λτ)‖v−w‖,∀v,w∈C, where τ=1−√1−α(2η−ακ2)∈(0,1].
Let the operator Sr be nonexpansive on H for all r=1,...,N and S:H→H be a ϖn-asymptotically nonexpansive operator. Let A:H→H be an L-Lipschitz pseudomonotone operator satisfying ‖Ax‖≤lim infn→∞‖Axn‖ when xn⇀x. Let Θ1,Θ2:C×C→R be two bifunctions fulfilling the hypotheses (H1)–(H4). Let B1:H→H be ρ-ism and B2:H→H be σ-ism. Let f:H→H be δ-contractive and F:H→H be κ-Lipschitz η-strongly monotone with δ<τ:=1−√1−α(2η−ακ2) for α∈(0,2ηκ2). Suppose that the sequences {εn}⊂[0,1],{ξn}⊂(0,1], and {ρn},{σn}⊂(0,1) satisfy
(ⅰ) limn→∞ρn=0 and ∑∞n=1ρn=∞;
(ⅱ) limn→∞ϖnρn=0 and supn≥1εnρn<∞;
(ⅲ) 0<lim infn→∞σn≤lim supn→∞σn<1;
(ⅳ) lim supn→∞ξn<1.
Let γ>0,ν∈(0,1),ℓ∈(0,1), α1∈(0,2ρ), and α2∈(0,2σ) be five constants. Set S0:=S and G:=TΘ1α1(I−α1B1)TΘ2α2(I−α2B2). Suppose that Δ:=⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A)≠∅.
Algorithm 3.1. Let x1,x0∈H be arbitrary. Let xn be known and compute xn+1 below:
Step 1. Set qn=Snxn+εn(Snxn−Snxn−1) and calculate
{pn=ξnqn+(1−ξn)un,vn=TΘ2α2(pn−α2B2pn),un=TΘ1α1(vn−α1B1vn). |
Step 2. Compute yn=PC(pn−ζnApn), where ζn is the largest ζ∈{γ,γℓ,γℓ2,...} s.t.
ζ‖Apn−Ayn‖≤ν‖pn−yn‖. | (3.1) |
Step 3. Compute tn=σnxn+(1−σn)zn with zn=PCn(pn−ζnAyn) and
Cn:={y∈H:⟨pn−ζnApn−yn,y−yn⟩≤0}. |
Step 4. Compute
xn+1=ρnf(xn)+(I−ρnαF)Sntn, | (3.2) |
where Sn is constructed as in Algorithm 1.1. Let n:=n+1 and return to Step 1.
Lemma 3.1. [21] min{γ,νℓ/L}≤ζn≤γ.
Lemma 3.2. Let p∈Δ and q=TΘ2α2(p−α2B2p). Then,
‖zn−p‖2≤‖qn−p‖2−α1(α1−2ρ)‖B1vn−B1q‖2−(1−ξn)[‖qn−pn‖2−(1−ν)[‖yn−zn‖2+‖yn−pn‖2]−α2(α2−2σ)‖B2pn−B2p‖2], |
where vn=TΘ2α2(pn−α2B2pn).
Proof. According to Lemma 2.2, there exists the unique point pn∈H satisfying pn=ξnqn+(1−ξn)Gpn. Since p∈Cn, we have
‖zn−p‖2≤⟨pn−ζnAyn−p,zn−p⟩=12(‖pn−p‖2−‖zn−pn‖2+‖zn−p‖2)−ζn⟨Ayn,zn−p⟩, |
which implies that
‖zn−p‖2≤‖pn−p‖2−‖zn−pn‖2−2ζn⟨Ayn,zn−p⟩. |
Noting that zn=PCn(pn−ζnAyn), we have ⟨pn−ζnApn−yn,zn−yn⟩≤0. Owing to the pseudomonotonicity of A, by (3.1), we get
‖zn−p‖2≤‖pn−p‖2−‖zn−pn‖2−2ζn⟨Ayn,yn−p+zn−yn⟩≤‖pn−p‖2−‖zn−pn‖2−2ζn⟨Ayn,zn−yn⟩=‖pn−p‖2+2⟨pn−‖yn−pn‖2−ζnAyn−yn,zn−yn⟩−‖zn−yn‖2=‖pn−p‖2−‖zn−yn‖2+2⟨pn−ζnApn−yn,zn−yn⟩−‖yn−pn‖2+2ζn⟨Apn−Ayn,zn−yn⟩≤‖pn−p‖2+2ν‖pn−yn‖‖zn−yn‖−‖zn−yn‖2−‖yn−pn‖2≤‖pn−p‖2−‖yn−pn‖2+ν(‖pn−yn‖2+‖zn−yn‖2)−‖zn−yn‖2=‖pn−p‖2−(1−ν)[‖yn−pn‖2+‖yn−zn‖2]. | (3.3) |
Observe that un=TΘ1α1(vn−α1B1vn), vn=TΘ2α2(pn−α2B2pn), and q=TΘ2α2(p−α2B2p). Then un=Gpn. Applying Lemma 2.1 to get
‖un−p‖2≤‖vn−q‖2+α1(α1−2ρ)‖B1vn−B1q‖2 |
and
‖vn−q‖2≤‖pn−p‖2+α2(α2−2σ)‖B2pn−B2p‖2. |
Then,
‖un−p‖2≤‖pn−p‖2+α1(α1−2ρ)‖B1vn−B1q‖2+α2(α2−2σ)‖B2pn−B2p‖2. |
Besides, thanks to pn=ξnqn+(1−ξn)un, we get ‖pn−p‖2≤ξn⟨qn−p,pn−p⟩+(1−ξn)‖pn−p‖2, which results in ‖pn−p‖2≤⟨qn−p,pn−p⟩=12[‖qn−p‖2+‖pn−p‖2−‖qn−pn‖2]. So,
‖pn−p‖2≤‖qn−p‖2−‖qn−pn‖2. | (3.4) |
Then,
‖pn−p‖2≤(1−ξn)‖un−p‖2+ξn‖qn−p‖2≤ξn‖qn−p‖2+(1−ξn)[‖pn−p‖2+α1(α1−2ρ)‖B1vn−B1q‖2+α2(α2−2σ)‖B2pn−B2p‖2]≤ξn‖qn−p‖2+(1−ξn)[‖qn−p‖2−‖qn−pn‖2+α2(α2−2σ)×‖B2pn−B2p‖2+α1(α1−2ρ)‖B1vn−B1q‖2]=‖qn−p‖2−(1−ξn)[‖qn−pn‖2−α1(α1−2ρ)‖B1vn−B1q‖2−α2(α2−2σ)‖B2pn−B2p‖2], |
which, together with (3.3), yields
‖zn−p‖2≤‖pn−p‖2−(1−ν)[‖yn−pn‖2+‖yn−zn‖2]≤‖qn−p‖2−(1−ξn)[‖qn−pn‖2−α1(α1−2ρ)‖B1vn−B1q‖2−α2(α2−2σ)‖B2pn−B2p‖2]−(1−ν)[‖yn−pn‖2+‖yn−zn‖2]. |
This ensures that the conclusion holds.
Lemma 3.3. Assume that
(i) the sequences {pn},{qn},{yn}, and {zn} are bounded;
(ii) limn→∞(xn+1−xn)=limn→∞(qn−zn)=limn→∞(xn−yn) =limn→∞(Sn+1xn−Snxn)=0.
Then ωw(xn)⊂Δ, where ωw(xn)={z∈H,andthereissome{xni}⊂{xn}suchthatxni⇀z}.
Proof. Take an arbitrary fixed z∈ωw({xn}). Then, there is some {ni}⊂{n} such that xni⇀z and yni⇀z∈H. Next, we show z∈Δ. Using Lemma 3.2, we deduce
(1−ξn)[‖qn−pn‖2−α1(α1−2ρ)‖B1vn−B1q‖2−α2(α2−2σ)‖B2pn−B2p‖2]+(1−ν)[‖yn−pn‖2+‖yn−zn‖2]≤‖qn−p‖2−‖zn−p‖2≤‖qn−zn‖(‖qn−p‖+‖zn−p‖). |
Because qn−zn→0,ν∈(0,1),α1∈(0,2ρ),α2∈(0,2σ), and 0<lim infn→∞(1−ξn), we deduce that
limn→∞‖B2pn−B2p‖=limn→∞‖B1vn−B1q‖=0, | (3.5) |
and
limn→∞‖yn−zn‖=limn→∞‖qn−pn‖=limn→∞‖yn−pn‖=0. |
Hence,
‖xn−qn‖≤‖xn−yn‖+‖yn−zn‖+‖zn−qn‖→0(n→∞), |
‖xn−pn‖≤‖xn−qn‖+‖qn−pn‖→0(n→∞), |
‖pn−zn‖≤‖pn−yn‖+‖yn−zn‖→0(n→∞), |
and
‖xn−zn‖≤‖xn−qn‖+‖qn−pn‖+‖pn−zn‖→0(n→∞). |
Note that
‖qn−Snxn‖=εn‖Snxn−Snxn−1‖≤(1+ϖn)‖xn−xn−1‖→0. |
Therefore,
‖xn−Snxn‖≤‖xn−qn‖+‖qn−Snxn‖→0(n→∞). |
Note that
‖xn−Sxn‖≤‖xn−Snxn‖+‖Sxn−Sn+1xn‖+‖Sn+1xn−Snxn‖≤‖xn−Snxn‖+‖Snxn−Sn+1xn‖+(1+ϖ1)‖Snxn−xn‖=(2+ϖ1)‖xn−Snxn‖+‖Snxn−Sn+1xn‖→0. | (3.6) |
Observe that
‖un−p‖2≤⟨vn−q,un−p⟩+α1⟨B1q−B1vn,un−p⟩≤12[‖vn−q‖2−‖vn−un+p−q‖2+‖un−p‖2]+α1‖B1q−B1vn‖‖un−p‖, |
which arrives at
‖un−p‖2≤‖vn−q‖2−‖vn−un+p−q‖2+2α1‖B1q−B1vn‖‖un−p‖. |
Similarly, we get
‖vn−q‖2≤‖pn−p‖2−‖pn−vn+q−p‖2+2α2‖B2p−B2pn‖‖vn−q‖. |
Combining the last two inequalities, we deduce that
‖un−p‖2≤‖pn−p‖2−‖pn−vn+q−p‖2−‖vn−un+p−q‖2+2α1‖B1q−B1vn‖‖un−p‖+2α2‖B2p−B2pn‖‖vn−q‖. |
Hence,
‖zn−p‖2≤‖pn−p‖2≤ξn‖qn−p‖2+(1−ξn)‖un−p‖2≤ξn‖qn−p‖2+(1−ξn)[‖qn−p‖2−‖pn−vn+q−p‖2−‖vn−un+p−q‖2+2α1‖B1q−B1vn‖‖un−p‖+2α2‖B2p−B2pn‖‖vn−q‖]≤‖qn−p‖2−(1−ξn)[‖pn−vn+q−p‖2+‖vn−un+p−q‖2]+2α2‖B2p−B2pn‖×‖vn−q‖+2α1‖B1q−B1vn‖‖un−p‖. |
This immediately implies that
(1−ξn)[‖pn−vn+q−p‖2+‖vn−un+p−q‖2]≤‖qn−p‖2−‖zn−p‖2+2α2‖B2p−B2pn‖‖vn−q‖+2α1‖B1q−B1vn‖‖un−p‖≤‖qn−zn‖(‖qn−p‖+‖zn−p‖)+2α2‖B2p−B2pn‖‖vn−q‖+2α1‖B1q−B1vn‖‖un−p‖. |
Since qn−zn→0, and 0<lim infn→∞(1−ξn), from (3.5) and the boundedness of {un},{vn}, {qn}, and {zn} we get that
limn→∞‖pn−vn+q−p‖=limn→∞‖vn−un+p−q‖=0, |
which hence yields
‖pn−Gpn‖=‖pn−un‖≤‖pn−vn+q−p‖+‖vn−un+p−q‖→0(n→∞). |
This immediately implies that
‖xn−Gxn‖≤‖xn−pn‖+‖pn−Gpn‖+‖Gpn−Gxn‖≤2‖xn−pn‖+‖pn−Gpn‖→0(n→∞). | (3.7) |
Next, we prove limn→∞‖xn−Snxn‖=0. In fact, we have
‖tn−xn‖≤‖xn−zn‖→0, |
and
‖Sntn−xn‖=‖xn+1−xn−ρn(f(xn)−αFSntn)‖≤‖xn+1−xn‖+ρn(‖f(xn)‖+‖αFSntn‖)→0. |
Hence,
‖xn−Snxn‖≤‖xn−Sntn‖+‖Snxn−Sntn‖≤‖xn−Sntn‖+‖xn−tn‖→0. |
Now, we show xn−Srxn→0,∀r∈{1,...,N}. For 1≤l≤N, it holds that
‖xn−Sn+lxn‖≤‖xn−xn+l‖+‖Sn+lxn+l−Sn+lxn‖+‖xn+l−Sn+lxn+l‖≤2‖xn−xn+l‖+‖xn+l−Sn+lxn+l‖. |
This, together with assumptions, implies that xn−Sn+lxn→0,1≤l≤N. So,
limn→∞‖xn−Srxn‖=0,1≤r≤N. | (3.8) |
Now, we show z∈VI(C,A). If Az=0, then z∈VI(C,A). Next, we suppose that Az≠0. By the condition, we conclude that 0<‖Az‖≤lim infi→∞‖Ayni‖ because yni⇀z. Observe that yn=PC(pn−ζnApn). It follows that ⟨pn−ζnApn−yn,y−yn⟩≤0,∀y∈C. Therefore,
1ζn⟨pn−yn,y−yn⟩+⟨Apn,yn−pn⟩≤⟨Apn,y−pn⟩,∀y∈C. | (3.9) |
Since {Apn} and {yn} are all bounded, from (3.9) and Lemma 3.1, we obtain lim infi→∞⟨Apni, y−pni⟩≥0,∀y∈C. Meanwhile, ⟨Ayn,y−yn⟩=⟨Ayn−Apn,y−pn⟩+⟨Apn,y−pn⟩+⟨Ayn,pn−yn⟩. Using pn−yn→0 and the uniform continuity of A, we get Apn−Ayn→0, which hence attains lim infi→∞⟨Ayni,y−yni⟩≥0,∀y∈C.
To attain z∈VI(C,A), let {λi}⊂(0,1) be a sequence such that λi↓0(i→∞). For every i≥1, let ki be the smallest positive integer satisfying
⟨Aynj,y−ynj⟩+λi≥0,∀j≥ki. | (3.10) |
Put υki=Ayki‖Ayki‖2, and hence get ⟨Ayki,υki⟩=1,∀i≥1. So, from (3.10), one gets ⟨Ayki,y+λiυki−yki⟩≥0, ∀i≥1. Since A is pseudomonotone, we have ⟨A(y+λiυki),y+λiυki−yki⟩≥0,∀i≥1. So,
⟨Ay,y−yki⟩≥⟨Ay−A(y+λiυki),y+λiυki−yki⟩−λi⟨Ay,υki⟩,∀i≥1. | (3.11) |
Let us show that limi→∞λiυki=0. In fact, note that {yki}⊂{yni} and λi↓0 as i→∞. So it follows that 0≤lim supi→∞‖λiυki‖=lim supi→∞λi‖Ayki‖≤lim supi→∞λilim infi→∞‖Ayni‖=0. Therefore, one gets λiυki→0 as i→∞. Thus, letting i→∞ in (3.11), we deduce that ⟨Ay,y−z⟩=lim infi→∞⟨Ay,y−yki⟩≥0,∀y∈C. We apply Lemma 2.3 to conclude z∈VI(C,A).
Finally, we prove z∈Δ. Note that xni⇀z and xni−Srxni→0,∀r∈{1,...,N} (due to (3.8)). Since I−Sr(1≤r≤N) is demiclosed by Lemma 2.5, we attain z∈⋂Nr=1Fix(Sr). By (3.6) and (3.7), we have xni−Sxni→0 and xni−Gxni→0, respectively. Similarly, I−S and I−G are all demiclosed at zero, and we have z∈Fix(S)∩Fix(G). Therefore, z∈⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A)=Δ.
Theorem 3.1. We have the following equivalent relation:
xn→u†∈Δ⇔{Sn+1xn−Snxn→0,xn+1−xn→0, |
where u†∈Δ solves the HVI: ⟨(αF−f)u†,p−u†⟩≥0,∀p∈Δ.
Proof. According to the condition, we assume that ϖn≤ρn(τ−δ)2 and {σn}⊂[a,b]⊂(0,1) for all n≥1. ∀x,y∈H, by Lemma 2.7, we obtain
‖PΔ(I−αF+f)(x)−PΔ(I−αF+f)(y)‖≤[1−(τ−δ)]‖x−y‖, |
which implies that PΔ(I−αF+f) is contractive. Set u†=PΔ(I−αF+f)(u†). Therefore, there is the unique solution u†∈Δ=⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A) of the HVI:
⟨(αF−f)u†,p−u†⟩≥0,∀p∈Δ. | (3.12) |
If xn→u†∈Δ, then we know that u†=Su† and
‖Snxn−Sn+1xn‖≤‖Snxn−u†‖+‖u†−Sn+1xn‖≤(1+ϖn)‖xn−u†‖+(1+ϖn+1)‖u†−xn‖=(2+ϖn+ϖn+1)‖xn−u†‖→0. |
Note that
‖xn+1−xn‖≤‖u†−xn+1‖+‖xn−u†‖→0. |
Now, we prove the sufficiency.
Step 1. Let p∈Δ. Then Gp=p, p=PC(p−ζnAp), Snp=p,∀n≥0, and the inequalities (3.3) and (3.4) hold, i.e.,
‖zn−p‖2≤‖pn−p‖2−(1−ν)‖yn−pn‖2−(1−ν)‖yn−zn‖2, | (3.13) |
and
‖pn−p‖2≤‖qn−p‖2−‖qn−pn‖2. | (3.14) |
Combining (3.13) and (3.14) guarantees that
‖zn−p‖≤‖pn−p‖≤‖qn−p‖. | (3.15) |
Observe that
‖qn−p‖≤‖Snxn−p‖+εn‖Snxn−Snxn−1‖≤(1+ϖn)[‖xn−p‖+εn‖xn−xn−1‖]=(1+ϖn)[‖xn−p‖+ρn⋅εnρn‖xn−xn−1‖]. | (3.16) |
Since supn≥1εnρn‖xn−xn−1‖<∞, there is a constant M1>0 satisfying
εnρn‖xn−xn−1‖≤M1. | (3.17) |
Combining (3.15)–(3.17), we get
‖zn−p‖≤‖pn−p‖≤‖qn−p‖≤(1+ϖn)[‖xn−p‖+ρnM1],∀n≥1. | (3.18) |
Also, it is readily known that
‖tn−p‖≤σn‖xn−p‖+(1−σn)‖zn−p‖≤(1+ϖn)[‖xn−p‖+ρnM1]. | (3.19) |
Thus, using (3.19) and ϖn≤ρn(τ−δ)2∀n≥1, from Lemma 2.7, we receive
‖xn+1−p‖=‖ρnf(xn)+(I−ρnαF)Sntn−p‖=‖ρn(f(xn)−f(p))+(I−ρnαF)Sntn−(I−ρnαF)p+ρn(f−αF)p‖≤ρnδ‖xn−p‖+(1−ρnτ)‖tn−p‖+ρn‖(f−αF)p‖≤ρnδ‖xn−p‖+(1−ρnτ)(1+ϖn)[‖xn−p‖+ρnM1]+ρn‖(f−αF)p‖≤ρnδ‖xn−p‖+[(1−ρnτ)+ϖn]‖xn−p‖+ρn(1+ϖn)M1+ρn‖(f−αF)p‖≤ρnδ‖xn−p‖+[(1−ρnτ)+ρn(τ−δ)2]‖xn−p‖+2ρnM1+ρn‖(f−αF)p‖=[1−ρn(τ−δ)2]‖xn−p‖+ρn(τ−δ)2⋅2(2M1+‖(f−αF)p‖)τ−δ. |
Hence,
‖xn−p‖≤max{‖x1−p‖,2(2M1+‖(f−αF)p‖)τ−δ},∀n≥1. |
We deduce that the sequences {xn}, {pn},{qn},{yn},{zn},{tn},{f(xn)}, {Sntn}, and {Snxn} are bounded.
Step 2. Observe that tn−p=σn(xn−p)+(1−σn)(zn−p) and
xn+1−p=ρn(f(xn)−f(p))+(I−ρnαF)Sntn−(I−ρnαF)p+ρn(f−αF)p. |
Utilizing Lemma 2.7, we attain
‖xn+1−p‖2≤‖ρn(f(xn)−f(p))+(I−ρnαF)Sntn−(I−ρnαF)p‖2+2ρn⟨(f−αF)p,xn+1−p⟩≤[ρnδ‖xn−p‖+(1−ρnτ)‖tn−p‖]2+2ρn⟨(f−αF)p,xn+1−p⟩≤ρnδ‖xn−p‖2+(1−ρnτ)‖tn−p‖2+2ρn⟨(f−αF)p,xn+1−p⟩=ρnδ‖xn−p‖2+(1−ρnτ)[σn‖xn−p‖2+(1−σn)‖zn−p‖2−σn(1−σn)‖xn−zn‖2]+2ρn⟨(f−αF)p,xn+1−p⟩≤ρnδ‖xn−p‖2+(1−ρnτ)[σn‖xn−p‖2+(1−σn)‖zn−p‖2]−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM2, | (3.20) |
where M2>0 is a constant such that supn≥12‖(f−αF)p‖‖xn−p‖≤M2. By (3.20) and Lemma 3.2, we have
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ)[σn‖xn−p‖2+(1−σn)‖zn−p‖2]−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM2≤ρnδ‖xn−p‖2+(1−ρnτ){σn‖xn−p‖2+(1−σn)[‖qn−p‖2−(1−ξn)‖qn−pn‖2−(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]}−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM2. | (3.21) |
Taking (3.18) into account, we obtain
‖qn−p‖2≤(1+ϖn)2(‖xn−p‖+ρnM1)2=(‖xn−p‖+ρnM1)2+ϖn(2+ϖn))(‖xn−p‖+ρnM1)2=‖xn−p‖2+ρn{M1(2‖xn−p‖+ρnM1)+ϖn(2+ϖn)ρn(‖xn−p‖+ρnM1)2}≤‖xn−p‖2+ρnM3, | (3.22) |
where M3>0 is a constant such that supn≥1{(2‖xn−p‖+ρnM1)M1+ϖn(2+ϖn)ρn(‖xn−p‖+ρnM1)2}≤M3. Based on (3.21) and (3.22), we get
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ){σn‖xn−p‖2+(1−σn)[‖xn−p‖2+ρnM3−(1−ξn)‖qn−pn‖2−(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]}−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM2≤[1−ρn(τ−δ)]‖xn−p‖2−(1−ρnτ)(1−σn)[(1−ξn)‖qn−pn‖2+ρnM3+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM2≤‖xn−p‖2−(1−ρnτ)(1−σn)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]−(1−ρnτ)σn(1−σn)‖xn−zn‖2+ρnM4, |
where M4:=M3+M2. This immediately implies that
(1−ρnτ)(1−σn)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]+(1−ρnτ)σn(1−σn)‖xn−zn‖2≤‖xn−p‖2−‖xn+1−p‖2+ρnM4. | (3.23) |
Step 3. Note that
‖qn−p‖2≤(1+ϖn)2(‖xn−p‖+εn‖xn−xn−1‖)2=(1+ϖn)2‖xn−p‖2+(1+ϖn)2εn‖xn−xn−1‖(2‖xn−p‖+εn‖xn−xn−1‖)=(1+ϖn(2+ϖn))‖xn−p‖2+(1+ϖn)2εn‖xn−xn−1‖(2‖xn−p‖+εn‖xn−xn−1‖). | (3.24) |
Combining (3.18), (3.20), and (3.24), we receive
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ)[σn‖xn−p‖2+(1−σn)‖zn−p‖2]+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)](1+ϖn)2(‖xn−p‖+εn‖xn−xn−1‖)2+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)]{(1+ϖn(2+ϖn))‖xn−p‖2+(1+ϖn)2εn‖xn−xn−1‖×(2‖xn−p‖+εn‖xn−xn−1‖)}+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)]‖xn−p‖2+εn‖xn−xn−1‖(1+ϖn)2(2‖xn−p‖+εn‖xn−xn−1‖)+ϖn(2+ϖn)‖xn−p‖2+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)]‖xn−p‖2+(εn‖xn−xn−1‖3(1+ϖn)2+ϖn(2+ϖn))M+2ρn⟨(f−αF)p,xn+1−p⟩=[1−ρn(τ−δ)]‖xn−p‖2+ρn(τ−δ)[2⟨(f−αF)p,xn+1−p⟩τ−δ+Mτ−δ(εnρn‖xn−xn−1‖×3(1+ϖn)2+ϖn(2+ϖn)ρn)], | (3.25) |
where M>0 is a constant such that supn≥1{‖xn−p‖,εn‖xn−xn−1‖,‖xn−p‖2}≤M.
Step 4. Taking p=u†, by (3.25), we have
‖xn+1−u†‖2≤[1−ρn(τ−δ)]‖xn−u†‖2+ρn(τ−δ)[2⟨(f−αF)u†,xn+1−u†⟩τ−δ+Mτ−δ(εnρn‖xn−xn−1‖3(1+ϖn)2+ϖn(2+ϖn)ρn)]. | (3.26) |
Set Γn=‖xn−u†‖2.
Case 1. There is an integer n0≥1 such that {Γn} is nonincreasing. In this case, limn→∞Γn=ℏ<+∞. From (3.23), we have
(1−ρnτ)(1−b)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]+(1−ρnτ)a(1−b)‖xn−zn‖2≤(1−ρnτ)(1−σn)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]+(1−ρnτ)σn(1−σn)‖xn−zn‖2≤‖xn−u†‖2−‖xn+1−u†‖2+ρnM4=Γn−Γn+1+ρnM4. |
Noticing 0<lim infn→∞(1−ξn), ρn→0 and Γn−Γn+1→0, for ν∈(0,1) one has limn→∞‖qn−pn‖=limn→∞‖yn−zn‖=0, and limn→∞‖yn−pn‖=limn→∞‖xn−zn‖=0. Thus, we get
‖xn−yn‖≤‖xn−zn‖+‖yn−zn‖→0, | (3.27) |
and
‖qn−zn‖≤‖qn−pn‖+‖pn−yn‖+‖yn−zn‖→0→0(n→∞). | (3.28) |
Since {xn} is bounded, there is a subsequence {xnk} of {xn} satisfying xnk⇀˜x and
lim supn→∞⟨(f−αF)u†,xn−u†⟩=limk→∞⟨(f−αF)u†,xnk−u†⟩. | (3.29) |
In the light of (3.29), one gets
lim supn→∞⟨(f−αF)u†,xn−u†⟩=limk→∞⟨(f−αF)u†,xnk−u†⟩=⟨(f−αF)u†,˜x−u†⟩. | (3.30) |
Since xn+1−xn→0,yn−xn→0,zn−qn→0 and Sn+1xn−Snxn→0, applying Lemma 3.3, we conclude that ˜x∈ωw({xn})⊂Δ. Combining (3.12) and (3.30), we get
lim supn→∞⟨(f−αF)u†,xn−u†⟩=⟨(f−αF)u†,˜x−u†⟩≤0. | (3.31) |
Since xn−xn+1→0, we have
lim supn→∞⟨(f−αF)u†,xn+1−u†⟩=lim supn→∞[⟨(f−αF)u†,xn+1−xn⟩+⟨(f−αF)u†,xn−u†⟩]≤lim supn→∞[‖(f−αF)u†‖‖xn+1−xn‖+⟨(f−αF)u†,xn−u†⟩]≤0. |
Note that
lim supn→∞[2⟨(f−αF)u†,xn+1−u†⟩τ−δ+Mτ−δ(εnρn‖xn−xn−1‖3(1+ϖn)2+ϖn(2+ϖn)ρn)]≤0. |
According to (3.26) and Lemma 2.4, we deduce that limn→∞‖xn−u†‖2=0.
Case 2. Suppose that ∃{Γnk}⊂{Γn} s.t. Γnk<Γnk+1,∀k∈N. Let ϕ:N→N be a mapping defined by
ϕ(n):=max{k≤n:Γk<Γk+1}. |
Based on Lemma 2.6, we have
Γϕ(n)≤Γϕ(n)+1andΓn≤Γϕ(n)+1. |
Putting p=u†, from (3.23), we have
(1−ρϕ(n)τ)(1−b)[(1−ξϕ(n))‖qϕ(n)−pϕ(n)‖2+(1−ν)(‖yϕ(n)−zϕ(n)‖2+‖yϕ(n)−pϕ(n)‖2)]+(1−ρϕ(n)τ)a(1−b)‖xϕ(n)−zϕ(n)‖2≤(1−ρϕ(n)τ)(1−σϕ(n))[(1−ξϕ(n))‖qϕ(n)−pϕ(n)‖2+(1−ν)(‖yϕ(n)−zϕ(n)‖2+‖yϕ(n)−pϕ(n)‖2)]+(1−ρϕ(n)τ)σϕ(n)(1−σϕ(n))‖xϕ(n)−zϕ(n)‖2≤‖xϕ(n)−u†‖2−‖xϕ(n)+1−u†‖2+ρϕ(n)M4=Γϕ(n)−Γϕ(n)+1+ρϕ(n)M4, | (3.32) |
which immediately yields limn→∞‖qϕ(n)−pϕ(n)‖=limn→∞‖yϕ(n)−zϕ(n)‖=0 and limn→∞‖yϕ(n) −pϕ(n)‖=limn→∞‖xϕ(n)−zϕ(n)‖=0. Therefore,
limn→∞‖xϕ(n)−yϕ(n)‖=limn→∞‖qϕ(n)−zϕ(n)‖=0, | (3.33) |
and
lim supn→∞⟨(f−αF)u†,xϕ(n)+1−u†⟩≤0. | (3.34) |
At the same time, by (3.26), we known that
ρϕ(n)(τ−δ)Γϕ(n)≤Γϕ(n)−Γϕ(n)+1+ρϕ(n)(τ−δ)[2⟨(f−αF)u†,xϕ(n)+1−u†⟩τ−δ+Mτ−δ(εϕ(n)ρϕ(n)‖xϕ(n)−xϕ(n)−1‖3(1+ϖϕ(n))2+ϖϕ(n)(2+ϖϕ(n))ρϕ(n))]≤ρϕ(n)(τ−δ)[2⟨(f−αF)u†,xϕ(n)+1−u†⟩τ−δ+Mτ−δ(εϕ(n)ρϕ(n)‖xϕ(n)−xϕ(n)−1‖×3(1+ϖϕ(n))2+ϖϕ(n)(2+ϖϕ(n))ρϕ(n))], |
which hence arrives at
lim supn→∞Γϕ(n)≤lim supn→∞[2⟨(f−αF)u†,xϕ(n)+1−u†⟩τ−δ+Mτ−δ(εϕ(n)ρϕ(n)‖xϕ(n)−xϕ(n)−1‖3(1+ϖϕ(n))2+ϖϕ(n)(2+ϖϕ(n))ρϕ(n))]≤0. |
Thus, limn→∞‖xϕ(n)−u†‖2=0. Also, note that
‖xϕ(n)+1−u†‖2−‖xϕ(n)−u†‖2=2⟨xϕ(n)+1−xϕ(n),xϕ(n)−u†⟩+‖xϕ(n)+1−xϕ(n)‖2≤2‖xϕ(n)+1−xϕ(n)‖‖xϕ(n)−u†‖+‖xϕ(n)+1−xϕ(n)‖2. | (3.35) |
Since Γn≤Γϕ(n)+1, we have
‖xn−u†‖2≤‖xϕ(n)+1−u†‖2≤‖xϕ(n)−u†‖2+2‖xϕ(n)+1−xϕ(n)‖‖xϕ(n)−u†‖+‖xϕ(n)+1−xϕ(n)‖2→0(n→∞). |
So, xn→u†.
According to Theorem 3.1, we have the following corollary.
Corollary 3.1. Suppose that S:C→C is a nonexpansive mapping. For two fixed points x1,x0∈H, let the sequence {xn} be defined by
{qn=Sxn+εn(Sxn−Sxn−1),pn=ξnqn+(1−ξn)un,vn=TΘ2α2(pn−α2B2pn),un=TΘ1α1(vn−α1B1vn),yn=PC(pn−ζnApn),zn=PCn(pn−ζnAyn),tn=σnxn+(1−σn)zn,xn+1=ρnf(xn)+(I−ρnαF)Sntn,∀n≥1, | (3.36) |
where Cn and ζn have the same form as in Algorithm 3.1. Then, xn→u†∈Δ⇔xn+1−xn→0, where u†∈Δ is the unique solution of the HVI: ⟨(αF−f)u†,p−u†⟩≥0,∀p∈Δ.
Next, we put forth another modification of the inertial composite subgradient extragradient implicit rule with line-search process.
Algorithm 3.2. Let x1,x0∈H be two fixed points. Let xn be given. Compute xn+1 via the following iterative steps:
Step 1. Set qn=Snxn+εn(Snxn−Snxn−1) and calculate
{pn=ξnqn+(1−ξn)un,vn=TΘ2α2(pn−α2B2pn),un=TΘ1α1(vn−α1B1vn). |
Step 2. Compute yn=PC(pn−ζnApn), with ζn being chosen to be the largest ζ∈{γ,γℓ,γℓ2,...} s.t.
ζ‖Apn−Ayn‖≤ν‖pn−yn‖. |
Step 3. Compute tn=σnzn+(1−σn)Sntn with zn=PCn(pn−ζnAyn) and
Cn:={y∈H:⟨pn−ζnApn−yn,y−yn⟩≤0}. |
Step 4. Compute
xn+1=ρnf(xn)+(I−ρnαF)Sntn, |
where Sn is constructed as in Algorithm 1.1. Set n:=n+1 and go to Step 1
Theorem 3.2. Let the sequence {xn} be generated by Algorithm 3.2. Then
xn→u†∈Δ⇔{Sn+1xn−Snxn→0,xn+1−xn→0, |
where u†∈Δ is the unique solution of the HVI: ⟨(αF−f)u†,p−u†⟩≥0,∀p∈Δ.
Proof. The necessity is obvious. Next, we prove the sufficiency.
Note that
‖tn−p‖≤σn‖zn−p‖+(1−σn)‖Sntn−p‖≤σn‖zn−p‖+(1−σn)‖tn−p‖. |
This, together with (3.18), ensures that
‖tn−p‖≤‖zn−p‖≤‖pn−p‖≤‖qn−p‖≤(1+ϖn)[‖xn−p‖+ρnM1],∀n≥1. | (3.37) |
By (3.37) and Lemma 2.7, we have
‖xn+1−p‖=‖ρn(f(xn)−f(p))+(I−ρnαF)Sntn−(I−ρnαF)p+ρn(f−αF)p‖≤ρnδ‖xn−p‖+(1−ρnτ)‖tn−p‖+ρn‖(f−αF)p‖≤ρnδ‖xn−p‖+(1−ρnτ)(1+ϖn)[‖xn−p‖+ρnM1]+ρn‖(f−αF)p‖≤[1−ρn(τ−δ)2]‖xn−p‖+ρn(τ−δ)2⋅2(2M1+‖(f−αF)p‖)τ−δ. |
It follows that ‖xn−p‖≤max{‖x1−p‖,2(2M1+‖(f−αF)p‖)τ−δ}∀n≥1. Therefore, {xn}, {pn}, {qn}, {yn}, {zn}, {tn}, {f(xn)}, {Sntn}, and {Snxn} are bounded.
According to Lemma 2.7, we get
‖xn+1−p‖2≤‖ρn(f(xn)−f(p))+(I−ρnαF)Sntn−(I−ρnαF)p‖2+2ρn⟨(f−αF)p,xn+1−p⟩≤ρnδ‖xn−p‖2+(1−ρnτ)‖tn−p‖2+2ρn⟨(f−αF)p,xn+1−p⟩=ρnδ‖xn−p‖2+(1−ρnτ)[σn‖zn−p‖2+(1−σn)‖Sntn−p‖2−σn(1−σn)‖zn−Sntn‖2]+2ρn⟨(f−αF)p,xn+1−p⟩≤ρnδ‖xn−p‖2+(1−ρnτ)[σn‖zn−p‖2+(1−σn)‖tn−p‖2]−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM2, | (3.38) |
where M2>0 is a constant such that supn≥12‖(f−αF)p‖‖xn−p‖≤M2. Using Lemma 3.2, from (3.37) and (3.38), we have
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ)‖zn−p‖2−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM2≤ρnδ‖xn−p‖2+(1−ρnτ){‖qn−p‖2−(1−ξn)‖qn−pn‖2−(1−ν)×(‖yn−zn‖2+‖yn−pn‖2)}−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM2. | (3.39) |
Also, using the same inferences as those of (3.22) of Theorem 3.1, we have
‖qn−p‖2≤‖xn−p‖2+ρnM3, | (3.40) |
where supn≥1{M1(2‖xn−p‖+ρnM1)+ϖn(2+ϖn)ρn(‖xn−p‖+ρnM1)2}≤M3 for some constant M3. By (3.39) and (3.40), we attain
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ){‖xn−p‖2+ρnM3−(1−ξn)‖qn−pn‖2−(1−ν)(‖yn−zn‖2+‖yn−pn‖2)}−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM2≤[1−ρn(τ−δ)]‖xn−p‖2−(1−ρnτ)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM3+ρnM2≤‖xn−p‖2−(1−ρnτ)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]−(1−ρnτ)σn(1−σn)‖zn−Sntn‖2+ρnM4, |
where M4:=M3+M2. Hence, we attain the assertion.
By the same argument as those of (3.24), we have
‖qn−p‖2≤(1+ϖn(2+ϖn))‖xn−p‖2+(1+ϖn)2εn‖xn−xn−1‖(2‖xn−p‖+εn‖xn−xn−1‖). | (3.41) |
By (3.37), (3.38), and (3.41), we obtain
‖xn+1−p‖2≤ρnδ‖xn−p‖2+(1−ρnτ)[(1−σn)‖tn−p‖2+σn‖zn−p‖2]+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)](1+ϖn)2(‖xn−p‖+εn‖xn−xn−1‖)2+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)]‖xn−p‖2+εn‖xn−xn−1‖(1+ϖn)2(2‖xn−p‖+εn‖xn−xn−1‖)+ϖn(2+ϖn)‖xn−p‖2+2ρn⟨(f−αF)p,xn+1−p⟩≤[1−ρn(τ−δ)]‖xn−p‖2+ρn(τ−δ)[2⟨(f−αF)p,xn+1−p⟩τ−δ+Mτ−δ(εnρn‖xn−xn−1‖×3(1+ϖn)2+ϖn(2+ϖn)ρn)], | (3.42) |
where supn≥1{‖xn−p‖,εn‖xn−xn−1‖,‖xn−p‖2}≤M for some constant M.
Setting p=u†, by (3.42), we have
‖xn+1−u†‖2≤[1−ρn(τ−δ)]‖xn−u†‖2+ρn(τ−δ)[2⟨(f−αF)u†,xn+1−u†⟩τ−δ+Mτ−δ(εnρn‖xn−xn−1‖3(1+ϖn)2+ϖn(2+ϖn)ρn)]. |
Set Γn=‖xn−u†‖2.
Case 1. Assume {Γn} is nonincreasing when n≥n0. Then, limn→∞Γn=ℏ<+∞. Choosing p=u†, from (3.38), we have
(1−ρnτ)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]+(1−ρnτ)a(1−b)‖zn−Sntn‖2≤(1−ρnτ)[(1−ξn)‖qn−pn‖2+(1−ν)(‖yn−zn‖2+‖yn−pn‖2)]+(1−ρnτ)σn(1−σn)‖zn−Sntn‖2≤‖xn−u†‖2−‖xn+1−u†‖2+ρnM4=Γn−Γn+1+ρnM4. |
Since Γn−Γn+1→0 for ν∈(0,1), limn→∞‖qn−pn‖=limn→∞‖yn−zn‖=0, and limn→∞‖yn−pn‖=limn→∞‖zn−Sntn‖=0. Observe that
‖zn−xn‖≤‖zn−Sntn‖+‖Sntn−xn‖=‖zn−Sntn‖+‖xn+1−xn−ρn(f(xn)−αFSntn)‖≤‖zn−Sntn‖+‖xn+1−xn‖+ρn(‖f(xn)‖+‖αFSntn‖)→0(n→∞). |
By the similar arguments as those in Theorem 3.1, we deduce limn→∞‖xn−u†‖2=0.
Case 2. Assume ∃{Γnk}⊂{Γn} s.t. Γnk<Γnk+1,∀k∈N. Let ϕ:N→N be a mapping defined by
ϕ(n)=max{k≤n:Γk<Γk+1}. |
By Lemma 2.6, we have
Γn≤Γϕ(n)+1andΓϕ(n)≤Γϕ(n)+1. |
Set p=u†. Then,
(1−ρϕ(n)τ)[(1−ξϕ(n))‖qϕ(n)−pϕ(n)‖2+(1−ν)(‖yϕ(n)−zϕ(n)‖2+‖yϕ(n)−pϕ(n)‖2)]+(1−ρϕ(n)τ)a(1−b)‖zϕ(n)−Sϕ(n)tϕ(n)‖2≤(1−ρϕ(n)τ)[(1−ξϕ(n))‖qϕ(n)−pϕ(n)‖2+(1−ν)(‖yϕ(n)−zϕ(n)‖2+‖yϕ(n)−pϕ(n)‖2)]+(1−ρϕ(n)τ)σϕ(n)(1−σϕ(n))‖zϕ(n)−Sϕ(n)tϕ(n)‖2≤‖xϕ(n)−u†‖2−‖xϕ(n)+1−u†‖2+ρϕ(n)M4=Γϕ(n)−Γϕ(n)+1+ρϕ(n)M4, |
which immediately yields limn→∞‖qϕ(n)−pϕ(n)‖=limn→∞‖yϕ(n)−zϕ(n)‖=0 and limn→∞‖yϕ(n) −pϕ(n)‖=limn→∞‖zϕ(n)−Sϕ(n)tϕ(n)‖=0. Therefore, limn→∞‖zϕ(n)−xϕ(n)‖=0. Finally, using the similar arguments to those in Theorem 3.1, we get the conclusion.
Remark 3.1. Compared with the corresponding results in Cai, Shehu, and Iyiola [2], Ceng and Shang [4], and Thong and Hieu [28], our results improve and extend them in the following aspects:
(ⅰ) The problem of finding an element of Fix(S)∩Fix(G) (with G=PC(I−μ1B1)PC(I−μ2B2)) in [2] is extended to develop our problem of finding an element of ⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A) where G=TΘ1μ1(I−μ1B1)TΘ2μ2(I−μ2B2) and S is an asymptotically nonexpansive mapping. The modified viscosity implicit rule for finding an element of Fix(S)∩Fix(G) in [2] is extended to develop our modified inertial composite subgradient extragradient implicit rules with line-search process for finding an element of ⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A), which is on the basis of the subgradient extragradient rule with line-search process, inertial iteration approach, viscosity approximation method, and hybrid deepest-descent technique.
(ⅱ) The problem of finding an element of Fix(S)∩VI(C,A) with a quasi-nonexpansive mapping S in [4] is extended to develop our problem of finding an element of ⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A) with an asymptotically nonexpansive mapping S. The inertial subgradient extragradient method with line-search process for finding an element of Fix(S)∩VI(C,A) in [28] is extended to develop our modified inertial composite subgradient extragradient implicit rules with line-search process for finding an element of ⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A), which is on the basis of the subgradient extragradient rule with line-search process, inertial iteration approach, viscosity approximation method, and hybrid deepest-descent technique.
(ⅲ) The problem of finding an element of Ω=⋂Nr=0Fix(Sr)∩VI(C,A) is extended to develop our problem of finding an element of Ω=⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A) with G=TΘ1μ1(I−μ1B1)TΘ2μ2(I−μ2B2). The hybrid inertial subgradient extragradient method with line-search process in [4] is extended to develop our modified inertial composite subgradient extragradient implicit rules with line-search process.
In this section, we give an example to show the feasibility of our algorithms. Put Θ1=Θ2=0, α=2, α1=α2=13,γ=1,ν=ℓ=12,σn=ξn=23, and εn=ρn=13(n+1), for all n≥0. Now, we construct an example of Δ=⋂Nr=0Fix(Sr)∩Fix(G)∩VI(C,A)≠∅ with S0:=S and G=TΘ1α1(I−α1B1)TΘ2α2(I−α2B2)=PC(I−α1B1)PC(I−α2B2), where A:H→H is pseudomonotone and a Lipschitz continuous mapping, B1,B2:H→H are two inverse-strongly monotone mappings, S:H→H is asymptotical nonexpansive, and each Sr:H→H is nonexpansive for r=1,...,N.
Let H=R and use ⟨a,b⟩=ab and ‖⋅‖=|⋅| to denote its inner product and induced norm, respectively. Set C=[−2,4] and the starting point x1 is arbitrarily chosen in C. Let f(x)=F(x)=12x, ∀x∈H with
δ=12<τ=1−√1−α(2η−ακ2)=1−√1−2(2⋅12−2(12)2)=1. |
Let B1x=B2x:=Bx=x−12sinx,∀x∈C. Let the operators A,S,Sr:H→H be defined by
Ax:=11+|sinx|−11+|x|,Sx:=34sinx,Srx:=S1x=sinx(r=1,⋯,N),∀x∈H. |
We have the following assertions:
(ⅰ) A is 2-Lipschitz continuous, in fact, for each x,y∈H, we have
|Ax−Ay|≤||y|−|x|(1+‖y|)(1+|x|)|+||siny|−|sinx|(1+|siny|)(1+|sinx|)|≤|x−y|(1+|x|)(1+|y|)+|sinx−siny|(1+|sinx|)(1+|siny|)≤|x−y|+|sinx−siny|≤2|x−y|. |
(ⅱ) A is pseudomonotone, in fact, for each x,y∈H, if
⟨Ax,y−x⟩=(11+|sinx|−11+|x|)(y−x)≥0, |
then
⟨Ay,y−x⟩=(11+|siny|−11+|y|)(y−x)≥0. |
(ⅲ) B is 29-inverse-strongly monotone. In fact, since B is 12-strongly monotone and 32-Lipschitz continuous, we know that B is 29-inverse-strongly monotone with ρ=σ=29.
Moreover, it is easy to check that S is asymptotically nonexpansive with ϖn=(34)n,∀n≥1, such that ‖Sn+1xn−Snxn‖→0 as n→∞. In fact, note that
‖Snx−Sny‖≤34‖Sn−1x−Sn−1y‖≤⋯≤(34)n‖x−y‖≤(1+ϖn)‖x−y‖, |
and
‖Sn+1xn−Snxn‖≤(34)n−1‖S2xn−Sxn‖=(34)n−1|34sin(Sxn)−34sinxn|≤2(34)n→0. |
It is obvious that Fix(S)={0} and
limn→∞ϖnρn=limn→∞(3/4)n1/3(n+1)=0. |
Accordingly, Δ=Fix(S)∩Fix(S1)∩Fix(G)∩VI(C,A)={0}≠∅. In this case, noticing G=PC(I−α1B1)PC(I−α2B2)=[PC(I−13B)]2, we rewrite Algorithm 3.1 as follows:
{qn=Snxn+13(n+1)(Snxn−Snxn−1),pn=23qn+13un,vn=PC(pn−13Bpn),un=PC(vn−13Bvn),yn=PC(pn−ζnApn),zn=PCn(pn−ζnAyn),tn=23xn+13zn,xn+1=13(n+1)⋅12xn+(1−13(n+1))S1tn,∀n≥1, |
where Cn and ζn are chosen as in Algorithm 3.1. Then, xn→0∈Δ.
In particular, since Sx:=34sinx is also nonexpansive, we consider the modified version of Algorithm 3.1, that is,
{qn=Sxn+13(n+1)(Sxn−Sxn−1),pn=23qn+13un,vn=PC(pn−13Bpn),un=PC(vn−13Bvn),yn=PC(pn−ζnApn),zn=PCn(pn−ζnAyn),tn=23xn+13zn,xn+1=13(n+1)⋅12xn+(1−13(n+1))S1tn,∀n≥1, |
where Cn and ζn are chosen as above. Then, xn→0∈Δ.
In a real Hilbert space, we have put forward two modified inertial composite subgradient extragradient implicit rules with line-search process for settling a generalized equilibrium problems system with constraints of a pseudomonotone variational inequality problem and a common fixed-point problem of finite nonexpansive mappings and an asymptotically nonexpansive mapping, respectively. Under the lack of the sequential weak continuity and Lipschitz constant of the cost operator A, we have demonstrated the strong convergence of the proposed algorithms to an element of the studied problem. In addition, an illustrated example was provided to demonstrate the feasibility of our proposed algorithms.
In the end, it is worthy to mention that part of our future research is aimed at acquiring the strong convergence results for the modifications of our proposed rules with a Nesterov inertial extrapolation step and adaptive stepsizes.
The authors declare that they have not used artificial intelligence (AI) tools in the creation of this article.
Yeong-Cheng Liou is partially supported by a grant from the Kaohsiung Medical University Research Foundation (KMU-M113001). Lu-Chuan Ceng is partially 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).
The authors declare that there are no conflicts of interest.
[1] | E. Blum, W. Oettli, From optimization and variational inequalities to equilibrium problems, Math. Student, 63 (1994), 123–145. |
[2] |
G. Cai, Y. Shehu, O. Iyiola, Strong convergence results for variational inequalities and fixed point problems using modified viscosity implicit rules, Numer. Algor., 77 (2018), 535–558. http://dx.doi.org/10.1007/s11075-017-0327-8 doi: 10.1007/s11075-017-0327-8
![]() |
[3] |
L. Ceng, A. Petrusel, X. Qin, J. Yao, Two inertial subgradient extragradient algorithms for variational inequalities with fixed-point constraints, Optimization, 70 (2021), 1337–1358. http://dx.doi.org/10.1080/02331934.2020.1858832 doi: 10.1080/02331934.2020.1858832
![]() |
[4] |
L. Ceng, M. Shang, Hybrid inertial subgradient extragradient methods for variational inequalities and fixed point problems involving asymptotically nonexpansive mappings, Optimization, 70 (2021), 715–740. http://dx.doi.org/10.1080/02331934.2019.1647203 doi: 10.1080/02331934.2019.1647203
![]() |
[5] |
L. Ceng, C. Wang, J. Yao, Strong convergence theorems by a relaxed extragradient method for a general system of variational inequalities, Math. Meth. Oper. Res., 67 (2008), 375–390. http://dx.doi.org/10.1007/s00186-007-0207-4 doi: 10.1007/s00186-007-0207-4
![]() |
[6] |
L. Ceng, J. Yao, A relaxed extragradient-like method for a generalized mixed equilibrium problem, a general system of generalized equilibria and a fixed point problem, Nonlinear Anal.-Theor., 72 (2010), 1922–1937. http://dx.doi.org/10.1016/j.na.2009.09.033 doi: 10.1016/j.na.2009.09.033
![]() |
[7] |
J. Chen, S. Liu, X. Chang, Extragradient method and golden ratio method for equilibrium problems on Hadamard manifolds, Int. J. Comput. Math., 98 (2021), 1699–1712. http://dx.doi.org/10.1080/00207160.2020.1846728 doi: 10.1080/00207160.2020.1846728
![]() |
[8] | P. Combettes, S. Hirstoaga, Equilibrium programming in Hilbert spaces, J. Nonlinear Convex Anal., 6 (2005), 117–136. |
[9] |
R. Cottle, J. Yao, Pseudomonotone complementarity problems in Hilbert space, J. Optim. Theory Appl., 75 (1992), 281–295. http://dx.doi.org/10.1007/BF00941468 doi: 10.1007/BF00941468
![]() |
[10] |
L. Deng, R. Hu, Y. Fang, Projection extragradient algorithms for solving nonmonotone and non-Lipschitzian equilibrium problems in Hilbert spaces, Numer. Algor., 86 (2021), 191–221. http://dx.doi.org/10.1007/s11075-020-00885-x doi: 10.1007/s11075-020-00885-x
![]() |
[11] |
S. Denisov, V. Semenov, L. Chabak, Convergence of the modified extragradient method for variational inequalities with non-Lipschitz operators, Cybern. Syst. Anal., 51 (2015), 757–765. http://dx.doi.org/10.1007/s10559-015-9768-z doi: 10.1007/s10559-015-9768-z
![]() |
[12] | K. Goebel, S. Reich, Uniform convexity, hyperbolic geometry, and nonexpansive mappings, New York: Marcel Dekker, 1983. |
[13] |
L. He, Y. Cui, L. Ceng, T. Zhao, D. Wang, H. Hu, Strong convergence for monotone bilevel equilibria with constraints of variational inequalities and fixed points using subgradient extragradient implicit rule, J. Inequal. Appl., 2021 (2021), 146. http://dx.doi.org/10.1186/s13660-021-02683-y doi: 10.1186/s13660-021-02683-y
![]() |
[14] |
L. Jolaoso, Y. Shehu, J. Yao, Inertial extragradient type method for mixed variational inequalities without monotonicity, Math. Comput. Simulat., 192 (2022), 353–369. http://dx.doi.org/10.1016/j.matcom.2021.09.010 doi: 10.1016/j.matcom.2021.09.010
![]() |
[15] | G. Korpelevich, The extragradient method for finding saddle points and other problems, Matecon, 12 (1976), 747–756. |
[16] |
R. Kraikaew, S. Saejung, Strong convergence of the Halpern subgradient extragradient method for solving variational inequalities in Hilbert spaces, J. Optim. Theory Appl., 163 (2014), 399–412. http://dx.doi.org/10.1007/s10957-013-0494-2 doi: 10.1007/s10957-013-0494-2
![]() |
[17] |
X. Li, Z. Liu, Sensitivity analysis of optimal control problems described by differential hemivariational inequalities, SIAM J. Comtrol Optim., 56 (2018), 3569–3597. http://dx.doi.org/10.1137/17M1162275 doi: 10.1137/17M1162275
![]() |
[18] |
T. Lim, H. Xu, Fixed point theorems for asymptotically nonexpansive mappings, Nonlinear Anal.-Theor., 22 (1994), 1345–1355. http://dx.doi.org/10.1016/0362-546X(94)90116-3 doi: 10.1016/0362-546X(94)90116-3
![]() |
[19] |
P. Maingé, Strong convergence of projected subgradient methods for nonsmooth and nonstrictly convex minimization, Set-Valued Anal., 16 (2008), 899–912. http://dx.doi.org/10.1007/s11228-008-0102-z doi: 10.1007/s11228-008-0102-z
![]() |
[20] | A. Moudafi, M. Théra, Proximal and dynamical approaches to equilibrium problems, In: Ill-posed variational problems and regularization techniques, Berlin: Springer, 1999. http://dx.doi.org/10.1007/978-3-642-45780-7_12 |
[21] |
X. Qin, A. Petrusel, B. Tan, J. Yao, Efficient extragradient methods for bilevel pseudomonotone variational inequalities with non-Lipschitz operators and their applications, Fixed Point Theor., 25 (2024), 309–332. http://dx.doi.org/10.24193/fpt-ro.2024.1.19 doi: 10.24193/fpt-ro.2024.1.19
![]() |
[22] |
Y. Shehu, Q. Dong, D. Jiang, Single projection method for pseudo-monotone variational inequality in Hilbert spaces, Optimization, 68 (2019), 385–409. http://dx.doi.org/10.1080/02331934.2018.1522636 doi: 10.1080/02331934.2018.1522636
![]() |
[23] |
Y. Shehu, O. Iyiola, Strong convergence result for monotone variational inequalities, Numer. Algor., 76 (2017), 259–282. http://dx.doi.org/10.1007/s11075-016-0253-1 doi: 10.1007/s11075-016-0253-1
![]() |
[24] |
Y. Song, O. Bazighifan, Two regularization methods for the variational inequality problem over the set of solutions of the generalized mixed equilibrium problem, Mathematics, 10 (2022), 2981. http://dx.doi.org/10.3390/math10162981 doi: 10.3390/math10162981
![]() |
[25] |
Y. Song, Y. Pei, A new viscosity semi-implicit midpoint rule for strict pseudo-contractions and (α,β)-generalized hybrid mappings, Optimization, 70 (2021), 2635–2653. http://dx.doi.org/10.1080/02331934.2020.1789640 doi: 10.1080/02331934.2020.1789640
![]() |
[26] | G. Stampacchia, Formes bilineaires coercivities sur les ensembles convexes, R. Acad. Scz. Paris, 258 (1964), 4413–4416. |
[27] |
B. Tan, S. Li, Modified inertial projection and contraction algorithms with non-monotonic step sizes for solving variational inequalities and their applications, Optimization, 73 (2024), 793–832. http://dx.doi.org/10.1080/02331934.2022.2123705 doi: 10.1080/02331934.2022.2123705
![]() |
[28] |
D. Thong, D. Hieu, Inertial subgradient extragradient algorithms with line-search process for solving variational inequality problems and fixed point problems, Numer. Algor., 80 (2019), 1283–1307. http://dx.doi.org/10.1007/s11075-018-0527-x doi: 10.1007/s11075-018-0527-x
![]() |
[29] |
P. Vuong, Y. Shehu, Convergence of an extragradient-type method for variational inequality with applications to optimal control problems, Numer. Algor., 81 (2019), 269–291. http://dx.doi.org/10.1007/s11075-018-0547-6 doi: 10.1007/s11075-018-0547-6
![]() |
[30] |
H. Xu, T. Kim, Convergence of hybrid steepest-descent methods for variational inequalities, J. Optim. Theory Appl., 119 (2003), 185–201. http://dx.doi.org/10.1023/B:JOTA.0000005048.79379.b6 doi: 10.1023/B:JOTA.0000005048.79379.b6
![]() |
[31] |
J. Yang, H. Liu, Z. Liu, Modified subgradient extragradient algorithms for solving monotone variational inequalities, Optimization, 67 (2018), 2247–2258. http://dx.doi.org/10.1080/02331934.2018.1523404 doi: 10.1080/02331934.2018.1523404
![]() |
[32] |
Y. Yao, O. Iyiola, Y. Shehu, Subgradient extragradient method with double inertial steps for variational inequalities, J. Sci. Comput., 90 (2022), 71. http://dx.doi.org/10.1007/s10915-021-01751-1 doi: 10.1007/s10915-021-01751-1
![]() |
[33] | Y. Yu, T. Yin, Weak convergence of a self-adaptive Tseng-type algorithm for solving variational inclusion problems, U.P.B. Sci. Bull., Series A, 85 (2023), 51–58. |
[34] | Y. Yu, T. Yin, Strong convergence theorems for a nonmonotone equilibrium problem and a quasi-variational inclusion problem, J. Nonlinear Convex Anal., 25 (2024), 503–512. |
[35] |
Z. Jing, Z. Liu, E. Vilches, C. Wen, J. Yao, Optimal control of an evolution hemivariational inequality involving history-dependent operators, Commun. Nonlinear Sci., 103 (2021), 105992. http://dx.doi.org/10.1016/j.cnsns.2021.105992 doi: 10.1016/j.cnsns.2021.105992
![]() |
1. | Nguyen Buong, A first order dynamical system and its discretization for a class of variational inequalities, 2025, 458, 03770427, 116341, 10.1016/j.cam.2024.116341 | |
2. | Prasit Cholamjiak, Zhongbing Xie, Min Li, Papinwich Paimsang, Double inertial subgradient extragradient algorithm for solving equilibrium problems and common fixed point problems with application to image restoration, 2024, 03770427, 116396, 10.1016/j.cam.2024.116396 |