Message Board

Dear readers, authors and reviewers,you can add a message on this page. We will reply to you as soon as possible!

2020 Volume 42 Issue 9
Article Contents

Qian-fen GONG, Jun AN. Convergence Analysis of Proximal Point Algorithm for a Strongly Pseudomonotone Equilibrium Problem[J]. Journal of Southwest University Natural Science Edition, 2020, 42(9): 101-105. doi: 10.13718/j.cnki.xdzk.2020.09.012
Citation: Qian-fen GONG, Jun AN. Convergence Analysis of Proximal Point Algorithm for a Strongly Pseudomonotone Equilibrium Problem[J]. Journal of Southwest University Natural Science Edition, 2020, 42(9): 101-105. doi: 10.13718/j.cnki.xdzk.2020.09.012

Convergence Analysis of Proximal Point Algorithm for a Strongly Pseudomonotone Equilibrium Problem

More Information
  • Received Date: 24/11/2018
    Available Online: 20/09/2020
  • MSC: O177.91

  • In this paper, the proximal point algorithm is improved by using the inertial technique for solving a strongly pseudomonotone equilibrium problem.The convergence speed of the modified iteration is accelerated and the stability of numerical method is improved.Moreover, the strong convergence theorems about the unique solution of the strongly pseudomonotone equilibrium problem are established under some suitable conditions defined on the iterative parameters.
  • 加载中
  • [1] BLUM E, OETTLI W.From Optimization and Variational Inequalities to Equilibrium Problems [J].Math Stndent, 1994, 63: 123-145.

    Google Scholar

    [2] FLÅM S D, ANTIPIN A S.Equilibrium Programming Using Proximal-Like Algorithms [J].Mathematical Programming, 1996, 78(1): 29-41.

    Google Scholar

    [3] KONNOV I V.Application of the Proximal Point Method to Nonmonotone Equilibrium Problems [J].Journal of Optimization Theory and Applications, 2003, 119(2): 317-333.

    Google Scholar

    [4] VAN HIEU D, MUU L D, ANH P K.Parallel Hybrid Extragradient Methods for Pseudomonotone Equilibrium Problems and Nonexpansive Mappings [J].Numerical Algorithms, 2016, 73(1): 197-217.

    Google Scholar

    [5] STRODIOT J J, VUONG P T, NGUYEN T T V.A Class of Shrinking Projection Extragradient Methods for Solving Non-Monotone Equilibrium Problems in Hilbert Spaces [J].Journal of Global Optimization, 2016, 64(1): 159-178. doi: 10.1007/s10898-015-0365-5

    CrossRef Google Scholar

    [6] 闻道君.关于伪单调平衡问题和不动点问题的粘滞-次梯度方法[J].数学进展, 2017, 46(2): 303-312.

    Google Scholar

    [7] 闻道君, 张蓉, 宋树枝.分层混合均衡问题和不动点问题的等价性[J].西南大学学报(自然科学版), 2018, 40(8): 110-113.

    Google Scholar

    [8] 严月月, 钟艳丽, 郭楠馨.求解双层弹性膜单侧接触问题的Uzawa算法[J].重庆工商大学学报(自然科学版), 2018, 35(6): 75-78.

    Google Scholar

    [9] DONG Q L, LU Y Y, YANG J F.The Extragradient Algorithm with Inertial Effects for Solving the Variational Inequality [J].Optimization, 2016, 65(12): 2217-2226. doi: 10.1080/02331934.2016.1239266

    CrossRef Google Scholar

    [10] VAN HIEU D.Convergence Analysis of a New Algorithm for Strongly Pseudomontone Equilibrium Problems [J].Numerical Algorithms, 2018, 77(4): 983-1001. doi: 10.1007/s11075-017-0350-9

    CrossRef Google Scholar

    [11] BAUSCHKE H H, COMBETTES P L.Convex Analysis and Monotone Operator Theory in Hilbert Spaces [M].Berlin: Springer, 2011.

    Google Scholar

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(1716) PDF downloads(17) Cited by(0)

Access History

Other Articles By Authors

Convergence Analysis of Proximal Point Algorithm for a Strongly Pseudomonotone Equilibrium Problem

Abstract: In this paper, the proximal point algorithm is improved by using the inertial technique for solving a strongly pseudomonotone equilibrium problem.The convergence speed of the modified iteration is accelerated and the stability of numerical method is improved.Moreover, the strong convergence theorems about the unique solution of the strongly pseudomonotone equilibrium problem are established under some suitable conditions defined on the iterative parameters.

  • H为一实Hilbert空间, 其内积和范数分别表示为〈·, ·〉和║·║.设CH的一个非空闭凸子集, fC×C→ℝ为一二元函数.考虑如下均衡问题:求一点x*C, 使

    Ω表示均衡问题(1)的解集, 即Ω={xC:f(x, y)≥0, ∀yC}.众所周知, 均衡问题是一个包含变分不等式和不动点问题的广义的数学模型, 也是非线性分析的重要组成部分[1-8].

    文献[9]介绍了一个惯性-近似点方法, 并应用于求解变分不等式问题的近似解, 数值实验显示惯性-近似点方法具有更好的收敛性和稳定性.文献[10]利用两点初始值改进传统数值方法的单点初始值, 介绍了求解强伪单调均衡问题的两步近似点方法, 该方法的优点是收敛性分析相对简单, 缺点是每一步迭代中的xnyn的交叉引用增加了近似点方法运算的复杂度.

    在文献[9-10]基础上, 本文提出了求解强伪单调均衡问题的惯性-近似点方法:设x0x1C, 定义序列{xn}和{yn}

    其中θn∈[0, 1], λn∈(0, +∞).本文的目的是利用惯性技巧改进文献[10]的近似点方法, 去掉迭代中xnyn的交叉引用以加快收敛速度, 提高数值方法的稳定性, 并在一定条件下建立了关于强伪单调均衡问题解的强收敛定理, 所得的方法和结果改进并推广了文献[9-10]的结论.

1.   预备知识
  • C为Hilbert空间H的非空闭凸子集, f:C×C→ℝ为二元函数, fC上单调, γ-强单调, 伪单调和γ-强伪单调的定义参见文献[1, 10].

    定义1[10]  设f:C×C→ℝ为二元函数, 称f满足Lipschtiz-型条件:如果存在常数c1>0, c2>0使得

    为了方便描述和证明, 假设f:C×C→ℝ满足下列条件:

    (A1) f(x, x)=0, ∀xC;

    (A2) f(x, y)是γ-强伪单调映象;

    (A3) f(x, y)在C上满足Lipschtiz-型条件;

    (A4) 对任意xC, f(x, ·)是凸的且下半连续, 并满足$ \mathop {\lim }\limits_{t \to 0} f\left( { \boldsymbol{\cdot} , tz + \left( {1 - t} \right)y} \right) = f\left( {\boldsymbol{\cdot} , \forall y, z \in C} \right) $.

    gC→ℝ为一适当的凸且下半连续泛函, 且常数λ>0.定义如下近似函数

    引理1[11]  对任意xH, yC和常数λ>0, 有下列不等式:

    由引理1, 如果x=Proxλg(x), 则有$ x \in \arg \;\rm{min}\left\{ {{\rm{g}}\left( y \right):y \in C} \right\}: = \left\{ {x \in C:g\left( x \right) = \mathop {\min }\limits_{y \in C} g\left( y \right)} \right\} $成立.

    引理2[11]  设H为一实Hilbert空间, 对λ∈[0, 1], 有下列不等式:

2.   主要结果
  • 定理1  设C为Hilbert空间H的非空闭凸子集, 且f:C×C→ℝ满足条件(A1)-(A4).如果记x*为均衡问题(1)的解, 且γc2, 则由式(2)定义的序列{xn}满足不等式

      由式(2)和近似函数的定义得

    其中

    再结合优化条件可知

    即存在

    使得

    因为f(wn, ·)是凸函数, 所以fn(x)为强凸函数, 则

    其中任意gn∈∂fn(yn).取x=x*, gn=gn*, 结合式(4)和式(5)得

    进一步结合fn(x)的定义和条件(A3)得

    又因为f(x, y)是强伪单调映射且

    所以

    由式(7)可得

    进一步整理得

    类似地, 由式(2)和(6)得

    并结合式(7)和条件(A3)得

    进一步整理得

    由条件γc2, 式(8)和(10)可得

    又因为

    由式(11)和(12)可得

    同时, 由式(2)和引理2得

    并且

    结合式(13),(14)和(15)得

    定理2  设C为Hilbert空间H的非空闭凸子集, 且f:C×C→ℝ满足条件(A1)-(A4).如果序列$ \left\{ {{\lambda _n}} \right\}_{n = 1}^\infty 和 \left\{ {{\theta _n}} \right\}_{n = 1}^\infty $满足条件λn-1λn, θnθn-1, 并且

    则由式(2)定义的迭代序列{xn}强收敛到强伪单调均衡问题(1)的唯一解x*Ω.

      由定理1, 式(16)整理得

    其中

    因为

    结合θnθn-1和式(17)得

    利用式(19),(20)得An-1Bn-Cn≥0, 结合式(20)进一步可得

    不难验证

    又因为

    所以

    即{xn}强收敛到强伪单调均衡问题(1)的唯一解x*Ω.

    定理3  设C为Hilbert空间H的非空闭凸子集, 且f:C×C→ℝ满足条件(A1)-(A4).对任意给定x0C, 如下定义近似点序列{xn}:

    其中$ 0 < {\lambda _{n - 1}} \le {\lambda _n} <\frac{1}{{2{c_1}}}且 \gamma \ge {c_2} $, 则序列{xn}强收敛到强伪单调均衡问题(1)的唯一解x*Ω.

      取θn=0, 即wn=xn, 惯性-近似方法(2)退化为近似点逼近算法(23), 结论由定理2类似可证.

Reference (11)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return