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 45 Issue 6
Article Contents

Xin-yu ZHANG, Ying LYU. A Generalization and Application of Lax-Milgram Theorem[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(6): 29-32. doi: 10.13718/j.cnki.xsxb.2020.06.006
Citation: Xin-yu ZHANG, Ying LYU. A Generalization and Application of Lax-Milgram Theorem[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(6): 29-32. doi: 10.13718/j.cnki.xsxb.2020.06.006

A Generalization and Application of Lax-Milgram Theorem

More Information
  • Corresponding author: Ying LYU
  • Received Date: 11/02/2020
    Available Online: 20/06/2020
  • MSC: O177.1

  • In this paper, the bilinear functional condition required in the classical Lax-Milgram theorem has been generalized to be linear with only one argument. Under the appropriate weak lower semi-continuous constraint, we prove that the variational functional l(x)=1/pa(x, x)-b(x)has a unique solution and apply it to the existence and uniqueness of the solutions of the p-Laplace boundary value problem.
  • 加载中
  • [1] 莫里斯·克莱因, 著.古今数学思想-第四册[M].邓东皋, 张恭庆, 译.上海: 上海科学技术出版社, 2002: 160-183.

    Google Scholar

    [2] 江泽坚, 孙善利.泛函分析[M].北京:高等教育出版社, 2005:58-70.

    Google Scholar

    [3] 张世清.泛函分析及其应用[M].北京:科学出版社, 2018:21-25.

    Google Scholar

    [4] LAX PD, MILGRAM A N. Parabolic Equations[M]//Contributions to the Theory of Partial Differential Equations. (AM-33).Princeton University Press, 1954: 167-190.

    Google Scholar

    [5] STAMPACCHIA G.Formes Bilinéaires Coercitives Sur Les Ensembles Convexes[J].CRAcadSciParis, 1964, 258:4413-4416.

    Google Scholar

    [6] BABUŠKA I.Error-Bounds for Finite Element Method[J].NumerMath, 1971, 16(4):322-333.

    Google Scholar

    [7] BREZZI F.On the Existence, Uniqueness and Approximation of Saddle Point Problems Arising from Lagrangian Multipliers[J].Revue Francaise Automatique, Informatiqe, Recherche Operationnelle Anal Numer, 1974, 8(R2):129-151. doi: 10.1051/m2an/197408R201291

    CrossRef Google Scholar

    [8] DUC D M, LOC N H, PHI L L. Nonlinear Versions of Stampacchia and Lax-Milgram Theorems and Applications to p-Laplace Equations[J].Nonlinear Anal, 2008, 68(4):925-931. doi: 10.1016/j.na.2006.11.048

    CrossRef Google Scholar

    [9] DHARA R N.Existence and Uniqueness of the Solution of Linear Degenerate PDEs in Weighted Sobolev Space with Nonhomogeneous Boundary Condition[J].MathMethods ApplSci, 2008, 41(2):544-558.

    Google Scholar

    [10] DANIEL R, SIMEON R.Fixed Points of Polarity Type Operators[J].JMathAnalAppl, 2018, 467(2):1208-1232.

    Google Scholar

    [11] CARUSO F, CEPARANO M C, MORGAN J.Uniqueness of Nash Equilibrium in Continuous Two-Player Weighted Potential Games[J].JMath Anal Appl, 2018, 459(2):1208-1221. doi: 10.1016/j.jmaa.2017.11.031

    CrossRef Google Scholar

    [12] COURTADET, FATHI M, PANANJADY A.Existence of Stein Kernels Under a Spectral Gap, and Discrepancy Bounds[J].AnnInstHenri Poincaré ProbabStat, 2019, 55(2):777-790.

    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(4776) PDF downloads(390) Cited by(0)

Access History

Other Articles By Authors

A Generalization and Application of Lax-Milgram Theorem

    Corresponding author: Ying LYU

Abstract: In this paper, the bilinear functional condition required in the classical Lax-Milgram theorem has been generalized to be linear with only one argument. Under the appropriate weak lower semi-continuous constraint, we prove that the variational functional l(x)=1/pa(x, x)-b(x)has a unique solution and apply it to the existence and uniqueness of the solutions of the p-Laplace boundary value problem.

  • 1907年,法国数学家Fréchet证明了:对于定义在平方可积函数空间L2上的每一个连续(有界)线性泛函U(f),存在L2中唯一的u(x),使得对L2中的每一个f都成立

    1909年,匈牙利数学家F.Riesz把上述结果推广到Lp空间(p≥1). 1927年,匈牙利数学家von Neumann引进了抽象的Hilbert公理化定义,自然地,就有相应的一般Hilbert空间上连续线性泛函的表示定理[1-3].文献[4]推广了Fréchet-Riesz表示定理.文献[5]推广了Lax-Milgram定理,证明了变分不等式中的基本定理.文献[6-7]各自独立地推广了Lax-Milgram定理,实际上,他们给出的两种不同形式的条件是等价的.关于Lax-Milgram定理的最新应用可见文献[8-12].特别地,文献[8]试图给出经典Lax-Milgram定理和Stampacchia定理的非线性观点,但是假设了一些很难验证的条件,即使退化到双线性的情况,也得不到最小值点的唯一性,因为证明过程中没有充分利用双线性性质和强制性.

    以上几个关于Fréchet-Riesz表示定理的推广都涉及双线性连续泛函,一个自然的问题是:当定义在Hilbert空间或者更一般的Banach空间的乘积空间上的二元泛函若仅对某一变量线性,而不是双线性时,能得到什么结果?我们研究p-Laplace算子相应的Dirichlet零边值问题

    其中Δpμ(x)→Δpμ=div(|▽μ|p-2μ),自然地会出现

    我们观察到:当p>1但p≠2时,aμ不是线性的,但对ν是线性的.

    本文先证明一个抽象的定理,然后应用于上述具体的问题.

    定理1  设X是实数域$\mathbb{R}$上的自反Banach空间,考虑二元映射a(xy):X×X$\mathbb{R}$.假设bX$\mathbb{R}$是连续线性泛函;a对第二个变量y是线性的,a(xx)在X上是弱下半连续的,即:

    (i) $\forall x\in X$,若{xn}⊂X${{x}_{n}}\xrightarrow{w}x$,则

    (ii) 存在cd>0,使得$\forall x\in X$,有

    则定义在X上的泛函

    X上达到下确界,即存在$\widetilde{x}\in X$,使得$I\left( \widetilde{x} \right)=\underset{x\in X}{\mathop{\inf }}\, \left\{ I\left( x \right) \right\}=e$,且$\widetilde{x}$满足方程

    进一步,如果a(xx)是严格凸泛函,则最小值点$\widetilde{x}$是唯一的.

      步骤1  证明I(x)在X上弱下半连续.由于b(x)是X上的连续线性泛函,故bX*.又由弱收敛的定义知,若${{x}_{n}}\xrightarrow{w}x$,则∀fX*,有f(xn)→f(x).自然地有b(xn)→b(x),故b是弱连续的.因此b既弱下半连续又弱上半连续,因此(-b)弱下半连续.又已知a(xx)在X上弱下半连续,故I(x)= $\frac{1}{p}a\left( x, x \right)-b\left( x \right)$X上是弱下半连续的.

    步骤2  证明I(x)在X上是强制的.事实上,由(ii)知

    步骤3  证明-∞ < e=inf{I(x):xX} < +∞.事实上,由步骤2中已证明的I(x)的强制性知,存在常数M>0,使当‖x‖≥M时,有I(x)≥1.又因当‖x‖≤M时成立

    e≥-‖b‖·M>-∞.又由已知假设a(xx)≤dxp

    I(x)≢+∞,e<+∞.

    步骤4  由Tonelli定理[3]知,I(x)在X上存在最小值点,记为$\tilde{x}$.令φ(ε)=I($\tilde{x}$ +εω),ε>0,ωX.因为$\tilde{x}$最小化I(x),故∀ε>0,有φ(ε)≥φ(0),故有一阶必要条件φ′(0)=0.下面具体计算φ′(0).注意到

    φ(ε)关于ε求导,有

    由此即得${\tilde{x}}$满足的方程.

    步骤5  若a(xx)在X上是严格凸的,下面进一步证明最小值点是唯一的.设${{\tilde{x}}_{1}}$${{\tilde{x}}_{2}}$是最小值点,由a(xx)严格凸,b(x)线性,故b也是凸的,I(x)在X上是严格凸的,则

    $I\left( \frac{{{{\tilde{x}}}_{1}}+{{{\tilde{x}}}_{2}}}{2} \right)$$\frac{{{{\tilde{x}}}_{1}}+{{{\tilde{x}}}_{2}}}{2}$也是最小值点.进一步由I的表示式有

    由于a是严格凸的,(1)式成立当且仅当${{{\tilde{x}}}_{1}}\text{=}{{{\tilde{x}}}_{2}}$.

    定理1的应用

    Ω${{\mathbb{R}}^{n}}$中的有界开集,x=(x1,…,xn)∈Ω,1<p<+∞,考虑p-Laplace算子

    给定fLq(Ω),其中qp的共轭指数,即$\frac{1}{p}+\frac{1}{q}=1$.考虑以下的Dirichlet边值问题:

    我们在上述假设下利用定理1来证明问题(2)有唯一解.

    定义Sobolev空间

    作为W1,p(Ω)的子空间,W01,p(Ω)的标准范数是‖μW01,p(Ω)=‖μLp+Lp.由Poincaré-Friedrics不等式[3]可知,‖Lp是‖μW01,p(Ω)的等价范数.设X=W01,p(Ω),则由1 < p < +∞,故W01,p(Ω)是自反的Banach空间.∀μνW01,p(Ω),定义

    下面用散度定理(或分部积分公式)可得

    因为p>1,且$a(\mu , \mu ) = \int_\mathit{\Omega } {{{\left| {{\nabla _\mu }} \right|}^p}} {\rm{d}}x = \mu _{W_0^{1, p}\left( \mathit{\Omega } \right)}^p$是弱下半连续严格凸泛函,而b(μ)是线性连续泛函,由定理1知,对定义在W01,p(Ω)上的泛函$I\left( \mu \right) = \frac{1}{p}a\left( {\mu , \mu } \right) - b\left( \mu \right)$,存在唯一的$\tilde \mu $W01,p(Ω),使得I($\tilde \mu $)=inf {I($\tilde \mu $):μW01,p(Ω)},且$\tilde \mu $满足

    注意到$\left\langle {{a_\mu }(\tilde \mu , \tilde \mu ), \nu } \right\rangle = {\left. {\frac{{\rm{d}}}{{{\rm{d}}\varepsilon }}a(\tilde \mu + \varepsilon \nu , \tilde \mu )} \right|_{\varepsilon = 0}}$,令$g(\varepsilon ) = a(\tilde \mu + \varepsilon \nu , \tilde \mu )$,则有

    $g'(0) = (p - 1)\int_\mathit{\Omega } | \nabla \tilde \mu {|^{p - 2}}\nabla \tilde \mu \cdot \nabla \nu {\rm{d}}x$.又因$a(\tilde \mu , \nu ) = \int_\mathit{\Omega } | \nabla \tilde \mu {|^{p - 2}}\nabla \tilde \mu \cdot \nabla \nu {\rm{d}}x$,故$\frac{1}{p}\left\langle {{a_n}(\tilde \mu , \tilde \mu ), \nu } \right\rangle + \frac{1}{p}a(\tilde \mu , \nu ) - b(\nu ) = 0$,即

    由散度定理可得

    由Reymond变分基本引理[3]

Reference (12)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return