数学物理学报, 2022, 42(6): 1611-1618 doi:

论文

一类反三角算子矩阵的本质谱

花蕊, 齐雅茹,

内蒙古工业大学理学院, 呼和浩特 010051

The Essential Spectrum of a Class of Anti-Triangular Operator Matrices

Hua Rui, Qi Yaru,

College of Sciences, Inner Mongolia University of Technology, Hohhot 010051

通讯作者: 齐雅茹, E-mail: qiyaru@imut.edu.cn

收稿日期: 2022-02-9  

基金资助: 国家自然科学基金.  12261065
内蒙古自然科学基金.  2021LHMS01004
内蒙古自然科学基金.  2022MS01005
自治区直属高校基本科研业务费项目.  JY20220151

Received: 2022-02-9  

Fund supported: the Natural Science Foundation of China.  12261065
the National Natural Science Foundation of Inner Mongolia.  2021LHMS01004
the National Natural Science Foundation of Inner Mongolia.  2022MS01005
the Basic Science Research Fund in the Universities Directly under the Inner Mongolia Autonomus Region.  JY20220151

Abstract

In this paper, the essential spectrum of a class of unbounded unself-adjoint anti-triangular operator matrices is studied. Firstly, we describe the essential spectrum of operator matrices by using the quadratic operator pencil and the properties of its operator entries, and estimate the essential spectrum of the whole operator matrix. On this basis, the accumulation point of the non-real spectrum of the operator matrix is analyzed.

Keywords: Anti-triangular operator matrices ; Essential spectrum ; Accumulation points of spectrum

PDF (335KB) 元数据 多维度评价 相关文章 导出 EndNote| Ris| Bibtex  收藏本文

本文引用格式

花蕊, 齐雅茹. 一类反三角算子矩阵的本质谱. 数学物理学报[J], 2022, 42(6): 1611-1618 doi:

Hua Rui, Qi Yaru. The Essential Spectrum of a Class of Anti-Triangular Operator Matrices. Acta Mathematica Scientia[J], 2022, 42(6): 1611-1618 doi:

1 引言

算子矩阵是以线性算子为元素的矩阵, 在数学物理问题中有广泛的应用. 例如, 流体力学、磁流体力学、弹性力学和量子力学等领域中涉及的微分方程的有关研究可转化为其相关算子矩阵的谱、半群、变分原理及特征值问题的研究. 而实际应用与高阶微分方程相关联的算子矩阵往往是一类非自伴无界$ 2\times 2 $阶反三角算子矩阵, 其主对角元中有一个元素为零, 且内部算子也多为微分算子. 因此, 无界非自伴反三角算子矩阵具有重要的研究意义[1-6]. 2007–2009年, 学者Jacob, Trunk在文献[79]中讨论了与弹性力学和流体力学相关的一类反三角算子矩阵, 研究了其本质谱、谱及非实谱的聚点问题, 并在此基础上给出了其半群的解析性和Riesz基的性质. 在2012年, Artamonov[10]分析了一类与二阶微分方程有关的反三角算子矩阵, 研究了其半群的指数衰减性, 并刻画了谱的范围. 2016年, Jacob, Langer等[11]作者对一类反三角算子矩阵的特征值的变分原理进行了研究. 2018年, Tretter, Trunk[12]等学者利用数值域和二次数值域刻画了一类反三角算子矩阵的谱范围, 并将其结果应用在流体力学中水平管道内理想的不可压缩流体的横向小振动方程中.

本质谱在线性算子的定量研究及控制论等领域有重要的应用. 对于自伴算子, 谱可以分解为本质谱和代数重数有限的离散点谱; 而对于预解集非空的非自伴算子, 当本质谱为空集时, 谱集也仅由代数重数有限的离散点谱构成, 这对于利用分离变量法研究其特征值问题提供了基础. 但本质谱难以利用数值的方法进行研究, 所以有必要从分析的角度刻画其性质和分布.

本文讨论一类无界非自伴的反三角算子矩阵的本质谱, 具体如下

$ \begin{equation} {\cal A}=\Bigg[\begin{array}{cc} 0\; &B \\ -B^{*}\; &-D\\ \end{array}\Bigg]: {\cal D}(B^{*})\times {\cal D}(B)\subset H_{1}\times H_{2}\rightarrow H_{1}\times H_{2}, \end{equation} $

其中$ H_{1}, H_{2} $为Hilbert空间, $ B:{\cal D}(B)\subset H_{2}\rightarrow H_{1} $为稠定闭算子且$ 0\in\rho(B) $, $ D:{\cal D}(D)\subset H_{2}\rightarrow H_{2} $为无界算子. 该类反三角算子矩阵可描述弹性力学中的弦振动方程, 例如[13-14]

$ \begin{equation} \left\{ \begin{array}{l} { } \frac{\partial^{2}u(x, t)}{\partial t^{2}}-\frac{\partial}{\partial x}\big(a(x)\frac{\partial} {\partial x}u(x, t)\big)-\frac{\partial^{2}}{\partial t\partial x}\big(b(x)\frac{\partial}{\partial x}u(x, t)\big)\\ { } +\int_{0}^{1}k(x, y)\frac{\partial^{2}}{\partial t\partial y}u(y, t){\rm d}y=0, (x\in(0, 1), t>0), \\ u(0, t)=u(1, t)=0, (t\geq 0), \\ { } u(x, 0)=u_{0}(x), \frac{\partial}{\partial t}u(x, 0)=u_{1}(x), (0\leq x\leq 1). \end{array} \right. \end{equation} $

$ a(x), b(x)\in C^{1}(0, 1), $对于$ x\in[0, 1] $$ a(x)\geq\lambda_{1}>0, b(x)\geq v_{1}>0, $$ \frac{\partial}{\partial y}k(x, y)\in L^{2}([0, 1]\times[0, 1]) $, $ \lambda_1, v_1 $的定义见下文. 对任意的$ p(x)\in L^{2}(0, 1) $满足$ \int^{1}_{0}\int^{1}_{0}\frac{\partial}{\partial y}k(x, y)p(x)p(y){\rm d}x{\rm d}y \leq 0. $因此弦振动方程(1.2)可理解为Hilbert空间$ L^{2}(0, 1) $中二阶微分方程

$ \begin{equation} \nonumber \left\{ \begin{array}{l} \ddot{z}(t)+T_{2}\dot{z}(t)+T_{3}\dot{z}(t)+T_{1}z(t)=0, \\ z(0)=z_{0}, \dot{z}(0)=z_{1} \end{array} \right. \end{equation} $

的Cauchy问题[15-17]. 此时方程的系数为

$ \begin{equation} T_{1}=-\frac{\partial}{\partial x}\big(a(x)\frac{\partial}{\partial x}\cdot\big), {\cal D}(T_{1})=\{u\in H^{2}(0, 1), u(0)=u(1)=0\}, \end{equation} $

$ \begin{equation} T_{2}=\int^{1}_{0}k(x, y)\frac{\partial}{\partial y}\cdot {\rm d}y, {\cal D}(T_{2})=\{u\in H^{1}(0, 1), u(0)=u(1)=0\}, \end{equation} $

$ \begin{equation} T_{3}=-\frac{\partial}{\partial x}\big(b(x)\frac{\partial}{\partial x}\cdot\big), {\cal D}(T_{3})={\cal D}(T_{1})\subset{\cal D}(T_{2}), \end{equation} $

其中$ H^{1}(0, 1) $, $ H^{2}(0, 1) $为1阶和2阶的Sobolev空间, 且可验证$ T_{1}, T_{3} $为正定算子[18], $ T_{2} $为非负算子. 我们记$ B=T_{1}^{\frac{1}{2}}, D=T_{2}+T_{3}, $则在Hilbert空间$ L^{2}(0, 1) $中, $ B\mbox{、}\ D $为无界正定算子, $ D $是相对于$ B $有界的算子. 上面提及的$ \lambda_{1}={\rm inf}\{\lambda| \lambda\in\sigma(B)\}, v_{1}={\rm inf}\{\lambda| \lambda\in\sigma(D)\}. $若令$ x_{1}=Bz, x_{2}=\dot{z}, $则二阶微分方程可转化为抽象的一阶方程

$ \begin{equation} \nonumber \left\{ \begin{array}{l} { } \frac{\rm d}{{\rm d}t}\left(\begin{array}{cc} x_{1} \\ x_{2} \\ \end{array}\right)= \Bigg[\begin{array}{cc} 0\; &B \\ -B^{*}\; &-D\\ \end{array}\Bigg]\left(\begin{array}{cc} x_{1} \\ x_{2} \\ \end{array}\right)={\cal B}\left(\begin{array}{cc} x_{1} \\ x_{2} \\ \end{array}\right), \\ x_{1}(0)=Bz_{0}, x_{2}(0)=z_{1}. \end{array} \right. \end{equation} $

因此原问题的研究可转化为形如无界算子$ {\cal B} $的研究. 本节所要讨论的算子$ {\cal A} $与算子$ {\cal B} $为同一类型, 即次对角线上的元素互为反共轭算子. 当$ D $为相对于$ B $有界的自伴算子且相对界小于1时, 反三角算子矩阵$ {\cal A} $$ {\cal J} $ -自伴算子, 其中$ {\cal J}=\left[\begin{array}{cccc}I\; &0\\ 0\; &-I \end{array} \right] $. 对于$ \left(\begin{array}{cccc} x_{1}\\y_{1} \end{array} \right), \left(\begin{array}{cccc} x_{2}\\y_{2}\end{array}\right)\in H_{1}\times H_{2} $, 定义Hilbert空间$ H_{1}\times H_{2} $中的二元运算

$ (H_{1}\times H_{2}, [\cdot, \cdot]) $是Krein空间[19]. 此时算子矩阵$ {\cal A} $为Krein空间$ (H_{1}\times H_{2}, [\cdot, \cdot]) $中的自伴算子.

$ T $为Hilbert空间$ H $中的线性算子, $ {\cal D}(T), {\cal R}(T) $分别表示线性算子$ T $的定义域和值域, $ T^{*} $$ T $的共轭算子. 并记$ {\cal N}(T)=\{x\in{\cal D}(T)|Tx=0\} $为线性算子$ T $的零空间, $ {\rm nul}(T) $$ {\rm def}(T) $分别表示零空间$ {\cal N}(T) $$ H/{\cal R}(T) $的维数.

定义1.1  设$ T $为Hilbert空间$ H $中的闭算子, 则$ T $的正则集$ \rho(T) $定义为

$ \begin{equation} \nonumber \rho(T):=\{\lambda\in {\Bbb C}| T-\lambda I\ \mbox{ 是单射}, \ {\cal R}(T-\lambda I)=H\}, \end{equation} $

并称集合$ \sigma(T):=\lambda\in{\Bbb C}\setminus \rho(T) $$ T $的谱. $ T $的点谱定义为

$ T $的近似点谱定义为

定义1.2[20]   设$ T $为Hilbert空间$ H $中的稠定算子. 若对于任意的$ x\in {\cal D}(T), $满足$ \langle Tx, x\rangle>0 $, 则称$ T $为正定算子. 若$ {\rm Re}\langle Tx, x\rangle\geq0 $, 则称$ T $为增生算子.

定义1.3[21]   设$ T $为Banach空间$ X $中的闭算子. 若$ {\cal R}(T) $是闭的且

则称$ T $是Fredholm算子, 并且称$ {\rm ind}(T)={\rm nul}(T)-{\rm def}(T) $为算子$ T $的指标. 闭算子$ T $的本质谱定义如下

引理1.1[21]   设$ T $为Banach空间$ X $上的闭算子, $ S $$ X $中相对于$ T $紧的线性算子, 则$ \sigma_{ess}(T+S)=\sigma_{ess}(T). $

引理1.2[22]   设$ T $为Banach空间$ X $上的闭算子, 则$ \sigma(T)=\sigma_{ess, 1}(T)\cup\sigma_{p, norm}(T), $其中$ \sigma_{ess, 1}(T)={\Bbb C}\backslash\triangle_{1}(T), $$ \triangle_{1}(T)=\{z\in\triangle_{2}(T)| z $的去心邻域包含于$ \rho(T) \}$, $ \triangle_{2}(T)=\{z\in{\Bbb C}| T-zI $是Fredholm算子且$ {\rm ind}(T-zI)=0\} $, $ \sigma_{p, norm}(T)=\{z\in{\Bbb C}| z $$ \sigma(T) $的孤立点, $ T-zI $为Fredholm算子且$ {\rm ind}(T-zI)=0\} $.

定义1.4[23]   设$ T $为Krein空间$ {\cal K} $中的自伴算子. 若$ \lambda\in\sigma_{ap}(T) $且存在线性流形$ {\cal K}_{0}\subset {\cal K}, {\rm dim} {\cal K}_{0}^{\perp}<\infty, $使得对任意序列$ (x_{n})\subset {\cal K}_{0}\cap{\cal D}(T) $满足

$ \liminf \limits _{ n\to \infty }[x_{n}, x_{n}]>0 , $则称$ \lambda $$ T $$ \pi_{+} $型谱点, 并将满足上述条件的$ \lambda $构成的集合记为$ \sigma_{\pi_{+}}(T) $.

引理1.3[23]   设$ T $为Krein空间$ {\cal K} $中的自伴算子且$ \rho(T)\neq\emptyset $. 若存在区间$ [a, b] $使得$ [a, b]\cap\sigma(T)\subset\sigma_{\pi+}(T) $成立, 并且$ [a, b] $中的每一点都是$ \rho(T) $的聚点, 则存在包含$ [a, b] $的开邻域$ \Omega\subset{\Bbb C} $使得$ \Omega\backslash {\Bbb R}\subset\rho(T) $.

2 主要结果及其证明

命题2.1   设$ {\cal A} $为(1.1) 式定义的反三角算子矩阵, 则$ {\cal A} $具有有界逆. 进一步, 当$ D $为相对于$ B $有界的自伴算子且相对界小于1时, $ \sigma({\cal A}) $关于实轴对称.

   由$ 0\in\rho(B) $可知$ B^{-1}, (B^{*})^{-1} $均为有界算子, 且

$ {\cal A} $具有有界逆. 进一步, 当$ D $为相对于$ B $有界的自伴算子且相对界小于1时, $ {\cal A} $在Krein空间$ (H_{1}\times H_{2}, [\cdot, \cdot]) $上是自伴算子. 根据文献[19, 推论6.3] 可知$ \sigma({\cal A}) $关于实轴对称.

证毕.

下面刻画反三角算子矩阵$ {\cal A} $的点谱、谱与本质谱.

定理2.1   设$ {\cal A} $为由(1.1) 式给定的分块算子矩阵, 若$ s\in {\Bbb C} $, 则$ {\cal R}({\cal A} - sI) $是闭的当且仅当$ {\cal R}(L(s)) $是闭的, 并且

$ s\in {\Bbb C}\backslash\sigma_{ess}({\cal A}) $, 则$ {\rm nul}({\cal A}-sI)={\rm nul} L(s) $$ {\rm def}({\cal A}-sI)={\rm def}L(s) $. 其中二次算子族$ L(s)=(B^{*})^{-1}(D+s)B^{-1}s+I. $

   当$ s=0 $时, 由命题2.1知算子$ {\cal A} $$ L(0) $具有有界逆, 故以上结论显然成立. 当$ s\neq0 $时,

$ \begin{equation} {\cal A}^{-1}-\frac{1}{s}I=\Bigg[\begin{array}{cc} I\; &s(B^{*})^{-1} \\ 0\; &I\\ \end{array}\Bigg] \left[\begin{array}{cc} { } M(\frac{1}{s})\; &0 \\ 0\; &{ } -\frac{1}{s}I\\ \end{array}\right] \Bigg[\begin{array}{cc} I\; &0 \\ -sB^{-1}\; &I\\ \end{array}\Bigg], \end{equation} $

其中$ M(\frac{1}{s})=-(B^{*})^{-1}(D+s)B^{-1}-s^{-1} I $.

由于$ s $属于$ \sigma({\cal A}) $$ \sigma_{p}({\cal A}) $$ \sigma_{ess}({\cal A}) $当且仅当$ \frac{1}{s} $属于$ \sigma({\cal A}^{-1}) $$ \sigma_{p}({\cal A}^{-1}) $$ \sigma_{ess}({\cal A}^{-1}) $. 因此由(2.1) 式可知, 0属于$ \sigma({\cal A}^{-1}-\frac{1}{s}I) $$ \sigma_{p}({\cal A}^{-1}-\frac{1}{s}I) $$ \sigma_{ess}({\cal A}^{-1}-\frac{1}{s}I) $等价于0属于$ \sigma(M(\frac{1}{s})) $$ \sigma_{p}(M(\frac{1}{s})) $$ \sigma_{ess}(M(\frac{1}{s})). $又由$ L(s)=-sM(\frac{1}{s})=(B^{*})^{-1}(D+s)B^{-1}s+I $, 故

$ s\in {\Bbb C}\backslash\sigma_{ess}({\cal A}) $, 由定义1.3得$ {\rm nul}({\cal A}-sI)={\rm nul} L(s) $$ {\rm def}({\cal A}-sI)={\rm def} L(s) $. 证毕.

定理2.2   设$ {\cal A} $为由(1.1) 式定义的反三角算子矩阵, 若$ B^{-1} $为紧算子, 则

进一步, 当$ D $为自伴算子时$ \sigma_{ess}({\cal A})\subset{\Bbb R} $.

   根据命题2.1可知$ {\cal A}^{-1}\ $存在, 且

$ B^{-1} $为紧算子, 根据引理1.1知

$ \begin{equation} \sigma_{ess}({\cal A})=\Big\{\lambda\in{\Bbb C}\backslash\{0\}\Big| \frac{1}{\lambda}\in \sigma_{ess}\Big(-(B^{*})^{-1}DB^{-1}\Big)\Big\}. \end{equation} $

进一步, 由$ B $为稠定闭算子且$ D $为自伴算子, 则$ \Big((B^{*})^{-1}DB^{-1}\Big)^{*}=(B^{*})^{-1}DB^{-1} $, 再结合(2.2) 式得$ \sigma_{ess}({\cal A})\subset {\Bbb R} $.证毕.

推论2.1   设$ {\cal A} $为由(1.1) 式定义的反三角算子矩阵, 其中$ B^{-1}, D $为紧算子, 则$ \sigma_{ess}({\cal A})=\emptyset $.

   由$ B^{-1}, D $均为紧算子可知$ (B^{*})^{-1}DB^{-1} $为紧算子. 再由文献[21, 定理III.6.26] 及上式(2.2) 得$ \sigma_{ess}({\cal A})=\emptyset $. 证毕.

下面为了估计$ {\cal A} $的本质谱的范围, 先给出如下的引理.

引理2.1   设$ {\cal A} $为由(1.1) 式定义的算子. 若$ B, DB^{-1} $均为自伴算子, 令$ \lambda=\mu+i\sigma, $$ \lambda\neq 0 $, 假定存在一个序列$ \{(x_{n}, y_{n})^{\top}\}_{n\in {\Bbb N}}\in D({\cal A}) $使得

$ \begin{equation} \|x_{n}\|^{2}+\|y_{n}\|^{2}=1, \end{equation} $

$ \begin{equation} \lim \limits _{ n\to \infty } \|({\cal A}-\lambda I )\left(\begin{array}{cc} x_{n} \\ y_{n} \\ \end{array}\right)\|=0 \end{equation} $

成立, 则

1. $ \|y_{n}-B^{-1}\lambda x_{n}\|\rightarrow 0, (n\rightarrow \infty) $;

2. $ \liminf \limits _{ n\to \infty }\|x_{n}\|> 0; $

3. 若$ \sigma\neq 0, $

4. 若$ \sigma=0, $

   由(2.4) 式有

$ \begin{equation} \lambda x_{n}-By_{n}\rightarrow 0, (n\rightarrow \infty), \end{equation} $

$ \begin{equation} Bx_{n}+(\lambda+D)y_{n}\rightarrow0, (n\rightarrow \infty). \end{equation} $

根据(2.3), (2.5) 式可知在Hilbert空间$ H_{1} $$ \{x_{n}\} $不存在收敛到0的子列, 故$ \liminf \limits _{ n\to \infty }\|x_{n}\|> 0. $在(2.6) 式两边与$ x_{n} $作内积, 并结合(2.5) 式有

$ \sigma\neq0 $时上式的实部与虚部分别趋近于零, 故

$ \begin{equation} \langle Bx_{n}, x_{n}\rangle +(\mu^{2}-\sigma^{2})\langle B^{-1}x_{n}, x_{n}\rangle + \mu\langle DB^{-1} x_{n}, x_{n}\rangle\rightarrow 0, (n\rightarrow \infty), \end{equation} $

$ \begin{equation} \langle DB^{-1}x_{n}, x_{n}\rangle+2\mu\langle B^{-1}x_{n}, x_{n}\rangle\rightarrow0, (n\rightarrow \infty). \end{equation} $

由(2.7), (2.8) 两式可得

$ \begin{equation} \langle Bx_{n}, x_{n}\rangle-(\mu^{2}+\sigma^{2})\langle B^{-1}x_{n}, x_{n}\rangle\rightarrow0, (n\rightarrow \infty). \end{equation} $

进一步结合(2.8), (2.9) 式有

$ \begin{equation} \langle DB^{-1}x_{n}, x_{n}\rangle+\frac{2\mu}{\mu^{2}+\sigma^{2}}\langle Bx_{n}, x_{n}\rangle\rightarrow0, (n\rightarrow \infty). \end{equation} $

$ \sigma=0 $

$ \begin{equation} \langle Bx_{n}, x_{n}\rangle+\mu^{2}\langle B^{-1}x_{n}, x_{n}\rangle+\mu\langle DB^{-1}x_{n}, x_{n}\rangle\rightarrow0, (n\rightarrow \infty). \end{equation} $

综上所述, 结论成立.

下面对于$ H_{1}=H_{2} $ (不妨记为$ H $) 的情形, 估计由(1.1) 式定义的反三角算子矩阵$ {\cal A} $的本质谱的范围.

定理2.3  设$ {\cal A} $为由(1.1) 式定义的反三角算子矩阵, 其中$ B, DB^{-1} $均为自伴算子且$ D $为增生算子. 令

特别地, 当$ \sigma_{ess}(DB^{-1})=\emptyset $时, 令$ \alpha_{1}=\infty $, 则$ \sigma_{ess}({\cal A})\subset (-\infty, 0)\cup\{\lambda\in{\Bbb C}|{\rm Re}\lambda\leq-\alpha_{1}\}. $

   根据文献[20, 定理1.2.12]可得$ \sigma({\cal A})\subset \overline{W({\cal A})} $. 任取$ \lambda\in W({\cal A}) $, 则存在$ (f, g)^{\top}\in{\cal D}({\cal A}), \|f\|^{2}+\|g\|^{2}=1, $满足

$ \begin{eqnarray} \lambda &=&\Bigg<\Bigg[\begin{array}{cc} 0&B \\ -B&-D\\ \end{array}\Bigg]\left(\begin{array}{cc} f \\ g \\ \end{array}\right), \left(\begin{array}{cc} f \\ g \\ \end{array}\right)\Bigg> =\Bigg<\Bigg(\begin{array}{cc} Bg \\ -Bf-Dg\\ \end{array}\Bigg), \left(\begin{array}{cc} f \\ g \\ \end{array}\right)\Bigg>{}\\ &=&\langle Bg, f\rangle-\langle Bf, g\rangle-\langle Dg, g\rangle =2{\rm i}{\rm Im}\langle g, Bf\rangle-\langle Dg, g\rangle. \end{eqnarray} $

$ D $为增生算子, 则$ {\rm Re}\lambda=-{\rm Re}\langle Dg, g\rangle\leq0. $从而$ \sigma({\cal A})\subset\{\lambda\in {\Bbb C}|{\rm Re}\lambda\leq 0\} $. 假设

$ \lambda\in M, \lambda=\mu+{\rm i}\sigma $, 则$ -2\mu\|B^{-1}\|<{\rm min}\{s\in{\Bbb R}|s\in\sigma_{ess}(DB^{-1})\} $, 并且

$ \begin{equation} G_{\lambda}:={\rm span}\Big\{x\in H|DB^{-1}x=vx, v\leq -2\mu\|B^{-1}\|\Big\} \end{equation} $

是Hilbert空间$ H $中的有限维子空间. 假设存在一个序列$ \{(x_{n}, y_{n})^{\top}\}_{n\in {\Bbb N}}\subset D({\cal A})\cap( G_{\lambda}\times G_{\lambda})^{\perp} $满足(2.3), (2.4)式. 根据引理2.1可知

$ \begin{equation} \langle DB^{-1}x_{n}, x_{n}\rangle+2\mu\langle B^{-1}x_{n}, x_{n}\rangle\rightarrow0 (n\rightarrow \infty). \end{equation} $

$ \{x_{n}\}\in G_{\lambda}^{\perp} $, 则存在$ \delta>0 $使得

$ \begin{equation} \langle DB^{-1}x_{n}, x_{n}\rangle\geq(-2\mu+\delta)\|B^{-1}\|\|x_{n}\|^{2}=-2\mu\|B^{-1}\|\|x_{n}\|^{2} +\delta\|B^{-1}\|\|x_{n}\|^{2}. \end{equation} $

$ \liminf \limits _{ n\to \infty }\|x_{n}\|> 0, $于是(2.15)式与(2.14) 式相矛盾. 故对任意的$ \lambda\in M, $存在一个有限维的子空间$ G_{\lambda} $和一个常数$ C_{\lambda}>0 $, 使得对于任意的$ (x, y)^{\top}\subset{\cal D}({\cal A})\cap( G_{\lambda}\times G_{\lambda})^{\perp} $

又因$ \rho({\cal A})\cap M\neq\emptyset $, 则由文献[21]中的[IV.5.6] 可知$ {\cal A}-\lambda I $是指标为0的Fredholm算子. 由引理1.2, 存在有限维的孤立点集$ K\subset M $使得$ K\subset\sigma_{p, norm}({\cal A}) $$ M\backslash K\subset\rho({\cal A}) $, 故$ \sigma_{ess}({\cal A})\subset (-\infty, 0) \cup\{\lambda\in{\Bbb C}|{\rm Re}\lambda\leq-\alpha_{1}\}. $证毕.

下面应用定理2.3, 对于方程(1.2) 中$ a(x)=b(x)=1 $, $ k(x, y)=0 $的特殊情况, 给出其相应算子矩阵本质谱的估计.

例2.1   当方程(1.2) 中$ a(x)=b(x)=1 $, $ k(x, y)=0 $时, (1.4), (1.6) 式中

而(1.5)式中的$ T_{2}=0 $. 显然$ T_{1}, T_{3} $是正定算子. 同前面的记法, 令$ B=T_{1}^{\frac{1}{2}}, D=T_{2}+T_{3} $, 则$ DB^{-1}=T_{1}^{\frac{1}{2}} $. 由于$ \sigma(T_{1})=\sigma_{p}(T_{1})=\{\lambda_{n}=\pi^{2}n^{2}, n\in{\Bbb N^{+}}\} $, 即$ \sigma_{e}(T_{1})=\emptyset, $进而可得[18]$ \sigma_{e}(DB^{-1})=\emptyset $. 根据定理2.3, 算子矩阵$ {\cal A}=\left[\begin{array}{cccc} 0\; &B\\ -B^{*}\; &-D \end{array} \right] $的本质谱满足$ \sigma_{e}({\cal A})\subset (-\infty, 0) $.

定理2.4   设$ {\cal A} $为由(1.1) 式给定的有界反三角算子矩阵, 其中$ B, DB^{-1} $均为自伴算子且$ D $为增生算子, 令

特别地, 当$ \sigma_{ess}(DB^{-1})=\emptyset $时, 令$ \gamma_{1}=0 $.$ 0\in\sigma_{ess}(DB^{-1}) $时, 令$ \gamma_{1}=\infty $. 如果$ \rho({\cal A})\cap\{\lambda\in{\Bbb C}|{\rm Re}\lambda<-\gamma_{1}, {\rm Im}\lambda\neq0\}\neq\emptyset, $$ \sigma_{ess}({\cal A})\subset(-\infty, 0) \cup \{\lambda\in{\Bbb C}|-\gamma_{1}\leq {\rm Re}\lambda\leq0 \}. $

   由定理2.3的证明可知$ \sigma({\cal A})\subset\{\lambda\in {\Bbb C}|{\rm Re}\lambda\leq 0\} $. 假定

$ \lambda\in N, $$ \lambda=\mu+{\rm i}\sigma $, 则有$ -\frac{2\mu\|B\|}{\mu^{2}+\sigma^{2}}<-\frac{2\|B\|}{\mu} < {\rm min}\Big\{s\in{\Bbb R}\Big|s\in\sigma_{ess}(DB^{-1})\Big\} $, 此时

为Hilbert空间$ H $中的有限维子空间. 假设存在一个序列$ \{(x_{n}, y_{n})^{\top}\}_{n\in {\Bbb N}}\subset ( G_{\lambda}\times G_{\lambda})^{\perp} $满足(2.3), (2.4)式. 由引理2.1知

$ \begin{equation} \langle DB^{-1}x_{n}, x_{n}\rangle+\frac{2\mu}{\mu^{2}+\sigma^{2}}\langle Bx_{n}, x_{n}\rangle\rightarrow0 (n\rightarrow \infty). \end{equation} $

$ \{x_{n}\}\in G_{\lambda}^{\perp} $, 那么存在$ \delta>0 $使得

$ \begin{equation} \nonumber \langle DB^{-1}x_{n}, x_{n}\rangle\geq\Big(-\frac{2\mu}{\mu^{2}+\sigma^{2}}+\delta\Big)\|B\|\|x_{n}\|^{2} =-\frac{2\mu}{\mu^{2}+\sigma^{2}}\|B\|\|x_{n}\|^{2}+\delta\|B\|\|x_{n}\|^{2}. \end{equation} $

由于$ \liminf \limits _{ n\to \infty }\|x_{n}\|> 0, $于是上式与(2.16) 式相矛盾. 故对任意的$ \lambda\in N, $ 存在一个有限维的子空间$ G_{\lambda} $和一个常数$ L_{\lambda}>0 $, 使得对于任意的$ (x, y)^{\top}\subset( G_{\lambda}\times G_{\lambda})^{\perp} $

$ \rho({\cal A})\cap N\neq\emptyset $, 根据文献[21]中的[IV.5.6] 可知$ {\cal A}-\lambda I $是指标为0的Fredholm算子. 再根据引理1.2可知, 存在有限维的孤立点集$ K\subset N $使得$ K\subset\sigma_{p, norm}({\cal A}) $$ N\backslash K\subset\rho({\cal A}) $. 故此结论成立.

基于以上本质谱的研究, 下面讨论反三角算子矩阵$ {\cal A} $的非实谱的聚点问题.

定理2.5  设$ {\cal A} $为由(1.1) 式定义的有界$ {\cal J} $ -自伴算子矩阵. 若$ B, DB^{-1} $均为自伴算子, $ D $为正定算子且$ \|B\|\|B^{-1}\|=1 $, 则在区间$ (-\alpha_{1}, 0) $内不存在$ {\cal A} $的非实谱的聚点, 其中$ \alpha_{1} $同定理2.3.

   设$ \lambda\in(-\alpha_{1}, 0), G_{\lambda} $满足(2.13) 式, 并且任意序列$ \{(x_{n}, y_{n})^{\top}\}_{n\in {\Bbb N}}\subset ( G_{\lambda}\times G_{\lambda})^{\perp} $满足(2.3), (2.4)式. 由引理2.1的(2.5), (2.11) 式知

$ \begin{eqnarray} \liminf \limits _{ n\to \infty }\Bigg[\left(\begin{array}{cc} x_{n} \\ y_{n} \\ \end{array}\right), \left(\begin{array}{cc} x_{n} \\ y_{n} \\ \end{array}\right)\Bigg]&=&\liminf \limits _{ n\to \infty }\Big(\langle x_{n}, x_{n}\rangle-\langle y_{n}, y_{n}\rangle\Big){}\\ &\geq&\liminf \limits _{ n\to \infty }\Big(\langle x_{n}, x_{n}\rangle-\lambda^{2}\|B^{-1}\|^{2}\|x_{n}\|^{2}\Big){}\\ &\geq&\liminf \limits _{ n\to \infty }\Big(-\frac{\lambda \langle DB^{-1}x_{n}, x_{n}\rangle+\lambda^{2}\langle B^{-1}x_{n}, x_{n}\rangle}{\|B\|}-\lambda^{2}\|B^{-1}\|^{2}\|x_{n}\|^{2}\Big){}\\ &=&-\lambda\liminf \limits _{ n\to \infty }\Big(\frac{\langle DB^{-1}x_{n}, x_{n}\rangle+2\lambda\|B^{-1}\|\|x_{n}\|^{2}}{\|B\|}\Big). \end{eqnarray} $

根据(2.15) 式可知$ \langle DB^{-1}x_{n}, x_{n}\rangle+2\lambda\|B^{-1}\|\|x_{n}\|^{2}> 0, $

由定义1.4和引理1.3可知, 在区间$ (-\alpha_{1}, 0) $内不存在$ {\cal A} $的非实谱的聚点. 证毕.

参考文献

Massatt P .

Limiting behavior for strongly damped nonlinear wave equations

J Differ Equ, 1983, 48 (3): 334- 349

DOI:10.1016/0022-0396(83)90098-0      [本文引用: 1]

Huang F L .

On the mathematical model for linear elastic systems with analytic damping

SIAM J Control Optim, 1988, 26 (3): 714- 724

DOI:10.1137/0326041     

Griniv R O , Shkalikov A A .

Exponential stability of semigroups related to operator models in mechanics

Math Notes, 2003, 73 (5): 618- 624

Weiss G , Tucsnak M .

How to get a conservative well-posed linear system out of thin air. Part Ⅱ. Controllability and stability

SIAM J Control Optim, 2003, 42 (3): 907- 935

DOI:10.1137/S0363012901399295     

Bátkai A , Engel K J .

Exponential decay of 2 × 2 operator matrix semigroups

J Comput Anal Appl, 2004, 6 (2): 153- 163

邱汶汶, 齐雅茹.

一类无界算子的二次数值域和谱

数学物理学报, 2020, 40A (6): 1420- 1430

DOI:10.3969/j.issn.1003-3998.2020.06.002      [本文引用: 1]

Qiu W W , Qi Y R .

The quadratic numerical range and the spectrum of some unbounded block operator matrices

Acta Math Sci, 2020, 40A (6): 1420- 1430

DOI:10.3969/j.issn.1003-3998.2020.06.002      [本文引用: 1]

Jacob B , Trunk C .

Location of the spectrum of operator matrices which are associated to second order equations

Oper Matrices, 2007, 1 (1): 45- 60

[本文引用: 1]

Jacob B , Trunk C , Winklmeier M .

Analyticity and riesz basis property of semigroups associated to damped vibrations

J Evol Equ, 2008, 8 (2): 263- 281

DOI:10.1007/s00028-007-0351-6     

Jacob B , Trunk C .

Spectrum and analyticity of semigroups arising in elasticity theory and hydromechanics

Semigroup Forum, 2009, 79 (1): 79- 100

DOI:10.1007/s00233-009-9148-y      [本文引用: 1]

Artamonov N V .

Estimate of the decay exponent of an operator semigroup associated with a second-order linear differential equation

Math Notes, 2012, 91 (5/6): 731- 734

[本文引用: 1]

Jacob B , Langer M , Tretter C .

Variational principles for self-adjoint operator functions arising from second order systems

Oper Matrices, 2016, 10 (3): 501- 531

[本文引用: 1]

Jacob B , Tretter C , Trunk C , Vogt H .

Systems with strong damping and their spectra

Math Methods Appl Sci, 2018, 41 (16): 6546- 6573

DOI:10.1002/mma.5166      [本文引用: 1]

Huang F L .

Some problems for linear elastic systems with damping

Acta Math Sci, 1990, 10 (3): 319- 326

DOI:10.1016/S0252-9602(18)30405-3      [本文引用: 1]

Mugnolo D. A variational approach to strongly damped wave equations//Amann W, Arendt M, Hieber I, Neubrander FM, Nicaise S, eds. Functional Analysis and Evolution Equations: The Günter Lumer Volume. Basel: Birkhäuser, 2008: 503-514

[本文引用: 1]

Francesca B .

A dirichlet boundary contral problem for the strongly damped wave equation

SIAM J Control Optim, 1992, 30 (5): 1092- 1100

DOI:10.1137/0330058      [本文引用: 1]

Chen S P , Liu K S , Liu Z Y .

Spectrum and stability for elastic systems with global or local Kelvin-Voigt damping

SIAM J Appl Math, 1998, 50 (2): 651- 668

Lasiecka I , Pandolfi L , Triggiani R .

A singular control approach to highly damped second-order abstract equations and applications

Appl Math Optim, 1997, 36, 67- 107

[本文引用: 1]

Weidmann J . Linear Operators in Hilbert Spaces. New York: Springer-Verlag, 1980

[本文引用: 2]

Bognar J . Indefinite Inner Product Spaces. New York: Springer-Verlag, 1974

[本文引用: 2]

吴德玉, 阿拉坦仓. 分块算子矩阵谱理论及其应用. 北京: 科学出版社, 2013

[本文引用: 2]

Wu D Y , Alatancang . The Spectral Theory of Block Operator Matrices and Its Applications. Beijing: Science Press, 2013

[本文引用: 2]

Kato T . Perturbation Theory for Linear Operators. Berlin: Springer-Verlag, 1976

[本文引用: 5]

Hundertmark D , Lee Y R .

Exponential decay of eigenfunctions and generalized eigenfunctions of a non-self-adjoint matrix Schrödinger operator related to NLS

Bull London Math Soc, 2007, 39 (5): 709- 720

DOI:10.1112/blms/bdm065      [本文引用: 1]

Azizov T Y , Jonas P , Trunk C .

Spectral points of type π+ and π- of self-adjoint operators in Krein spaces

J Funct Anal, 2005, 226, 114- 137

DOI:10.1016/j.jfa.2005.03.009      [本文引用: 2]

/