\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small \emph{Electronic Journal of Differential Equations}, Vol. 2015 (2015), No. 54, pp. 1--20.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu \newline ftp ejde.math.txstate.edu} \thanks{\copyright 2015 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2015/54\hfil Riesz basis and exponential stability] {Riesz basis and exponential stability for Euler-bernoulli beams with variable coefficients and indefinite damping under a force control in position and velocity} \author[K. A. Tour\'e, A. Coulibaly, A. A. H. Kouassi \hfil EJDE-2015/54\hfilneg] {K. Augustin Tour\'e, Adama Coulibaly, Ayo A. Hermith Kouassi} \address{K. Augustin Tour\'e \newline Institut National Polytechnique Houphou\"et-Boigny de Yamoussoukro, BP 2444 Yamoussoukro, C\^ote d'Ivoire} \email{latourci@yahoo.fr} \address{Adama Coulibaly \newline Universit\'e Felix Houphou\"et-Boigny and UFR Math\'ematiques Appliqu\'ees et Informatique, C\^ote d'Ivoire} \email{couliba@yahoo.fr} \address{Ayo A. Hermith Kouassi \newline Universit\'e Felix Houphou\"{e}t-Boigny \and UFR Math\'ematiques Appliqu\'ees et Informatique, C\^ote d'Ivoire} \email{hermithkouassi@gmail.com} \thanks{Submitted January 23, 2015. Published February 26, 2015.} \subjclass[2000]{93C20, 93D15, 35B35, 35P10} \keywords{Beam equation; asymptotic analysis; Riesz basis; exponential stability} \begin{abstract} This article concerns the Riesz basis property and the stability of a damped Euler-Bernoulli beam with nonuniform thickness or density, that is clamped at one end and is free at the other. To stabilize the system, we apply a linear boundary control force in position and velocity at the free end of the beam. We first put some basic properties for the closed-loop system and then analyze the spectrum of the system. Using the modern spectral analysis approach for two-points parameterized ordinary differential operators, we obtain the Riesz basis property. The spectrum-determined growth condition and the exponential stability are also concluded. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{definition}[theorem]{Definition} \allowdisplaybreaks \section{Introduction} We study the Riesz basis property and the stability of a flexible beam with nonuniform thickness or density, that is clamped at one end and is submitted to a linear boundary control force in position and velocity at the free end. The equations of motion of the system are given by \begin{gather} m(x) u_{tt}(x,t) +(EI(x)u_{xx}(x,t) ) _{xx}+\gamma (x) u_{t}( x) =0, \quad 00, \label{1a} \\ u(0,t) =u_{x}(0,t) =u_{xx}(1,t) =0, \quad t>0, \label{1b} \\ (EI(\cdot) u_{xx}(.,t) ) _{x}(1) =\alpha u(1,t) +\beta u_{t}(1,t) , \quad t>0, \label{1c}\\ u(x,0) =u_0(x) ,\quad u_{t}(x,0)=u_1(x) , \quad 00. \label{1} \end{equation} With the assumption \eqref{1}, we shall prove that system \eqref{1a}--\eqref{1d} is a Riesz spectral system in the sense that the generalized eigenfunctions of the system form a Riesz basis on the suitable Hilbert space (see \cite{4}). The Riesz basis property, meaning that the generalized eigenvectors of the system form an unconditional basis for the state Hilbert space, is one of the fundamental properties of a linear vibrating system. For this kind of system, the stability is usually determined by the spectrum of the associated operator and one can also use the theory of nonharmonic Fourier series to obtain important properties such as the optimal decay rate of the energy. There are two steps usually found in the study of linear systems with variable coefficients. The first is to transform the "dominant term" of the system under study into a uniform "dominant equation" by space scaling and state transformation where no variable coefficient is involved any longer. The second step is to approximate the eigenfunctions of the system by those of uniform "dominant equation". This fundamental idea comes essentially from Birkhoff's works \cite{1} and \cite{2} and Naimark \cite{12} to estimate the eigenvalues. This approach has been used in dealing with the beam equations of variable coefficients (see Guo \cite{7,8}, Wang \cite{19} or \cite{20} and the references therein). Moreover, one of the methods on the verification of Riesz basis property well developed recently and applied successfully, is the classical Bari's theorem \cite{3}. When $\alpha =0 $, the undamped case ($\gamma =0$) has been studied in \cite{7}, where the author used a corollary of Bari's theorem on the Riesz basis property in \cite{3}. Our work shall make use a result due to Wang \cite{20}, which deals with the eigenvalue problem of beams in the form of an ordinary differential equation $L( f) =\lambda f$ with $\lambda $-polynomial boundary conditions (Shkalikov \cite{14}, Tretter \cite{16}, Wang \cite{17}, \cite{8} and the references therein). We establish conditions on the both positive feedback parameters $\alpha $ and $\beta $ in order to get the Riesz basis property and the exponential stability for system \eqref{1a}--\eqref{1d}. The content of this article is as follows: in the next section, we convert system \eqref{1a}--\eqref{1d} into an abstract Cauchy problem in Hilbert state space, and discuss some basic properties of system. We show that system \eqref{1a}--\eqref{1d} can be associated to a $\mathcal{C}_0$ -semigroup, and the generator $A_{\gamma }$ of $C_0$-semigroup has compact resolvents. Furthermore, we obtain an asymptotic expression for eigenvalues. In section 3, we discuss the Riesz basis property of the eigenfunctions as well as the exponential stability of the system. Through a bounded invertible transform $\mathcal{L}$, we establish the relationship between $A_{\gamma }$ and $\mathbb{A}$ defined in \eqref{01} and obtain the Riesz basis property from the strong regularity of boundary conditions that has been verified in section 2. Incidentally, we also obtain conditions for the exponential stability of the system for indefinite damping. \section{Basic properties of the problem \eqref{1a}--\eqref{1d}} Let us introduce the following spaces: \begin{gather} H_{E}^2(0,1) =\{ u(x)\in H^2( 0,1) |u(0) =u_{x}(0) =0\} , \label{10} \\ H=H_{E}^2(0,1) \times L^2(0,1), \label{11} \end{gather} The superscript $T$ stands for the transpose and the spaces $L^2(0,1) $ and $H^{k}(0,1) $ are defined as \begin{gather} L^2(0,1) =\big\{ u:[ 0,1] \to \mathbb{C} :\int_0^{1}| u| ^2dx<\infty \big\}, \label{12} \\ H^{k}(0,1) =\{ u:[ 0,1] \to \mathbb{C} : u,u^{(1)},\ldots ,u^{(k)}\in L^2(0,1) \} . \label{13} \end{gather} In the space $H$, we define the inner-product \begin{equation} \langle u,v\rangle _{H}=\int_0^{1}( m(x)f_2(x) \overline{g_2(x) }+EI(x) f_1''(x) \overline{g_1''(x) }) dx +\alpha f_1(1) \overline{g_1(1) }, \label{14} \end{equation} where $u=(f_1,f_2) ^{T}\in H$ and $v=(g_1,g_2) ^{T}\in H$ and we denote by $\|\cdot\| _{H}$ the associated norm. Next, we define an unbounded linear operator $A_{\gamma }:D(A_{\gamma}) \subset H\to H$ as follows: \begin{equation} A_{\gamma }(f,g) =\Big(g(x) ,-\frac{1}{m(x) }((EI(x) f''( x) ) ''+\gamma (x) g(x) )\Big) ^{T}, \label{15} \end{equation} where $D(A_{\gamma }) $, the domain of operator $A_{\gamma }$ is \begin{equation} \begin{aligned} D(A_{\gamma }) =\Big\{&(f,g) ^{T}\in (H^{4}(0,1) \cap H_{E}^2(0,1) )\times H_{E}^2(0,1) : \\ &f''(1) =0, \; (EI(\cdot)f''(\cdot) ) ''(1) =\alpha f(1)+\beta g(1)\big\}. \end{aligned} \label{16} \end{equation} With this notation, the set of equations \eqref{1a}--\eqref{1d} can be formally written as \begin{equation} \begin{gathered} \frac{dY(t) }{dt}=A_{\gamma }Y(t) \\ Y(0) =Y_0\in H, \end{gathered} \label{17} \end{equation} where $Y(t) =(u(.,t) ,u_{t}(.,t)) ^{T}$, $Y(0) =(u_0,u_1) ^{T}$. Here, it is clear that $A_0$ denotes the undamped case $\gamma (x) =0$ which was studied in \cite{23} and that \[ \Gamma _{\gamma }(f,g)=A_{\gamma }-A_0=\Big(0,-\frac{ \gamma (x) g(x)}{m(x) }\Big) \] is a boundary linear operator on $H$. Therefore the following result follows immediately from the theory of operator semigroups (see Pazy \cite[theorem 1.1]{13}). \begin{theorem} \label{theo1} Let operators $A_{\gamma }$ and $A_0$ be defined as before.Then $A_0$ is a m-dissipative operator and generates a $C_0$-group on $H$, and hence $A_{\gamma }$ generates a $C_0$-group $e^{A_{\gamma }t}$ on $H$. \end{theorem} \begin{proof} In \cite{23} we applied the Lumer-Phillips theorem, (see, e.g., \cite[p.14]{13}) to prove that operator $A_0$ is m-dissipative. Then using Hille-Yosida-Phillips theorem, we also obtained that operator $A_0 $ is infinitesimal generator of a $C_0$-semigroup $S(t)=e^{A_0t}$on $H$, satisfying \[ \| S(t) \| \leq Me^{\omega t}. \] Moreover we obtain $A_{\gamma }=\Gamma _{\gamma }+A_0$ where $\Gamma _{\gamma }$ is a boundary linear operator on $H$. Then using the perturbation by bounded linear operator, we deduce that $A_{\gamma }=\Gamma _{\gamma }+A_0$ is infinitesimal generator of a $C_0$-semigroup $T(t) =e^{A_{\gamma}t}$, satisfying \[ \| T(t) \| \leq Me^{(\omega +M\| \Gamma _{\gamma }\| ) t} \] (see A. Pazy \cite[Theorem 1.1]{13}). \end{proof} \begin{theorem} \label{theo2} $A_{\gamma }$ has compact resolvents and $0\in \rho (A_{\gamma }) $. Therefore, the spectrum $\sigma (A_{\gamma }) $ consists entirely of isolated eigenvalues. \end{theorem} \begin{proof} Clearly, we only need to prove that $0\in \rho (A_{\gamma}) $ and $A_{\gamma }^{-1}$ is compact on $H$. For any $G=(g_1,\text{ }g_2) \in \mathit{H,}$ we need to find a unique $F=(f_1, f_2) \in D(A_{\gamma}) $ such that \[ A_{\gamma }F=G. \] In other words such that the following equations are satisfied: \begin{gather} f_2(x) = g_1(x) ,\quad g_1\in H _{E}^2(0,1) \label{2a} \\ -\frac{1}{m(x) }((EI(x) f_1''(x) ) ''+\gamma (x) f_2(x) ) = g_2(x) ,\quad g_2\in L^2(0,1) \label{2b} \\ f_1(0) = f_1'(0) =f_1''(1) =0 \label{2c} \\ (EI(\cdot) f_1''(\cdot) )'(1) =\alpha f_1(1) +\beta f_2(1) . \label{2d} \end{gather} Using \eqref{2b} we obtain \[ (EI(x) f_1''(x) )''=-m(x) g_2(x) -\gamma (x) g_1(x) \] By integrating we obtain \begin{gather*} \int_{x}^{1}(EI(r) f_1''(r)) ''dr=-\int_{x}^{1}m(r) g_2( r) +\gamma (r) g_1(r) \,dr\,, \\ (EI(\cdot) f_1''(\cdot) )'(1) -(EI(x) f_1''(x) ) '=-\int_{x}^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,. \end{gather*} Using the boundary condition \eqref{2d} we obtain \begin{gather*} \alpha f_1(1) +\beta g_1(1) -(EI( x) f_1''(x) ) '=-\int_{x}^{1}m(r) g_2(r) +\gamma (r)g_1(r) \,dr\,, \\ (EI(x) f_1''(x) ) '-\alpha f_1(1) =\int_{x}^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr+\beta g_1(1)\,. \end{gather*} By integrating again we obtain \begin{align*} &\int_{x}^{1}(EI(\eta ) f_1''(\eta ) ) 'd\eta -\alpha f_1(1)\int_{x}^{1}d\eta \\ &=\int_{x}^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta +\beta g_1(1) \int_{x}^{1}d\eta\,, \end{align*} \begin{align*} &EI(1) f_1''(1) -EI(x) f_1''(x) -\alpha (1-x) f_1(1) \\ &=\int_{x}^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta +\beta (1-x) g_1(1)\,. \end{align*} Since $f_1''(1) =0$, we obtain \begin{align*} &f_1''(x) +\alpha \frac{(1-x) }{EI(x) }f_1(1) \\ &=-\frac{1}{EI(x) } \int_{x}^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta -\beta \frac{(1-x) }{EI(x) }g_1(1)\,, \end{align*} \begin{align*} &\int_0^{x}f_1''(\xi ) d\xi +\alpha f_1(1) \int_0^{x}\frac{(1-\xi ) }{EI(\xi) }d\xi \\ &=-\beta g_1(1) \int_0^{x}\frac{(1-\xi ) }{EI(\xi ) }d\xi -\int_0^{x}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta d\xi\,, \end{align*} \begin{align*} &f_1'(x) +\alpha f_1(1) \int_0^{x} \frac{(1-\xi ) }{EI(\xi ) }d\xi \\ &= \int_0^{x}-\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta d\xi -\beta g_1(1) \int_0^{x}\frac{(1-\xi ) }{ EI(\xi ) }d\xi\,, \end{align*} \begin{align*} &\int_0^{x}f_1'(s) ds+\alpha f_1(1) \int_0^{x}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds \\ &=-\int_0^{x}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1(1) \int_0^{x}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\,. \end{align*} Using the boundary condition \eqref{2c} we have \begin{align*} &f_1(x) +\alpha f_1(1) \int_0^{x}\int_0^{s} \frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\\ &=-\int_0^{x}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1(1) \int_0^{x}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\,. \end{align*} Next we determine $f(1)$. We obtain \begin{align*} &f_1(1) +\alpha f_1(1) \int_0^{1}\int_0^{s} \frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds \\ &=-\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1(1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\,, \end{align*} \begin{align*} &f_1(1) (1+\alpha \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds) \\ &=-\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1(1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\,, \end{align*} \begin{align*} f_1(1) &=\Big(-\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2( r) +\gamma (r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1(1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{ EI(\xi ) }\,d\xi\,ds\Big)\Big/ \Big(1+\alpha \int_0^{1}\int_0^{s}\frac{( 1-\xi ) }{EI(\xi ) }\,d\xi\,ds\Big)\,, \end{align*} then \begin{align*} f_1(x) &=-K\int_0^{x}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds-\beta g_1(1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) } \,d\xi\,ds \\ &\quad -\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds\,, \end{align*} with \begin{align*} K&=\alpha \Big(-\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) } \int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma (r) g_1(r) \,dr\,d\eta\,d\xi\,ds\\ &\quad -\beta g_1( 1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\Big)\Big/\Big(1+\alpha \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds\Big). \end{align*} Obviously, $(f_1,\text{ }f_2) \in D(A_{\gamma })$, therefore \[ F=(f_1,\text{ }f_2) =A_{\gamma }^{-1}G=( B(x) ,\text{ }g_1) \] where \begin{align*} B(x) &=-K\int_0^{x}\int_0^{s}\frac{(1-\xi ) }{ EI(\xi ) }\,d\xi\,ds-\beta g_1(1) \int_0^{1}\int_0^{s}\frac{(1-\xi ) }{EI(\xi ) }\,d\xi\,ds \\ &\quad -\int_0^{1}\int_0^{s}\frac{1}{EI(\xi ) }\int_{\xi }^{1}\int_{\eta }^{1}m(r) g_2(r) +\gamma ( r) g_1(r) \,dr\,d\eta\,d\xi\,ds. \end{align*} Finally we obtain that $0\in \rho (A_{\gamma }) $ and Sobolev's embedding theorem implies that $A_{\gamma }^{-1}$ is a compact operator on $H$. Therefore, the spectrum $\sigma (A_{\gamma }) $ consists entirely of isolated eigenvalues. \end{proof} \section{Spectral analysis and the Riesz basis property} \subsection{Spectral analysis of operator $A_{\gamma }$} In this section, we study the basic properties of system \eqref{1a}--\eqref{1d}. Our work shall make use of the following result from \cite{20}, which deals with the eigenvalue problem of beams in the form of an ordinary differential equation $L(f)=\lambda f$ with $\lambda $-polynomial boundary conditions (see Shkalikov \cite{14}; Tretter \cite{16}). To begin, we recall some notations and definitions. Let $L(f) $ be an ordinary differential operator of order $n=2m\in \mathbb{N}$, \begin{equation} L(f) =f^{(n) }(x) +\sum_{\nu =1}^{n}f_{\nu }(x) f^{(n-\nu ) }(x) , \quad 00$. Suppose that the coefficient functions $f_{\nu }(x) $ $(1\leq \nu \leq n) $ in \eqref{3} are sufficiently smooth in $(0,1) $, and that the boundary conditions are normalized in the sense that $\kappa =\sum_{j=1}^{n}k_j$ is minimal with respect to all equivalent boundary conditions (see Naimark \cite{12}). Let $f_{k}(x,\rho ) $ $(k=1,2,\ldots ,n) $ be the fundamental solutions for the equation: \begin{equation} L(f) +\rho ^{n}f+\rho ^{m}\mu (x) f(x)=0,\quad \rho \in \mathbb{C} \label{5} \end{equation} where $\mu (x) $ being continuous in $[ 0,1] $, and let $\omega _{k}$ $(k=1,2,\ldots ,n) $ be the n-th roots of $\omega ^{n}+1=0$. If we denote by $\Delta (\rho ) $ the characteristic determinant of \eqref{5} with respect to \eqref{4} \[ \Delta (\rho ) =\det \big[B_j(f_{k}(.,\rho ) ) \big]_{j,k=1,2,\ldots ,n}\,, \] then $\Delta (\rho ) $ can be expressed asymptotically in the form, for $(r\geq 1)$, \begin{equation} \Delta (\rho ) =\rho ^{k}\sum_{\mathbb{K} _{k}}e^{\rho \mu \mathbb{K} _{k}} [ F^{\mathbb{K} _{k}}] _{r}\,, \label{6} \end{equation} whenever $\rho $ is large enough (see Shkalikov \cite{14} and Naimark \cite{12}). Here, $\mathbb{K}_{k}$ is a $k$-elements subset of $\{1,2,\ldots ,n\} $, $\mu _{\mathbb{K} _{k}}=\sum_{j\in \mathbb{K}_{k}}\omega _j$, \[ [ F^{\mathbb{K} _{k}}] _{r}=F_0^{\mathbb{K} _{k}}+\rho ^{-1}F_1^{\mathbb{K} _{k}}+\ldots +\rho ^{-r+1}F_{r-1}^{\mathbb{K} _{k}}+\mathcal{O}(\rho ^{-r}) , \] and the sum runs over all possible selections of $\mathbb{K}_{k}$. Here and henceforth, $\mathcal{O}(\rho ^{-r}) $ means that $|\rho ^{r}\times \mathcal{O}(\rho ^{-r}) | $ is bounded as $| \rho | \to \infty$. \begin{definition}[{\cite[p. 461]{20}}] \label{definition1} \rm The boundary problem \eqref{5} with \eqref{4} is said to be regular if the coefficients $F_0^{\mathbb{K} _{k}}$ in \eqref{6} are nonzero. Furthermore, the regular boundary problem \eqref{5} with \eqref{4} is said to be strongly regular if the zeros of $\Delta (\rho ) $ are asymptotically simple and isolated one from another. \end{definition} Let $W_2^{m}(0,1) $ be the usual Sobolev space of order $m$ and let \[ V_{E}^{m}(0,1)=\{ f(x)\in W_2^{m}(0,1) |B_j(f) =0, \quad k_j0$ there are only finitely many eigenvalues of $A_{\gamma }$ in the following half-plane: \begin{equation} \Sigma :\operatorname{Re}\lambda >-\frac{1}{2h}\int_0^{1}\frac{\gamma ( x) }{m(x) }\Big(\frac{m(x) }{EI( x) }\Big) ^{1/4}dx-\frac{2\beta }{hEI(1) } \Big(\frac{m(1) }{EI(1) }\Big) ^{-3/4}+\varepsilon . \end{equation} The following are two stability results that describe how stability depend on the damping function $\gamma $. \section{Exponential stability} Following the idea in \cite[Theorem 2.4]{7}, all the properties of operator $A_{\gamma }$ found above, allow us to claim that for the semigroup $e^{A_{\gamma }t}$ generated by $A_{\gamma }$ the spectrum-determined growth condition holds: \[ \omega (A_{\gamma }) =s(A_{\gamma }) , \] where \[ \omega (A_{\gamma }) =\underset{t\to +\infty }{ \lim }\frac{1}{t}\| e^{A_{\gamma }t}\| _{H} \] is the growth order of $e^{A_{\gamma }t}$ and \[ s(A_{\gamma }) =\sup \big\{ \operatorname{Re}\lambda :\lambda \in \sigma (A_{\gamma }) \big\} \] is the spectral bound of $A_{\gamma }$. The Theorem \ref{theo7} is one of the fundamental properties of the evolutive system \eqref{1a}--\eqref{1d}. Many other important properties of this system can be concluded from Theorem \ref{theo7}. The exponential stability stated below is one of such important property. \begin{theorem}\label{theo8} If $\gamma $ is continuous and nonnegative, the system \eqref{1a}--\eqref{1d} is exponential stable for any $\beta >0$ and $\alpha \geq 0$. That is, there are nonnegative constants $M$, $\omega $ such that the energy $E(t) =\frac{1}{2}\| (u,u_{t})^{T}\| _{H}^2$ of system \eqref{1a}--\eqref{1d} satisfies \[ E(t) \leq ME(0) e^{-\omega t},\quad \forall t\geq 0, \] for any initial condition $(u(x,0) ,u_{t}(x,0)) \in H$. \end{theorem} \begin{proof} We have $\gamma (x) \geq 0$, and for any $(f,g) \in D(A_{\gamma }) $, \begin{align*} &\langle A_{\gamma }(f,g) ,(f,g)\rangle \\ &=\big\langle (g(x) ,-\frac{1}{m( x) }((EI(x) f''(x) ) ''+\gamma (x) g(x) )) ,(f,\text{ }g) \big\rangle \\ &=\int_0^{1}[ EI(x) g''(x) \overline{f''(x) }-(EI(x) f''(x) ) ''\overline{g(x) }-\gamma (x) | g(x) |^2] dx +\alpha g(1) \overline{f(1) } \\ &= \int_0^{1}EI(x) [ g''(x)\overline{f''(x) }-f''( x) \overline{g(x) }] dx + \alpha \big(g(1) \overline{f(1) }-f( 1) \overline{g(1) }\Big) \\ &\quad -\beta | g(1) | ^2-\int_0^{1}\gamma (x) |g(x) | ^2dx; \end{align*} further \[ \operatorname{Re}\langle A_{\gamma }(f,g) ,(f,g)\rangle =-\beta | g(1) |^2-\int_0^{1}\gamma (x) | g(x)| ^2dx\leq 0. \] Thus $A_{\gamma }$ is dissipative and $e^{A_{\gamma }t}$ is a contraction semigroup on $H$. Moreover, the spectrum of $\mathit{ A}_{\gamma \text{ }}$ has an asymptote \[ \operatorname{Re}\lambda \sim-\frac{1}{2h}\int_0^{1}\frac{\gamma ( x) }{m(x) }(\frac{m(x)}{EI(x) }) ^{1/4}dx -\frac{2\beta }{hEI(1) }(\frac{m(1) }{EI(1) }) ^{-3/4}. \] If we can show that there is no eigenvalue on the imaginary axis, then the exponential stability holds. Let $\lambda =ir$ with $r\in \mathbb{R}^{\ast }$ be an eigenvalue of operator $A_{\gamma}$ on the imaginary axis and $\Psi =(\phi ,\psi ) ^{T}$ be the corresponding eigenfunction, then $\psi =\lambda \phi $. We have \begin{gather*} \operatorname{Re}(\langle A_{\gamma }\Psi ,\Psi \rangle _{ H})=-\beta | \psi (1) | ^2-\int_0^{1}\gamma (x) | \psi (x) | ^2dx, \\ 0=\| \Psi \| _{H}^2\operatorname{Re}(\lambda) =\operatorname{Re}(\langle A_{\gamma }\Psi ,\Psi \rangle_{H}) =-\beta | \psi (1) |^2-\int_0^{1}\gamma (x) | \psi (x)| ^2dx, \end{gather*} since $\beta >0$, $\gamma (x) \geq 0$ and $\psi (x) $ are continuous, we obtain \begin{equation} \psi (1) =0\quad\text{and}\quad \gamma (x) | \psi (x) | ^2=0,\quad \forall x\in [0,1]. \label{009} \end{equation} Then $\phi (1) =0$ because $\psi =\lambda \phi$. We have the following equation satisfied by $\phi $, \begin{equation} \begin{gathered} \lambda ^2m(x)\phi (x) +(EI(x) \phi ''(x) ) ''+\lambda \gamma ( x) \phi (x) =0,\quad 00 \\ 0&\text{otherwise}, \end{cases} \\ \gamma ^{-}(x) =\max \{ -\gamma (x),0\} =\begin{cases} -\gamma (x) , &\text{if }\gamma (x) <0 \\ 0&\text{otherwise}, \end{cases} \end{gather*} and let \[ A_{\gamma ^{+}}(f,g) =\Big(g(x) ,-\frac{1}{m(x) }((EI(x) f''( x) ) ''+\gamma ^{+}(x) g( x) )\Big) ^{T}, \] for all $(f,g) \in D(A _{\gamma ^{+}}) =D(A_{\gamma })$, and \[ \Gamma _{-}(f,g) =(0,\frac{\gamma ^{-}(x) }{ m(x) }g(x) ) ^{T},\quad \forall (f,g)\in H. \] Then $A_{\gamma }$ can be written as $A_{\gamma }=A_{\gamma ^{+}}+\Gamma _{-}$. \begin{theorem} Let $s(A_{\gamma ^{+}}) =\sup \{ \operatorname{Re}\lambda :\lambda \in \sigma (A_{\gamma ^{+}}) \} $. If \[ \underset{x\in [ 0,1] }{\max }\big\{ \frac{\gamma ^{-}( x) }{m(x) }\big\} <| s(A_{\gamma ^{+}}) | , \] then system \eqref{17} is exponentially stable. \end{theorem} \begin{proof} It is easy to verify that $\Gamma _{-}$ is self-adjoint operator and \begin{equation} \| \Gamma _{-}\| =\max_{x\in [ 0,1] } \big\{ \frac{\gamma ^{-}(x) }{m(x) }\big\} . \end{equation} By Theorem \ref{theo8} and definition of operator $A_{\gamma ^{+}}$, $e^{A_{\gamma ^{+}}t}$ is a contraction semigroup and $s(A_{\gamma ^{+}}) <0$. Applying the perturbation theory of linear operators semigroup (see Pazy \cite[Theorem 1.1 page 76]{13}), we have $\lambda \in \rho (A_{\gamma }) $ whenever $\operatorname{Re}\lambda >s(A_{\gamma ^{+}}) +\| \Gamma _{-}\| $. Again, Theorem \ref{theo7} gives \[ \omega (A_{\gamma }) =s(A_{\gamma }) \leq s(A_{\gamma ^{+}}) +\| \Gamma_{-}\| <0, \] where $\omega (A_{\gamma }) $ denotes the growth bound of semigroup $e^{A_{\gamma }t}$. Therefore, system \eqref{17} is exponentially stable. \end{proof} \subsection*{Acknowledgments} We want to thank the anonymous referee for the throughout reading of the manuscript and several suggestions that help us to improve the presentation of this article. \begin{thebibliography}{99} \bibitem{1} G. D. Birkhoff; \emph{On the asymptotic character of the solutions of certain linear differential equations containing a parameter}. Trans. Amer. Math. Soc. 9 (1908), pp. 219-231. \bibitem{2} G. D. Birkhoff; \emph{Boundary value and expansion problems of ordinary linear differential equations}. Trans. Amer. Math, 9 (1908), pp. 373-395. \bibitem{3} B. Z. Guo, R. Yu; \emph{On Riesz basis property of discrete operators with application to an Euler-Bernoulli beam equation with boundary linear feedback control}, IMA J. Math. Control Information. 18 (2001), pp 241-251. \bibitem{4} R. F. Curtain, H. J. Zwart; \emph{An Introduction to Infinite Dimensional Linear System Theory}. New York: Springer Verlag, 1995. \bibitem{5} Fatima-Zahra Saouri; \emph{Stabilisation de quelques syst\`{e}mes \'elastiques. Analyse spectrale et comportement asymptotique}, Th\`{e}se de Doctorat, Universit\'e Henri Poincar\'e, Nancy 1, 2000. \bibitem{6} B. Z. Guo; \emph{On the boundary control for a hybrid system with variable coefficients}. J. Optim: Theor. Appl, 114 (2002), pp. 373-395. \bibitem{7} B. Z. Guo; \emph{Riesz basis property and exponential stability of controlled Euler-Bernoulli beam equation with variable coefficients}. SIAM J. Control Optim, Vol. 40, Number 6 (2002), pp. 1905-1923. \bibitem{8} B. Z. Guo, J. M. Wang; \emph{Riesz basis generation of abstract second-order partial differential equation systems with general non-separated boundary conditions}. Numerical Functional Analysis and Optimization, vol 27 (2006), pp. 291-328. \bibitem{10} V. Komornik; \emph{Exact Controllability and Stabilization (the Multiplier Method)}. Masson, Paris: Wiley, 1995. \bibitem{11} B. R. Li; \emph{The pertubation theory of a class of linear operators with applications}. Acta Math. Sin. (Chinese), 21 (1978), pp. 206-222. \bibitem{12} M. A. Naimark; \emph{Linear Differential Operators}, Vol. 1. New York: Ugar, 1967. \bibitem{13} A. Pazy; \emph{Semigroups of Linear Operators and Applications to Partial Differential Equations}, Springer-Verlag, New York, 1983. \bibitem{14} A. A. Shkalikov; \emph{Boundary value problems for ordinary differential equations with a parameter in the boundary conditions}, J. Sov. Math., 33 (1986), pp. 1311-1342. \bibitem{15} K. A. Tour\'e, E. Mensah, M. Taha; \emph{Riesz Basis Property and Exponential Stability of a Flexible Euler-Bernoulli Beam with a Force Control in Position and Velocity}. Far East Journal of Applied Mathematics, Volume 69, Number 1 (2012), pp. 29-56. \bibitem{16} C. Tretter; \emph{On $\lambda$-Nonlinear Boundary Eigenvalue Problem}, Mathematical Research, Vol. 71 (1993), Berlin: Akademie. \bibitem{17} J. M. Wang; \emph{Spectral operators generated by partial differential equations of one dimensional spacial variable}. Proc. Chinese. Control Conf. (Yichang) (2003), pp. 400-404. \bibitem{18} J. M. Wang, G. Q. Xu, S. P. Yung; \emph{Exponential stability of variable coefficients Rayleigh beams under boundary feedback control: a Riesz basis approach}. Syst. Control Lett., 51 (2004), pp. 33-50. \bibitem{19} J. M. Wang; \emph{Riesz basis property of some infinite-dimensional control problems and its applications}. The HKU Scholars Hub The University of Hong Kong, 2004. \bibitem{20} J. M. Wang, G. Q. Xu, S. P. Yung; \emph{Riesz basis property, exponential stability of variable coefficient Euler-Bernoulli beams with indefinite damping}. IMA J. Appl. Math, 70 (2005), pp. 459-477. \bibitem{21} H. Tanabe; \emph{Equations of evolution}, Pitman, London, 1979. \bibitem{23} A. Tour\'e, A. Coulibaly, A. A. H. Kouassi; \emph{Riesz basis, exponential stability of variable coefficients Euler-Bernoulli beams with a force control in position and vellocity}. Far East Journal of Applied Mathematics. Volume 88, Number 3 (2014), pp. 193-220. \end{thebibliography} \end{document}