Global existence and optimal decay of solutions to the dissipative Timoshenko system Naofumi Mori Graduate School of Mathematics, Kyushu University Shuichi Kawashima Faculty of Mathematics, Kyushu University 1 Introduction In this paper we consider the nonlinear version of the dissipative Timoshenko system $\{\begin{array}{l}\varphi_{tt}-(\varphi_{x}-\psi)_{x}=0,\psi_{tt}-\sigma(\psi_{x})_{x}-(\varphi_{x}-\psi)+\gamma\psi_{t}=0\end{array}$ (1.1) with the initial data $(\varphi, \varphi_{t}, \psi, \psi_{t})(x, 0)=(\varphi_{0}, \varphi_{1}, \psi_{0}, \psi_{1})(x)$ in the one-dimensional whole space. The original Timoshenko system $(\gamma=0)$ was first introduced by S.P. Timoshenko in [6, 7] as a model system which describes the vibration of the beam called Timoshenko beam: It considers not only transversal movement but also shearing deformation. Here $t\geq 0$ is a time variable and $x\in \mathbb{R}$ is a spacial variable which denotes a point on the center line of the beam; $\varphi$ and $\psi$ are unknown functions of $(x, t)$ , which denote the transversal displacement and the rotation angle of the beam, respectively. Note that $\varphi_{x}-\psi$ denotes the shearing stress. The nonlinear term $\sigma(\eta)$ is assumed to be a smooth function of $\eta$ such that $\sigma’(\eta)>0$ for any $\eta$ under considerations; the coefficient $\gamma$ in the frictional damping term $\gamma\psi_{t}$ is a positive constant. The Timoshenko system is very important as a model system of symmetric hyperbolic systems or symmetric hyperbolic-parabolic systems because the system has new dissipative structures which can not be characterized by the general theory established by S. Kawashima in [5, 8] in terms of the Kawashima condition. In this paper we investigate the nonlinear version of the system by introducing frictional damping as the dissipative mechanism, and first prove the global existence and uniqueness of solutions under smallness assumption on the initial data in the Sobolev space $H^{2}$ (with the critical regularity-index). Also, for initial data in $H^{2}\cap L^{1}$ , we show that the solutions decay in $L^{2}$ at the the optimal rate $t^{-1/4}$ for $arrow\infty.$ 数理解析研究所講究録 第 1971 巻 2015 年 150-164 150
15
Embed
dissipative the Timoshenko system - Research Institute …kyodo/kokyuroku/contents/pdf/...Global existence and optimal decay ofsolutions to the dissipative Timoshenko system Naofumi
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Global existence andoptimal decay of solutions to
the dissipative Timoshenko system
Naofumi Mori
Graduate School of Mathematics, Kyushu University
Shuichi Kawashima
Faculty of Mathematics, Kyushu University
1 Introduction
In this paper we consider the nonlinear version of the dissipative Timoshenko system
with the initial data$(\varphi, \varphi_{t}, \psi, \psi_{t})(x, 0)=(\varphi_{0}, \varphi_{1}, \psi_{0}, \psi_{1})(x)$
in the one-dimensional whole space. The original Timoshenko system $(\gamma=0)$ was firstintroduced by S.P. Timoshenko in [6, 7] as a model system which describes the vibrationof the beam called Timoshenko beam: It considers not only transversal movement butalso shearing deformation. Here $t\geq 0$ is a time variable and $x\in \mathbb{R}$ is a spacial variablewhich denotes a point on the center line of the beam; $\varphi$ and $\psi$ are unknown functionsof $(x, t)$ , which denote the transversal displacement and the rotation angle of the beam,respectively. Note that $\varphi_{x}-\psi$ denotes the shearing stress. The nonlinear term $\sigma(\eta)$ isassumed to be a smooth function of $\eta$ such that $\sigma’(\eta)>0$ for any $\eta$ under considerations;the coefficient $\gamma$ in the frictional damping term $\gamma\psi_{t}$ is a positive constant. The Timoshenkosystem is very important as a model system of symmetric hyperbolic systems or symmetrichyperbolic-parabolic systems because the system has new dissipative structures which cannot be characterized by the general theory established by S. Kawashima in [5, 8] in termsof the Kawashima condition. In this paper we investigate the nonlinear version of thesystem by introducing frictional damping as the dissipative mechanism, and first provethe global existence and uniqueness of solutions under smallness assumption on the initialdata in the Sobolev space $H^{2}$ (with the critical regularity-index). Also, for initial data in$H^{2}\cap L^{1}$ , we show that the solutions decay in $L^{2}$ at the the optimal rate $t^{-1/4}$ for $arrow\infty.$
we transform our system (1.1) into the first order hyperbolic system
$v_{t}-u_{x}+y=0$ , (1.2a)
$y_{t}-\sigma(z/a)_{x}-v+\gamma y=0$ , (1.2b)
$u_{t}-v_{x}=0$ , (1.2c)
$z_{t}-ay_{x}=0$ , (1.2d)
where $a:=\sqrt{\sigma’(0)}$ . The corresponding initial data are given by
$(v, y, u, z)(x, O)=(v_{0}, y_{0}, u_{0}, z_{0})(x)$ , (1.3)
where $v_{0}:=\varphi_{0,x}-\psi_{0},$ $y_{0}:=\psi_{1},$ $u_{0}:=\varphi_{1},$ $z_{0}:=a\psi_{0,x}$ . Here we remark that thenonlinearity of the system (1.2) depends on the component $z$ only.
Our system (1.2) is a symmetric hyperbolic system with non-symmetric relaxation. Infact, we can write (1.2) as
$A^{0}(z)W_{t}+A(z)W_{x}+LW=0$ , (1.4)
where $W=(v, y, u, z)^{T},$ $A^{0}(z)=diag(1,1,1, b(z)/a)$ with $b(z)=\sigma’(z/a)/a$ , and
In our system (1.4) or (1.5) the relaxation matrix $L$ is not symmetric such that $kerL\neq$
$kerL_{1}$ , where $L_{1}$ denotes the symmetric part of $L$ . This is the reason why the generaltheory on the dissipative structure developed in [5, 8] can not be applicable to our system.
151
1.2 Known results for linear system
The decay property of the linear system (1.6) was first investigated by J.E. Munoz Riveraand R. Racke in [4]. They studied (1.6) in a bounded region and with simple boundaryconditions and showed that the energy of the solution decays exponentially when $a=1,$
but polynomially when $a\neq 1$ as $tarrow\infty.$
To explain this interesting decay property, K. Ide, K. Haramoto and S. Kawashima[2] considered the system (1.6) in one-dimensional whole space and showed that the dis-sipative structure of the system (1.6) can be described as
${\rm Re}\lambda(i\xi)\leq-c\eta(\xi)$ , $\eta(\xi)=\{\begin{array}{ll}\xi^{2}/(1+\xi^{2}) for a=1,\xi^{2}/(1+\xi^{2})^{2} for a\neq 1,\end{array}$
where $\lambda(i\xi)$ denotes the eigenvalues of the system (1.6) in the Fourier space, and $c$ is apositive constant. We note that the dissipative structure for $a=1$ is the same as thatin the general theory developed in [5, 8]. However, the dissipative strudture for $a\neq 1$
is much weaker in the high frequency region and causes regularity-loss in the dissipationterm of the energy estimate and also in the decay estimate.
In fact, by using the energy method in the Fourier space, the authors in [2] derivedthe following pointwise estimate for the linear solution $W=(v, y, u, z)^{T}$ of (1.6):
where $W_{0}=(v_{0}, y_{0}, u_{0}, z_{0})^{T}$ is the corresponding initial data. Moreover, based on thispointwise estimate, they showed the optimal time decay estimates of the linear solution:
$\Vert\partial_{x}^{k}W(t)\Vert_{L^{2}}\leq C(1+t)^{-1/4-k/2}\Vert W_{0}\Vert_{L^{1}}+\{\begin{array}{ll}Ce^{-ct}\Vert\partial_{x}^{k}W_{0}\Vert_{L^{2}} for a=1,C(1+t)^{-\ell/2}\Vert\partial_{x}^{k+\ell}W_{0}\Vert_{L^{2}} for a\neq 1,\end{array}$
where $k$ and $l$ are nonnegative integers, and $C$ and $c$ are positive constants. We note thatwhen $a\neq 1$ , in order to obtain the optimal decay rate $(1+t)^{-1/4-k/2}$ we have to assumethe additional $\ell$-th order regularity on the initial data to make the decay rate $(1+t)^{-\ell/2}$
better than $(1+t)^{-1/4-k/2}$ . Therefore the regularity-loss can not be avoided in the decayestimate for $a\neq 1.$
1.3 Known results for nonlinear system
Based on these linear results in [2], K. Ide and S. Kawashima [1] proved the global existenceand decay of solutions to the nonlinear system (1.2). To state the result, we introducethe following time-weighted norms $\tilde{E}(t)$ and $\tilde{D}(t)$ :
Theorem 1.1 ([1]). Assume that the initial data satisfy $W_{0}\in H^{s}\cap L^{1}$ for $s\geq 6$ and put$\tilde{E}_{1}$
$:=\Vert W_{0}\Vert_{H^{s}}+\Vert W_{0}\Vert_{L^{1}}$ . Then there exists a positive constant $\tilde{\delta}_{1}$ such that if $\tilde{E}_{1}\leq\tilde{\delta}_{1}$ , theCauchy problem (1.2), (1.3) has a unique global solution $W(t)$ with $W\in C([O, \infty);H^{s})\cap$
$C^{1}([0, \infty);H^{s-1})$ . Moreover the solution $W(t)$ verifies the energy estimate
where $0\leq k\leq[s/2]-1$ , and $C>0$ is a constant.
Remark 1.1. The result in Theorem 1.1 requires the regularity $s\geq 6$ and $L^{1}$ property onthe initial data. Also, the norms $\tilde{E}(t)$ and $\tilde{D}(t)$ contains the time weights with negativeexponents. These were crucial in [1] to overcome the difficulty caused by the regularity-lossproperty.
1.4 Main results
The main purpose of this paper is to refine Theorem 1.1 under less regularity assumptionon the initial data. To state our results, we introduce the energy norm $E(t)$ and thecorresponding dissipation norm $D(t)$ by
Notice that in the dissipation norm $D(t)$ we have 1 regularity-loss for $(v, u)$ but noregularity-loss for $(y, z)$ . Our first result is then stated as follows.
Theorem 1.2 (Global existence). Assume that the initial data satisfy $W_{0}\in H^{S}$ for$s\geq 2$ and put $E_{0}:=\Vert W_{0}\Vert_{H^{s}}$ . Then there exists a positive constant $\delta_{0}$ such that if$E_{0}\leq\delta_{0}$ , the Cauchy problem (1.2), (1.3) has a unique global solution $W(t)$ with $W\in$
$C([O, \infty);H^{s})\cap C^{1}([0, \infty);H^{s-1})$ . Moreover the solution $W(t)$ verifies the energy estimate
$E(t)^{2}+D(t)^{2}\leq CE_{0}^{2},$
where $C>0$ is a constant.
Remark 1.2. Our global existence result holds true under less regularity assumption$s\geq 2$ and without $L^{1}$ property on the initial data. This refinement is based on thebetter Lyapunov function constructed in [3]. Our Lyapunov function produces the optimaldissipation estimate for $z$ without any regularity-loss (see $D(t)$ ), which enables us tocontrol the nonlinearity depending only on $z.$
Next we state the result on the optimal time decay estimate.
153
Theorem 1.3 (Optimal $L^{2}$ decay estimate). Assume that the initial data satisfy $W_{0}\in$
$H^{2}\cap L^{1}$ and put $E_{1}:=\Vert W_{0}\Vert_{H^{2}}+\Vert W_{0}\Vert_{L^{1}}$ . Then there is a positive constant $\delta_{1}$ such that
if $E_{1}\leq\delta_{1}$ , then the solution $W(t)$ obtained in Theorem 1.2 satisfies the following optimal$L^{2}$ decay estimate:
$\Vert W(t)\Vert_{L^{2}}\leq CE_{1}(1+t)^{-1/4},$
where $C>0$ is a constant.
Remark 1.3. In order to show the above decay estimate, we first estimate the nonlinearsolution by using the energy method in the Fourier space and then apply the refined decayestimate of $U-L^{q_{-}}L^{r}$ type which was established in [9]. For the details, see Section 3.
Notations. Let $\hat{f}=\mathcal{F}[f]$ be the Fourier transform of $f$ :
For $1\leq p\leq\infty$ , we denote by $L^{p}=If(\mathbb{R})$ the usual Lebesgue space on $\mathbb{R}$ with the norm$\Vert\cdot\Vert_{Lr}$ . Also, for nonnegative integer $s$ , we denote by $H^{s}=H^{s}(\mathbb{R})$ the Sobolev space of$L^{2}$ functions, equipped with the norm $\Vert\cdot\Vert_{H^{s}}$ . In this paper, every positive constant isdenoted by the same symbol $C$ or $c$ without confusion.
2 Energy method
The aim of this section is to prove the global existence result in Theorem 1.2. Our globalexistence result can be shown by the combination of a local existence result and thedesired a priori estimate. Since our system (1.2) is a symmetric hyperbolic system, it isnot difficult to show a local existence result by the standard method, and we omit thedetails. To state our result on the a priori estimate, we consider a solution $W(t)$ of theproblem (1.2), (1.3) satisfying $W\in C([O, T];H^{s})\cap C^{1}([0, T];H^{s-1})$ for $s\geq 2$ and
where $\delta$ is a fixed positive constant. Our a priori estimate is now given as follows.
Proposition 2.1 (A priori estimate). Suppose that $W_{0}\in H^{s}$ for $s\geq 2$ and put $E_{0}=$
$\Vert W_{0}\Vert_{H^{S}}$ . Let $T>0$ and let $W(t)$ be a solution to the Cauchy problem (1.2), (1.3) satisfying(2.1). Then there exists a positive constant $\delta_{2}$ independent of $T$ such that if $E_{0}\leq\delta_{2}$ , wehave the a priori estimate
To prove the above a priori estimate in Proposition 2.1, we need to show the followingenergy inequality by applying the energy method.
154
Proposition 2.2 (Energy inequality). Suppose that $W_{0}\in H^{s}$ for $s\geq 2$ and put $E_{0}=$
$\Vert W_{0}\Vert_{H^{s}}$ . Let $T>0$ and let $W(t)$ be a solution to the Cauchy problem (1.2), (1.3) satisfying(2.1). Then we have the following energy inequality:
We note that the desired a priori estimate (2.2) easily follows from the energy inequality(2.3), provided that $E_{0}$ is suitably small. Therefore it is sufficient to prove (2.3) for ourpurpose.
2.1 Proof of Proposition 2.2
In this subsection we prove the energy inequality (2.3) in Proposition 2.2 by using theenergy method. Our energy method is based on the refined Lyapunov function constructedin [3] and gives the optimal dissipation estimate for $z$ without any regularity-loss, whichcan control the nonlinearity of the system (1.2). Our proof is divided into 4 steps.
Step 1: (Basic energy and dissipation for y) We calculate as $(1.2a)\cross v+(1.2b)\cross y+$
$(1.2c)\cross u+(1.2d)\cross\{\sigma(z/a)-\sigma(O)\}/a$ . This yields
Step 2: (Dissipation for v) We rewrite the system (1.2) in the form
$v_{t}-u_{x}-y=0,$
$y_{t}-az_{x}-v+\gamma y=g(z)_{x},$
(2.12)$u_{t}-v_{x}=0,$
$z_{t}-ay_{x}=0,$
where $g(z)$ $:=\sigma(z/a)-\sigma(O)-\sigma’(O)z/a=O(z^{2})$ as $zarrow 0$ . We apply $\partial_{x}^{k}$ to (2.12). Letting$(V, Y, U, Z)=\partial_{x}^{k}(v, y, u, z)$ as before, we have
for $0\leq k\leq s-2$ . We integtate this inequality with respect to $t$ and add for $k$ with$0\leq k\leq s-2$ . Noting that $\sum_{k=0}^{s-1}|E_{1}^{(k)}|\leq C\Vert W\Vert_{H^{s-1}}^{2}$ and u\’{s}ing the Young inequality, weobtain
This combined with the basic estimate (2.11) yields the desired inequality $E(t)^{2}+D(t)^{2}\leq$
$CE_{0}^{2}+CE(t)D(t)^{2}$ . Thus the proof of Proposition 2.2 is comptlete. $\square$
158
3 $L^{2}$ decay estimate
The aim of this section is to show the optimal decay estimate stated in Theorem 1.3. Forthis purpose we derive the pointwise estimate of solutions in the Fourier space. We recallthat the system (1.2) is written in the form of (2.12) or in the vector notation as
$W_{t}+AW_{x}+LW=G_{x}$ , (3.1)
where $G=(0, g(z), 0,0)^{T}$ with $g(z)=\sigma(z/a)-\sigma(0)-\sigma’(0)z/a=O(z^{2})$ for $zarrow 0$ ; thecoefficient matrices $A$ and $L$ are given in (1.5).
Proposition 3.1 (Pointwise estimate). Let $W$ be a solution of (3.1) with the initial data$W_{0}$ . Then the Fourier image $\hat{W}$ satisfies the pointwise estimate
where $k\geq 0,$ $1\leq q,$ $r\leq 2,$ $\ell>\frac{1}{r}-\frac{1}{2}$ $(\ell\geq 0 if r=2)$ .
Remark 3.1. The first (resp. the second) term on the right hand side of (3.4) iscorresponding to the low frequency region $|\xi|\leq 1$ (resp. high frequency region $|\xi|\geq 1$ ).
When $m=0,$ $q=1$ and $r=2$ , the estimate (3.4) is reduced to
We divide the last integral into two parts corresponding to $|\xi|\leq 1$ and $|\xi|\geq 1$ , respec-tively, and estimate each part by applying the H\"older inequality and the Hausdorff-Young
inequality. This yields the desired estimate (3.4). We omit the details and refer to [9].
where $g=g(z)$ . We construct a Lyapunov function of the system (3.5) in the Fourierspace. The computations below are essentially the same as in [3] and correspond to thosein the proof of Proposition 2.2. We divide the proof into 4 steps.
Step 1: (Basic energy and dissipation for $\hat{y}$ ) We compute as (3.5a) $\cross\overline{\hat{v}}+(3.5b)\cross\overline{\hat{y}}+$
$(3.5c)\cross\overline{\hat{u}}+(3.5d)\cross\overline{\hat{z}}$ and take the real part. This yields
Step 2: (Dissipation for $\hat{v}$ ) To create the dissipation term for $\hat{v}$ , we compute as $(3.5b)\cross$
$(-\overline{\hat{v}})+(3.5a)\cross(-\overline{\hat{y}})+(3.5c)\cross(-a\overline{\hat{z}})+(3.5d)\cross(-a\overline{\hat{u}})$ and take the real part. This gives
where $E_{1}$ $:=-{\rm Re}(\hat{v}\overline{\hat{y}}+a\hat{u}\hat{z})-$ . We multiply this equality by $1+\xi^{2}$ . Then, using the Younginequality, we obtain
for any $\epsilon>0$ , where $c_{1}$ is a positive constant with $c_{1}<1$ and $C_{\epsilon}$ is a constant dependingon $\epsilon.$
Step 3: (Dissipation for $\hat{u}$ and $\hat{z}$ ) To create the dissipation term $|\hat{u}|^{2}$ , we compute as(3.5a) $\cross i\xi\overline{\hat{u}}-(3.5c)\cross i\xi\overline{\hat{v}}$ and take the real part. The result is
where $C_{\epsilon,\alpha_{1}}$ and $C_{\alpha_{1}}$ are constants depending on $(\epsilon, \alpha_{1})$ and $\alpha_{1}$ , respectively. Also, letting$\alpha_{2}>0$ , we combine (3.6) and (3.11) such that $(3.6)+(3.11) \cross\frac{\alpha_{2}}{(1+\xi^{2})^{2}}$ . Then, putting
where $c_{0}$ and $C_{0}$ are positive constants. Futhermore, we choose $\alpha_{1}\in(0, \alpha_{0}$ ] such that$c_{1}-\alpha_{1}C>0$ and take $\epsilon>0$ so small as $\alpha_{1}c_{1}-\epsilon>$ O. Finally, we choose $\alpha_{2}\in(0, \alpha_{0}$ ]such that $\gamma-\alpha_{2}C_{\epsilon,\alpha_{1}}>0$ . Then (3.13) becomes to
This suggests that $E$ in (3.12) is the desired Lyapunov function of the system (3.5). Noting(3.14), we find that $F\geq c\rho(\xi)E$ , where $\rho(\xi)=\xi^{2}/(1+\xi^{2})^{2}$ . Therefore (3.15) becomesto $E_{t}+c\rho(\xi)E\leq C\xi^{2}|\hat{g}|^{2}$ . Solving this ordinary differential inequality for $E$ and using(3.14), we arrive at the desired estimate (3.2) in the form
For the high frequency part $J_{2}$ , using $\Vert\partial_{x}^{2}G\Vert_{L^{1}}\leq C\Vert z\Vert_{L^{2}}\Vert\partial_{x}^{2}z\Vert_{L^{2}}$ , we have
Thus we have the inequality $N(t)^{2}\leq CE_{1}^{2}+CN(t)^{4}+CE_{1}^{2}N(t)^{2}$ . This inequality canbe solved as $N(t)\leq CE_{1}$ , provided that $E_{1}$ is suitably small. Thus we have proved thedesired decay estimate $\Vert W(t)\Vert_{L^{2}}\leq CE_{1}(1+t)^{-1/4}$ . This completes the proof of Theorem1.3. $\square$
References
[1] K. Ide and S. Kawashima, Decay property of regularity-loss type and nonlinear effectsfor dissipative Timoshenko system, Math. Models Meth. Appl. Sci., 18 (2008), 1001-1025.
[2] K. Ide, K. Haramoto and S. Kawashima, Decay property of regularity-loss type fordissipative Timoshenko system, Math. Models Meth. Appl. Sci., 18 (2008), 647-667.
[3] N. Mori and S. Kawashima, Decay property for the Timoshenko system with Fourier’stype heat conduction, J. Hyperbolic Differential Equations, 11 (2014), 135-157.
[4] J.E. Munoz Rivera and R. Racke, Global stability for damped Timoshenko systems,Discrete and Continuous Dynamical Systems, 9 (2003), 1625-1639.
[5] Y. Shizuta and S. Kawashima, Systems of equations of hyperbolic-parabolic typewith applications to the discrete Boltzmann equation, Hokkaido Math. J., 14 (1985),249-275.
[6] S.P. Timoshenko, On the correction for shear of the differential equation for trans-verse vibrations of prismatic bars, Philosophical Magazine, 41 (1921), 744-746.
[7] S.P. Timoshenko, On the transverse vibrations of bars of uniform cross-section, Philo-sophical Magazine, 43 (1922), 125-131.
[8] T. Umeda, S. Kawashima and Y. Shizuta, On the decay of solutions to the linearizedequations of electro-magneto-fluid dynamics, Japan J. Appl. Math., 1 (1984), 435-457.
163
[9] J. Xu, N. Mori and S. Kawashima, $L^{p}-L^{q}-L^{r}$ estimates and minimal decay regularityfor compressible Euler-Maxwall equations, preprint.