数学杂志  2014, Vol. 34 Issue (4): 723-728   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
徐红梅
吴笑天
BBM-Burgers方程解得衰减估计
徐红梅, 吴笑天    
河海大学理学院, 江苏 南京 210098
摘要:本文研究了一维空间的BBM-Burgers方程.利用时频分解和能量估计等工具, 在解整体存在的前提下, 得到了本方程柯西问题解的某些衰减估计.
关键词BBM-Burgers方程    衰减估计    时频分解    能量估计    
THE DECAY ESTIMATE OF SOLUTIONS OF BBM-BURGERS EQUATION
XU Hong-mei, WU Xiao-tian    
College of Science, Hohai University, Nanjing 210098, China
Abstract: In this paper we study BBM-Burgers equation in one space dimension.Using tools of time-frequency decomposition and energy estimate, when the global solution of this equation with initial data exists, we get some decay estimate of this solution.
Key words: BBM-Burgers equation     decay estimate     time-frequency decomposition     energy estimate    
1 引言

本文中, 我们研究BBM-Burgers方程

$ \begin{eqnarray*} u_{t}+f(u)_{x}=\delta u_{xxt}+\sum^{N}_{n=1}(-1)^{n+1}\upsilon_{n}\partial_{x}^{2n}u, \end{eqnarray*} $ (1.1)

在初值

$ \begin{equation}u(x, 0)=u_{0}(x), x\in R\end{equation} $ (1.2)

时的解的衰减估计, 这里 $\delta>0, \upsilon_{n}>0$是常数, $f(u)=u^{1+\alpha_{0}}, \alpha_{0}\geq1, t, x$分别代表时间和空间变量.著名的Bennjamin-Bona-Mahony (BBM)方程

$ u_{t}-u_{xxt}+u_{x}+uu_{x}=0, -\infty<x<\infty, t>0 $

和kortewey-Devries (KDV)方程

$ u_{t}-u_{xxx}+u_{x}+uu_{x}=0, -\infty<x<\infty, t>0 $

是研究非线性色散媒介中长波传播的模型方程. BBM方程在由Bennjamin, Bona, Mahony于1972年文献[1]提出后, 关于BBM方程的各种变形的各种问题, 如初值问题, 边界问题, 解的存在性, 周期解, 行波解, 解的衰减估计等都受到各国数学家的关注, 可参看文献[1-3, 5-7].方程(1.1) 称为BBM-Burgers方程.如文献[1, 4]指出, 此方程中粘性项 $\upsilon_{1}u_{xx}$和耗散项 $\upsilon_{2}u_{xxxx}$有不同的物理背景.当 $N=1$时, Schonbek[8]讨论了解的存在性和 ${u(x, t;\delta, \upsilon_{1})}$ $\delta\rightarrow 0, \upsilon_{1}\rightarrow 0$时的收敛性.在文献[9]中, 赵和玄得到了当 $N=2$时解的存在性和 $L_{p}(0\leq p<2)$衰减估计.在文献[10]中, Kondo和Webler研究了式(1.1) 和(1.2) 在特定条件下解的整体存在性和收敛性.本文中, 我们延续文献[10]的结论, 在解整体存在的情况下研究解的衰减估计.

本文中, 我们用C表示一般常数, $W^{s, p}(IR), s\in Z_{+}, p\in[1, \infty)$表示通常的索伯列夫空间, 它的模定义为 $||f||_{{W}^{S, P}}:=\sum^{s}\limits_{|\alpha|\geq 0}||\partial_{x}^{\alpha}f||_{L^{p}}$.特别地, $W^{s, 2}=H^{s}$. $F(f)$ $\hat{f}$代表函数 $f$的傅里叶变换, 且 $F(f)=\hat{f}=\int e^{-ix\xi}f(x)dx$. $F^{-1}(\hat{f})$代表函数 $\hat{f}$的逆傅里叶变换, 且 $F^{-1}(\hat{f})=\frac{1}{2\pi}\int e^{ix\xi}\hat{f(\xi)}d\xi$.

本文安排如下, 在第二节中, 我们给出一些准备工作, 如式(1.1), (1.2) 解的存在性结论, 解的表达式等.在第三节中给出格林函数低频部分的估计以及由能量积分等工具得到解的某些估计.我们将在第四节给出本文结论.

2 准备工作

形式上, 在方程(1.1) 两边对变量 $x$作傅里叶变换, 得

$ \begin{eqnarray*} \hat{u_{t}}+F(f(u)_{x})&=& \delta(i\xi)^{2}\hat{u_{t}}+\sum^{N}_{n=1}(-1)^{n+1}\upsilon_{n}(i\xi)^{2n}\hat{u} = -\delta\xi^{2}\hat{u}_t+\sum^{N}_{n=1}(-1)^{n+1}(-1)^{n}\upsilon_{n}\xi^{2n}\hat{u}. \end{eqnarray*} $

所以 $ (1+\delta\xi^{2})\hat{u_{t}}+\sum^{N}\limits_{n=1}\upsilon_{n}\xi^{2n}\hat{u}+F(f(u)_{x})=0.$直接计算得(1.1), (1.2) 式的解形式上可表示为

$ \begin{equation} u(x, t)=G\ast u_{0}-\int^{t}_{0}H(t-s, \cdot)\ast f(u(s))_{x}ds, \end{equation} $ (2.1)

其中

$ \begin{eqnarray} \hat{G}=e^{-\frac{\sum^{N}_{n=1}\upsilon_{n}\xi^{2n}}{1+\xi^{2}\delta}t}, \end{eqnarray} $ (2.2)
$ \begin{eqnarray} \hat{H}=\frac{1}{1+\delta\xi^{2}}\hat{G}, \end{eqnarray} $ (2.3)

$G$称为(1.1), (1.2) 式的格林函数.

定义截断函数 $ \chi(\xi)=\left\{\begin {array}{ll} 1, \ \ &\hbox{如果}\ |\xi|\leq R, \\ 0, \ \ &\hbox{如果}\ |\xi|>R-1. \end{array} \right. $ $\hat{G}_{1}(\xi, t)=\chi(\xi)\hat{G}(\xi, t)$, $\hat{H}_{1}(\xi, t)=\chi(\xi)\hat{H}(\xi, t).$

下面我们先对低频部分的 $G$ $H$做出衰减估计.

定理2.1  当 $2\leq p\leq+\infty$, 存在常数C, 有

$ \begin{eqnarray} ||\partial_{x}^{\alpha}G_{1}||_{L^{p}}\leq C(1+t)^{-\frac{\alpha}{2}-\frac{1}{2}(1-\frac{1}{p})}, \end{eqnarray} $ (2.4)
$ \begin{eqnarray} ||\partial_{x}^{\alpha}H_{1}||_{{L}^{p}}\leq C(1+t)^{-\frac{\alpha}{2}-\frac{1}{2}(1-\frac{1}{p})}.\end{eqnarray} $ (2.5)

  由Hausdorff-Young不等式和(2.2) 式, 当 $p\geq2, q$满足 $\frac{1}{p}+\frac{1}{q}=1$时, 有

$ \begin{eqnarray*} ||\partial_{x}^{\alpha}G_{1}||_{L_{p}}&\leq & C||\xi^{\alpha}\hat{G}_{1}||_{{L}_{q}} \leq C(\int_{|\xi|\leq R+1}|\xi|^{q\alpha}e^{-q\frac{\sum^{N}_{n=1}\upsilon_{n}\xi^{2n}}{1+\delta\xi^{2}}t}d\xi)^{\frac{1}{q}}\\ &\leq& C(\int_{|\xi|\leq R+1}|\xi|^{q\alpha}e^{-q\frac{\upsilon_{1}\xi^{2}}{1+\delta(R+1)^{2}}t}d\xi)^{\frac{1}{q}}\\ &\leq& C((1+t)^{-\frac{q\alpha}{2}-\frac{1}{2}})^{\frac{1}{q}} = C(1+t)^{-\frac{\alpha}{2}-\frac{1}{2}(1-\frac{1}{p})}. \end{eqnarray*} $

因为 $\frac{\xi^{\alpha}}{1+\delta\xi^{2}}\hat{G}_{1}\leq\xi^{\alpha}\hat{G}_{1}$, 所以得到(2.5) 式.

本文工作延续了文献[10]的结论, 是在文献[10]中得到了解的整体存在性基础上得到的衰减估计, 为了方便讨论, 我们列出文献[10]的结论.

定理2.2  当 $f$充分光滑, 假设 $u(t)$, $u_{0}\in H^{1}(IR)$ $||u(t)-G\ast u_{0}||_{H^{1}(IR)}\leq||u_{0}||_{{H}^{1}(IR)}$, $\forall t\in(0, T]$, 则式(1.1), (1.2) 存在整体解 $u(x, t)\in C([0, \infty); H^{1}(IR))$.

在解整体存在的基础上, 在下节中我们用能量积分和时频分解等工具得到解的某些衰减估计.

3 衰减估计

用能量积分, 我们可得到下述定理.

定理3.1  当 $u_{0}\in H^{1}$, 有 $u\in H^{1}$, 且 $\int^{t}_{0}||\partial_{x}^{n}u||^{2}_{L^{2}}ds\leq C$, 其中 $ 1\leq n\leq N.$

  在(1.1) 式两边同时乘以 $u$, 再对变量 $x$ $IR$上积分, 有

$\begin{eqnarray*} \int uu_{t}dx+\int u(u^{1+\alpha_{0}})_{x}dx=\int\delta u_{xxt}udx+\int\sum^{N}_{n=1}(-1)^{n+1}\upsilon_{n}\partial_{x}^{2n}uudx.\end{eqnarray*} $

因为

$ \int u(u^{1+\alpha_{0}})_{x}dx=\int u(1+\alpha_{0})u^{\alpha_{0}}u_{x}dx=(1+\alpha_{0})\int u^{\alpha_{0}+1}u_{x}dx, $

$ \int u(u^{1+\alpha_{0}})_{x}dx=-\int u_{x}u^{1+\alpha_{0}}dx, $

所以

$ \int(u^{1+\alpha_{0}})_{x}udx=0. $

于是

$ \begin{eqnarray*} \frac{1}{2}\frac{d||u||^{2}_{L^{2}}}{dt}=-\frac{\delta}{2}\frac{d||u_{x}||_{L^{2}}^{2}}{dt} +\sum^{N}_{n=1}(-1)^{n+1}\upsilon_{n}(-1)^{n}||\partial^{n}_{x}u||^{2}_{L_{2}}. \end{eqnarray*} $ (3.1)

所以

$ \begin{equation} ||u(\cdot, t)||^{2}_{L_{2}}+\delta||u_{x}(\cdot, t)||^{2}_{L_{2}}+2\int^{t}_{0}\upsilon_{n}||\partial^{n}_{x}u(\cdot, s)||^{2}_{L_{2}}ds =||u_{0}||_{L_{2}}^{2}+\delta||u_{0x}||_{L_{2}}^{2}. \end{equation} $ (3.2)

于是我们得到了本定理

定义时频算子 $\chi(t, D)$, 它的特征为 $\chi_0(\frac{1+t}{\mu}\xi^{2})$, $\mu$为待定常数, 其中 $\chi_0=\left\{\begin{array}{cc} 1, \quad |\eta|\leq 1, \\ 0, \quad |\eta|>2 \end{array}\right.$为光滑截断函数.

$P=||u||^{2}_{L^{2}}+||u_{x}||_{L^{2}}^{2}$, $\eta(t)=\sqrt{\frac{\mu}{1+t}}, $

$ \begin{equation} ||u_{x}||_{L^{2}}^{2} = ||\xi\hat{u}||^{2}_{L^{2}} \geq\int_{|\xi|\geq \eta(t)}|\xi\hat{u}|^{2}d\xi \geq \eta^{2}(t)(||u||^{2}_{L^{2}}-\int_{|\xi|\leq\eta(t)}|\hat{u}|^{2}d\xi). \end{equation} $ (3.3)

同理

$ \begin{equation}||u_{xx}||^{2}_{L^{2}}\geq\eta^{2}(t)(||u_{x}||^{2}_{L^{2}}-\int_{|\xi|\leq\eta(t)}|F(u_x)|^{2}d\xi). \end{equation} $ (3.4)

(3.3) 和(3.4) 式相加得

$ \eta^{2}(t)(||u||^{2}_{L^{2}}+||u_{x}||_{L^{2}}^{2})\leq||u_{x}||_{L^{2}}^{2}+||u_{xx}||_{L^{2}}^{2}+\eta^{2}(t)\\ (\int_{|\xi|\leq\eta(t)}|\hat{u}|^{2}d\xi+\int_{|\xi|\leq\eta(t)}|F(u_x)|^{2}d\xi). $

$\displaystyle R=\int_{|\xi|\leq\eta(t)}|\hat{u}_{x}|^{2}d\xi+\int_{|\xi|\leq\eta(t)}|F(u_x)|^{2}d\xi.$所以

$ \begin{eqnarray*} &&\frac{1}{2}\frac{d}{dt}(||u||^{2}_{L_{2}}+\delta||u_{x}||_{L_{2}}^{2})+\eta^{2}(t)(||u||^{2}_{L^{2}}+||u_{x}||_{L^{2}}^{2})\\ &\leq& \frac{1}{2}\frac{d}{dt}(||u||^{2}_{L_{2}}+\delta||u_{x}||^2_{L_{2}})+||u_{x}||_{L_{2}}^{2}+||u_{xx}||_{L^{2}}^{2}+\eta^{2}(t)R\\ &\leq& C(\frac{1}{2}\frac{d}{dt}(||u||^{2}_{L_{2}}+\delta||u_{x}||^2_{L_{2}})+\sum^{N}_{n=1}\upsilon_{n}||\partial^{n}_{x}u||^{2}_{L_{2}})+\eta^{2}(t)R. \end{eqnarray*} $

由(3.1) 式, 得到

$ \begin{equation}\frac{dP}{dt}+C\eta^{2}(t)P\leq C\eta^{2}(t)R.\end{equation} $ (3.5)

下面我们估计 $R.$

$u_{L}(x, t)=F^{-1}(\chi_0(\frac{1+t}{\mu}\xi^{2})\hat{u}), G_{L}(x, t)=F^{-1}(\chi_0(\frac{1+t}{\mu}\xi^{2})\hat{G}), H_{L}(x, t)$= $F^{-1}(\chi_0(\frac{1+t}{\mu}\xi^{2})\hat{H}), $由(2.1) 式, 有

$ u_{L}(x, t)=G_{L}\ast u_{0}-\int^{t}_{0}H_{Lx}(t-s, \cdot)\ast f(u(s))ds. $

由Minkowski不等式, 得

$ \begin{equation} ||u_{L}||_{L^{2}}\leq||G_{L}\ast u_{0}||_{L^{2}}+(\int^{t}_{0}||H_{L_x}^1(t-s, \cdot)\ast f(u)||^{2}_{L^{2}}ds)^{\frac{1}{2}}. \end{equation} $ (3.6)

由定理2.1, 当 $u_{0}\in L_{1}$, 有

$ \begin{equation} ||G_{L}\ast u_{0}||_{L^{2}}\leq||G_{L}||_{L^{2}}||u_{0}||_{L^{1}}\leq C(1+t)^{-\frac{1}{4}}. \end{equation} $ (3.7)

由Sobolev嵌入不等式和定理3.1, 有

$ \begin{equation} ||u||_{L^{\infty}}\leq C||u||_{H^{1}}\leq C. \end{equation} $ (3.8)

所以当 $\alpha_0\geq1$, 有

$ ||f(u)||_{L^{1}}=||u^{1+\alpha_{0}}||_{L^{1}}\leq||u||^{2}_{L^{2}}||u||^{\alpha_{0}-1}_{L^{\infty}}\leq C. $

由定理2.1, 式(3.6) 有

$ \|H_{Lx}\|_{L_2}^2=\int_{\sqrt{\frac{1+t}{\mu}}|\xi|\leq 1}\xi^2\hat H^2d\xi\leq C(1+t)^{-1}\|H\|_{L_2}^2\leq C(1+t)^{-1-\frac{1}{2}}. $

所以

$ \begin{equation} \int^{t}_{0}||H_{Lx}(t-s, \cdot)\ast f(u(s))||^{2}_{L_{2}}ds \leq\\ \int^{t}_{0}||f(u)||_{L_{1}}^2\|H_{Lx}(t-s, \cdot)\|_{L_{2}}^2ds\leq C(1+t)^{-\frac{1}{2}}.\end{equation} $ (3.9)

由(3.6), (3.7), (3.9) 式得

$ ||u_{L}||_{L_{2}}\leq C(1+t)^{-\frac{1}{4}}. $

由(2.1) 式,

$ u_{Lx}=G_{Lx}\ast u_{0}-\int^{t}_{0}H_{Lxx}(t-s, \cdot)\ast f(u(s))ds. $

用同样的方法可以得到 $||u_{Lx}||_{L^{2}}\leq C(1+t)^{-\frac{1}{4}}.$所以由(3.5) 式, 得到

$ \frac{dP}{dt}+2\eta^{2}(t)P\leq C\eta^{2}(t)(1+t)^{-\frac{1}{2}}, $

$ \frac{dP}{dt}+\frac{2\mu}{1+t}P\leq C\frac{\mu}{1+t}(1+t)^{-\frac{1}{2}}. $

所以

$ \begin{eqnarray*} \frac{d(e^{\int^{t}_{0}\frac{2\mu}{1+s}ds}P)}{dt}&=& e^{\int^{t}_{0}\frac{2\mu}{1+s}ds}P_{t}+e^{\int^{t}_{0}\frac{2\mu}{1+s}ds}\cdot \frac{2\mu}{1+t}P \leq C\mu(1+t)^{-\frac{3}{2}}\cdot e^{\int^{t}_{0}\frac{2\mu}{1+s}ds}\\ &=& C\mu(1+t)^{-\frac{3}{2}}(1+t)^{2\mu}. \end{eqnarray*} $

所以只要 $\mu>\frac{1}{2}$就有 $(1+t)^{2\mu}P(t)-P(0)\leq C\mu(1+t)^{2\mu-\frac{3}{2}+1}.$所以

$ ||u||^{2}_{L^{2}}+||u_{x}||^{2}_{L^{2}}\leq C(1+t)^{-\frac{1}{2}}. $

于是得到下面定理

定理3.2  当 $u_{0}\in L^{1}\cap H^{1}$, 有 $||u||_{L^{2}}\leq C(1+t)^{-\frac{1}{4}}$, $||u_{x}||_{L^{2}}\leq C(1+t)^{-\frac{1}{4}}.$

4 结论

综合定理3.1, 定理3.2, (3.8) 式, 得到本文结论

定理4.1  当 $f=u^{1+\alpha_{0}}, \alpha_{0}\geq 1$, 假设 $u(t), u_{0}\in H^{1}(IR), u_{0}\in L^{1}, $

$ ||u(t)-G\ast u_{0}||_{H^{1}(IR)}\leq||u_{0}||_{H^{1}(IR)}, $

则式(1.1), (1.2) 存在整体解 $u(x, t)\in C([0, \infty);H^{1}(IR)), $

$ \begin{eqnarray*}&& ||u||_{H^{1}}\leq C(1+t)^{-\frac{1}{4}}, \ \ \ \ ||u||_{L_{\infty}}\leq C(1+t)^{-\frac{1}{4}}, \\ && \int^{t}_{0}||\partial_{x}^{n}u||^{2}_{L_{2}}ds\leq \max(1, \delta)||u_{0}||^{2}_{H^{1}}. \end{eqnarray*} $
参考文献
[1] Benjamin T B, Bona J L, Mahony J J. Model equations for long waves in nonlinear dispersive systems[J]. Phil. R. Soc. London, 1972, 272: 47–78. DOI:10.1098/rsta.1972.0032
[2] Avrin J, Goldstein J A. Global existence for the Benjamin-Bona-Mahony equations[J]. Nonlinear Analysis, 1985, 9: 861–865. DOI:10.1016/0362-546X(85)90023-9
[3] Goldstein J A, Wichnoski B J. On the Benjamin-Bona-Mahony equation in higher dimensions[J]. Nonlinear Analysis, 1980, 4: 861–865.
[4] Guo B. The vanishing viscosity method and the viscosity of the difference schemne[M]. Beijing: Science press, 1972.
[5] Albert J P. On the decay of solutions of the generalized Benjamin-Bona-Mahony equation[J]. Math. Anal. Appl., 1989, 141: 527–537. DOI:10.1016/0022-247X(89)90195-9
[6] Biler P. Long time behavior of solutions of the generalized Benjamin-Bona-Mahony equation in two space dimensions[J]. Diff. Integral Eqns., 1992, 5: 891–901.
[7] Fang Shaomei, Guo Boling. Long time behavior for solution of initial-boundary value problem for one class of system with multidimensional inhomogeneous GBBM equations[J]. Appl. Math. Mech., 2005, 26(6): 665–675.
[8] Schonbek M E. Convergence of solutions to nonlinear dispersive equations[J]. Comm. Partial Differential Equations, 1982, 7(8): 959–1000. DOI:10.1080/03605308208820242
[9] Zhao H J, Xuan B J. Existence and convergence of solutions for the generalized Benjamin-BonaMahony equations with dissipative term[J]. Nonlinear Analysis: TMA, 1997, 28(11): 1835–1849. DOI:10.1016/S0362-546X(95)00237-P
[10] Cezar I, Kondo, Claudete M, Webler. Higher order for the generalized Benjamin-Bona-Mahony equation: existence and convergence results[J]. Applicable Analysis, 2009, 88: 977–995. DOI:10.1080/00036810903114759