数学杂志  2016, Vol. 36 Issue (2): 328-334   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
李婕
一维对流扩散方程解的逐点衰减估计
徐红梅, 李婕     
河海大学理学院, 江苏 南京 210098
摘要:本文研究一维空间对流扩散方程柯西问题.利用格林函数, 频谱分解, 傅立叶变换等方法, 得到了方程解的逐点估计.结果显示解沿特征线作传播.
关键词对流扩散方程    逐点衰减估计    频谱分析    格林函数    
POINTWISE ESTIMATE OF SOLUTION OF CONVECTION-DIFFUSION EQUATION IN ONE DIMENSION
XU Hong-mei, LI Jie     
College of Science, Hohai University, Nanjing 210098, China
Abstract: In this paper, we study the Cauchy problem of convection-diffusion equation in one dimension. By using Green function, Fourier analysis, frequency decomposition, we get some pointwise estimate of this solution, and the solution decay along the characteristic line with heat kernel.
Key words: convection-diffusion equation     pointwise estimate     frequency decomposition     Green function    
1 引言

本文我们考虑一维对流扩散方程

$\begin{equation} \left\{ \begin{gathered} \frac{{\partial c}}{{\partial t}} + u\frac{{\partial c}}{{\partial x}} = D{c_{xx}} + {c_{xt}} - {({c^2})_x}, \hfill \\ \frac{{\partial u}}{{\partial x}} = 0, \hfill \\ c( {0,x} ) = {c_0}( x ) \hfill \\ \end{gathered} \right. \label{eq:1.1}\end{equation}$ (1.1)

解的逐点衰减估计.此方程可模拟溶质中溶液浓度的衰减情况, 在水利工程, 环境工程及化工, 冶金, 航空等领域都有很重要的应用.方程中$ x \in \mathbb{R} $代表空间变量, $ t $是时间, $ D $是扩散常系数, $ u $是流速, 因为$ \frac{{\partial u}}{{\partial x}} = 0, $所以$ u $只与时间有关, 和空间变量没关系. $ c $可代表溶液浓度.本文中我们研究$ c $随着时间和空间的衰减变化情况.

数学家们对对流扩散方程的研究由来已久. Escobedo和Zuazua在文[1]中研究了方程$ {u_t} - \Delta u = a \cdot ( {{{\left| u \right|}^{q - 1}}u} ) $的柯西问题.由Banach空间不动点理论他们证明了当$ q > 1 $时, 此方程存在唯一经典解$ u \in ( {[ {0,\infty } );{L^1}( {{R^n}} )})$.在文[2]中, Kirane和Qafsaoui考虑了

$\begin{cases} {u_t} + L_0^ * b{L_0}u = a\nabla u, \\ u(0) = u_0 \in L^1 \end{cases}\;\mbox{在}\;\mathbb{R} \times (0,\infty) \mbox{上}$

的解, 此处$ b $是正的有界函数, 且

$b( {x,t} ) = b( {x + at} ), {L_0} = {( { - 1} )^m}\sum\limits_{\left| \alpha \right| = \left| \beta \right| = m} {{a_{\alpha \beta }}} \partial _x^{\alpha + \beta },$

其中$ {a_{\alpha \beta }} $是正的常系数, $ L_0^ * $$ {L_0} $的伴随算子.他们得到了解的整体存在性.徐红梅和马慧玲[3]研究了式(1.1) 解的整体存在性.她们通过构造一个Banach空间柯西序列的方法, 得到了此方程解的整体存在性和某些衰减估计.还有很多计算工作者也研究了对流扩散方程的数值解, 这可参看文献[4-6].

本文中, 用$C$表示一般常数, $ {W^{m,p}}( \mathbb{R} ),m \in {Z_ + }, p \in [ {1,\infty } ) $表示Sobolev空间, 模定义为$ {\left\| f \right\|_{{w^{m,p}}}}: = \sum\limits_{k = 0}^m {{{\left\| {\partial _x^kf} \right\|}_{{L_P}}}} $, 特别的$ {H^S}: = {W^{s,2}} $. $ m,n,N $为非负整数. $ F( f ) = \mathop f\limits^ \wedge = \displaystyle\int {{e^{ - ix\xi }}f( x )dx} $表示函数$ f $的傅立叶变换.本文中所有卷积都是关于空间变量$ x $的.本文是在[3]中解整体存在基础上得到的解的逐点衰减估计, 所以先列出文献[3]的结论.

定理1.1 当$ D > u,l \geqslant 1,{\left\| {{c_0}} \right\|_{{H^l}}} = E $充分小, 方程(1.1) 有整体解$ c( {x,t} ) $存在, 且$ {\left\| c \right\|_{{H^l}}} \leqslant CE $.

在定理1.1的基础上, 得到本文结论.

定理1.2 当$ D > u,l \geqslant 1,E = \max ({\left\| {{c_0}} \right\|_{{H^l}}},{\left\| {{c_0}} \right\|_{{L_1}}}) $充分小, $ {c_0} $有紧支集且$ {\left| {{c_0}} \right|_{}} \leqslant C{(1 + {\left| x \right|^2})^{ - 1}} $, 则式(1.1) 的解$ c(x,t) $满足

$\begin{equation} \left| {\partial _x^nc(x,t)} \right| \leqslant C{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{( {1 + \frac{{{{( {x - ut} )}^2}}}{{1 + t}}} )^{ - 1}}, \quad\forall n \leqslant l. \label{eq:1.2}\end{equation}$ (1.2)

为更好的理解(1.2) 式, 令$ x = kut, $

$\left| {\partial _x^nc(x,t)} \right| \leqslant C{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{( {1 + (k - 1){u^2}t} )^{ - 1}},$

$ k = 1, $即沿特征线方向$ c(x,t) $以热核的速度衰减, 否则以$ {t^{ - \frac{{1 + n}}{2} - 1}} $的速度衰减.所以可得

注1 方程(1.1) 是典型的双曲-抛物耦合方程, 定理1.2既反映了方程(1.1) 的双曲性质, 即解沿着特征线传播, 也反映了方程的抛物性质, 即解有与热核算子相同的衰减速度.

注2 溶质中溶液浓度在流体流动的方向最大, 这也与物理现象相符.

本文安排如下, 在第二章中, 由duhamel原理给出方程(1.1) 解的表达式, 并给出一些不等式, 解的逐点衰减估计将在第三章给出.

2 解的表达式

对方程$ \frac{{\partial c}}{{\partial t}} + u\frac{{\partial c}}{{\partial x}} = D{c_{xx}} + {c_{xt}} $两边关于变量$ x $作傅立叶变换, 考虑到$ u $只与时间有关, 和空间变量没关系, 得到

$\begin{equation} \frac{{\partial \hat c}}{{\partial t}} + ui\xi \hat c = D{(i\xi )^2}\hat c + i\xi {\hat c_t}. \label{eq:3}\end{equation}$ (2.1)

由常微分方程求解得

$\hat c{(\xi ,t)^{}} = {\hat c_0}{e^{ - \frac{{( {D - u} ){\xi ^2}}}{{1 + {\xi ^2}}}t - \frac{{i\xi ( {D{\xi ^2} + u} )}}{{1 + {\xi ^2}}}t}}.$

$\begin{equation} \hat G(\xi ,t) = {e^{ - \frac{{( {D - u} ){\xi ^2}}}{{1 + {\xi ^2}}}t - \frac{{i\xi ( {D{\xi ^2} + u} )}}{{1 + {\xi ^2}}}t}},\quad\hat H = \frac{{1 + i\xi }}{{1 + {\xi ^2}}}\hat G, \label{eq:2.2}\end{equation}$ (2.2)

由duhamel原理知方程(1.1) 的解$ c( {x,t} ) $可表示为

$\begin{equation} c(x,t) = G * {c_0} - \int_0^t {H( {t - s} ) * {{({c^2})}_x}( s )ds}. \label{eq:2.3}\end{equation}$ (2.3)

为得到$ c(x,t) $的衰减估计, 需要对$ G,H $做详细分析.由(2.2) 式知, 当$ \xi $有界时, $ |\hat G( {\xi ,t} )| \leqslant {e^{ - b{\xi ^2}t}},|\hat H( {\xi ,t} )| \leqslant {e^{ - b{\xi ^2}t}},b \geqslant 0 $.知道$ {F^{ - 1}}({e^{ - b{\xi ^2}t}}) = {(2\pi )^{ - 1}}{t^{ - \frac{1}{2}}}{e^{ - \frac{{{x^2}}}{t}}} $, 当然这里对$ G,H $不可能得到如$ {e^{ - \frac{{{x^2}}}{t}}} $这么好的估计, 但可用$ {B_N}( {x,t} ) = {(1 + \frac{{{x^2}}}{{1 + {t^{}}}})^{ - N}} $来近似代替$ {e^{ - \frac{{{x^2}}}{t}}} $, 对这部分处理结果见引理2.1.当$ \xi $充分大时, 有$ \left| {\hat G} \right| \leqslant {e^{ - bt}},\left| {\hat H} \right| \leqslant {e^{ - bt}},b \geqslant 0 $, 所以这一部分$ G,H $衰减是没有问题的, 但$ \hat G,\hat H $$ {L_1} $模不存在, 所以在处理高频部分的卷积时, 需要分析$ G,H $的结构, 对这部分处理结果见引理2.2.由以上分析, 需要对频谱分情况讨论, 所以作光滑截断函数$ {\chi _1}(\xi ) = \left\{ \begin{gathered} 1,\quad \left| \xi \right| \leqslant 1 \hfill \\ 0,\quad \left| \xi \right| \geqslant 2 \hfill \\ \end{gathered} \right., $ $ {\chi _2}(\xi ) = \left\{ \begin{gathered} 1\quad \left| \xi \right| \geqslant 2 \hfill \\ 0\quad \left| \xi \right| \leqslant 1 \hfill \\ \end{gathered} \right. $$ {\chi _1}(\xi ) + {\chi _2}(\xi ) = 1 $.令$ {\hat G_i} = {\chi _i}\hat G,{\hat H_i} = {\chi _i}\hat H,i = 1,2 $.

引理2.1 当$ D > u $时, 对任意的正整数N, 存在仅依赖于N和n的常数$ {C_{N,n}} $, 有

$\begin{eqnarray*} &&\left| {\partial _x^n{G_1}(x,t)} \right| \leqslant {C_{N,n}}{(1 + t)^{ - \frac{{1 + n}}{2}}}{B_N}( {x - ut,t} ),\\ &&\left| {\partial _x^n{H_1}(x,t)} \right| \leqslant {C_{N,n}}{(1 + t)^{ - \frac{{1 + n}}{2}}}{B_N}( {x - ut,t} ). \end{eqnarray*}$

 当$ \left| \xi \right| $充分小时, 由Taylor展开, 有

$\begin{eqnarray*} &&-\frac{{( {D - u} ){\xi ^2}}}{{1 + {\xi ^2}}} = - ( {D - u} ){\xi ^2}( {1 - {\xi ^2} + O( {{\xi ^4}} )} ) = - ( {D - u} ){\xi ^2} + O( {{\xi ^4}} ),\\ &&- \frac{{( {u + D{\xi ^2}} )}}{{1 + {\xi ^2}}}i\xi = - i\xi ( {u + D{\xi ^2}} )( {1 - {\xi ^2} + O( {{\xi ^4}} )} ) = - iu\xi + O( {{\xi ^3}} ). \end{eqnarray*}$

所以$ {\hat G_1}( {\xi ,t} ){e^{iu\xi t}} = {e^{( { - ( {D - u} ){\xi ^2} + O( {{\xi ^3}} )} )t}} $, 于是

$\left| {\partial _\xi ^m{\xi ^n}( {{{\hat G}_1}( {\xi ,t} ){e^{iu\xi t}}} )} \right| \leqslant C{\left| \xi \right|^{n - m}}{( {1 + {\xi ^2}t} )^{1 + m}}{e^{ - ( {D - u} ){\xi ^2}t}}.$

由文献[7]的引理1得

$\left| {\partial _x^n{F^{ - 1}}( {{{\hat G}_1}( {\xi ,t} ){e^{iu\xi t}}} )} \right| \leqslant {C_{N,n}}{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_N}( {x,t} ),$

所以

$\begin{aligned} \left| {\partial _x^n{G_1}( {x,t} )} \right| &= \left| {{F^{ - 1}}( {{e^{ - iu\xi t}}{e^{iu\xi t}}{\xi ^n}{{\hat G}_1}( {\xi ,t} )} )} \right|\\ &= \left| {\partial _x^n{F^{ - 1}}( {{e^{iu\xi t}}{{\hat G}_1}( {\xi ,t} )} )( {x - ut,t} )} \right|\\ &\leqslant {C_{N,n}}{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_N}( {x - ut,t} ). \end{aligned}$

同理可得

$\left| {\partial _x^n{H_1}( {x,t} )} \right| \leqslant {C_{N,n}}{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_N}(x - ut,t).$

引理2.2 存在正常数$ b $和函数$ g_1^i( x ) $, $ g_2^i( x ) $, $ ( {i = 1,2} ) $, 有

$\begin{eqnarray*} &&\left| {{G_2}(x,t)} \right| \leqslant C{e^{ - bt}}( {g_1^1(x) + g_2^1(x) + \delta (x)} ),\\ &&\left| {{\partial _x}{H_2}(x,t)} \right| \leqslant C{e^{ - bt}}( {g_1^2(x) + g_2^2(x) + \delta (x)} ). \end{eqnarray*}$

此处$ \delta ( x ) $是Dirac函数, 且

$\begin{eqnarray*} &&\left| {\partial _x^ng_1^i( x )} \right| \leqslant {C_{N,n}}{( {1 + {{\left| x \right|}^2}} )^{ - N}},\quad N \geqslant 1 \mbox{正整数}.\\ &&{{\left\| g_{2}^{i}(x) \right\|}_{{{L}_{1}}}}\le {{C}_{1}}\mbox{且}{\rm supp} g_{2}^{i}( x )\subset \left\{ x,\left| x \right|<2\varepsilon \right\},\quad \varepsilon \mbox{充分小}. \end{eqnarray*}$

 当$ |\xi | $充分大, 由泰勒展开得

$\begin{eqnarray*} &&-\frac{{( {D - u} ){\xi ^2}}}{{1 + {\xi ^2}}} = - ( {D - u} )( {1 - \frac{1}{{{{\left| \xi \right|}^2}}} + O( {\frac{1}{{{{\left| \xi \right|}^4}}}} )} ),\\ &&\frac{1}{{1 + {\xi ^2}}} = \frac{1}{{{\xi ^2}}}( {1 - \frac{1}{{{\xi ^2}}} + O( {\frac{1}{{{\xi ^4}}}} )} ),\\ &&-\frac{{( {u + D{\xi ^2}} )}}{{1 + {\xi ^2}}}i\xi = - i\xi \frac{{D + \frac{u}{{{\xi ^2}}}}}{{1 + \frac{1}{{{\xi ^2}}}}} = - i\xi ( {D + \frac{u}{{{\xi ^2}}}} )( {1 - \frac{1}{{{\xi ^2}}} + O( {\frac{1}{{{\xi ^4}}}} )} ) = - i\xi D + O( {\frac{1}{\xi }} ). \end{eqnarray*}$

于是

$\begin{eqnarray*} &&{\hat G_2}(\xi ,t) = {e^{ - ( {D - u} )t}}{e^{(( {D - u} )\frac{1}{{{{\left| \xi \right|}^2}}} + O( {{\xi ^{ - 4}}} ))t}}{e^{ - i\xi Dt + O( {\frac{1}{\xi }} )t}},\\ &&\xi {\hat H_2}(\xi ,t) = {e^{ - ( {D - u} )t}}{e^{( {D - u} )\frac{t}{{{{\left| \xi \right|}^2}}} + O( {\frac{1}{{{{\left| \xi \right|}^4}}}} )t}}{e^{ - i\xi Dt + O( {\frac{1}{\xi }} )t}}\frac{1}{\xi } ( 1 - \frac{1}{{{\xi ^2}}} + O( {\frac{1}{{{\xi ^4}}}} ) ). \end{eqnarray*}$

所以当$ t $充分大时, 存在正常数$ b $

$\begin{eqnarray*} &&|{\hat G_2}(\xi ,t)| \leqslant C,\quad |\xi {\hat H_2}(\xi ,t)| \leqslant C,\\ &&\left| {\partial _\xi ^m{{\hat G}_2}} \right| \leqslant C{\left| \xi \right|^{ - 1 - m}}{e^{ - bt}},\quad \left| {\partial _\xi ^m(\xi {{\hat H}_2})} \right| \leqslant C{\left| \xi \right|^{ - 1 - m}}{e^{ - bt}}, \end{eqnarray*}$

其中$ m \geqslant 1 $.由文献[7]的引理3.2, 得到结论.

3 解的衰减估计

有了解的表达式和对格林函数的分析, 下面给出解的逐点衰减估计.

引理3.1 若$ {c_0} $有紧支集, $ E = \max ({\left\| {{c_0}} \right\|_{{H^l}}},{\left\| {{c_0}} \right\|_{{L_1}}}) $充分小, 当t充分大时, 有

$\left| {\partial _x^n{G_1} * {c_0}} \right| \leqslant {C_{N,n}}E{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_N}( {x - ut,t} ).$

 由文献[8]的引理2.4及引理2.1有

$\begin{aligned} \left| {\partial _x^n{G_1} * {c_0}} \right| &\leqslant C{( {1 + t} )^{ - \frac{{1 + n}}{2}}}\int {\frac{1}{{{{( {1 + \frac{{|x - y - ut{|^2}}}{{1 + t}}} )}^N}}}{c_0}(y)dy} \\ &\leqslant CE{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_N}( {x - ut,t} ). \\ \end{aligned}$

引理3.2 当$ l \geqslant 1 $, $ E = \max ({\left\| {{c_0}} \right\|_{{H^l}}},{\left\| {{c_0}} \right\|_{{L_1}}}) $, 有

${\left\| {\partial _x^n{G_2} * {c_0}} \right\|_{{L_\infty }}} \leqslant CE{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ).$

 由(2.2) 式,

$\left| {{x^m}\partial _x^n{G_2} * {c_0}} \right| \leqslant {\left\| {{x^m}\partial _x^n{G_2}} \right\|_{{L_\infty }}}{\left\| {{c_0}} \right\|_{{L_1}}} \leqslant CE{e^{ - bt}}\int {\left| {\partial _\xi ^m{\xi ^n}{{\hat G}_2}} \right|} d\xi \leqslant CE{e^{ - bt}}{\int {\left| \xi \right|} ^{n - m - 1}}d\xi.$

$ m = 2N $充分大, 则

$\begin{equation} \left| {{x^m}\partial _x^n{G_2} * {c_0}} \right| \leqslant CE{e^{ - bt}}. \label{eq:3.1}\end{equation}$ (3.1)

由引理2.2,

$\begin{equation} \left| {\partial _x^n{G_2} * {c_0}} \right| \leqslant C{e^{ - bt}} (|{c_0}(x)| + {\left\| {\partial _x^ng{{_1^1}_{}}} \right\|_{{L_1}}}{\left\| {{c_0}} \right\|_{{L_\infty }}} + {\left\| {\partial _x^ng{{_2^1}_{}}} \right\|_{{L_1}}}{\left\| {{c_0}} \right\|_{{L_\infty }}}) \leqslant C{e^{ - bt}}. \label{eq:3.2}\end{equation}$ (3.2)

$ l \geqslant 1, $由Sobolev嵌入不等式$ ||{c_0}|{|_{{L^\infty }}} \leqslant ||{c_0}|{|_{{H^l}}} $和(3.2) 式得到

$\begin{equation} \left| {\partial _x^n{G_2} * {c_0}} \right| \leqslant CE{e^{ - bt}}. \label{eq:3.3}\end{equation}$ (3.3)

由(3.1)、(3.3) 式得

${\left\| {\partial _x^n{G_2} * {c_0}} \right\|_{{L_\infty }}} \leqslant CE{e^{ - bt}}{( {1 + {{\left| x \right|}^2}} )^{ - N}} \leqslant CE{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ).$

引理得证.

在对非线性部分进行估计时, 需先有下面准备工作.

$ M(t) = \mathop{\sup}\limits_{\begin{subarray}{l} 0 \leqslant s \leqslant t\\ \left| n \right| \leqslant l \end{subarray}} {( {1 + s} )^{\frac{{1 + n}}{2}}}\left| {\partial _x^nc(x,s)} \right|(1 + \frac{{{{\left| {x - us} \right|}^2}}}{{1 + s}}), $

$\left| {\partial _x^nc(x,s)} \right| \leqslant M(t){( {1 + s} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - us,s} ),\quad \left| n \right| \leqslant l,$ (3.4)
$\left| {\partial _x^n{c^2}(x,s)} \right| = \left| {\sum\limits_{{n_1} + {n_2} = n} {\partial _x^{{n_1}}c\partial _x^{{n_2}}c} } \right| \leqslant {M^2}(t){( {1 + s} )^{ - 1 - \frac{n}{2}}}{B_2}( {x - us,s} ),\quad |n| \leqslant l.$ (3.5)

引理3.3 当$ \left| n \right| \leqslant l $, 有

$\int_0^t {\partial _x^nH( {t - s} ) * {{({c^2})}_x}( s )ds} \leqslant C{M^2}(t){( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}(x - ut,t).$

 由引理2.1, (3.5) 式, 得

$\begin{aligned}&\int_0^t {\partial _x^nH( {t - s} ) * {{({c^2})}_x}( s )ds} \\ \leqslant &\int_0^{\frac{t}{2}} {\int_\mathbb{R} {{{( {1 + t - s} )}^{ - \frac{{1 + n + 1}}{2}}}{B_2}(x - y - u(t - s),t - s){M^2}(t){{(1 + s)}^{ - 1}}{B_2}( {y - us,s} )} } dyds \\ &+ \int_{\frac{t}{2}}^t {\int_\mathbb{R} {{{( {1 + t - s} )}^{ - \frac{{1 + 1}}{2}}}{B_2}(x - y - u(t - s),t - s){M^2}(t){{(1 + s)}^{ - 1 - \frac{n}{2}}}{B_2}( {y - us,s} )} } dyds. \\ \end{aligned}$

为证明引理, 仅需证明

$\begin{equation} \begin{aligned} P &= \int_0^t {\int_\mathbb{R} {{{( {1 + t - s} )}^{ - 1}}{{(1 + s)}^{ - 1}}{B_2}(x - y - u(t - s),t - s){B_2}( {y - us,s} )} } dyds \\&\leqslant C{( {1 + t} )^{ - \frac{1}{2}}}{B_1}(x - ut,t). \\ \end{aligned}\end{equation}$ (3.6)

这可由文[9]的引理3.5得到.

引理3.4 当$ \left| n \right| \leqslant l $, 有

$\int_0^t {\partial _x^n{H_2}( {t - s} ) * {{({c^2})}_x}( s )ds} \leqslant C{M^2}(t){( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}(x - ut,t).$

 由引理2.2和(3.5) 式得

$\begin{aligned} R: &= \int_0^t {\partial _x^n{H_2}( {t - s} ) * {{({c^2})}_x}( s )ds} \quad \\ &\leqslant \int_0^t {\left| {{\partial _x}{H_2}( {t - s} ) * \partial _x^n({c^2})( s )} \right|ds} \\ &\leqslant C\int_0^t {{e^{ - b(t - s)}}} ({g^2}_1(x) + {g^2}_2(x) + \delta (x)) * {M^2}(t){(1 + s)^{ - 1 - \frac{n}{2}}}{B_2}( {x - us,s} )ds. \\ \end{aligned}$

由文[9]的引理3.4, 引理3.6得

$\begin{equation} {g^2}_1(x) * {B_2}( {x - us,s} ) \leqslant C{(1 + \frac{{{{\left| {x - us} \right|}^2}}}{{1 + s}})^{ - 1}} \leqslant C(1 + t - s){B_1}( {x - ut,t} ). \end{equation}$ (3.7)

由(3.7) 式,

$\begin{equation} \begin{gathered} \quad \int_0^t {{e^{ - b(t - s)}}} {(1 + s)^{ - 1 - \frac{n}{2}}}{g^2}_1(x) * {B_2}( {x - us,s} )ds \hfill \\ \leqslant {C_{}}{(1 + t)^{ - 1 - \frac{n}{2}}}{B_1}( {x - ut,t} ) \leqslant {C_{}}{(1 + t)^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ). \hfill \\ \end{gathered} \end{equation}$ (3.8)

由文[9]的引理3.4, 引理3.6得

$g_2^2( x ) * {B_2}( {x - us,s} ) \leqslant C{\left\| {g_2^2} \right\|_{{L_1}}}{B_2}( {x - us,s} ) \leqslant C( {1 + t - s} ){B_2}( {x - ut,t} ).$

于是

$\begin{equation} \begin{aligned}&\int_0^t {{e^{ - b( {t - s} )}}} {( {1 + s} )^{ - 1 - \frac{n}{2}}}g_2^2( x ) * {B_2}( {x - us,s} )ds \\ \quad \leqslant&C{( {1 + t} )^{ - 1 - \frac{n}{2}}}{B_2}( {x - ut,t} ) \leqslant C{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ). \\ \end{aligned} \end{equation}$ (3.9)

由文[9]的引理3.6, 得

$\begin{equation} \int_0^t {{e^{ - b( {t - s} )}}} {( {1 + s} )^{ - 1 - \frac{n}{2}}}{B_2}( {x - us,s} )ds \leqslant C{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ). \label{eq:3.10}\end{equation}$ (3.10)

由(3.8), (3.9), (3.10) 式得到引理.

由引理3.1, 引理3.2, 引理3.3, 引理3.4和(2.3) 式, 得到

$ \left| {\partial _x^nc( {x,t} )} \right| \leqslant CE{( {1 + t} )^{ - \frac{{1 + n}}{2}}}{{\text{B}}_1}( {x - ut,t} ) + C{M^2}( t ){( {1 + t} )^{ - \frac{{1 + n}}{2}}}{B_1}( {x - ut,t} ),\quad \left| n \right| \leqslant l, $

所以

$ \left| {\partial _x^nc( {x,t} )} \right|{( {1 + t} )^{\frac{{1 + n}}{2}}}B_1^{ - 1}( {x - ut,t} ) \leqslant CE + C{M^2}( t ) ,$

$ M(t) $的定义得到$ M( t ) \leqslant CE + C{M^2}( t ) $.因为$ E $充分小, 所以$ M( t ) $有界.由(3.4) 式, 得到本文结论, 即定理1.2.

参考文献
[1] Escobedo M, Zuazua E. Large time behavior for convection-diffusion equations in $ {\Re.n} $[J]. J. Funct. Anal., 1991, 100(1): 119–161. DOI:10.1016/0022-1236(91)90105-E
[2] Kirane M, Qfsaoui M. On the asymptotic behavior for convection-diffusion equations associated to higher order elliptic operators under divergence form[J]. Rev. Mat. Complut., 2002, 15(2): 585–598.
[3] Xu Hongmei, Ma Huiling. Global existence of solution to one dimensional convection-diffusion equation and decay estimate[J]. J. Wuhan Univ. (Nat. Sci.), 2013, 18(6): 461–465. DOI:10.1007/s11859-013-0957-1
[4] 王同科. 一维对流扩撒方程Crank-nicholson特征差分格式[J]. 应用数学, 2001, 14(4): 55–60.
[5] 曾晓艳, 陈建业, 孙乐林. 对流扩散方程的一种新型差分格式[J]. 数学杂志, 2003, 23(1): 37–42.
[6] 徐红梅, 吴笑天. BBM-Burgers方程解得衰减估计[J]. 数学杂志, 2014, 04: 723–728.
[7] Wang Weike, Yang Tong. The pointwise estimate of solutions of Euler equation with damping in multi-dimensions[J]. J. D. E., 2001, 173: 410–450. DOI:10.1006/jdeq.2000.3937
[8] Liu Taiping, Wang Weike. The pointwise estimate of diffusion wave for the Navier-stokes systems in odd multi-dimensions[J]. Commum. Math. Phys., 1998, 196: 145–173. DOI:10.1007/s002200050418
[9] Xu Hongmei, Liang Yan. Global existence and pointwise estimate of solutions of generalized Benjamin-Bona-Mahony equations in multi-dimensions[J]. Chin. Ann. Math., 2014, 35(4): 659–668. DOI:10.1007/s11401-014-0841-0