数学物理学报  2018, Vol. 38 Issue (4): 750-769   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
赵元章
马相如
一类具有变扩散系数的非局部反应-扩散方程解的爆破分析
赵元章, 马相如     
中国海洋大学数学科学学院 山东 青岛 266100
摘要:该文考虑了具有变扩散系数的反应-扩散方程Dirichlet初边值问题解的爆破现象.利用辅助函数法和修正微分不等式技巧,对变扩散系数和非线性项给出适当的条件,以保证解整体存在或有限时刻发生爆破,并在整体空间中(N ≥1)导出了爆破时间的界.同时,给出几个应用举例.
关键词反应-扩散方程    变扩散系数    爆破时间的界    
Blow-Up Analysis for a Nonlocal Reaction-Diffusion Equation with Variant Diffusion Coefficient
Zhao Yuanzhang, Ma Xiangru     
School of Mathematical Sciences, Ocean University of China, Shandong Qingdao 266100
Abstract: In this paper, blow-up phenomena for the Dirichlet initial boundary value problem of a reaction-diffusion equation with variant diffusion coefficient is considered. By virtue of the auxiliary function method and the modified differential inequality, we established appropriate conditions on variant diffusion coefficient and nonlinearities to guarantee existence of global solution or blow-up solution at finite time. Moreover, lower bounds for the blow-up time of the solution are derived in all dimensional spaces (N ≥ 1). In the meantime, several examples are presented to illustrate applications of our results.
Key words: Reaction-diffusion equation     Variant diffusion coefficient     Bounds for the blow-up time    
1 引言

我们考虑具有变扩散系数和非局部源项的反应-扩散方程

$ u_{t}={\rm{div}}(a(x)\nabla u(x, t))+f(u), \;\;\; (x, t)\in\Omega\times(0, t^{\ast}), $ (1.1)

给出齐次Dirichlet边界条件和初始条件

$ u(x, t)=0, \;\;\; (x, t)\in\partial\Omega\times(0, t^{\ast}), $ (1.2)
$ u(x, 0)=u_{0}(x), \;\;\; x\in\Omega, $ (1.3)

其中$\Omega\subset{\Bbb R}^{N}(N\geq1)$为具有光滑边界$\partial\Omega$的有界区域, $t^{\ast}<+\infty$表示可能发生爆破的时间, 反之$t^{\ast}=+\infty$.变扩散系数$a(x)$为正的适当光滑函数, 非线性项$f(u)$为非负连续函数并满足非局部条件, 比如, 包含$(u(x, t))^{p}\int_{\Omega}(u(x, t))^{q}{\rm d}x$型非局部项, 其中$p+q>1.$初值$u_{0}(x)$为非负$C^{1}$类函数且满足适当的相容性条件.因此, 由经典抛物型理论知, 问题(1.1)-(1.3)存在唯一的非负局部解且充分光滑.

方程(1.1)出现在许多物理现象和生物种群理论.比如, 热传导现象中温度, 流体的流动中浓度及某种生物种群密度的扩散等, 见文献[1-3]及相关文献.

近几十年来, 已有许多学者致力于非线性抛物型反应-扩散方程解的爆破现象的研究.有许多文献研究了半线性抛物方程整体解的存在性与非存在性, 解的爆破, 爆破时间的界, 爆破速率, 爆破集和解的渐进行为, 读者可参考专著[4-6]以及综述性文献[7-8].粗略地讲, 半线性抛物方程整体解和非整体解的存在性以及解的行为依赖于非线性项, 维数, 初始值以及非线性边界流.特别地, Quittner和Souplet[5, 第五章]详细介绍了具有Dirichlet边界条件的非局部反应-扩散方程解的定性性质.从某种意义上讲, 非局部模型比局部模型更贴近实际问题, 而局部理论在非局部模型中不再成立, 使我们必须改进和探索现有的研究方法及理论.本文中, 我们的兴趣在于讨论具有变扩散系数和非局部项半线性抛物方程解的爆破现象中爆破时间界的估计问题.实际上, 此类问题中估计爆破时间上界的方法较多(见Levine[9]介绍的六种方法), 但是爆破时间下界的估计值很难得到.最近, 关于爆破时间下界估计问题的研究也有了新的进展.对具有常数扩散系数的局部反应-扩散方程, 读者可参看文献[10-11] (三维情形)和文献[12] (高维情形).

对具有常数扩散系数的非局部反应-扩散方程的研究方面, Song[13]在齐次Dirichlet或齐次Neumann边界条件下, 研究了具有非局部源项和局部吸收项的半线性抛物方程

$ u_{t}=\Delta u+\int_{\Omega}u^{q}{\rm d}x-ku^{s}, \;\;\; (x, t)\in\Omega\times(0, t^{\ast}). $ (1.4)

他在三维空间中得到了问题解的爆破时间$t^{\ast}$的下界.之后, 文献[14]将文献[13]中的结果推广至高维空间$(N\geq 3).$ Liu[15]考虑了在非线性边界条件下的方程(1.4)的解在三维空间中爆破时间下界. Song等[16]和Ma等[17]考虑了具有空变系数非局部源项的反应扩散方程, 并在高维空间中$(N\geq 3)$得到了解的爆破时间的上下界估计值.此外, 关于非局部拟线性抛物型方程中爆破时间的下界估计问题, 请参见文献[18] $(N=1, 2)$和文献[19-21] $(N\geq 3).$

对具有变扩散系数的局部反应-扩散方程的研究方面, Li等[22]考虑了非齐次Neumann边界条件下具有内部吸收项的半线性散度型偏微分方程

$ u_{t}=\sum\limits^{N}_{i, j=1}(a^{ij}(x)u_{x_{i}})_{x_{j}}-f(u), \;\;\; (x, t)\in\Omega\times(0, t^{\ast}), $

其中$(a^{ij}(x))$为可微的$N \times N$阶正定矩阵, 非负函数$f$满足适当的局部条件.他们得到了解的整体存性和爆破的充分条件, 并给出了在适当测度意义下的爆破时间的上界估计及三维空间中爆破时间下界估计. Baghaei等[23]在文献[22]的基础上, 将爆破时间下界的估计推广至高维$(N\geq 3)$情形. Fang和Wang[24]及Ma和Fang[25]考虑了具有时变系数或空变系数吸收项的局部问题, 并给出了在高维空间中$(N\geq 3)$变系数对爆破时间上下界的影响.对具有变扩散系数的非局部反应-扩散方程的研究方面, Ma和Fang[26]研究了具有空变系数非局部吸收项和非线性边界流的反应-扩散方程解的爆破时间上下界.但是, 前述的文献中变扩散系数矩阵$(a^{ij}(x))$是为了保证一致椭圆型且主要研究了吸收项的系数对爆破时间界的影响.

特别地, 我们关注于Wang和Song[27]最新研究进展.他们考虑了一类具有变扩散系数和非局部源项的耦合方程组, 并对变扩散系数适当限制后讨论了爆破时间下界及非同时爆破现象.本文中, 我们的目的在于对具有变扩散系数的非局部模型(1.1)-(1.3)建立若干不同测度意义下解的整体存在性与爆破的充分条件, 并在整体空间上$(N\geq 1)$讨论变扩散系数对爆破时间界的影响.

本文的剩余部分结构如下:第二节, 我们给出整体解存在的充分条件.第三节, 利用修正不等式技巧, 找出变扩散系数$a(x)$对爆破的影响且在两种不同的测度意义下得到爆破发生的条件及爆破时间的上界.第四节, 在整体空间上$(N\geq 1), $给出若干个不同测度意义下的爆破时间的下界.第五节, 给出几个例子来验证我们的主要结论.

2 整体存在性

本节中, 我们建立问题(1.1)-(1.3)解的整体存在性结论.

定理2.1  假设非负函数$f(u)$满足非局部条件

${\rm(F_{1})}$ $f(s(x, t))\leq (s(x, t))^{p}\int_{\Omega}(s(x, t))^{q}{\rm d}x, \;\;\; s(x, t)\geq0, $

其中函数$s(x, t)\in C(\Omega\times(0, t^{\ast})), $常数$p\geq0$, $q>0$且满足$p+q>1.$同时, 扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{1}(\Omega)$满足

${\rm (a_{1})}$ $a(x)\geq c>0, \; \; \; x\in\bar{\Omega}$,

其中$c>0$为常数.若初值充分小使得$\kappa-\mu<0, $其中$\kappa$, $\mu>0$$(2.6)$式给出.则问题(1.1)-(1.3)的解$u(x, t)$不发生爆破, 即$u(x, t)$$\forall t>0$整体存在.

  定义辅助函数

$ \psi(t)=\theta^{-1}\int_{\Omega}u^{2n(p+q-1)}{\rm d}x, \;\;\; n>\max \bigg\{\frac{1}{2(p+q-1)}, \frac{N}{4}\bigg\}, $

其中$\theta=\int_{\Omega}u_{0}^{2n(p+q-1)}{\rm d}x.$

$\psi(t)$求导并由(1.1)式, (1.2)式以及Green公式得

$ \begin{eqnarray*} \psi'(t)&=&2n(p+q-1)\theta^{-1}\int_{\Omega}u^{2n(p+q-1)-1}[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x\\ &\leq&2n(p+q-1)\theta^{-1}\int_{\partial\Omega}a(x)u^{2n(p+q-1)-1}\frac{\partial u}{\partial n}{\rm d}s\\ &&-2n(p+q-1)[2n(p+q-1)-1]\theta^{-1} \int_{\Omega}a(x)u^{2n(p+q-1)-2}|\nabla u|^{2}{\rm d}x\\ &&+2n(p+q-1)\theta^{-1}\int_{\Omega}u^{2n(p+q-1)+p-1}{\rm d}x\int_{\Omega}u^{q}{\rm d}x\\ &\leq&-\frac{2[2n(p+q-1)-1]\theta^{-1}c}{n(p+q-1)}\int_{\Omega}|\nabla u^{n(p+q-1)}|^{2}{\rm d}x\\ &&+2n(p+q-1)\theta^{-1}|\Omega|\int_{\Omega} u^{(2n+1)(p+q-1)}{\rm d}x. \end{eqnarray*} $

$u^{n(p+q-1)}=v, $则上式可改写为

$ \begin{equation} \psi'(t)\leq-\frac{2[2n(p+q-1)-1]\theta^{-1}c}{n(p+q-1)}\int_{\Omega}|\nabla v|^{2}{\rm d}x+2n(p+q-1)\theta^{-1}|\Omega|\int_{\Omega} v^{2+\frac{1}{n}}{\rm d}x. \end{equation} $ (2.1)

对(2.1)式右端第二项应用Hölder不等式, 我们得到

$ \begin{equation} \int_{\Omega}v^{2+\frac{1}{n}}{\rm d}x\leq\bigg(\int_{\Omega}v^{2}{\rm d}x\bigg)^{\frac{4n-N+2}{4n}}\bigg(\int_{\Omega}v^{\frac{2N}{N-2}}{\rm d}x\bigg)^{\frac{N-2}{4n}}, \end{equation} $ (2.2)

其中$n>\frac{N-2}{4}.$

又由${\Bbb R}^{N}$ $(N\geq3)$中的Sobolev嵌入不等式$(L^{\frac{2N}{N-2}}(\Omega)\hookrightarrow W_{1, 2}(\Omega))$

$ \begin{equation} \int_{\Omega}v^{\frac{2N}{N-2}}{\rm d}x\leq C_{s}^{\frac{2N}{N-2}}\bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{N-2}}, \end{equation} $ (2.3)

其中$C_{s}$为Sobolev最优常数.

将(2.3)式代入(2.2)式中得

$ \begin{equation} \int_{\Omega}v^{2+\frac{1}{n}}{\rm d}x\leq C_{s}^{\frac{N}{2n}} \bigg(\int_{\Omega}v^{2}{\rm d}x\bigg)^{\frac{4n-N+2}{4n}} \bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}}. \end{equation} $

其中$n>\frac{N}{4}.$

再把(2.4)式代入(2.1)式中, 我们可以导出

$ \begin{eqnarray*} \psi'(t)&\leq&-\frac{2[2n(p+q-1)-1]\theta^{-1}c}{n(p+q-1)}\int_{\Omega}|\nabla v|^{2}{\rm d}x\\ &&+2n(p+q-1)\theta^{-\frac{N-2}{4n}}|\Omega|C_{s}^{\frac{N}{2n}}(\psi(t))^{\frac{4N-N+2}{4N}} \bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}} \\ &=&\bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}}\bigg\{2n(p+q-1)|\Omega|C_{s}^{\frac{2N}{N-2}} \theta^{\frac{2-N}{4n}}(\psi(t))^{\frac{4n-N+2}{4n}}\\ &&-\frac{2[2n(p+q-1)-1]\theta^{-1}c}{n(p+q-1)} \bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{4n-N}{4n}}\bigg\}, \end{eqnarray*} $

接下来, 对上式大括号中第二项应用膜不等式

$ \int_{\Omega}|\nabla v|^{2}{\rm d}x\geq \lambda_{1}\int_{\Omega}v^{2}{\rm d}x, $

其中$\lambda_{1}$为如下固定膜问题的第一特征值

$ \Delta\phi_{1}+\lambda_{1}\phi_{1}=0, \;\;\; x\in\Omega;\;\;\; \phi_{1}=0, \;\;\; x\in\partial\Omega. $

因此, 我们可得

$ \begin{eqnarray} \psi'(t)&\leq&\bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}}\bigg\{2n(p+q-1)|\Omega|C_{s}^{\frac{2N}{N-2}}\theta^{\frac{2-N}{4n}}(\psi(t))^{\frac{4n-N+2}{4n}} \\ &&-\frac{2[2n(p+q-1)-1]\theta^{-\frac{N}{4n}}c\lambda_{1}^{\frac{4n-N}{4n}}}{n(p+q-1)} (\psi(t))^{\frac{4n-N}{4n}}\bigg\} \\ &=&\bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}}(\psi(t))^{\frac{4n-N}{4n}} \bigg\{2n(p+q-1)|\Omega|C_{s}^{\frac{2N}{N-2}}\theta^{\frac{2-N}{4n}} (\psi(t))^{\frac{1}{2n}}\\ &&-\frac{2[2n(p+q-1)-1]\theta^{-\frac{N}{4n}}c\lambda_{1}^{\frac{4n-N}{4n}}}{n(p+q-1)} \bigg\} \\ &=&\theta^{-\frac{N}{4n}}\bigg(\int_{\Omega}|\nabla v|^{2}{\rm d}x\bigg)^{\frac{N}{4n}}(\psi(t))^{\frac{4n-N}{4n}}[\kappa(\psi(t))^{\frac{1}{2n}}-\mu], \end{eqnarray} $ (2.5)

其中

$ \kappa=2n(p+q-1)|\Omega|C_{s}^{\frac{2N}{N-2}}\theta^{\frac{1}{2n}}, \;\;\;\mu=\frac{2c\lambda_{1}^{\frac{4n-N}{4n}}[2n(p+q-1)-1]}{n(p+q-1)}. $ (2.6)

假设初值充分小使得

$ \kappa-\mu<0, $

则对$\forall t>0, \kappa(\psi(t))^{\frac{1}{2n}}-\mu<0$一直成立, 或存在第一时刻$t_{0}$使得

$ \kappa(\psi(t_{0}))^{\frac{1}{2n}}-\mu=0. $ (2.7)

$t\in(0, t_{0}), $若有$\kappa(\psi(t))^{\frac{1}{2n}}-\mu<0, $则由(2.5)式知

$ \psi'(t)\leq0, \;\;\; t\in(0, t_{0}), $ (2.8)

且与(2.7)式矛盾, 即(2.8)式对$\forall t>0$都成立且问题(1.1)-(1.3)解整体存在.

3 爆破时间的上界

本节中, 我们对变扩散系数$a(x)$给出适当的条件, 使问题(1.1)-(1.3)的解在有限时刻发生爆破, 并在两种不同测度意义下给出爆破时间$t^{\ast}$的上界.

首先, 我们得到如下结论.

定理 3.1  假设$u(x, t)$为问题(1.1)-(1.3)的非负经典解, 变扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{1}(\Omega)$满足条件(a$_{{1}}), $非负可积函数$f(u)$满足条件

${\rm(F_{2})}$ $\int_{\Omega}\xi f(\xi){\rm d}x\geq2(1+\alpha)\int_{\Omega}F(\xi){\rm d}x, \;\;\; s(x, t)\geq0, $

其中$F(\xi)=\int_{0}^{\xi}f(\eta){\rm d } \eta, $且有$\alpha\geq0.$

$ H(t)=-\int_{\Omega}a(x)|\nabla u|^{2}{\rm d}x+2\int_{\Omega}F(u){\rm d}x, $

并令

$ H(0)>0. $

则问题(1.1)-(1.3)的解$u(x, t)$$\varphi_{1}(t)=\int_{\Omega}u^{2}{\rm d}x$的意义下有限时刻

$ t^{\ast}\leq T=\frac{\varphi_{1}(0)}{2\alpha(1+\alpha)H(0)}, \;\;\; \alpha>0 $

爆破, 其中$\varphi_{1}(0)=\int_{\Omega}u_{0}^{2}{\rm d}x.$$\alpha=0, $$T=\infty.$

  对$\varphi_{1}(t)$求导, 并由定理3.1中的假设可得

$ \begin{eqnarray} \varphi'_{1}(t)&=&2\int_{\Omega}u[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x \\ &=&2\int_{\partial\Omega}a(x)u\frac{\partial u}{\partial n}{\rm d}s-2\int_{\Omega}a(x)|\nabla u|^{2}{\rm d}x+2\int_{\Omega}uf(u){\rm d}x \\ &\geq&-2\int_{\Omega}a(x)|\nabla u|^{2}{\rm d}x+4(1+\alpha)\int_{\Omega}F(u){\rm d}x \\ &\geq&-2(1+\alpha)\int_{\Omega}a(x)|\nabla u|^{2}{\rm d}x+4(1+\alpha)\int_{\Omega}F(u){\rm d}x \\ &=&2(1+\alpha)H(t). \end{eqnarray} $ (3.1)

$H(t)$求导并由Green公式, 我们导出

$ H'(t)=-2\int_{\Omega}a(x)\nabla u\cdot \nabla u_{t}{\rm d}x+2\int_{\Omega}f(u)u_{t}{\rm d}x =2\int_{\Omega}u_{t}^{2}{\rm d}x\geq0, $

$H(0)>0$$H(t)>0$, $t\in(0, t^{\ast}).$

由Schwarz不等式, 我们得到

$ \begin{equation} (\varphi'_{1}(t))^{2}=4 \bigg(\int_{\Omega}u u_{t}{\rm d}x\bigg)^{2}\leq4\int_{\Omega}u^{2}{\rm d}x \int_{\Omega}u_{t}^{2}{\rm d}x=2H'(t)\varphi_{1}(t). \end{equation} $ (3.2)

结合(3.1)式和(3.2)式, 我们导出如下微分不等式

$ \begin{equation} 2(1+\alpha)\varphi'_{1}(t)H(t)\leq(\varphi'_{1}(t))^{2}\leq2H'(t)\varphi_{1}(t), \end{equation} $ (3.3)

$ \begin{equation} (H\varphi_{1}^{-(1+\alpha)})'\geq0. \end{equation} $ (3.4)

对(3.4)式从$0$$t$积分得

$ \begin{equation} H(t)(\varphi_{1}(t))^{-(1+\alpha)}\geq H(0)(\varphi_{1}(0))^{-(1+\alpha)}. \end{equation} $ (3.5)

将(3.5)式代入(3.1)式中, 我们得到如下微分不等式

$ \begin{equation} \varphi'_{1}(t)\geq2(1+\alpha)H(0)(\varphi_{1}(0))^{-(1+\alpha)}(\varphi_{1}(t))^{1+\alpha}. \end{equation} $ (3.6)

再对(3.6)式从$0$$t$积分得如下不等式

$ \begin{equation} (\varphi_{1}(t))^{-\alpha}\leq(\varphi_{1}(0))^{-\alpha}-2\alpha(1+\alpha)H(0)(\varphi_{1}(0))^{-(1+\alpha)}t. \end{equation} $ (3.7)

显然, (3.7)式不可能对$t>0$总是成立.因此, 对$\alpha>0, $我们有$t^{\ast}\leq T=\frac{\varphi_{1}(0)}{2\alpha(1+\alpha)H(0)}.$

$\alpha=0, $则由(3.6)式我们得

$ \varphi_{1}(t)\geq \varphi_{1}(0) e^{2H(0)(\varphi_{1}(0))^{-(1+\alpha)}t}, $

$t>0$成立, 这说明$t^{\ast}=\infty, $$T=\infty$且发生无限爆破.定理3.1证毕.

下面, 我们在加权$L^{1}$意义下得到解在有限时刻发生爆破及爆破解的爆破时间的上界.

定理3.2  假设$\Omega\subset{\Bbb R}^{N}(N\geq2)$为具有光滑边界的有界区域, $u(x, t)$为问题$(1.1)-(1.3)$的非负经典解, 变扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{2}(\Omega)$满足条件

${\rm (a_{2})}$ $a''(x)\geq c'>0, \;\;\; x\in\bar{\Omega}$,

其中$c'$为常数.同时, 非负函数$f$满足非局部条件

${\rm(F_{3})}$ $f(s(x, t))\geq (s(x, t))^{p}\int_{\Omega}(s(x, t))^{q}{\rm d}x, \;\;\; s(x, t)\geq0, $

其中函数$s(x, t)\in C(\Omega\times(0, t^{\ast})), $常数$p\geq0$ $, q>0$且满足$p+q\geq1.$定义辅助函数

$ \varphi_{2}(t)=\int_{\Omega}a(x)u(x, t){\rm d}x, $

则当$p+q>1$时, 问题(1.1)-(1.3)的解$u(x, t)$在加权测度$\varphi_{2}(t)$的意义下有限时刻$t^{\ast}$发生爆破, 且满足

$ t^{\ast}\leq T_{1}=\frac{p+q-1}{C\varphi_{2}^{p+q-1}(0)}, $

其中$C$为下面证明过程中给出的可计算常数.当$p+q=1$时, $t^{\ast}=\infty, $即解无限爆破.

  当$p+q>1$时, 我们对$\varphi_{2}(t)$求导并由(1.1)式, 条件(F$_{3})$以及Green公式得

$ \begin{eqnarray} \varphi'_{2}(t)&=&\int_{\Omega}a(x)[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x \\ &\geq&\int_{\partial\Omega}(a(x))^{2}\frac{\partial u}{\partial n}{\rm d}s-\int_{\Omega}a(x)\nabla a(x)\cdot\nabla u {\rm d}x+\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &=&-\frac{1}{2}\int_{\Omega}\nabla(a(x))^{2}\cdot \nabla u {\rm d}x+\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &=&-\frac{1}{2}\int_{\partial\Omega}u \frac{\partial(a(x))^{2}}{\partial n}{\rm d}s+\frac{1}{2}\int_{\Omega}u \Delta (a(x))^{2}{\rm d}x+\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &=&\frac{1}{2}\int_{\Omega}u(a(x)a''(x)+(a(x))^{2}){\rm d}x+\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &\geq&\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x. \end{eqnarray} $ (3.8)

对(3.8)式利用Cauchy不等式以及反向Hölder不等式, 我们导出

$ \begin{eqnarray} \varphi'_{2}(t)&\geq& \bigg(\int_{\Omega}(a(x))^{\frac{1}{2}}u^{\frac{p+q}{2}}{\rm d}x\bigg)^{2} \\ &\geq& \bigg[\bigg(\int_{\Omega}a(x)u{\rm d}x\bigg)^{\frac{p+q+1}{2}} \bigg(\int_{\Omega}(a(x))^{\frac{p+q}{p+q-1}}u^{\frac{1}{p+q-1}}{\rm d}x \bigg)^{\frac{1-(p+q)}{2}}\bigg]^{2} \\ &=&\varphi_{2}^{p+q+1}(t) \bigg[\int_{\Omega}(a(x))^{\frac{p+q}{p+q-1}}u^{\frac{1}{p+q-1}}{\rm d}x\bigg]^{1-(p+q)}. \end{eqnarray} $ (3.9)

由于$p+q>1, $对(3.9)式中积分项应用Hölder不等式, 我们得

$ \begin{equation} \int_{\Omega}(a(x))^{\frac{p+q}{p+q-1}}u^{\frac{1}{p+q-1}}{\rm d}x\leq \bigg(\int_{\Omega}a(x)u{\rm d}x\bigg)^{\frac{1}{p+q}} \bigg[\int_{\Omega}(a(x))^{-\frac{(p+q)^{2}-(p+q-1)}{(p+q-1)^{2}}}u^{-\frac{1}{(p+q-1)^{2}}}{\rm d}x \bigg]^{\frac{p+q-1}{p+q}}. \end{equation} $ (3.10)

将(3.10)式代入(3.9)式并由Hölder不等式, 我们有

$ \begin{eqnarray*} \varphi'_{2}&\geq&\varphi_{2}^{^{p+q+1-\frac{p+q-1}{p+q}}}(t) \bigg[\int_{\Omega}(a(x))^{\frac{(p+q)^{2}-(p+q-1)}{(p+q-1)^{2}}}u^{-\frac{1}{(p+q-1)^{2}}{}}{\rm d}x \bigg]^{\frac{(1-p-q)^{2}}{p+q}} \\ &\geq&\varphi_{2}^{^{p+q+1-\frac{p+q-1}{p+q}}}(t) \bigg[\varphi_{2}^{-\frac{1}{(p+q-1)^{2}}}(t) \bigg(\int_{\Omega}(a(x))^{-\frac{(p+q)(p+q-1)}{(p+q-1)^{2}+1}}{\rm d}x \bigg)^{\frac{(p+q-1)^{2}+1}{(p+q-1)^{2}}}\bigg]^{\frac{(p+q-1)^{2}}{p+q}}. \end{eqnarray*} $

$C:=(\int_{\Omega}(a(x))^{-\frac{(p+q)(p+q-1)}{(p+q-1)^{2}+1}}{\rm d}x)^{\frac{(p+q-1)^{2}+1}{p+q}}, $则我们得到如下微分不等式

$ \begin{equation} \varphi'_{2}(t)\geq C\varphi_{2}^{p+q}(t), \end{equation} $ (3.11)

对上式在$[0, t]$上积分

$ \frac{1}{1-(p+q)}[\varphi_{2}^{1-(p+q)}(t)-\varphi_{2}^{1-(p+q)}(0)]\geq Ct, $

$ \begin{equation} (\varphi_{2}(t))^{1-(p+q)}\leq(\varphi_{2}(0))^{1-(p+q)}-\frac{Ct}{p+q-1}. \end{equation} $ (3.12)

显然, (3.12)式不可能对$t>0$总是成立.因此, $u(x, t)$在加权$L^{1}$意义下在有限时刻发生爆破且满足

$ t^{\ast}\leq T_{1}=\frac{p+q-1}{C\varphi_{2}^{p+q-1}(0)}, $

$p+q>1$成立.

$p+q=1$时, 对辅助函数$\varphi_{2}(t)$求导并由(1.1)和(1.2)式, Green公式得

$ \begin{eqnarray} \varphi'_{2}(t)&\geq&\int_{\partial\Omega}(a(x))^{2}\frac{\partial u}{\partial n}{\rm d}s-\frac{1}{2}\int_{\Omega}\nabla (a(x))^{2}\cdot\nabla u {\rm d}x+\int_{\Omega}a(x)u^{p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &\geq&-\frac{1}{2}\int_{\Omega}\nabla (a(x))^{2}\cdot\nabla u {\rm d}x \\ &=&\frac{1}{2}\int_{\Omega}(a(x)a''(x)+(a'(x))^{2})u{\rm d}x \\ &\geq& c'\varphi_{2}(t). \end{eqnarray} $ (3.13)

对(3.13)式从$0$$t$上积分得$\varphi_{2}(t)\geq\varphi_{2}(0)e^{c't}, $$t>0$成立, 其中$\varphi_{2}(0)=\int_{\Omega}a(x)u_{0}{\rm d}x. $因此, 我们得到$t^{\ast}=\infty$且发生无限爆破.定理3.2证毕.

4 爆破时间的下界

本节中, 我们在整体空间$(N\geq1)$上导出当问题(1.1)-(1.3)解发生爆破时若干个适当的测意义下的爆破时间下界.

4.1 $N\geq3$的情形

我们利用修正不等式技巧, 得到爆破时间的下界.因为我们用到了Sobolev不等式, 所以要求$\Omega\subset{\Bbb R}^{N}(N\geq3)$为具有光滑边界$\partial\Omega$的有界区域.

定理 4.1  假设$u(x, t)$为问题(1.1)-(1.3)的非负经典解, $u(x, t)$在有限时刻$t^{\ast}$发生爆破, 非线性项$f(u)$满足非局部条件(F$_{1}), $其中常数$p\geq0$, $q>0$且满足$p+q>1. $若扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{2}(\Omega)$满足条件(a$_{1})$

${\rm (a_{3})}$ $|\nabla a^{\frac{1}{2}}(x)|^{2}\leq M, \;\;\; x\in\bar{\Omega}$,

其中$M\geq0$为常数且定义辅助函数

$ \phi_{1}(t)=\int_{\Omega}u^{l+1}{\rm d}x, $

其中

$ l+1>\max\bigg\{1, (2N-4)(p-1), 1-p+q, 1-p+\frac{Nq}{3N-8}\bigg\}. $

则爆破时间$t^{\ast}$满足

$ t^{\ast}\geq\int_{\phi_{1}(0)}^{\infty}\frac{{\rm d}\eta}{J_{1}+J_{2}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+J_{3}\eta^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}}, $

其中$\phi_{1}(0)=\int_{\Omega}u_{0}^{l+1}{\rm d}x, $ $J_{1}$, $J_{2}$, $J_{3}$为后面证明过程中给出.

  对$\phi_{1}(t)$求导并由(1.1)和(1.2)式以及Green公式得

$ \begin{eqnarray} \phi'_{1}(t)&=&(l+1)\int_{\Omega}u^l[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x \\ &\leq&(l+1)\int_{\partial\Omega}a(x)u^l\frac{\partial u}{\partial n}{\rm d}s-l(l+1)\int_{\Omega}a(x)u^{l-1}|\nabla u|^2{\rm d}x+(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &=&-\frac{4l}{l+1}\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^{2}{\rm d}x+(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x. \end{eqnarray} $ (4.1)

对(4.1)式中右端第二项应用Hölder不等式, 我们得

$ \begin{eqnarray} (l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x\leq (l+1)|\Omega|^{\frac{l+p-q}{l+p}}\bigg[\int_{\Omega}u^{l+p}{\rm d}x\bigg]^{\frac{l+p+q}{l+p}}, \end{eqnarray} $ (4.2)

其中$l+p-q>0$.

由条件知$l+1>(2N-4)(p-1)$且对(4.2)式右端积分项应用Young不等式, 我们导出

$ \begin{eqnarray} \int_{\Omega}u^{l+p}{\rm d}x\leq \bigg(\int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x\bigg)^{q_{1}}|\Omega|^{1-q_{1}} \leq q_{1}\int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x+(1-q_{1})|\Omega|, \end{eqnarray} $ (4.3)

其中

$ \begin{eqnarray} q_{1}=\frac{2(N-2)(l+p)}{(l+1)(2N-3)}\in(0, 1). \end{eqnarray} $ (4.4)

将(4.3)式代入(4.2)式中, 并由基本不等式$(a_{1}+a_{2})^{r}\leq2^{r-1}(a_{1}^{r}+a_{2}^{r})$, $a_{1}$, $a_{2}>0$, $r>1$可得

$ \begin{eqnarray*} &&(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &\leq&(l+1)|\Omega|^{\frac{l+p-q}{l+p}} \bigg[q_{1}\int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x+(1-q_{1})|\Omega|\bigg]^{\frac{l+p+q}{l+p}} \\ & \leq&(l+1)|\Omega|^{\frac{l+p-q}{l+p}}2^{\frac{q}{l+p}}\bigg[{q_{1}}^{\frac{l+p+q}{l+p}} \bigg(\int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x\bigg)^{\frac{l+p+q}{l+p}} +((1-q_{1})|\Omega|)^{\frac{l+p+q}{l+p}}\bigg]. \end{eqnarray*} $

$ K_{1}=(l+1)|\Omega|^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}((1-q_{1})|\Omega|)^{\frac{l+p+q}{l+p}}, \;\;\; K_{2}=(l+1)|\Omega|^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}{q_{1}}^{\frac{l+p+q}{l+p}}, $

则上式可改写为

$ \begin{eqnarray} (l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^q{\rm d}x\leq K_{1}+K_{2} \bigg(\int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x\bigg)^{\frac{l+p+q}{l+p}}. \end{eqnarray} $ (4.5)

再对(4.5)式右端积分项应用Hölder不等式并由条件(a$_{1})$

$ \begin{eqnarray} \int_{\Omega}u^{\frac{(l+1)(2N-3)}{2(N-2)}}{\rm d}x &\leq &\bigg[\int_{\Omega}((a(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x\bigg]^{\frac{1}{4}} \bigg[\int_{\Omega}(a(x))^{-\frac{N}{3(N-2)}}u^{l+1}{\rm d}x\bigg]^{\frac{3}{4}} \\ &\leq& c^{-\frac{N}{4(N-2)}}\bigg[\int_{\Omega}((a(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x \bigg]^{\frac{1}{4}}[\phi_{1}(t)]^{\frac{3}{4}}. \end{eqnarray} $ (4.6)

对(4.6)式中积分项应用$\Omega\subset{\Bbb R}^{N}(N\geq3)$中的加权Sobolev嵌入不等式$(L^{\frac{2N}{N-2}}(\Omega)\hookrightarrow W_{1, 2}(\Omega))$

$ \|(a(x))^{\frac{1}{2}}u^{\frac{l+1}{2}}\|_{L^{\frac{2N}{N-2}}(\Omega)}\leq C_{s}\|(a(x))^{\frac{1}{2}}u^{\frac{l+1}{2}}\|_{w^{1, 2}(\Omega)}, $

其中$C_{s}$为Sobolev最优常数, 并由Jensen不等式可得

$ \begin{eqnarray} \bigg[\int_{\Omega}((a(x))^{\frac{1}{2}}u^{\frac{l+1}{2} })^{\frac{2N}{N-2}}{\rm d}x\bigg]^{\frac{1}{4}} &\leq& {C_{s}}^{\frac{N}{2(N-2)}} \bigg[\int_{\Omega}(a(x))|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x+\int_{\Omega}|\nabla(a(x))^{\frac{1}{2}}|^2 u^{l+1}{\rm d}x\bigg]^{\frac{N}{4(N-2)}} \\ &\leq& {C_{s}}^{\frac{N}{2(N-2)}}\bigg[\int_{\Omega}(a(x))|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x+M\int_{\Omega}u^{l+1}{\rm d}x\bigg]^{\frac{N}{4(N-2)}} \\ &\leq& C_{b}\bigg[M^{\frac{N}{4(N-2)}}(\phi_{1}(t))^{\frac{N}{4(N-2)}}+ \bigg(\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x \bigg)^{\frac{N}{4(N-2)}}\bigg], \end{eqnarray} $ (4.7)

其中

$ C_{b}= \left\{\begin{array}{ll} 2^{\frac{1}{2}}(C_{s})^{\frac{3}{2}} , \;\;\;& N=3, \\ (C_{s})^{\frac{N}{2(N-2)}} , \;\;\;&N>3. \end{array}\right. $

由(4.5)和(4.7)式以及带$\epsilon$的Young不等式, 我们导出

$ \begin{eqnarray} &&(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^q {\rm d}x\\ &\leq& K_{1}+K_{2}\{c^{-\frac{N}{4(N-2)}} \bigg[\int_{\Omega}((a(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x \bigg]^{\frac{1}{4}}(\phi_{1}(t))^{\frac{3}{4}}\}^{\frac{l+p+q}{l+p}} \\ &\leq &K_{1}+K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}{C_{b}}^{\frac{l+p+q}{l+p}} \bigg[M^{\frac{N}{4(N-2)}}(\phi_{1}(t))^{\frac{2N-3}{2(N-2)}}\\ &&+ \bigg(\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x \bigg)^{\frac{N}{4(N-2)}}(\phi_{1}(t))^{\frac{3}{4}}\bigg]^{\frac{l+p+q}{l+p}} \\ &\leq &K_{1}+K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}} 2^{\frac{q}{l+p}}M^{\frac{N(l+p+q)}{4(N-2)(l+p)}}(\phi_{1}(t))^{ \frac{(2N-3)(l+p+q)}{2(N-2)(l+P)}}\\ &&+K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}} \bigg(\epsilon\int_{\Omega}a|\nabla u^{\frac{l+1}{2}}|^{2}{\rm d}x \bigg)^{\frac{N(l+p+q)}{4(N-2)(l+p)}}\\ &&\times\Big[\epsilon^{-\frac{N(l+p+q)} {4(N-2)(l+p)-N(l+p+q)}}(\phi_{1}(t))^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}} \Big]^{\frac{4(N-2)(l+p)-N(l+p+q)}{4(n-2)(l+p)}} \\ &\leq& K_{1}+K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}M^{\frac{N(l+p+q)}{4(N-2)(l+p)}}(\phi_{1}(t))^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}} \\ &&+K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}} \bigg[{\frac{N(l+p+q)\epsilon}{4(N-2)(l+p)}}\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x \\ &&+{\frac{4(N-2)(l+p)-N(l+p+q)}{4(N-2)(l+p)}}\epsilon^{-\frac{N(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}(\phi_{1}(t))^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}} \bigg], \end{eqnarray} $ (4.8)

其中$l>-p+{\frac{Nq}{3N-8}}$$\epsilon>0$为待定常数.

将(4.8)式代入(4.1)式并整理得

$ \begin{equation} \phi'_{1}(t)\leq J_{1}+J_{2}(\phi_{1}(t))^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+J_{3}(\phi_{1}(t))^{\frac{3(N-2)(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}+J_{4}\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x, \end{equation} $ (4.9)

其中

$ \begin{eqnarray*} &&J_{1}=K_{1}, \\ &&J_{2}=K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}M^{\frac{N(l+p+q)}{4(N-2)(l+p)}}, \\ &&J_{3}=K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}{\frac{4(N-2)(l+p)-N(l+p+q)}{4(N-2)(l+p)}}\epsilon^{-\frac{N(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}, \\ &&J_{4}=K_{2}c^{-\frac{N(l+p+q)}{4(N-2)(l+p)}}C_{b}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}{\frac{N(l+p+q)\epsilon}{4(N-2)(l+p)}}-{\frac{4l}{l+1}}. \end{eqnarray*} $

取适当的$\epsilon>0, $使得$J_{4}=0, $则(4.8)式可改写为

$ \begin{eqnarray} \phi'_{1}(t)\leq J_{1}+J_{2}(\phi_{1}(t))^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+J_{3}(\phi_{1}(t))^{\frac{3(N-2)(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}, \end{eqnarray} $ (4.10)

且由条件知$J_{1}, J_{2}, J_{3}>0.$

$\displaystyle\lim_{t\rightarrow t^{\ast}}\phi_{1}(t)=\infty$, 我们对(4.9)式在$[0, t^{\ast}]$上积分得

$ t^{\ast}\geq \int_{\phi_{1}(0)}^{\infty}\frac{{\rm d}\eta}{J_{1}+J_{2}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+J_{3}\eta^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}}, $

其中${\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}>1.$

下面, 我们考虑加权测度意义下的爆破时间的下界.

定理 4.2  假设$u(x, t)$为问题(1.1)-(1.3)的非负经典解, $u(x, t)$在有限时刻$t^{\ast}$发生爆破, 非负函数$f$满足如下非局部条件

${\rm(F_{4})}$ $f(s(x, t))\leq (s(x, t))^{p}\int_{\Omega}b(x, t)(s(x, t))^{q}{\rm d}x, \;\;\; s(x, t)\geq0, $

其中函数$s(x, t)\in C(\Omega\times(0, t^{\ast})), $常数$p\geq0$, $q>0$$p+q>1.$若加权函数$b(x)\in C^{0}(\bar\Omega)\cap C^{1}(\Omega)$满足如下条件

${\rm(B_{1})}$ $b(x)>0, x\in\Omega$$b(x)=0, x\in\partial\Omega, $

${\rm(B_{2})}$ $b(x)\geq c_{0}, x\in\bar\Omega, $

${\rm(B_{3})}$ $-b(x)B\leq\nabla b(x)\leq b(x)B\Leftrightarrow|\frac{\partial b(x)}{\partial x_{i}}|\leq B_{i}b(x), x\in\Omega, $

其中$B_{i}>0$, $B=(B_{1}, B_{2}, \cdots, B_{N})$为正常数向量.

同时, 扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{1}(\Omega)$满足条件(a$_{1}).$定义辅助函数

$ \phi_{2}(t)=\int_{\Omega}b(x)u^{l+1}{\rm d}x, $

其中

$ l+1>\max\bigg\{1, (2N-4)(p-1), 1-p+q, 1-p+\frac{Nq}{3N-8}\bigg\}. $

则爆破时间$t^{\ast}$满足

$ t^{\ast}\geq\int_{\phi_{2}(0)}^{\infty}\frac{{\rm d}\eta}{L_{1}+L_{2}\eta+L_{3}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+L_{4}\eta^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}}, $

其中$\phi_{2}(0)=\int_{\Omega}b(x)u_{0}^{l+1}{\rm d}x$.

  对$\phi_{2}(t)$求微分并且由(1.1)和(1.2)式以及Green公式, 条件(a$_{1})$, (F$_{4})$可得

$ \begin{eqnarray} \phi'_{2}(t)&=&(l+1)\int_{\Omega}b(x){u^l}[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x \\ &=&(l+1)\int_{\partial\Omega}a(x)b(x)u^{l}\frac{\partial u}{\partial n}{\rm d}s -(l+1)\int_{\Omega}a(x)\nabla u\cdot\nabla(b(x)u^{l}){\rm d}x\\ &&+(l+1)\int_{\Omega}b(x)u^{l}f(u){\rm d}x\\ &=&-(l+1)c\int_{\Omega}u^{l}\nabla u\cdot\nabla b(x){\rm d}x-l(l+1)c\int_{\Omega}b(x)u^{l-1}|\nabla u|^{2}{\rm d}x \\ &&+(l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^{q}{\rm d}x\\ &=&(l+1)c|B|\int_{\Omega}b(x){u^l}|\nabla u|{\rm d}x-{\frac{4lc}{l+1}} \int_{\Omega}b(x)|\nabla {u^{\frac{l+1}{2}}}|^2{\rm d}x\\ &&+(l+1)\int_{\Omega}b(x){u^{l+p}}{\rm d}x\int_{\Omega}b(x){u^q}{\rm d}x, \end{eqnarray} $ (4.11)

其中$\epsilon_{1}>0$为待定常数.

对(4.11)式右端第一项应用Young不等式, 我们得

$ \begin{equation} (l+1)c|B|\int_{\Omega}b(x){u^l}|\nabla u|{\rm d}x\leq {\frac{{(l+1)^2}c^2|B|^2}{2\epsilon_{1}}}\int_{\Omega}b(x)u^{l+1}{\rm d}x+{\frac{2\epsilon_{1}}{(l+1)^2}}\int_{\Omega}b(x)|\nabla {u^{\frac{l+1}{2}}}|^2{\rm d}x. \end{equation} $ (4.12)

将(4.12)式代入(4.11)式整理得

$ \begin{eqnarray} \phi'_{2}(t)&\leq& {\frac{{(l+1)^2}c^2|B|^2}{2\epsilon_{1}}}\phi_{2}(t)+ \bigg[{\frac{2\epsilon_{1}}{(l+1)^2}}-{\frac{4lc}{l+1}}\bigg]\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x\\ &&+(l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^q{\rm d}x. \end{eqnarray} $ (4.13)

对(4.13)式右端最后一项应用Hölder不等式, 我们导出

$ \begin{eqnarray} (l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^q{\rm d}x\leq (l+1) \bigg[\int_{\Omega}b(x){\rm d}x\bigg]^{\frac{l+p-q}{l+p}} \bigg[\int_{\Omega}b(x)u^{l+p}{\rm d}x\bigg]^{\frac{l+p+q}{l+p}}, \end{eqnarray} $ (4.14)

其中$l+p-q>0.$

再对(4.14)式右端第二个积分项应用Young不等式可得

$ \begin{eqnarray} \int_{\Omega}b(x)u^{l+p}{\rm d}x&\leq& \bigg(\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}} )^{\frac{2N-3}{N-2}}{\rm d}x\bigg)^{q_{1}} \bigg(\int_{\Omega}(b(x))^{\frac{(1-p)(2N-3)}{l+1+(1-p)(2N-4)}}{\rm d}x\bigg)^{1-q_{1}} \\ &\leq &q_{1}\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N-3}{N-2}} {\rm d}x+(1-q_{1})\int_{\Omega}(b(x))^{\frac{(1-p)(2N-3)}{l+1+(1-p)(2N-4)}}{\rm d}x, \end{eqnarray} $ (4.15)

其中$q_{1}\in(0, 1)$如(4.4)式中所定义.

将(4.15)式代入(4.14)式中, 并由基本不等式$(a_{1}+a_{2})^{r}\leq2^{r-1}(a_{1}^{r}+a_{2}^{r})$, $a_{1}$, $a_{2}>0$, $r>1$

$ \begin{eqnarray*} &&(l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^q{\rm d}x \\ &\leq&(l+1)\bigg[\int_{\Omega}b(x){\rm d}x\bigg]^{\frac{l+p-q}{l+p}} \bigg[q_{1}\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}} )^{\frac{2N-3}{N-2}}{\rm d}x\\ &&+(1-q_{1})\int_{\Omega}(b(x ))^{\frac{(1-p)(2N-3)}{l+1+(1-p)(2N-4)}}{\rm d}x\bigg]^{\frac{l+p+q}{l+p}} \\ &\leq&(l+1)\bigg[\int_{\Omega}b(x){\rm d}x\bigg]^{\frac{l+p-q}{l+p}}2^{\frac{q}{l+p}} \bigg[q_{1}^{\frac{l+p+q}{l+p}}\bigg(\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N-3}{N-2}}{\rm d}x \bigg)^{\frac{l+p+q}{l+p}} \\ &&+((1-q_{1})\int_{\Omega}(b(x))^{\frac{(1-p)(2N-3)}{l+1+(1-p)(2N-4)}}{\rm d}x)^{\frac{l+p+q}{l+p}}\bigg]. \end{eqnarray*} $

$ \begin{eqnarray*} &&K'_{1}=(l+1)\bigg[\int_{\Omega}b(x){\rm d}x\bigg]^{\frac{l+p-q}{l+p}} 2^{\frac{q}{l+p}}\bigg[(1-q_{1})\int_{\Omega}(b(x))^{\frac{(1-p)(2N-3)}{l+1+(1-p)(2N-4)}}{\rm d}x\bigg]^{\frac{l+p+q}{l+p}}, \\ &&K'_{2}=(l+1)\bigg[\int_{\Omega}b(x){\rm d}x\bigg]^{\frac{l+p-q}{l+p}}2^{\frac{q}{l+p}}{q_{1}}^{\frac{l+p+q}{l+p}}, \end{eqnarray*} $

则上式可改写为

$ \begin{eqnarray} (l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^q{\rm d}x\leq K'_{1}+K'_{2} \bigg[\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N-3}{N-2}}{\rm d}x \bigg]^{\frac{l+p+q}{l+p}}. \end{eqnarray} $ (4.16)

再对(4.16)式右端积分项应用Hölder不等式, 我们导出

$ \begin{eqnarray} \int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N-3}{N-2}}{\rm d}x &\leq&\bigg[\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x \bigg]^{\frac{1}{4}}\bigg[\int_{\Omega}b(x)u^{l+1}{\rm d}x \bigg]^{\frac{3}{4}} \\ &\leq&\bigg[\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x\bigg]^{\frac{1}{4}}[\phi_{2}(t)]^{\frac{3}{4}}. \end{eqnarray} $ (4.17)

对(4.17)式右端积分项应用${\Bbb R}^{N}(N\geq3)$中的加权Sobolev嵌入不等式$(L^{\frac{2N}{N-2}}(\Omega)\hookrightarrow W_{1, 2}(\Omega))$

$ \|(b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}}\|_{L^{\frac{2N}{N-2}}(\Omega)} \leq C_{s}\|(b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}}\|_{w^{1, 2}(\Omega)}, $

其中$C_{s}$为Sobolev最优常数, 并由Jensen不等式以及条件(B$_{3})$我们得到

$ \begin{eqnarray} \bigg[\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x \bigg]^{\frac{1}{4}} &\leq& C_{s}^{\frac{N}{2(N-2)}} \bigg[\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^{2}{\rm d}x+ \int_{\Omega}|\nabla(b(x))^{\frac{1}{2}}|^{2}u^{l+1}{\rm d}x\bigg]^{\frac{N}{4(N-2)}} \\ &\leq& C_{s}^{\frac{N}{2(N-2)}} \bigg[\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^{2}{\rm d}x+ |B|^{2}\int_{\Omega}u^{l+1}{\rm d}x\bigg]^{\frac{N}{4(N-2)}} \\ &\leq& C_{B}\bigg[(\phi_{2}(t))^{\frac{N}{4(N-2)}}+ \bigg(\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x\bigg)^{\frac{N}{4(N-2)}}\bigg], \end{eqnarray} $ (4.18)

其中$C_{B}=\max\left\{[1+(\frac{1}{2}|B|^{2})^{\frac{N}{4(N-2)}}]C_{b}, 2^{\frac{N}{4(N-2)}}C_{b}\right\}$, $C_{b}$如(4.7)式中所定义.

由条件知$l>-p+\frac{Nq}{3N-8}$且将(4.17)和(4.18)式带入(4.16)式, 再应用带$\epsilon$的Young不等式和基本不等式$(a_{1}+a_{2})^{r}\leq 2^{r-1}(a_{1}^{r}+a_{2}^{r})$, $a_{1}$, $a_{2}>0$, $r>1$, 我们可得

$ \begin{eqnarray} &&(l+1)\int_{\Omega}b(x)u^{l+p}{\rm d}x\int_{\Omega}b(x)u^q{\rm d}x \\ &\leq& K'_{1}+K'_{2}\bigg\{\bigg[\int_{\Omega}((b(x))^{\frac{1}{2}}u^{\frac{l+1}{2}})^{\frac{2N}{N-2}}{\rm d}x \bigg]^{\frac{1}{4}}(\phi_{2}(t))^{\frac{3}{4}} \bigg\}^{\frac{l+p+q}{l+p}} \\ &\leq& K'_{1}+K'_{2}C_{B}^{\frac{l+p+q}{l+p}} \bigg\{\bigg[(\phi_{2}(t))^{\frac{2N-3}{2(N-2)}}+ \bigg(\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x\bigg)^{\frac{N}{4(N-2)}} \bigg](\phi_{2}(t))^{\frac{3}{4}}\bigg\}^{\frac{l+p+q}{l+p}} \\ &\leq &K'_{1}+K'_{2}C_{B}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}(\phi_{2}(t))^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+K'_{2}C_{B}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}} \\ &&\times\bigg[{\frac{N(l+p+q)\epsilon_{2}}{4(N-2)(l+p)}}\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x+{\frac{4(N-2)(l+p)-N(l+p+q)}{4(N-2)(l+p)}} \\ &&\times\epsilon_{2}^{-\frac{N(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}(\phi_{2}(t))^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}\bigg], \end{eqnarray} $ (4.19)

其中$\epsilon_{2}>0$为待定常数.

将(4.19)式代入(4.13)式整理得

$ \begin{eqnarray} \phi'_{2}(t)&\leq &L_{1}+L_{2}\phi_{2}(t)+L_{3}(\phi_{2}(t))^{\frac{(2N-3)(l+p+q)} {2(N-2)(l+p)}}\\ &&+L_{4}(\phi_{2}(t))^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}+L_{5}\int_{\Omega}b(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x, \end{eqnarray} $ (4.20)

其中

$ \begin{eqnarray*} && L_{1}=K'_{1}, \\ &&L_{2}=\frac{(l+1)^2c^2|B|^2}{2\epsilon_{2}}, \\ && L_{3}=K'_{2}{C_{B}}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}, \\ &&L_{4}=K'_{2}{C_{B}}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}{\frac{4(N-2)(l+p)-N(l+p+q)}{4(N-2)(l+p)}}{\epsilon_{2}}^{-\frac{N(l+p+q)}{4(N-2)(l+p)-N(l+p+q)}}, \\ &&L_{5}=K'_{2}{C_{B}}^{\frac{l+p+q}{l+p}}2^{\frac{q}{l+p}}{\frac{N(l+p+q)\epsilon_{2}}{4(N-2)(l+p)}}+{\frac{2\epsilon_{2}}{(l+1)^2}}-{\frac{4lc}{l+1}}. \end{eqnarray*} $

取充分小的$\epsilon_{2}>0$以及适当的$\epsilon_{1}>0$使得$L_{5}=0, $则(4.20)式可改写为

$ \begin{eqnarray} \phi'_{2}(t)\leq L_{1}+L_{2}\phi_{2}(t)+L_{3}(\phi_{2}(t))^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+L_{4}(\phi_{2}(t))^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}, \end{eqnarray} $ (4.21)

并由条件知$L_{1}$, $L_{2}$, $L_{3}$, $L_{4}>0. $

$\displaystyle\lim_{t\rightarrow t^{\ast}}\phi_{2}(t)=\infty$, 我们对(4.21)式在$[0, t^{\ast}]$上积分得

$ t^{\ast}\geq\int_{\phi_{2}(0)}^{\infty}\frac{{\rm d}\eta}{L_{1}+L_{2}\eta+L_{3}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+L_{4}\eta^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}}, $

其中$\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}>1. $定理4.2证毕.

4.2 $N=1$的情形

$N=1$时, 我们假设$\Omega=(0, a), $其中$a>0$为常数.问题(1.1)-(1.3)可改写为

$ \begin{eqnarray} \left\{\begin{array}{ll} u_{t}=(a(x)u_{x})_{x}+f(u), &\;\;\; (x, t)\in(0, a)\times(0, t^{\ast}), \\ u(0, t)=u(a, t)=0, &\;\;\; t\in(0, t^{\ast}), \\ u(x, 0)=u_{0}(x), &\;\;\; x\in(0, a). \end{array}\right. \end{eqnarray} $ (4.22)

我们导出如下结论.

定理 4.3  假设$u(x, t)$为问题$(4.22)$的非负经典解, $u(x, t)$在有限时刻$t^{\ast}$爆破且定义辅助函数

$ \phi_{1}(t)=\int_{0}^{a}u^{l+1}{\rm d}x, $

其中

$ l+1>\max\left\{1, 1-p+q\right\}. $

若扩散系数$a(x)\in C^{0}(\Omega)\cap C^{1}(\Omega)$满足条件(a$_{1}), $非线性项$f(u)$满足条件(F$_{1})$, 其中常数$0\leq p<1$, $q>0$且满足$p+q>1.$则爆破时间$t^{\ast}$满足

$ t^{\ast}\geq{\frac{l+1}{C(p+q-1)(\phi_{1}(0))^{\frac{p+q-1}{l+1}}}}, $

其中$\phi_{1}(0)=\int_{0}^{a}u_{0}^{l+1}{\rm d}x$, 且$C$为可计算常数.

  对辅助函数$\phi_{1}(t)$求导并由(1.1)和(1.2)式以及Green公式得

$ \begin{eqnarray} \phi'_{1}(t)&=&2(l+1)\int_{0}^{a}u^{l}[(a(x)u_{x})_{x}+f(u)]{\rm d}x \\ &\leq&-{\frac{4l}{l+1}}\int_{0}^{a}a(x)|(u_{x})^{\frac{l+1}{2}}|^2{\rm d}x+(l+1)\int_{0}^{a}u^{p+l}{\rm d}x\int_{0}^{a}u^q{\rm d}x \\ &\leq&(l+1)\int_{0}^{a}u^{p+l}{\rm d}x\int_{0}^{a}u^q{\rm d}x \\ &\leq&(l+1)a^{\frac{l+p-q}{l+p}} \bigg[\int_{0}^{a}u^{l+p}{\rm d}x\bigg]^{\frac{l+p+q}{l+1}} \\ &:=&C(\phi_{1}(t))^{\frac{l+p+q}{l+1}}, \end{eqnarray} $ (4.23)

其中$l+p-q>0$, $C=(l+1)a^{\frac{(l+p-q)(l-p+2)}{(l+p)(l+1)}}>0.$

对(4.23)式在$[0, t^{\ast}]$上积分得

$ t^{\ast}\geq{\frac{l+1}{C(p+q-1)(\phi_{1}(0))^{\frac{p+q-1}{l+1}}}}, $

其中$\phi_{1}(0)=\int_{0}^{a}u_{0}^{l+1}{\rm d}x$.定理4.3证毕.

4.3 $N=2$的情形

本节中$\Omega\subset{\Bbb R}^{N}(N=2)$为具有光滑边界$\partial\Omega$的有界区域.

定理 4.4  假设$u(x, t)$为问题(1.1)-(1.3)的非负经典解, $u(x, t)$在有限时刻$t^{\ast}$爆破, 扩散系数$a(x)\in C^{0}(\bar{\Omega})\cap C^{1}(\Omega)$满足条件(a$_{1})$和条件(a$_{3})$, 非线性项$f(u)$满足非局部条件(F$_{1}), $其中常数$1>p\geq0$, $q>0$且满足$p+q>1.$定义辅助函数

$ \phi_{1}(t)=\int_{\Omega}u^{l+1}{\rm d}x, $

其中

$ l+1> \max\left\{1, 1-p+q\right\}. $

则爆破时间$t^{\ast}$有下界, 且

$ t^{\ast}\geq\int_{\Phi(0)}^{\infty}{\frac{{\rm d}\eta}{C_{1}\eta+C_{2}\eta^{\frac{l+p+q}{l+p}}}}, $

其中$\phi_{1}(0)=\int_{\Omega}u_{0}^{l+1}{\rm d}x$, $C_{1}$, $C_{2}$为可计算常数.

  与定理4.3的证明过程类似, 对辅助函数求导并由(1.1)和(1.2)式得

$ \begin{eqnarray} \phi'_{1}(t)&=&2(l+1)\int_{\Omega}u^{l}[{\rm{div}}(a(x)\nabla u)+f(u)]{\rm d}x \\ &\leq&(l+1)\int_{\partial\Omega}a(x)u^{l}\frac{\partial u}{\partial n}{\rm d}s-l(l+1)\int_{\Omega}a(x)u^{l-1}|\nabla u|^{2}{\rm d}x+\int_{\Omega}u^{p+l}{\rm d}x\int_{\Omega}u^{p}{\rm d}x \\ &=&-{\frac{4l}{l+1}}\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x+(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^q{\rm d}x. \end{eqnarray} $ (4.24)

由加权的Poincare不等式以及条件(a$_{3})$

$ \begin{eqnarray} C_{p}\int_{\Omega}a(x)u^{l+1}{\rm d}x\leq\int_{\Omega}a(x)|\nabla u^{\frac{l+1}{2}}|^2{\rm d}x+M\int_{\Omega}u^{l+1}{\rm d}x, \end{eqnarray} $ (4.25)

其中$C_{p}>0.$

将(4.25)式代入(4.24)式得

$ \phi'_{1}(t)\leq-{\frac{4lC_{p}}{l+1}}\int_{\Omega}a(x)u^{l+1}{\rm d}x+{\frac{4lM}{l+1}}\int_{\Omega}u^{l+1}{\rm d}x+(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^q{\rm d}x. $

又由条件(a$_{1})$以及Hölder不等式知

$ \begin{eqnarray*} \phi'_{1}(t)&\leq&\frac{4lM}{l+1}\int_{\Omega}u^{l+1}{\rm d}x+(l+1)\int_{\Omega}u^{l+p}{\rm d}x\int_{\Omega}u^{q}{\rm d}x \\ &\leq&{\frac{4lM}{l+1}}\int_{\Omega}u^{l+1}{\rm d}x+(l+1)|\Omega|^{\frac{l+p-q}{l+p}} \bigg[\int_{\Omega}u^{l+p}{\rm d}x\bigg]^{\frac{l+p+q}{l+p}} \\ &\leq&{\frac{4lM}{l+1}}\int_{\Omega}u^{l+1}{\rm d}x+(l+1)|\Omega|^{\frac{(1-p) (l+p+q)+(l+1)(l+p-q)}{(l+1)(l+p)}} \bigg(\int_{\Omega}u^{l+1}{\rm d}x\bigg)^{\frac{l+p+q}{l+p}}, \end{eqnarray*} $

其中$l+p-q>0, $

$ \begin{eqnarray} \phi'_{1}(t)\leq C_{1}\phi_{1}(t)+C_{2}(\phi_{1}(t))^{\frac{l+p+q}{l+p}}, \end{eqnarray} $ (4.26)

其中$C_{1}={\frac{4lM}{l+1}}$, $C_{2}=(l+1)|\Omega|^{\frac{(1-p)(l+p+q)+(l+1)(l+p-q)}{(l+1)(l+p)}}>0.$

对(4.26)式, 在$[0, t^{\ast}]$上积分得

$ t^{\ast}\geq\int_{\phi_{1}(0)}^{\infty}{\frac{{\rm d}\eta}{C_{1}\eta +C_{2}\eta^{\frac{l+p+q}{l+p}}}}, $

定理4.4证毕.

注 4.1  如果问题(1.1)-(1.3)中齐次Dirichlet边界条件换成齐次Neumann边界条件, 则本节中所有结论仍成立.

5 应用

本节中, 我们给出五个例子来验证定理3.1, 3.2, 4.1, 4.2, 4.3, 4.4.

例 5.1  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}={\rm{div}}\Big(\frac{(1+|x|^{2})}{100}\nabla u\Big)+u^{2}\int_{\Omega}u{\rm d}x, &\;\;\;(x, t)\in\Omega\times(0, t^{\ast}), \\ u(x, t)=0, &\;\;\;(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=1-|x|^{2}, &\;\;\; x\in\Omega, \end{array}\right. $

其中$\Omega=\{x=(x_{1}, x_{2}, x_{3})| |x|^{2}=x_{1}^{2}+x_{2}^{2}+x_{3}^{2}<1\}$${\Bbb R}^{3}$中的单位球.即我们有

$ a(x)=\frac{1}{100}(1+|x|^{2}), \;\;\; f(u)=u^{2}\int_{\Omega}u{\rm d}x, \;\;\; u_{0}(x)=1-|x|^{2}. $

易验证(a$_{1}$), (F$_{2})$成立.由$H(t)$在定理$3.1$中的定义, 我们有

$ H(0)=-\int_{\Omega}a(x)|\nabla u_{0}|^{2}{\rm d}x+2\int_{\Omega} \bigg(\int_{0}^{u_{0}}\xi^{2}{\rm d}\xi\bigg){\rm d}x\int_{\Omega}u_{0}{\rm d}x=0.5407>0. $

由定理$3.1$知, 若$\alpha>0, $ $u(x, t)$在有限时刻$t^{\ast}$爆破.取$\alpha=\frac{1}{10}, $我们得爆破时间的上界

$ t^{\ast}\leq\frac{\varphi_{1}(0)}{2\alpha(1+\alpha)H(0)}=8.0501, $

其中$\varphi_{1}(0)=\int_{\Omega}(1-|x|^{2})^{2}{\rm d}x=0.9576.$$\alpha=0, $$t^{\ast}=\infty, $这说明解无限爆破.

例 5.2  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}={\rm{div}}(e^{|x|^{2}}\nabla u)+u\int_{\Omega}u{\rm d}x, &\;\;\;(x, t)\in\Omega\times(0, t^{\ast}), \\ u(x, t)=0, &\;\;\;(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=\frac{\sqrt{2}}{2}-|x|, &\;\;\; x\in\Omega, \end{array}\right. $

其中$\Omega=\{x=(x_{1}, x_{2}, x_{3})| |x|^{2}=x_{1}^{2}+x_{2}^{2}+x_{3}^{2}<\frac{1}{2}\}$${\Bbb R}^{3}$中的球.即我们有

$ a(x)=e^{|x|^{2}}, \;\;\; f(u)=u\int_{\Omega}u{\rm d}x, \;\;\; u_{0}(x)=\frac{\sqrt{2}}{2}-|x|. $

易验证(a$_{1})$, (a$_{2}$), (F$_{2})$成立.则由定理$3.2$知, $u(x, t)$在有限时刻$t^{\ast}$爆破, 且当$p+q>1$时爆破时间的上界为

$ t^{\ast}\leq\frac{p+q-1}{C\varphi_{2}^{p+q-1}(0)}=2.8074, $

其中$C=\int_{\Omega}(e^{|x|^{2}})^{-1}{\rm d}x=1.1071, $ $\varphi_{2}(0)=\int_{\Omega}e^{|x|^{2}}(\frac{\sqrt{2}}{2}-|x|){\rm d}x=0.3217.$$p+q=1$时, $t^{\ast}=\infty, $即解$u(x, t)$无限爆破.

例 5.3  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}={\rm{div}}((1+|x|^{2}+|x|^{3})\nabla u)+u\int_{\Omega}u^{\frac{1}{6}}{\rm d}x, &\;\;\;(x, t)\in\Omega\times(0, t^{\ast}), \\ u(x, t)=0, &\;\;\;(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=\frac{1}{2}-|x|, &\;\;\; x\in\Omega, \end{array}\right. $

其中$\Omega=\{x=(x_{1}, x_{2}, x_{3})| |x|^{2}=x_{1}^{2}+x_{2}^{2}+x_{3}^{2}<\frac{1}{2}\}$${\Bbb R}^{3}$中的球.即我们有

$ a(x)=1+|x|^{2}+|x|^{3}, \;\;\; f(u)=u\int_{\Omega}u^{\frac{1}{6}}{\rm d}x, \;\;\; u_{0}(x)=\frac{1}{2}-|x|^{2}. $

并且, 在三维空间中Sobolev常数为$C_{s}=3^{-\frac{1}{2}}4^{\frac{1}{3}}\pi^{-\frac{2}{3}}.$

我们取$l=1, $则易验证(a$_{1})$, (a$_{3}), $ (F$_{1})$成立, 其中$c=1$, $M=1.$将以上参数代入定理$4.1$得到$q_{1}=\frac{2}{3}, $ $K_{1}=1.4136, $ $K_{2}=1.9574, $ $\epsilon=2.9820, $ $J_{1}=1.4136, $ $J_{2}=0.8255, $ $J_{3}=0.0014, $从而得到

$ t^{\ast}\geq\int_{\phi_{1}(0)}^{\infty}\frac{{\rm d}\eta}{J_{1}+J_{2}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+J_{3}\eta^{\frac{3(l+p+q)(N-2)}{4(N-2)(l+p)-N(l+p+q)}}}=1.5129, $

其中$\phi_{1}(0)=\int_{\Omega}(\frac{1}{2}-|x|^{2})^{2}{\rm d}x=0.0842.$

例 5.4  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}={\rm{div}}((1+\sin|x|)\nabla u)+u\int_{\Omega}u^{\frac{1}{3}}{\rm d}x, &\;\;\;(x, t)\in\Omega\times(0, t^{\ast}), \\ u(x, t)=0, &\;\;\;(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=\sin2\pi|x|, &\;\;\; x\in\Omega, \end{array}\right. $

其中$\Omega=\{x=(x_{1}, x_{2}, x_{3})| |x|^{2}=x_{1}^{2}+x_{2}^{2}+x_{3}^{2}<\frac{1}{4}\}$${\Bbb R}^{3}$中的球.即我们有

$ a(x)=1+\sin|x|, \;\;\; f(u)=u\int_{\Omega}u^{\frac{1}{3}}{\rm d}x, \;\;\; u_{0}(x)=\sin2\pi|x|. $

并且, 在三维空间中Sobolev常数为$C_{s}=3^{-\frac{1}{2}}4^{\frac{1}{3}}\pi^{-\frac{2}{3}}.$

我们取$l=1, $ $b(x)=1+|x|, $则易验证(a$_{1})$, (B$_{2}), $ (F$_{3})$成立, 并由条件(B$_{3})$$B=(1, 1, 1).$将以上参数代入定理$4.2$得到$c=1, $ $q_{1}=\frac{2}{3}, $ $K'_{1}=0.0259, $ $K'_{2}=0.4165, $ $\epsilon_{1}=3.7568, $ $\epsilon_{2}=0.9, $ $L_{1}=0.0259, $ $L_{2}=1.5971, $ $L_{3}=0.1734, $ $L_{4}=0.1088, $从而得到

$ t^{\ast}\geq\int_{\phi_{2}(0)}^{\infty}\frac{{\rm d}\eta}{L_{1}+L_{2} \eta+L_{3}\eta^{\frac{(2N-3)(l+p+q)}{2(N-2)(l+p)}}+L_{4}\eta^{\frac{3(l+p+q) (N-2)}{4(N-2)(l+p)-N(l+p+q)}}}=1.6723, $

其中$\phi_{2}(0)=\int_{\Omega}(1+|x|)\sin2\pi|x|{\rm d}x=0.0792.$

例 5.5  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}=((1+|\ln\frac{2x}{3}|)u_{x})_{x}+u^{\frac{1}{2}} \int_{0}^{\frac{3}{2}}u{\rm d}x, \;\;\;&(x, t)\in(0, \frac{3}{2})\times(0, t^{\ast}), \\ u(0, t)=u(\frac{3}{2}, t)=0, \;\;\;&(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=\sin\frac{2\pi x}{3}, \;\;\;&x\in(0, \frac{3}{2}), \end{array}\right. $

即我们有

$ a(x)=1+|\ln\frac{2x}{3}|, \;\;\; f(u)=u^{\frac{1}{2}}\int_{0}^{\frac{3}{2}} u{\rm d}x, \;\;\; u_{0}(x)=\sin\frac{2\pi x}{3}. $

易验证(a$_{1})$, (F$_{1})$成立.

我们取$l=2, $则由定理$4.3$中的定义得$C=3.9846.$由定理$4.3$知, $u(x, t)$在有限时刻$t^{\ast}$爆破, 且

$ t^{\ast}\geq{\frac{l+1}{C(p+q-1)(\phi_{1}(0))^{\frac{p+q-1}{l+1}}}}=1.6236, $

其中$\phi_{1}(0)=\int_{0}^{\frac{3}{2}}\sin^{3}\frac{2\pi x}{3}{\rm d}x=0.6366.$

例 5.6  令$u(x, t)$为如下问题的非负经典解

$ \left\{\begin{array}{ll} u_{t}={\rm{div}}((1+|x|)\nabla u)+u^{\frac{1}{4}}\int_{\Omega}u^{2}{\rm d}x, &\;\;\;(x, t)\in\Omega\times(0, t^{\ast}), \\ u(x, t)=0, &\;\;\;(x, t)\in\partial\Omega\times(0, t^{\ast}), \\ u(x, 0)=(\frac{1}{5})^{2}-|x|^{2}, &\;\;\; x\in\Omega, \end{array}\right. $

其中$\Omega=\{x=(x_{1}, x_{2})| |x|^{2}=x_{1}^{2}+x_{2}^{2}<\frac{1}{5}\}$${\Bbb R}^{2}$中的圆.即我们有

$ a(x)=1+|x|, \;\;\; f(u)=u^{\frac{1}{4}}\int_{\Omega}u^{2}{\rm d}x, \;\;\; u_{0}(x)=\frac{1}{5}-|x|^{2}. $

则易验证(a$_{1})$, (a$_{3}), $ (F$_{1})$成立且$M=1.$

$l=2$并将以上参数代入定理$4.4$得到$C_{1}=2.6667, $ $C_{2}=2.2877, $从而得到

$ t^{\ast}\geq\int_{\phi_{1}(0)}^{\infty}{\frac{{\rm d}\eta}{C_{1}\eta+C_{2}\eta^{\frac{l+p+q}{l+p}}}}=0.6788, $

其中$\phi_{1}(0)=\int_{\Omega}(\frac{1}{5}-|x|^{2})^{3}{\rm d}x=0.0013.$

参考文献
[1] Bebernes J, Bressan A. Thermal behavior for a confined reactive gas. J Differential Equations, 1982, 44(1): 118–133. DOI:10.1016/0022-0396(82)90028-6
[2] Bebernes J, Eberly D. Mathematical Problems from Combustion Theory. New York: Springer-Verlag, 1989.
[3] Pao C V. Nonlinear Parabolic and Elliptic Equations. New York: Plenum Press, 1992.
[4] Straughan B. Explosive Instabilities in Mechanics. Berlin: Springer, 1998.
[5] Quittner R, Souplet P. Superlinear Parabolic Problems:Blow-up, Global Existence and Steady States. Basel: Birkhauser, 2007.
[6] Hu B. Blow-up Theories for Semilinear Parabolic Equations. Berlin: Springer, 2011.
[7] Bandle C, Brunner H. Blow-up in diffusion equations:A survey. J Comput Appl Math, 1998, 97(1): 3–22.
[8] Levine H A. The role of critical exponents in blow-up theorems. SIAM Rev, 1990, 32(2): 262–288. DOI:10.1137/1032046
[9] Levine H A. Nonexistence of global weak solutions to some properly and improperly posed problems of mathematical physics:the method of unbounded Fourier coefficients. Math Ann, 1975, 214(3): 205–220. DOI:10.1007/BF01352106
[10] Payne L E, Schaefer P W. Lower bounds for blow-up time in parabolic problems under Neumann boundary conditions. Appl Anal, 2006, 85(10): 1301–1311.
[11] Payne L E, Schaefer P W. Lower bounds for blow-up time in parabolic problems under Dirichlet boundary conditions. J Math Anal Appl, 2007, 328(2): 1196–1205. DOI:10.1016/j.jmaa.2006.06.015
[12] Quittner P. On global existence and stationary solutions for two classes of semilinear parabolic problems. Comment Math Univ Carolin, 1993, 34(1): 105–124.
[13] Song J C. Lower bounds for the blow-up time in a non-local reaction-diffusion problem. Appl Math Lett, 2011, 24(5): 793–796. DOI:10.1016/j.aml.2010.12.042
[14] Tang G S, Li Y F, Yang X T. Lower bounds for the blow-up time of the nonlinear non-local reaction diffusion problems in $\mathbb{R}$N(N ≥ 3). Bound Value Probl, 2014. DOI:10.1186/s13661-014-0265-5
[15] Liu Y. Lower bounds for the blow-up time in a non-local reaction diffusion problem under nonlinear boundary conditions. Math Comput Modelling, 2013, 57(3/4): 926–931.
[16] Song X F, Lv X S. Bounds for the blowup time and blowup rate estimates for a type of parabolic equations with weighted source. Appl Math Comput, 2014, 236: 78–92.
[17] Ma L W, Fang Z B. Blow-up analysis for a nonlocal reaction-diffusion equation with robin boundary conditions. Taiwanese J Math, 2017, 21(1): 131–150. DOI:10.11650/tjm.21.2017.7380
[18] Wang N. A remark on bounds for the blowup time of the solutions to quasilinear parabolic problems. Math Appl (Wuhan), 2015, 28(2): 299–302.
[19] Liu D M, Mu C L, Xin Q. Lower bounds estimate for the blow-up time of a nonlinear nonlocal porous medium equation. Acta Math Sci, 2012, 32B(3): 1206–1212.
[20] Fang Z B, Yang R, Chai Y. Lower bounds estimate for the blow-up time of a slow diffusion equation with nonlocal source and inner absorption. Math Probl Eng, 2014, 2014(2): 1–6.
[21] Bao A G, Song X F. Bounds for the blow-up time of the solutions to quasi-linear parabolic problems. Z Angew Math Phys, 2014, 65(1): 115–123. DOI:10.1007/s00033-013-0325-1
[22] Li F S, Li J L. Global existence and blow-up phenomena for nonlinear divergence form parabolic equations with inhomogeneous Neumann boundary conditions. J Math Anal Appl, 2012, 385(2): 1005–1014. DOI:10.1016/j.jmaa.2011.07.018
[23] Baghaei K, Hesaaraki M. Lower bounds for the blow-up time in the higher-dimensional nonlinear divergence form parabolic equations. C R Math Acad Sci Paris, 2013, 351(s19/20): 731–735.
[24] Fang Z B, Wang Y X. Blow-up analysis for a semilinear parabolic equation with time-dependent coefficients under nonlinear boundary flux. Z Angew Math Phys, 2015, 66(5): 1–17.
[25] Ma L W, Fang Z B. Blow-up phenomena for a semilinear parabolic equation with weighted inner absorption under nonlinear boundary flux. Math Methods Appl Sci, 2016, 40(1): 1–14.
[26] Ma L W, Fang Z B. Blow-up analysis for a reaction-diffusion equation with weighted nonlocal inner absorptions under nonlinear boundary flux. Nonlinear Anal Real World Appl, 2016, 32: 338–354. DOI:10.1016/j.nonrwa.2016.05.005
[27] Wang N, Song X F, Lv X S. Estimates for the blow-up time of a combustion model with nonlocal heat sources. J Math Anal Appl, 2015, 436(2): 1180–1195.