The non-Newtonian polytropic filtration equation with a convection term
$$ v_{t}= \operatorname{div} \bigl(a(x) \vert v \vert ^{\alpha }{ \vert {\nabla v} \vert ^{p-2}}\nabla v \bigr)+ \sum_{i=1}^{N}\frac{\partial a_{i}(v,x,t)}{\partial x_{i}} $$
is considered, where \(p>1\), \(\alpha >0\), \(a(x)\geq 0\) with \(a(x) | _{x\in \partial \varOmega }=0\). This kind of equation is degenerate on the boundary, the usual boundary value condition may be overdetermined. Some conditions depending on \(a(x)\) and \(a_{i}(\cdot ,x,t)\), which can take place of the boundary value condition, are found. Moreover, how the nonlinear term \(|v|^{\alpha }\) affects the stability of weak solutions is revealed.
Hinweise
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
Consider non-Newtonian polytropic filtration equation with a convection term
where \(p>1\), \(\alpha \geq 0\), \(a(x)\in C^{1}(\overline{\varOmega })\), \(a_{i}(s,x,t)\in C^{1}(\mathbb{R}\times \overline{\varOmega }\times [0,T])\) for every \(i=1,2,\ldots , N\), \(\varOmega \subset \mathbb{R}^{N}\) is a bounded domain with a smooth boundary ∂Ω. The equations like (1.1) arise from a variety of diffusion phenomena, such as soil physics, fluid dynamics, combustion theory and reaction chemistry, one can see [1‐4] and the references therein. When \(\alpha >0\), \(a(x)\equiv 1\), the well-posedness problem of Eq. (1.1) has been studied thoroughly. Many scholars had been interested in the polytropic infiltration equations of the form
$$ u_{t}=\operatorname{div} \bigl( \vert u \vert ^{r} \vert \nabla u \vert ^{p-2}\nabla u \bigr)+f(x,t,u, \nabla u), $$
To be more specific: If \(f(x,t,u, \nabla u)=0\), the Cauchy problem of this kind of equation can be traced back to Gilding–Peletier [5]. Later, if the initial value \(u_{0}(x)\in L^{1}(\mathbb{R}^{N})\), the existence and the uniqueness of weak solution were studied and \(u_{t}\in L^{1}(\mathbb{R}^{N}\times (\tau , T))\) was proved for any \(\tau >0\) [6]. If the initial value \(u_{0}(x)\) is just a measure, a similar problem was considered in [7, 8]. A more general equation was studied in [9] based on a \(L^{1}\) initial value condition. If \(f(x,t,u, \nabla u)\) is a source term, the Cauchy problem was studied in [10] etc. For the initial-boundary value problem, when \(f(x,t,u, \nabla u)=0\) and \(u_{0}(x)\in L^{\infty }(\varOmega )\), the well-posedness problem was studied in [11, 12] etc. When \(f(x,t,u, \nabla u)=\nabla A(u)\), by modifying the usual Moser iteration and imposing some restrictions on the convection function \(A(s)\), the local \(L^{\infty }\)-estimates were made and \(u_{t}\in L^{2}(\mathbb{R}^{N} \times (\tau , T))\) was proved, provided that \(u_{0}(x)\in L^{q}( \varOmega )\) with \(q\geq 1\) [13]. The large time behavior of solutions had been studied in [14‐16] etc. The extinction, positivity and the blow-up of solutions had been studied in [17, 18] etc. The finite speed propagation was studied in [19] provided that \(f(x,t,u, \nabla u)=- \vec{\beta }(x)\cdot \nabla u^{q}\) is with orientated convection. Of course, there are a great deal of papers to study various subjects on these kinds of equations, for examples, one can refer to [20‐26] and the references therein.
If we regard Eq. (1.1) as the generalization of the heat conduction equation, then \(a(x)|v|^{\alpha }\) has the meaning of nonlinear thermal conductivity depending on the temperature \(v = v(x, t)\) [20]. If we consider Eq. (1.1) as the generalization of the reaction-diffusion equation, \(a(x)\) is the diffusion coefficient [27]. When \(a(x)=d(x)^{ \beta }\), \(\alpha =0\) and \(a_{i}\equiv 0\) in Eq. (1.1), Yin–Wang’s work [28] reveals that the uniqueness of the weak solution is independent of the boundary value condition when \(\beta >p-1\). Moreover, they generalized [27]’s work to the following equation:
$$ \frac{\partial u}{\partial t}-\operatorname{div} \bigl(a(x) \vert \nabla u \vert ^{p-2} \nabla u \bigr)-\sum_{i=1}^{N}b_{i}(x)D_{i}u+c(x,t)u=f(x,t), $$
where \(a(x)\geq 0\), and divided the boundary ∂Ω into three parts: the non-degenerate part, the weakly degenerate part and the strongly degenerate part. On the first two parts, the boundary value condition can be imposed in the sense of the trace. While, in the strongly degenerate part, the boundary value condition only can be explained in a much weaker sense than the trace. Recently, using some ideas of [27], the existence and the uniqueness of weak solution to some special cases of Eq. (1.1) had been studied by the author in [29‐31], where \(a(x)\) satisfies
is required when the stability of weak solutions is discussed, where \(\varSigma _{p}\subseteq \partial \varOmega \) is a relatively open subset of ∂Ω. The problem lies in that, by choosing different test functions, we only can find the different \(\varSigma _{p}\subset \partial \varOmega \) in (1.5). From this we observe that how to find an exact partial boundary \(\varSigma _{p}\) in (1.5) to match up with the nonlinear equations with the type (1.1) seems almost impossible. In this paper, we may ponder about and tackle the problem from another perspective. We are ready to find some conditions, which are related to \(a(x)\) and \(a_{i}(\cdot ,x,t)\), to take place of the boundary value condition (1.3) (or (1.5)). Technically, since \(\alpha >0\), how the nonlinear term \(|v|^{\alpha }\) affects the stability of weak solutions is the main difficulty we should overcome. Let us give the definition of weak solution.
Anzeige
Definition 1.1
A function \(u(x,t)\) is said to be a weak solution of Eq. (1.1) with the initial value (1.2), if
If v satisfies the boundary value condition (1.3) (or the partial boundary value condition (1.5)) in the sense of the trace, then we say v is a weak solution of the initial-boundary value problem of Eq. (1.1).
Here \(p'=\frac{p}{p-1}\). By the parabolically regularized method, we can prove the following existence theorem.
where\(\delta >0\)satisfied\(\frac{1}{\delta }\leq p-1\), then the boundary value condition (1.3) (or the partial boundary value condition (1.5)) is true in the sense of the trace.
We do not pay so much attention on the existence of the weak solution. Theorem 1.2 is given for the completeness of the paper. Actually, when \(a_{i}\equiv 0\), \(i=1,2,\ldots , N\), only if \(p>1\), the existence of weak solution with \(\frac{\partial v}{\partial t}\in L^{2}(Q_{T})\) had been proved in [30]. At the same time, the existence of weak solution to Eq. (1.1) can be proved in the other sense [31]. Moreover, the condition \(v_{0}\in W^{1,p}(\varOmega )\) can be weaken to \(a(x) \vert {\nabla {v_{0}}} \vert ^{p} \in {L^{1}}(\varOmega )\). In addition, if \(1< p\leq 2\) and \(a_{i}(v,x,t)\equiv 0\), how the degeneracy of \(a(x)\) on the boundary affects the stability of weak solutions had first been studied in [30]. If \(a_{i}(v,x,t)=a_{i}(v)\), the stability of a class of solutions satisfying
had been originally studied in [31], where \(m>1\) and \(\varOmega _{\lambda }=\{x\in \varOmega : a(x)>\lambda \}\). Different from the method used in [30, 31], in order to prove the stability of weak solutions, we transfer Eq. (1.1) to another type of equation. Let \(v=|u|^{\beta -1}u, \beta =\frac{(p-1)}{(\alpha +p-1)}\), \(\delta =\beta ^{p-1}\). Then the main equation (1.1) becomes
$$ \bigl( \vert u \vert ^{\beta -1}u \bigr)_{t}= \delta \operatorname{div} \bigl(a(x){ \vert \nabla u \vert ^{p-2}}\nabla u \bigr)+\sum_{i=1}^{N} \frac{\partial b_{i}(u,x,t)}{ \partial x_{i}},\quad (x,t) \in \varOmega \times (0,T), $$
(1.9)
and with the usual initial-boundary value conditions
Roughly speaking, if \(u(x,t)\) is a weak solution of Eq. (1.9) with the initial value (1.10), if condition (1.12) is true, then \(\int _{\varOmega }|\nabla u|\,dx<\infty \), and the boundary value condition (1.11) can be imposed in the sense of the trace. Accordingly, similar as the usual evolutionary p-Laplacian equation, we can prove Theorem 1.3.
The main aim of this paper is to find the other conditions to replace the boundary value condition (1.11) (or equivalently, (1.3)). At first, even condition (1.12) is true and one can impose the boundary value condition (1.11), we still hope the stability of weak solutions can be obtained without the boundary value condition (1.11).
Theorem 1.4
Let\(a(x)\)satisfy (1.4), \(u(x,t)\)and\(v(x,t)\)be two solutions of Eq. (1.9) with the initial values\(u_{0}(x)\), \(v_{0}(x)\), respectively, but without any boundary value condition. If
$$ \frac{1}{\lambda } \biggl(\lambda \int _{\varOmega \setminus \varOmega _{\lambda }} \vert \nabla a \vert ^{p}\,dx \biggr)^{ \frac{1}{p}}\leq c $$
then the stability of the weak solutions is true in the sense of (1.13).
If there is not condition (1.12), then one cannot impose the boundary value condition (1.11) generally. In this case, there is only way to go. We have to try to prove the stability of weak solutions without any boundary value condition. The following theorem partially achieves our goals.
Theorem 1.5
Let\(u(x,t)\), \(v(x,t)\)be two weak nonnegative solutions of Eq. (1.1) with the initial values\(u_{0}(x)\), \(v_{0}(x)\), respectively. If\(a(x)\)satisfies (1.4), and
One can see that the condition \(\int _{\varOmega }a(x)^{-\frac{1}{p-1}}\,dx< \infty \) and the condition \(\int _{\varOmega }a(x)^{-(p-1)}\,dx<\infty \) cannot be true at the same time unless \(p=2\). This fact shows that Theorem 1.4 and Theorem 1.5 complement each other.
At last, since the nonlinear convection term \(\sum_{i=1}^{N}\frac{ \partial a_{i}(v,x,t)}{\partial x_{i}}\) depends on the spatial variable x, we will use some ideas in [29‐31] to prove the stability of weak solutions based on the partial boundary value condition.
Theorem 1.6
Let\(u(x,t)\)and\(v(x,t)\)be two solutions of Eq. (1.9) with the initial values\(u_{0}(x)\), \(v_{0}(x)\), respectively. If\(a(x)\)satisfies (1.4), and there is a function\(h_{i}(x)\)such that
is imposed, then the stability of the weak solutions is true in the sense of (1.13).
Here, we give a simple comment on Theorem 1.6. One can find that condition (1.12) does not appear in this theorem. But, in general, under condition (1.12), \(\int _{\varOmega }|\nabla u|\,dx<\infty \) is true, and the partial boundary value condition (1.23) can be imposed in the sense of the trace. Accordingly, we can say that condition (1.12) is hidden in condition (1.23). However, as we have said before, we cannot judge whether the partial boundary value condition (1.23) is the optimal one for Eq. (1.9) or not.
2 Existence of the weak solutions
The existence of weak solutions can be obtained by the parabolically regularized method. Consider the approximate problem
Here, \(0\leq u_{\varepsilon 0} \in C^{\infty }_{0}(\varOmega )\), \(\|u_{\varepsilon 0}\|_{L^{\infty }(\varOmega )}\leq \|u_{0}\|_{L^{ \infty }(\varOmega )}\), \(u_{\varepsilon 0} \rightarrow u_{0}(x)\) in \(W^{1,{p}}(\varOmega )\). It is well-known that the above problem has a unique nonnegative solution \(u_{\varepsilon }\in L^{\infty }(0,T; W ^{1,p}_{0}(\varOmega ))\) [20], provided that \(\frac{\partial a_{i}}{ \partial x_{i}}\leq 0\).
For a fixed s such that \(s>\frac{N}{2}+1\), one has \(H_{0}^{s}( \varOmega )\hookrightarrow W^{1, p}(\varOmega )\). Consequently, \(W^{-1, {p'}}(\varOmega )\hookrightarrow H^{-s}(\varOmega )\). As a result, one has
Noticing that \(W^{1,p}_{0}(\varOmega )\hookrightarrow L^{p}(\varOmega ) \hookrightarrow H^{-s}(\varOmega )\), one can employ Aubin’s compactness theorem in [32] to obtain \(\varphi u_{1\varepsilon }\rightarrow \varphi u_{1}\) strongly in \(L^{p'}(0,T;L^{p}(\varOmega ))\). Thus \(\varphi u_{1\varepsilon }\rightarrow \varphi u_{1}\) a.e. in \(Q_{T}\). In particular, due to the arbitraries of φ, \(u_{1\varepsilon }\rightarrow u_{1}\) a.e. in \(Q_{T}\). Accordingly, \(u_{\varepsilon }\rightarrow u\) a.e. in \(Q_{T}\).
Moreover, by (2.6), there is a n-dimensional vector function \(\overrightarrow{\zeta }= ({\zeta _{1}}, \ldots ,{\zeta _{n}})\), \(\vert \zeta _{i} \vert \in L^{\frac{p}{p - 1}}(Q_{T})\), such that
for any given function \(\varphi \in C_{0}^{1} ({Q_{T}})\). One can refer to [33] for the details. Also by [33], the initial value (1.2) can be proved in the sense of (1.7). Then u is a nonnegative weak solution of Eq. (1.1) with the initial value (1.2). □
Letube a solution of Eq. (1.1) with the initial value (1.2). For the constantss, βsatisfying\(s> \alpha \delta +1\), \(\frac{1}{\delta }\leq p-1\), such that\(\int _{\varOmega } a^{-\delta }(x)\,dx<\infty \), we have
$$ \int _{Q_{T}} \bigl\vert \nabla u^{s} \bigr\vert \,dx\,dt\leq c. $$
By Lemma 2.1, the conclusion of Theorem 1.2 follows easily.
If \(u(x,t)\), \(v(x,t)\) are two nonnegative solutions of Eq. (1.9) with the same homogeneous boundary value and with the different initial values \(u_{0}(x)\), \(v_{0}(x)\), respectively.
From the definition of the weak solution, for \(\varphi \in L^{\infty }(0,T; W_{0}^{1,p}(\varOmega ))\), one has
Since \(u(x,t)\) and \(v(x,t)\) have the same homogeneous boundary value condition, by taking the limit, one can choose \(\varphi ={S_{\eta }}(u - v)\) as the test function, then
$$ \lim_{\lambda \rightarrow 0}\lim_{\eta \rightarrow 0} \biggl\vert \int _{\varOmega } a(x) \bigl( \vert \nabla u \vert ^{p-2}\nabla u - \vert \nabla v \vert ^{p-2}\nabla v \bigr) \cdot \nabla \phi _{\lambda }(x) S_{\eta }(u-v)\,dx \biggr\vert =0. $$
(4.7)
Secondly, for any given \(\lambda >0\), since \(a(x)>0\), \(x\in \varOmega \) and \(|\nabla u|\in L_{\mathrm{loc}}^{p}(\varOmega )\), for the first part of the fourth term on the left hand side of (4.3), one has
Once more, for the second part of the fourth term on the left hand side of (4.3), one can deal with it in two cases according to the conditions (1.15) and (1.16), respectively. In detail, if \(|\nabla a|=0\), \(x\in \partial \varOmega \), then
If (1.16) is true, when \(x\in \partial \varOmega \), \(b_{i}(\cdot , x,t)=0\) is reasonable (we would like to suggest that we do not require \(u|_{x\in \partial \varOmega }=0\) here), then
Since \(\int _{\varOmega }a(x)^{- \frac{1}{p-1}}\,dx<\infty \), one can define the trace of weak solution u on the boundary. Accordingly, the partial boundary value condition (1.23) is reasonable. Now,
The rest of the proof Theorem 1.6 can be completed just in the same way as that of Theorem 1.4. □
From the proofs of Theorem 1.4 and Theorem 1.6, we easily have the following corollary, no matter whether \(\int _{\varOmega }a^{- \frac{1}{p-1}}(x)\,dx<\infty \) or not.
Corollary 4.1
Let\(u(x,t)\)and\(v(x,t)\)be two solutions of Eq. (1.9) with the initial values\(u_{0}(x)\)and\(v_{0}(x)\), respectively, but without any boundary value condition. If\(a(x)\)satisfies (1.14), \(b_{i}(\cdot , x,t)\)satisfies (1.22), and
$$ \vert \nabla a \vert h_{i}(x)=0,\quad i=1,2,\ldots , N, x\in \partial \varOmega , $$
then the stability of weak solutions is true in the sense of (1.13).
Let \(u(x,t)\), \(v(x,t)\) be two solutions of Eq. (1.1) with the initial values \(u_{0}(x)\), \(v_{0}(x)\). By taking the limit, one can choose \(S_{\eta }(a(u^{m}-v^{m}))\) as the test function;
The authors would like to thank everyone for help.
Availability of data and materials
No applicable.
Competing interests
The authors declare that they have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.