- Research
- Open access
- Published:
The global existence of a parabolic cross-diffusion system with discontinuous coefficients
Boundary Value Problems volume 2016, Article number: 128 (2016)
Abstract
This paper deals with a one-dimensional parabolic cross-diffusion system describing two-species models. The system under consideration is strongly coupled and the coefficients of the equations are allowed to be discontinuous. The existence and uniqueness of nonnegative global solutions are given when one of the cross-diffusion pressures is zero. Our approach to the problem is by an approximation method and various estimates.
1 Introduction
In the previous work [1], we presented a parabolic cross-diffusion system describing two-species models on a bounded domain with different natural conditions, and studied the corresponding steady-state problem, an elliptic cross-diffusion system. In this paper, we deal with this parabolic cross-diffusion system for the one-dimensional case under the homogeneous Neumann boundary conditions.
Assume that \((0,l)\) is partitioned into two intervals \(I^{(1)}:=(0,l_{0})\) and \(I^{(2)}:=(l_{0},l)\) separated by \(x=l_{0}\), and the natural conditions of \(I^{(i)}\) are different. For any \(T>0\), set
Let \(u=u(x,t)\), \(v=v(x,t)\) be the population densities of the two species. Assume that the species share the same habitat, the interval \([0,l]\), and that for each species, the density and the flux of the density are all continuous across the inner boundary \(\Gamma_{T}\). According to [1], when the cross-diffusion pressure for the second species is zero, u, v are governed by the parabolic cross-diffusion system
where \(u_{t}:=\partial u/\partial t\), \(u_{x}:=\partial u/\partial x\), \(u^{-}(l_{0},t)\) and \(u^{+}(l_{0},t)\) represent the limits of u from left and right for the space variable, respectively,
\(k_{j,1}\) and \(k_{j,2}\) (\(j=1,2\)) are positive constants, and β, γ and δ are nonnegative constants. In (1.1), \(k_{1}(x)\) and \(k_{2}(x)\) are the diffusion rates, \(\beta k_{1}(x)\) and \(\delta k_{2}(x)\) are the self-diffusion rates, and \(\gamma k_{1}(x)\) is cross-diffusion rate (see [1]).
In 1979 Shigesada et al. [2] proposed a mathematical model with cross-diffusion to investigate the spatial segregation phenomena of two competing species under inter- and intra-species population pressures. It is a strongly coupled quasilinear parabolic system. Over the past 36 years, quasilinear elliptic and parabolic systems with cross-diffusion have been treated extensively in the literature both in theory and in applications, and most of the treatments are for systems with continuous coefficients (see [3–16] and the references therein). In [1], by Schauder’s fixed point theorem, we discussed the existence of nonnegative solutions for the elliptic system with cross-diffusion and discontinuous coefficients. In this paper, problem (1.1) is a strongly coupled parabolic system with cross-diffusion. In addition, the diffusion rates, self-diffusion rates, cross-diffusion rates and reaction functions are all allowed to be discontinuous. Based on the result of Lou et al. [6] for a parabolic system with continuous coefficients, we shall use approximation method and various estimates to show the existence and uniqueness of global solutions for problem (1.1).
The paper is organized as follows. In the next section we introduce the notations, hypotheses, and main result. In Section 3 we construct an approximation problem of (1.1) and establish the uniform estimates of solutions for the approximation problem. Section 4 is devoted to the existence and uniqueness of solutions for problem (1.1).
2 The notations, hypotheses, and main result
For any set S, S̄ denotes the closure of S. The symbol \(I'\subset\subset I \) means that I and \(I'\) are open intervals and \(\bar{I}'\) is a subset of I. Denote
where \(s,r\geq1\). \(W_{2}^{1,0}(Q_{T})\) and \(W_{2}^{1,1}(Q_{T})\) are the Hilbert spaces with scalar products
respectively. \(V_{2}(Q_{T})\) is the Banach space consisting of all elements of \(W^{1,0}_{2}(Q_{T})\) having a finite norm
We let
equipped with the norm
Definition 2.1
A nonnegative vector function \((u,v)\) is called a solution of (1.1) if \((u,v)\) possesses the following properties:
-
(i)
For any \(T>0\), \(u,v\in W^{1,1}(Q_{T})\cap C^{2,1}(Q^{(i)}_{T})\), \(i=1,2\). For any \(I'\subset\subset (0,l)\), there exists \(\alpha'\in(0,1)\), such that \(u,v\in C^{\alpha'}(\bar{I}'\times[0,T])\). For any fixed \(i\in\{ 1,2\}\) and for any \(I''\subset\subset I^{(i)}\), \(\tau\in(0,T)\), there exists \(\alpha''\in(0,1)\), such that \(u,v\in C^{2+\alpha'',1+\alpha''/2}(\bar{I}''\times[\tau,T])\).
-
(ii)
For any \(I'\subset\subset (0,l)\) and \(\tau\in(0,T)\), there exists \(\bar{\alpha}'\in(0,1)\), such that \(u_{t},v_{t}\in C^{\bar{\alpha}'}(\bar{I}'\times[\tau,T])\) and \(u_{x},v_{x}\in C^{\bar{\alpha}'}((\bar{I}'\cap\bar{I}^{(i)})\times[\tau,T])\), \(i=1,2\).
-
(iii)
\((u,v)\) satisfies pointwise the equations in (1.1) on \(Q^{(i)}_{T}\) (\(i=1,2\)), the initial conditions on \(\{(x,t):x\in (0,l), t=0\}\) and the inner boundary conditions on \(\Gamma_{T}\), and satisfies the homogeneous Neumann boundary conditions on \(S_{T}\) for almost all t.
We see from property (ii) in Definition 2.1 that if \((u,v)\) is a solution of (1.1), then \(u_{x}\), \(v_{x}\) are Hölder continuous up to the inner boundary \(\Gamma_{T}\), and \(u_{t}\), \(v_{t}\) are Hölder continuous across \(\Gamma_{T}\).
Throughout the paper we make the following hypotheses on the various functions in (1.1):
-
(H)
For each \(j,i=1,2\), \(f_{j,i}(u,v)\in C^{1}(\mathbb{R}_{+}^{2} )\), where \(\mathbb{R}_{+}:=[0,+\infty)\). There exist positive constants \(\underline{d}_{1}\), \(\bar{d}_{1}\), \(\underline{e}_{1}\), \(\underline{d}_{2}\), \(\underline{e}_{2}\), \(\bar{e}_{2}\), such that, for any \(u,v\geq0\),
$$ \left \{ \textstyle\begin{array}{l} -\underline{d}_{1}\leq\frac{\partial f_{1,i}(u,v)}{\partial u}\leq -\bar{d}_{1}, \qquad -\underline{e}_{1}\leq\frac{\partial f_{1,i}}{\partial v}\leq0, \\ -\underline{d}_{2}\leq\frac{\partial f_{2,i}(u,v)}{\partial u}\leq 0,\qquad -\underline{e}_{2}\leq\frac{\partial f_{2,i}}{\partial v}\leq -\bar{e}_{2}. \end{array}\displaystyle \right . $$(2.1)Assume that \(u_{0}=u_{0}(x)\), \(v_{0}=v_{0}(x)\) possess the properties
$$ \left \{ \textstyle\begin{array}{l} u_{0}(x),v_{0}(x)\in W^{1}_{r_{0}}(0,l)\cap C^{1}((0,l_{0}])\cap C^{1}([l_{0},l)) \quad \text{for some }r_{0}>2, \\ u_{0}(x),v_{0}(x)\geq0\quad \text{for } x\in[0,l], \end{array}\displaystyle \right . $$(2.2)and satisfy the compatibility conditions
$$ \left \{ \textstyle\begin{array}{l} {[k_{1}(x) ((1+\beta u_{0}+\gamma v_{0})u_{0} )_{x} ]}^{-}(l_{0})= [k_{1}(x) ((1+\beta u_{0}+\gamma v_{0})u_{0} )_{x} ]^{+}(l_{0}), \\ {[k_{2}(x) ((1+\delta v_{0})v_{0} )_{x} ]}^{-}(l_{0})= [k_{2}(x) ((1+\delta v_{0})v_{0} )_{x} ]^{+}(l_{0}). \end{array}\displaystyle \right . $$(2.3)
From (2.2) it follows that \(u_{0}(x),v_{0}(x)\in C^{1/2}([0,l])\).
The main result of this paper is the following theorem.
Theorem 2.1
Let hypothesis (H) hold. Then problem (1.1) has a unique solution \((u,v)\).
Since T is an arbitrary positive number, the solution \((u,v)\) given by Theorem 2.1 is global.
3 An approximation problem of (1.1)
In order to show the existence and uniqueness of solutions for problem (1.1), in this section we shall construct an approximation problem and establish the uniform estimates of its solutions.
3.1 The global solution for the approximation problem
To construct an approximation problem of (1.1), we first construct some approximation functions. For an arbitrary \(\varepsilon>0\), let \(\zeta_{\varepsilon}=\zeta_{\varepsilon}(x)\) a smooth function with values between 0 and 1, such that \(\zeta_{\varepsilon}=1\) for \(x\leq l_{0}\), \(\zeta _{\varepsilon}=0\) for \(x\geq l_{0}+\varepsilon\) and \(|\zeta_{\varepsilon x}|\leq C/\varepsilon\) for all \(x\in \mathbb{R}\). For each \(j=1,2\), define
Then (2.1) in hypothesis (H) implies that, for all \((x,u,v)\in[0,l]\times\mathbb{R}_{+}^{2}\),
and
By employing \(k_{j\varepsilon}(x)\), \(f_{j\varepsilon}(x,u,v)\), we consider the following approximation problem of (1.1):
Proposition 3.1
Let hypothesis (H) be satisfied. Then problem (3.5) has a unique nonnegative global solution \((u_{\varepsilon}, v_{\varepsilon})=(u_{\varepsilon}(x,t), u_{\varepsilon}(x,t))\) satisfying \(u_{\varepsilon}(\cdot, t), v_{\varepsilon}(\cdot, t)\in C([0,\infty), W^{1}_{r_{0}}(0,l))\cap C^{\infty}((0,\infty),C^{\infty}(0,l))\).
Proof
By (2.2) and (3.3), and by a minor modification, we can use the arguments in [6] to prove this proposition. The detailed proofs are omitted here. □
3.2 Uniform estimates of \(\max_{Q_{T}}u_{\varepsilon}\), \(\max_{Q_{T}}v_{\varepsilon}\)
In order to investigate the limit of vector sequence \(\{(u_{\varepsilon},v_{\varepsilon})\}\) governed by (3.5), in the rest of this sections we derive some uniform estimates of \(u_{\varepsilon}\), \(v_{\varepsilon}\). We shall use the following relations several times:
For notational simplicity, we shall use \(\alpha'\), \(\alpha''\), ᾱ, \(C_{0}\), \(C_{1}\) and \(C(\cdots)\) to denote constants depending only on T, β, γ, δ, \(k_{j,m}\), \(f_{j,m}(0,0)\), \(\underline{d}_{j}\), \(\underline{e}_{j}\) (\(j,m=1,2\)), \(\bar{d}_{1}\), \(\bar{e}_{2}\), \(\|u_{0}\|_{W^{1}_{r_{0}}(0,l)}\), \(\|v_{0}\|_{W^{1}_{r_{0}}(0,l)}\) and the quantities appearing in parentheses, independent of ε. The same letter C will be used to denote different constants depending on the same set of arguments.
Lemma 3.2
We have
Proof
By using the maximum principle (see [17]) and (3.2)-(3.4), the proofs similar to those of Lemmas 2.1, 2.2, 2.4 in [6] imply (3.6)-(3.8). □
Lemma 3.3
The following estimates hold:
Proof
Let \(w_{\varepsilon}=(1+\delta v_{\varepsilon})v_{\varepsilon}\). Then
and
Recall that \(v_{\varepsilon}(\cdot, t)\in C([0,\infty ),W^{1}_{r_{0}}(0,l))\cap C^{\infty}((0,\infty),C^{\infty}(0,l))\). Multiplying (3.12) by \((k_{2\varepsilon}(x)w_{\varepsilon x} )_{x}\), integrating it over \(Q_{t}\) and then using (3.4), (3.6), (3.7), and Cauchy’s inequality, we see that, for any \(\vartheta>0\),
We choose \(\vartheta=1/2\) and employ (3.12) to find
Furthermore, from (3.13), (3.11), and (3.8) it follows that (3.9) holds, and
and from Chapter II, equation (3.8) in [18] that
Combining this inequality and (3.11) leads us to estimate (3.10). □
Based on Lemma 3.3, let us estimate \(\max_{Q_{T}}u_{\varepsilon}\).
Lemma 3.4
We have
Proof
Choose \(\hat{\sigma}>\max\{\max_{x\in [0,l]}u_{0}(x),1\}\), and let
Multiplying the first equation of (3.5) by \(u^{(\sigma)}_{\varepsilon}\) and then integrating by parts over \(Q_{t_{1}}\), we see from (3.2), (3.4), and Cauchy’s inequality that, for any \(\vartheta>0\),
Setting \(\vartheta=\mu_{0}/2\), we have
for \(\bar{s}=\bar{r}=2\), where \(\mathcal{D}(x,t):=v^{2}_{\varepsilon x}+1\), \(Q_{t_{1}}(\sigma):=\{(x,t):x\in A_{\sigma}(t),t\in(0,t_{1}] \}\). Thus \(1/\bar{r}+1/(2\bar{s})=1-\chi_{1}\) for \(\chi_{1}=1/4\). In view of (3.10), we find that (3.15) has the same property as [18], Chapter III, equation (7.8). By using Chapter II, Theorem 6.1 and Remark 6.2 in [18], the proof similar to that of Chapter III, equation (7.15) in [18] leads to (3.14). □
3.3 Uniform Hölder estimates
In this subsection, we establish uniform Hölder estimates of \(u_{\varepsilon}\), \(v_{\varepsilon}\) and their derivatives.
Lemma 3.5
For any open interval \(I'\subset\subset (0,l)\), there exists \(\alpha'=\alpha'(d')\in(0,1)\) such that
Here and below, \(d':=\min\{\operatorname{dist}(0,I'),\operatorname{dist}(l,I')\}\).
Proof
It is obvious that \(u_{\varepsilon}\) is a bounded generalized solution of equation \(u_{t}-(\bar{a}_{\varepsilon}(x,t,u_{x}))_{x}+a_{\varepsilon}(x,t)=0\) and \(v_{\varepsilon}\) is a bounded generalized solution of equation \(v_{t}-(\bar{b}_{\varepsilon}(x,t,v_{x}))_{x}+b_{\varepsilon}(x,t)=0\) in the sense of Chapter V, Section 1 in [18], where
By (3.2), (3.6), and (3.14), we get
where \(\varphi_{0}(x,t):= v^{2}_{\varepsilon x}(x,t)\) and \(\varphi _{1}(x,t):= |v_{\varepsilon x}(x,t)|\). Choosing \(s=r=2\), we find from (3.10) that
In view of \(u_{0}(x),v_{0}(x)\in C^{1/2}([0,l])\), by employing Chapter V, Section 1, Theorem 1.1 in [18], we obtain (3.16). □
We next give the Hölder estimates of derivatives.
Lemma 3.6
For any fixed \(i\in\{1,2\}\) and for any \(I''\subset\subset I^{(i)}\), \(\tau\in(0,T)\), there exists \(\alpha''=\alpha''(d'',\tau )\in (0,1)\) such that
Hereafter, \(d'':=\operatorname{dist}(\partial I^{(i)} ,I'')\).
Proof
Choose open intervals \(I''_{j}\) and positive numbers \(\tau_{j}\), \(j=1,\ldots,4\), such that \(I''\subset\subset I''_{4}\subset\subset \cdots\subset\subset I''_{1}\subset\subset I^{(i)}\) and \(\tau_{1}<\cdots <\tau_{4}<\tau\). By (3.1), for small enough ε, we have
We first derive uniform Hölder estimates of derivatives of \(v_{\varepsilon}\). Set
Then \(v_{\varepsilon}\) is a solution of equation
Note that
Thus when \((x,t,v,q)\in I''_{1}\times(\tau_{1},T]\times[0,C_{0}]\times \mathbb{R}\),
According to Chapter V, Theorem 3.1 in [18], it follows that, for some \(\alpha_{1}''=\alpha_{1}''(d'',\tau)\in(0,1)\),
In addition, \(v_{\varepsilon}\) is a solution of the linear equation
where \(\bar{h}_{\varepsilon}(x,t)=k_{2,i}(1+2\delta v_{\varepsilon}(x,t))\), \(\hat{h}_{\varepsilon}(x,t)=-2k_{2,i}\delta v_{\varepsilon x}\) and \(h_{\varepsilon}(x,t)=v_{\varepsilon}f_{2,i}(u_{\varepsilon},v_{\varepsilon})\). Estimates (3.16), (3.18) imply that, for some \(\alpha''_{2}=\alpha''_{2}(d'',\tau)\in(0,1)\),
The Schauder estimate for linear parabolic equation further yields
We next use (3.19) to show the uniform Hölder estimates of derivatives of \(u_{\varepsilon}\). Let
Thus \(u_{\varepsilon}=u_{\varepsilon}(x,t)\) is a solution of the following equation:
Since
then we see from (3.6), (3.14), and (3.19) that when \((x,t,u,p)\in I''_{3}\times(\tau_{3},T]\times[0,C_{1}]\times \mathbb{R}\),
Again by Chapter V, Theorem 3.1 in [18] we see that, for some \(\alpha''_{3}=\alpha''_{3}(d'',\tau)\in(0,1)\),
Furthermore, \(u_{\varepsilon}\) is a solution of the linear equation
where
Using (3.16), (3.19), and (3.20) leads to
for some \(\alpha''_{4}=\alpha''_{4}(d'',\tau)\in(0,1)\). Again by Schauder estimate we have
which along with (3.19) yields (3.17). □
3.4 Estimates of \(\|u_{\varepsilon_{1}}-u_{\varepsilon_{2}}, v_{\varepsilon_{1}}-v_{\varepsilon_{2}}\|_{V_{2}(Q_{T})}\)
We first give the following uniform estimates of derivatives of \(u_{\varepsilon}\).
Lemma 3.7
We have
Proof
Let \(z_{\varepsilon}=(1+\beta u_{\varepsilon}+\gamma v_{\varepsilon})u_{\varepsilon}\). Then
Multiplying (3.24) by \((k_{1\varepsilon}(x)z_{\varepsilon x} )_{x}\) and then integrating it over \(Q_{t}\), from (3.4), (3.9), (3.14), and Cauchy’s inequality we see that, for any \(\vartheta>0\),
Choosing \(\vartheta=1/2\) and using (3.24) and (3.9), we further get
As in the proof of Lemma 3.3, from (3.23), (3.24), (3.9), (3.10), and Chapter II, equation (3.8) in [18] we conclude that (3.21) and (3.22) hold. □
Lemma 3.8
For any \(\varepsilon_{1}, \varepsilon _{2}>0\), we have
Proof
Let
Then it follows from the second equation of (3.5) that
Multiplying this equation by v̄ and integrating by parts over \(Q_{t}\), from (3.6), (3.14), and Cauchy’s inequality we find that, for any \(\vartheta>0\),
where \(\psi:=|u_{\varepsilon_{2}x}|+|u_{\varepsilon _{1}x}|+|v_{\varepsilon_{1}x}|+|v_{\varepsilon_{2}x}|+1\). Taking \(\vartheta=\mu_{0}/2\), we get
Similarity, we see from the first equation of (3.5) that
As we have done in the derivation of (3.26), by a direct computation we have
Combining (3.26) and (3.27) leads us to the inequality
where \(y(t):=\int_{0}^{t} \{ \|\psi(\cdot,t)\|^{2}_{L^{\infty}(0,l)}\int_{0}^{l} |\bar{\mathbf {w}}|^{2}\,\mathrm{d} x \}\,\mathrm{d}t\). In view of (3.10) and (3.22), it follows from (3.28) that
Moreover, Gronwall’s inequality leads to
which along with (3.28) implies (3.25). □
3.5 Uniform estimates of \(\|u_{\varepsilon xt},v_{\varepsilon xt}\|_{L^{2}(I'\times(\tau,T])}\)
To get the regularity of the limit function of sequence \(\{(u_{\varepsilon},v_{\varepsilon})\}\), we need to derive the uniform estimates of \(\|u_{\varepsilon xt},v_{\varepsilon xt}\|_{L^{2}(I'\times(\tau,T])}\).
Lemma 3.9
For any open interval \(I'\subset\subset (0,l)\) and any number \(\tau\in(0,T)\), we have
Proof
Choose open intervals \(I'_{j}\) and positive number \(\tau_{j}\), \(j=1,2,3\), such that \(I'\subset\subset I'_{3}\subset \subset I'_{2}\subset\subset I'_{1}\subset\subset(0,l)\) and \(\tau_{1}<\tau_{2}<\tau_{3}<\tau\). Differentiating the second equation of (3.5) with respect to t and noting that \(v_{\varepsilon xt}=v_{\varepsilon tx}\) we get
Let \(\lambda=\lambda(x,t)\) be an arbitrary smooth function taking values in \([0,1]\) such that \(\lambda=1\) for \((x,t)\in I'_{2}\times(\tau_{2},T]\), \(\lambda=0\) for \(x\notin I'_{1}\) or \(t\leq\tau_{1}\), and \(|\lambda_{x}|+|\lambda_{t}|\leq C/d'+C/\tau\) for all \((x,t)\). Multiplying equation (3.31) by \(v_{\varepsilon t}\lambda^{2}\) and then integrating by parts over \(I'_{1}\times(\tau_{1},t_{1}]\) for \(t_{1}\in(\tau,T]\), from (3.6), (3.9), (3.14), (3.21), and Cauchy’s inequality we deduce that, for any \(\vartheta>0\),
Choosing \(\vartheta=\mu_{0}/4\), we have
As in the proof of Lemma 3.8, employing Gronwall’s inequality and (3.10), from this inequality we get (3.30) and the uniform estimate of \(\|v_{\varepsilon t}\|_{ V_{2}(I'_{2}\times(\tau_{2},T])}\). Chapter II, equation (3.8) in [18] further implies that
Based on (3.30) and (3.32), we next show that (3.29) holds. Let \(\xi=\xi(x,t)\) be a smooth function with values between 0 and 1 such that \(\xi=1\) for \((x,t)\in I'_{3}\times(\tau_{3},T]\), \(\xi=0\) for \(x\notin I'_{2}\) or \(t\leq \tau_{2}\), and \(|\xi_{x}|+|\xi_{t}|\leq C/d'+C/\tau\) for all \((x,t)\). Similarly, differentiating the first equation of (3.5) with respect to t, multiplying the resulting equation by \(u_{\varepsilon t}\xi^{2}\) and integrating it over \(I'_{2}\times(\tau_{2},T]\), we find from (3.6), (3.9), (3.14), (3.21), and Cauchy’s inequality that, for any \(\vartheta>0\),
Setting \(\vartheta=\mu_{0}/4\) and using (3.10), (3.22), (3.30), and (3.32), we have
Again by Gronwall’s inequality we obtain (3.29). □
4 The proof of Theorem 2.1
To prove Theorem 2.1, let us discuss the behavior of vector sequence \(\{(u_{\varepsilon},v_{\varepsilon})\}\) governed by (3.5) as \(\varepsilon\to0\).
Lemma 4.1
The sequences \(\{u_{\varepsilon}\}\) and \(\{ v_{\varepsilon}\}\) converge to nonnegative bounded functions u and v in \(V_{2}(Q_{T})\), respectively. Moreover, \((u, v)\) possesses property (i) in Definition 2.1, and satisfies pointwise the equations in (1.1) on \(Q^{(i)}_{T}\) (\(i=1,2\)) and the initial conditions on \(\{(x,t): x\in (0,l), t=0\}\). For any \(\eta\in W^{1,0}_{2}(Q_{T})\), the following integral identities hold:
Proof
We see from estimate (3.25) that there exist functions \(u, v\in V_{2}(Q_{T})\) such that
and from (3.9), (3.21) that there exist subsequences (hereafter we retain the same notations for them) \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\), such that
In addition, it follows from (3.6), (3.14), (3.16), and the Arzela-Ascoli theorem that, for any given open interval \(I'\subset\subset(0,l)\), there exist subsequences \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\), such that \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\) converge to u, v in \(C(\bar{I}'\times[0,T] )\), respectively, and from (3.17) that, for any given \(i\in\{1,2\}\), \(I''\subset\subset Q^{(i)}_{T}\) and \(\tau\in(0,T)\), there exist subsequences \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\), such that \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\) converge to u, v in \(C^{2,1}(\bar{I}''\times[\tau,T] )\), respectively. Then u, v are in \(W^{1,1}_{2}(Q_{T})\cap C^{\alpha'}(\bar{I}'\times[0,T])\cap C^{2,1}(Q^{(i)}_{T})\cap C^{2+\alpha'',1+\alpha''/2}(\bar{I}''\times [\tau,T])\). Furthermore, (3.5) implies that \((u,v)\) satisfies pointwise the equations in (1.1) on \(Q^{(i)}_{T}\) (\(i=1,2\)) and the initial conditions on \(\{(x,t):x\in (0,l), t=0\}\). By (3.6), (3.9), (3.14), and (3.21) we have
For any \(\eta\in W^{1,0}_{2}(Q_{T})\), multiply the equations in (3.5) by η and then integrate by parts over \(Q_{t}\) to get
We next investigate the limit of each term in (4.7), (4.8). It follows from (3.1), (4.3), and (4.5) that
and
Furthermore, (3.22) and (4.10) yield
Consequently,
Similar arguments show that
Letting \(\varepsilon\to0\) and using (4.4) and (4.9)-(4.13), we find from (4.7), (4.8) that the integral identities (4.1), (4.2) hold. □
We next investigate the regularity of \((u,v)\) in the neighborhood of inner boundary \(\Gamma_{T}\).
Lemma 4.2
For any \(I'\subset\subset (0,l)\) and \(\tau\in(0,T)\), \(v_{t}\in C^{\bar{\alpha}'}(\bar{I}'\times[\tau ,T])\) and \(v_{x}\in C^{\bar{\alpha}'}((\bar{I}'\cap\bar{I}^{(i)})\times[\tau,T])\) (\(i=1,2\)) for some \(\bar{\alpha}'=\bar{\alpha}'(d',\tau)\in(0,1)\).
Proof
It follows from (3.29), (3.30), (4.3), and (4.6) that there exist subsequences \(\{u_{\varepsilon}\}\), \(\{v_{\varepsilon}\}\), such that
and
Let \(\tilde{B}:= \tilde{B}(x,t,w, q)=k_{2}(x)(1+2\delta w)q\), \(\hat{B}:= \hat{B}(x,t,w)=-wf_{2}(x,u(x,t),w)\). Then \(v(x,t)\) is a generalized solution of the single equation \(w_{t}- (\tilde{B}(x,t,w, w_{x}) )_{x}+\hat{B}(x,t,w)=0\) in the sense of Definition 1.1 in [19]. We find that when \((x,t,w,q)\in(I'\cap I^{(i)})\times(\tau,T]\times[0,C_{0}]\times\mathbb{R}\),
Hence when \((x,t,w,q)\in(I'\cap I^{(i)})\times(\tau,T]\times [0,C_{0}]\times\mathbb{R}\),
where \(\psi_{1}(x,t):=|u_{x}(x,t)|+|u_{t}(x,t)|\). By (4.6) and (4.14), we have
According to Theorem 1.1 in [19], there exists \(\bar{\alpha}'_{1}=\bar{\alpha}'_{1}(d',\tau)\in(0,1)\) such that
Note that \(v\in C^{\alpha'}(\bar{I}'\times[0,T])\) and \((k_{2,i}(1+2\delta v)v_{x})_{x}=v_{t}-vf_{2,i}(u,v)\) on \(Q^{(i)}_{T}\). We see that \((k_{2,i}(1+2\delta v)v_{x})_{x}\in C^{\bar{\alpha}'_{2}}((\bar{I}'\cap \bar{I}^{(i)})\times[\tau,T])\), and \(k_{2,i}(1+2\delta v)v_{x}\in C^{1+\bar{\alpha}'_{2},\bar{\alpha}'_{2} }((\bar{I}'\cap \bar{I}^{(i)})\times[\tau,T])\) for some \(\bar{\alpha}'_{2}\in(0,1)\). Thus \(v_{x}\in C^{\bar{\alpha}'_{3} }((\bar{I}'\cap \bar{I}^{(i)})\times[\tau,T])\). Since \((k_{2,i}(1+2\delta v)v_{x})_{x}=k_{2,i}(1+2\delta v)v_{xx}+2k_{2,i}\delta v^{2}_{x}\) on \(Q^{(i)}_{T}\),
This completes the proof of Lemma 4.2. □
Lemma 4.3
For any \(I'\subset\subset (0,l)\) and \(\tau\in(0,T)\), \(u_{t}\in C^{\tilde{\alpha}'}(\bar{I}'\times[\tau,T])\) and \(u_{x}\in C^{\tilde{\alpha}'}((\bar{I}'\cap \bar{I}^{(i)})\times[\tau,T])\) (\(i=1,2\)) for some \(\tilde{\alpha}'=\tilde{\alpha}'(d',\tau)\in(0,1)\). Furthermore, \((u,v)\) satisfies pointwise the inner boundary conditions in (1.1) on \(\Gamma_{T}\), and satisfies the homogeneous Neumann boundary conditions on \(S_{T}\) for almost all \(t\in[0,T]\).
Proof
To investigate the regularity of u, we need to estimate \(\sup_{t\in[\tau,T]}\|v_{xt}(\cdot,t)\|_{L^{2}(I')}\). Choose open intervals \(I'_{1}\), \(I'_{2}\), \(I'_{3}\) and positive numbers \(\tau_{1}\), \(\tau_{2}\), \(\tau_{3}\), such that \(I'\subset\subset I'_{3}\subset \subset I'_{2}\subset\subset I'_{1}\subset\subset(0,l)\) and \(\tau_{1}<\tau_{2}<\tau_{3}<\tau\). According to Lemma 4.1, \((u,v)\) satisfies pointwise the equations in (1.1) on \(Q^{(i)}_{T}\) (\(i=1,2\)). Furthermore, from (4.2) in Lemma 4.1 and the regularity of v in Lemma 4.2 we conclude that v satisfies pointwise the inner boundary condition \([k_{2}(x) ((1+\delta v)v )_{x} ]^{-}(l_{0},t)= [k_{2}(x) ((1+\delta v)v )_{x} ]^{+}(l_{0},t) \) for \(t\in(0,T]\).
Let \(w=(1+\delta v)v\), \(w_{(t)}:=(w(x,t+\Delta t)-w(x,t))/\Delta t\). Then
and
Let \(\xi=\xi(x,t)\) be a smooth function taking values between 0 and 1 such that \(\xi(x,t)=1\) for \((x,t)\in I'_{2}\times(\tau_{2},T]\), \(\xi=0\) for \(x\notin I'_{1}\) or \(t\leq\tau_{1}\), and \(|\xi_{x}|+|\xi_{t}|\leq C(d',\tau)\) for all \((x,t)\). Since \(v_{\varepsilon xt}=v_{\varepsilon tx}\), then \(v_{ xt}=v_{ tx}\) and \(w_{ xt}=w_{ tx}\) on \(I'_{1}\times(\tau_{1},T] \). Multiplying the equations in (4.18) by \((k_{2}(x)w_{x(t)} )_{x}\xi^{2}\) and integrating by parts over \(I'_{1}\times(\tau_{1},t)\), by direct computation we have
Employing Cauchy’s inequality we see that, for any \(\vartheta_{1},\vartheta_{2}>0\),
where \(J_{2}=\int_{\tau_{1}}^{t}\int_{I'_{1}} [w^{2}_{x(t)}+v^{2}_{(t)} (k_{2}(x)w_{x} )^{2}_{x} +u^{2}_{(t)}+v^{2}_{(t)} ]\,\mathrm{d}x\,\mathrm{d}t\). Moreover, it follows from the equations in (4.18) that
Choosing \(\vartheta_{1}\), \(\vartheta_{2}\), such that \(\vartheta _{1}+C^{*}\vartheta_{2}=1/2\), we find from (4.19)-(4.21) that
Then
which along with (4.17) further implies that
We next use (4.22) to prove that u possesses property (ii) in Definition 2.1. It is obvious that \(u(x,t)\) is a generalized solution of the single equation \(z_{t}- (\tilde{A}(x,t,z, z_{x}) )_{x}+\check{A}(x,t,z)=0\) in the sense of Definition 1.1 in [19], where
From (4.15) and (4.16) it follows that when \((x,t,z,p)\in(I'_{2}\cap I^{(i)})\times(\tau_{2},T]\times[0,C_{1}]\times \mathbb{R}\),
where \(\phi_{1}:=|v_{xt}|\), and from (4.22) that
Consequently, the conditions of Theorem 1.1 in [19] are fulfilled. Then Theorem 1.1 in [19] shows that
Since \([k_{1,i} ((1+\beta u+\gamma v)u )_{x} ]_{x}=uf_{1,i}(u,v)-u_{t}\) on \((I'_{2}\cap I^{(i)})\times (\tau_{2},T]\), then by (4.23) we get \([k_{1,i} ((1+\beta u+\gamma v)u )_{x} ]_{x}\in C^{\tilde{\alpha}'_{2} }((\bar {I}'_{3}\cap \bar{I}^{(i)})\times[\tau_{3},T])\). Thus \((1+2\beta u+\gamma v )u_{ x}+\gamma u v_{ x}\in C^{1+\tilde{\alpha}'_{2},\tilde{\alpha}'_{2} }((\bar {I}'\cap \bar{I}^{(i)})\times[\tau,T])\). In view of \(u,v\in C^{\alpha' }(\bar{I}'\times[0,T])\) and \(v_{x}\in C^{\bar{\alpha}'_{1} }((\bar {I}'\cap \bar{I}^{(i)})\times[\tau,T])\), we have \(u_{x}\in C^{\tilde{\alpha}'_{3} }((\bar{I}'\cap\bar{I}^{(i)})\times[\tau,T])\). Since \((u,v)\) satisfies pointwise the equations in (1.1) on \(Q^{(i)}_{T}\) (\(i=1,2\)), then we further use integral identities (4.1), (4.2), the regularity of \((u,v)\) in the neighborhood of \(\Gamma_{T}\) and the compatibility conditions in (2.3) to conclude that \((u,v)\) satisfies pointwise the inner boundary conditions in (1.1) on \(\Gamma_{T}\), and it satisfies homogeneous Neumann boundary conditions on \(S_{T}\) for almost all t (see Chapter III, Section 13 in [18]). □
From Lemmas 4.1-4.3 we see that problem (1.1) has at least one solution.
Lemma 4.4
The solution of problem (1.1) is unique.
Proof
Suppose that \((u_{1},v_{1})\), \((u_{2},v_{2})\) are solutions of (1.1). Let
Set \(\eta=\tilde{v}\) in (4.2). Then by a subtraction of the two integral identities (4.2) for \(v_{1}\), \(v_{2}\), we have
Using Cauchy’s inequality we get
where \(\bar{\psi}:=1+|u_{1x}|+|u_{2x}| +|v_{1x}|+|v_{2x}|\).
Similarly, set \(\eta=\tilde{u}\) in (4.1). By a subtraction of the two integral identities (4.1) for \(u_{1}\), \(u_{2}\), we obtain
which, together with Cauchy’s inequality, implies that
Furthermore, (4.24), (4.25) yield
Again by using Gronwall’s inequality we deduce that \(\tilde{\mathbf{w}}=0\). Then the solution of (1.1) is unique. □
References
Tan, QJ: An elliptic cross-diffusion system describing two-species models on a bounded domain with different natural conditions. J. Math. Anal. Appl. 437, 853-869 (2016)
Shigesada, N, Kawasaki, K, Teramoto, E: Spatial segregation of interacting species. J. Theor. Biol. 79, 83-99 (1979)
Chen, X, Qi, Y, Wang, M: A strongly coupled predator-prey system with non-monotonic functional response. Nonlinear Anal. 67, 1966-1979 (2007)
Li, H, Pang, PYH, Wang, M: Global solutions for a general strongly coupled prey-predator model. Appl. Math. Lett. 22, 1508-1512 (2009)
Ling, Z, Zhang, L, Lin, Z: Turing pattern formation in a predator-prey system with cross diffusion. Appl. Math. Model. 38, 5022-5032 (2014)
Lou, Y, Ni, WM, Wu, Y: On the global existence of a cross-diffusion system. Discrete Contin. Dyn. Syst. 4, 193-203 (1998)
Lou, Y, Ni, WM: Diffusion, self-diffusion and cross-diffusion. J. Differ. Equ. 131, 79-131 (1996)
Lou, Y, Ni, WM, Yotsutani, S: Pattern formation in a cross-diffusion system. Discrete Contin. Dyn. Syst. 35, 1589-1607 (2015)
Ko, W, Ryu, K, Ahn, I: Coexistence of three competing species with non-negative cross-diffusion rate. J. Dyn. Control Syst. 20, 229-240 (2014)
Ni, WM, Wu, Y, Xu, Q: The existence and stability of nontrivial steady states for S-K-T competition model with cross diffusion. Discrete Contin. Dyn. Syst. 34, 5271-5298 (2014)
Pao, CV: Strongly coupled elliptic systems and applications to Lotka-Volterra models with cross-diffusion. Nonlinear Anal. 60, 1197-1217 (2005)
Ryu, K, Ahn, I: Coexistence theorem of steady states for nonlinear self-cross diffusion systems with competitive dynamics. J. Math. Anal. Appl. 283, 46-65 (2003)
Shim, SA: Uniform boundedness and convergence of solutions to cross-diffusion systems. J. Differ. Equ. 185, 281-305 (2002)
Tuộc, PV: Global existence of solutions to Shigesada-Kawasaki-Teramoto cross-diffusion systems on domains of arbitrary dimensions. Proc. Am. Math. Soc. 135, 3933-3941 (2007)
Tuoc, PV: On global existence of solutions to a cross-diffusion system. J. Math. Anal. Appl. 343, 826-834 (2008)
Wu, Y, Zhao, Y: The existence and stability of traveling waves with transition layers for the S-K-T competition model with cross-diffusion. Sci. China Math. 53, 1161-1184 (2010)
Protter, MH, Weinberger, HF: Maximum Principles in Differential Equations, 2nd edn. Springer, New York (1984)
Ladyzenskaya, OA: Linear and Quasilinear Equations of Parabolic Type. Am. Math. Soc., Providence (1968)
Tan, QJ, Leng, ZJ: The regularity of generalized solutions for the n-dimensional quasi-linear parabolic diffraction problems. Nonlinear Anal. 69, 4624-4642 (2008)
Acknowledgements
The authors would like to thank the reviewers and the editors for their valuable suggestions and comments. The work was supported by the research fund of Department of Education of Sichuan Province (15ZB0343) and the research fund of Chengdu Normal University (CS14ZD01).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors read and approved the final manuscript.
Rights and permissions
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.
About this article
Cite this article
Tan, QJ., Su, J. & Pan, CY. The global existence of a parabolic cross-diffusion system with discontinuous coefficients. Bound Value Probl 2016, 128 (2016). https://doi.org/10.1186/s13661-016-0636-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13661-016-0636-1