Abstract
This paper is concerned with existence and stability of V-shaped traveling fronts for a class of nonlocal dispersal equations with unbalanced bistable nonlinearity. The main tool is sub- and supersolution technique combined with a comparison principle.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
We consider the following equation
with the nonlocal dispersal operator \( (J*u-u)(\varvec{x},t)=\int \limits _{\mathbb {R}^{2}}J(\varvec{x}-\varvec{y})\left( u(\varvec{y},t)-u(\varvec{x},t)\right) \mathrm{d}\varvec{y}. \) The kernel \(J\in C^{1}(\mathbb {R}^{2})\) satisfies
The nonlinearity \(f\in C^{2}(\mathbb {R})\) has only three zeros 0, a and 1, and satisfies
Obviously, if \(J(\varvec{x}) = \frac{1}{4\pi \lambda } e^{-\frac{|\varvec{x}|^{2}}{4\lambda }}\) for any given \(\lambda > 0\) or \(J(\varvec{x})\) is compactly supported with symmetric property, then it satisfies (J1)–(J2). Condition (F2) guarantees that the solution of the corresponding Cauchy problem of (1.1) has the same regularity with its initial function [13].
Traveling waves of (1.1) are widely used to model nonlocal diffusion phenomena in fields such as physics, chemistry, ecology and epidemiology. In one-dimensional space, traveling wave solutions have the form \(u(x,t)=U(\xi )\), \(\xi =x+ct\), where U is the wave profile and c is the wave speed. It is referred to [1, 6, 8, 9, 19, 29] for the mathematical study on traveling waves of (1.1).
Let
Then under the condition (J1), \(J_{1}\) is nonnegative and even, with unit integral on \(x\in \mathbb {R}\). It is also not difficult to prove that \(J'_{1}(x)\in L^{1}(\mathbb {R})\) with the aid of condition (J2). Then under the condition (F1), the following equation
admits a unique (up to translation) solution U connecting 0 and 1. Moreover, U is of class \(C^{k+1}\) if f is of class \(C^{k}\) for some \(k\ge 1\), and its speed c is given by
which can be positive or negative [1]. We assume that \(c>0\) in the present paper, and the case \(c<0\) can be dealt with by a same way. Furthermore, the wave profile U and its derivative \(U'\) have exponential behaviors near \(\pm \infty \):
where \(A_{i},B_{i},\lambda _{i},\delta _{i}(i\in \{1,2,3\})\) are positive constants, see [9, 10]. Following the technique as in [9, Section 1.5], we can also get that \(U''\) has exponential behavior near \(\pm \infty \).
Studies on the existence and stability of nonplanar traveling waves for the classical reaction diffusion equations or systems are already quite a lot, see [2,3,4,5, 11, 12, 16, 17, 20,21,22,23, 26] and references therein for scalar equations, and see [15, 18, 24, 25, 27, 28] and references therein for reaction–diffusion systems. While there are still few studies on nonplanar traveling waves of nonlocal dispersal equations. Chan and Wei proved the existence of pyramidal traveling wave solutions for the fractional bistable equation [7], and Li et al. proved the existence of pyramidal traveling wave solutions for the bistable nonlocal equation [14]. However, to the best of our knowledge, there is still no result about the stability of nonplanar traveling wave solutions for the nonlocal dispersal equations. In the current paper, we aim to prove the existence and stability of V-shaped traveling fronts for (1.1).
Since the curvature accelerates propagation of waves, it is naturally to assume that the speed s of nonplanar traveling waves satisfies \(s>c\). Without loss of generality, we also assume that the solutions travel towards the \(x_{2}-\)direction; thus, they have the form \(u(\varvec{x},t)=\widehat{u}(x_{1},x_{2}+st,t)\), and \(\widehat{u}\) satisfies
Throughout this paper, we denote the solution of (1.5) by \(\widehat{u}(\varvec{x},t;u_{0})\). In this paper, we first find a nontrivial steady-state function \(V(\varvec{x})\) of (1.5), i.e., \(V(\varvec{x})\) satisfies
and then prove its stability.
Let \( m_{*}=\sqrt{s^2-c^2}/c\) and
Then, \(v^{-}(\varvec{x})\) is a subsolution to (1.6). Actually, denote \(v_{j}^{1}(\varvec{x}):= U(A_{j}\cdot \varvec{x})\) with \(A_{j}=\frac{c}{s}(m_{*}, 1)\). If we let \({\varvec{\xi }}=\varvec{A}\varvec{y}\) with \(\varvec{A}\) a \(2\times 2\) orthonormal matrix whose first row is \(A_{j}\), then we have
See (1.2) for the definition of \(J_{1}\). It follows that
which means that \(v^{1}_{j}(\varvec{x})\) is a planar wave to (1.6). Similarly, denote \(v_{j}^{2}(\varvec{x}):= U(A_{j}\cdot \varvec{x})\) with \(A_{j}=\frac{c}{s}(-m_{*}, 1)\), and then, we can get \( \mathscr {L}[v_{j}^{2}(\varvec{x})]=0\). Thus, \(v^{-}(\varvec{x})\) is a subsolution.
Now we give the main results.
Theorem 1.1
(Existence) Assume (J1)–(J2) and (F1)–(F2) hold. For each \(s>c\), (1.6) admits a solution \(V_{*}(\varvec{x})\) with \(\partial _{x_{2}}V_{*}(\varvec{x})>0\) in \(\mathbb {R}^{2}\) and
Then \(u(\varvec{x},t)=V_{*}(x_{1},x_{2}+st)\) is a traveling front of (1.1), whose global average speed tends to c along with time, i.e.,
where \(L_{t}\) represents the level set of \(u(\varvec{x},t)\) at time t.
Theorem 1.2
(Stability) Let \(v_{0}\in C(\mathbb {R}^{2},[0,1])\) satisfy \(v_{0}-v^{-}\in L^{1}(\mathbb {R}^{2})\) and
Then, we have
Remark 1.3
(1.8) implies that \(V_{*}(\varvec{x})\) has V-shaped level sets and behaves like planar traveling waves far away in the space. The speed of \(V_{*}(\varvec{x})\) is a semi-continuum, that is, \(s\in (c,+\infty )\), which is quite different to classical bistable case, while (1.10) tells that the average speed of \(V_{*}(\varvec{x})\) is unique and always equals the planar wave’s speed c.
In the next section, we establish the existence result. The proof looks simpler than that of [14] but is a little different. In Sect. 3, we obtain the stability result.
2 Existence of V-shaped traveling fronts
For any \(s>c\), the equation
admits a unique solution \(\varphi (x)\) whose asymptotic line is \(y=m_{*}|x|\) satisfying \(m_{*}|x|\le \varphi (x)\), see [16]. Furthermore, there exist positive constants \(K_{0},K_{1},K_{2},K_{3}\) such that for all \(x\in \mathbb {R}\),
where \(\mu _{\pm }>0\) are constants and \(\gamma =sm_{*}=\frac{s\sqrt{s^{2}-c^{2}}}{c}>0.\)
2.1 Construction of a supersolution
By the assumption (F1), there exists a positive constant \(\delta _{0}\in (0,\frac{1}{4})\) such that
where \( \kappa _{1}:=\frac{1}{2}\min \{-f'(0),-f'(1)\}>0. \)
Lemma 2.1
There exist a positive constant \(\varepsilon _{0}^{+}\) and a positive function \(\alpha _{0}^{+}(\varepsilon )\) such that for any \(\varepsilon \in (0, \varepsilon _{0}^{+})\) and \(\alpha \in (0, \alpha _{0}^{+}(\varepsilon ))\), the function
is a supersolution to (1.6) and
Proof
Assume \(\alpha \in (0,1)\) and write \(v^{+}(\varvec{x})\) instead of \(v^{+}(\varvec{x};\varepsilon ,\alpha )\) throughout the proof. Denote
then \(v^{+}(\varvec{x})\) can be rewritten as \( v^{+}(\varvec{x})=U(\zeta (\varvec{x}))+\varepsilon \sigma (x_{1}). \) By the equation (1.3) and the definition of \(\mathscr {L}\), we have
Denote
Now we estimate the four terms.
(1) Estimate of term I. Since J is radially symmetric, it is easy to see that
Let
and \({{\varvec{\xi }}}=A\varvec{y}\), where \({{\varvec{\xi }}}=(\mu ,\nu )^{T}\). Such an orthogonal transformation gives that \(\mu =\frac{\varphi '(\alpha x_{1})}{\sqrt{1+\left( \varphi '(\alpha x_{1})\right) ^{2}}}y_{1}+\frac{1}{\sqrt{1+\left( \varphi '(\alpha x_{1})\right) ^{2}}}y_{2}\). Then
Let
and define \(F(t)=-U(\mu ^{*}(t)), t\in (0,1)\), and then, (2.9) can be written as
Denote \(y_{1}(t)=\alpha (x_{1}-ty_{1})\). Then, \(\mu ^{*}_{t}(t)=A(t)+B(t)\mu ^{*}(t)\), where \(B(t)=\frac{\alpha \varphi '(y_{1}(t))\varphi ''(y_{1}(t))y_{1}}{1+\left( \varphi '(y_{1}(t))\right) ^{2}}\) and
Furthermore, \(A_{t}(t)=A(t)B(t)\), and thus, \(\mu ^{*}_{tt}(t)=2A(t)B(t)+(B_{t}(t)+B^{2}(t))\mu ^{*}(t)\), where
Following from (2.1)–(2.2), we have
Since \(F(1)-F(0)=F'(0)+\int \limits _{0}^{1}(1-t)F''(t)\mathrm{d}t\) and
we have
where \(K^{*}=\max \{K_{1},K_{1}^{2},K_{0}K_{1},K_{0}K_{1}^{2},K_{0}^{2}K_{1}^{2},K_{1}(1+K_{0})(1+K_{0}^{2})\}\) and
Here, \(||\cdot ||_{\infty }\) is the supremum norm about \(\mu \in \mathbb {R}\). And in the above estimate we use the inequality \(\mathrm{sech}(x_{1}+y_{1})\le \mathrm{sech}x_{1}\cdot e^{|y_{1}|}\). Combining (2.9)–(2.10) and the above estimates, we have
Under the condition (J2), the integral \(\int \limits _{\mathbb {R}^{2}}J(\varvec{y})\left( |y_{1}|+4y_{1}^{2}+4|y_{1}|^{3}+y_{1}^{4}\right) e^{\gamma |y_{1}|}\mathrm{d}\varvec{y}\) is bounded for some \(\lambda >0\), and thus, there exists a constant \(C_{1}>0\) such that
(2) Estimate of term II.
where \(\theta \in (0,1)\). Since \(\mathrm{sech}'x=\mathrm{sech} x\cdot \frac{e^{-x}-e^{x}}{e^{x}+e^{-x}}\), we have
Again the assumption (J2) guarantees that \(\int \limits _{\mathbb {R}^{2}}J(\varvec{y})|y_{1}|e^{\gamma |y_{1}|}\mathrm{d}\varvec{y}\) is bounded for some \(\lambda >0\), and thus, there exists a constant \(C_{2}>0\) such that
(3) Estimate of terms III and IV. By (2.3), we have
About the fourth term, we have
where \(\tilde{\theta }\in (0,1).\)
In order to prove \(\mathscr {L}[v^{+}(\varvec{x})]\ge 0\), we consider two cases.
Case 1. \(U(\zeta (\varvec{x}))\ge 1-\delta _{0}\) or \(U(\zeta (\varvec{x}))\le \delta _{0}\).
Then, \( -f'\left( U(\zeta (\varvec{x}))+\tilde{\theta }\varepsilon \sigma (x_{1})\right) \ge \kappa _{1} \) by (2.5), provided that \(0<\varepsilon <\delta _{0}\). And thus
for any \(\alpha \) satisfying \(\alpha \le \frac{\kappa _{1}\varepsilon }{C_{1}+C_{2}\gamma }\).
Case 2. \( \delta _{0}\le U(\zeta (\varvec{x}))\le 1-\delta _{0}\).
Denote \(U_{*}=\min _{U(x)\in \left[ \delta _{0},1-\delta _{0}\right] }U'(x)\) and \(f^{*}=\max _{x\in [-1,2]}|f'(x)|\). We have
provided that \( \left( C_{1}+C_{2}\gamma \right) \alpha +\varepsilon f^{*}\le K_{2}U_{*}. \) Let
then \(v^{+}(\varvec{x})\) is a supersolution if \(0<\varepsilon <\varepsilon _{0}^{+}\) and \(0<\alpha <\alpha _{0}^{+}(\varepsilon )\). A similar argument to that of Taniguchi [20, Lemma 7] yields (2.7)-(2.8). The proof is complete. \(\square \)
2.2 Proof of the existence result
First, we establish the comparison principle. Define
Theorem 2.2
Assume that (J1) and (F1)–(F2) hold. Let \(u_{0}(\varvec{x})\) and \(\partial _{x_{2}}u_{0}(\varvec{x})\) belong to \(BUC(\mathbb {R}^{2})\), and then, the following Cauchy problem
has a unique solution \(\widehat{u}(\varvec{x},t;u_{0})\in C(\mathbb {R}^{2}\times [0,\infty ),[0,2])\), which is also differentiable with respect to \(x_{2}\). Here, \(b\in \mathbb {R}\) is a nonzero constant. Moreover, if \(u_{0}(\varvec{x})\) is globally Lipschitz continuous, then \(\widehat{u}(\varvec{x},t;u_{0})\) is also a globally Lipschitz solution which is uniform in time.
Lemma 2.3
(Maximum principle) Assume that (J1) hold and that \(\widehat{u}\in C(\mathbb {R}^{2}\times [0,\infty ))\) is bounded and differentiable with respect to \(x_{2}\). If \(\widehat{u}\) satisfies
where \(K(\varvec{x},t):\mathbb {R}^{2}\times [0,\infty )\rightarrow \mathbb {R}\) is continuous and uniformly bounded, and b is a nonzero constant, then \(\widehat{u}(\varvec{x},t)\ge 0\) for \((\varvec{x},t)\in \mathbb {R}^{2}\times [0,\infty )\). Furthermore, if \(\widehat{u}(\varvec{x},0)\not \equiv 0\) for \(\varvec{x}\in \mathbb {R}^{2}\), then \(\widehat{u}(\varvec{x},t)> 0\) for \((\varvec{x},t)\in \mathbb {R}^{2}\times (0,\infty )\).
Lemma 2.4
(Comparison principle) Assume that (J1) holds and \(u_{1},u_{2}\in C(\mathbb {R}^{2}\times [0,\infty ))\) are both bounded and differentiable with respect to \(x_{2}\). Denote \(\mathscr {L}_{t}[u]=u_{t}-(J*u-u)+b u_{x_{2}} -f(u)\), where f is continuously differentiable with respect to u, \(f'(u)\) is uniformly bounded and b is a nonzero constant. If \(u_{1},u_{2}\) satisfy
then \(u_{1}\ge u_{2}\) on \((\varvec{x},t)\in \mathbb {R}^{2}\times [0,\infty )\). Furthermore, if \(u_{1}(\varvec{x},0)\not \equiv u_{2}(\varvec{x},0)\) for \(\varvec{x}\in \mathbb {R}^{2}\), then \(u_{1}> u_{2}\) for \((\varvec{x},t)\in \mathbb {R}^{2}\times (0,\infty )\).
The proof of the above three results can be referred to [14]. Now we prove Theorem 1.1.
Proof of Theorem 1.1
After making a slight modification of the proof of [14, Theorem 1.1], we can prove the existence of \(V_{*}\) and (1.8)–(1.9). Now we focus on the proof of (1.10).
By the comparison principle, we have
where \(v^{-}\) and \(v^{+}\) are defined by (1.7) and (2.6), respectively. Fix a constant \(\delta \in (0,1)\) and denote the level sets of \(v^{\pm }(x_{1},x_{2}+st)=\delta \) at time t by \(L^{\pm }_{t}\). Due to (2.11), the level sets \(L^{\pm }_{t}\) and \(L_{t}\) do not intersect each other at any time t. We know
and
For convenience, we also denote
Since
we know \(g^{-}(x_{1},t)>g^{+}(x_{1},t)\) for all \(x_{1}\in \mathbb {R}\) and \(t\in \mathbb {R}\) by the monotonicity of \(v^{-}(\varvec{x})\) on \(x_{2}\). Similarly, there hold \(g^{-}(x_{1},t)>g(x_{1},t)\) and \(g(x_{1},t)>g^{+}(x_{1},t)\) for all \(x_{1}\in \mathbb {R}\) and \(t\in \mathbb {R}\). In summary, there is
Moreover, we know that
holds for each fixed \(\alpha >0\). Consequently, \(\mathrm{dist}(L^{\pm }_{t},L_{t})=0\). Define
Obviously, \(M^{*}<+\infty \) for each fixed \(\alpha \). Now we take two moments \(t_{1}\), \(t_{2}\in \mathbb {R}\) and assume \(s(t_{2}-t_{1})>M_{*}\) without loss of generality. Under the assumption \(s(t_{2}-t_{1})>M_{*}\), there holds \(g^{-}(x_{1},t_{2})<g^{+}(x_{1},t_{1})\), and thus, \(L^{+}_{t_{1}}\) does not intersect \(L^{-}_{t_{2}}\).
The inequality (2.12) means that the level set \(L_{t}\) of \(u(\varvec{x},t)\) is between those of \(L^{\pm }_{t}\) for all \(x_{1}\in \mathbb {R}\) at any time \(t\in \mathbb {R}\). Thus by the definition of \(M^{*}\) and the choice of \(t_{i}~(i=1,2)\), it is straightforward that
To obtain \(\mathrm{dist}(L^{-}_{t_{1}},L^{+}_{t_{2}})\), it is sufficient to consider all the perpendicular segments from \(L^{+}_{t_{2}}\) to \(L^{-}_{t_{1}}\) except those intersecting \(L^{+}_{t_{2}}\) once more. Now take an arbitrary point \(\varvec{x}^{+}_{2}\in L^{+}_{t_{2}}\) and find the corresponding point \(\varvec{x}^{-}_{1}\in L^{-}_{t_{1}}\) such that the segment generated by \(\varvec{x}^{+}_{2}\) and \(\varvec{x}^{-}_{1}\) is perpendicular to \(L^{-}_{t_{1}}\). We denote this perpendicular segment by \(\textit{l}_{\varvec{x}^{-}_{1}\varvec{x}^{+}_{2}}\). Obviously, \(\textit{l}_{\varvec{x}^{-}_{1}\varvec{x}^{+}_{2}}\) is also perpendicular to \(L^{-}_{t_{2}}\), and we denote their intersection point by \(\varvec{x}^{-}_{2}\). It follows immediately that \(\left| \varvec{x}^{-}_{2}-\varvec{x}^{-}_{1}\right| =c|t_{2}-t_{1}|\). Then for any \(\varvec{x}^{+}_{2}\in L^{+}_{t_{2}}\) and \(\varvec{x}^{-}_{i}~(i=1,2)\) chosen by this way, we have
Here, \(|\cdot |\) denotes the Euclidean norm in \(\mathbb {R}^{N}~(N\ge 1)\). This fact and (2.13) yield that
Now we prove that the average speed is not less than c. For any given point \(\varvec{x}_{i}\in L_{t_{i}}~(i=1,2)\), draw a line passing through \(\varvec{x}_{i}\) and parallel to the \(x_{2}\) axis. Necessarily, this line intersects \(L_{t_{i}}^{-}\) at a point, which is still denoted by \(\varvec{x}^{-}_{i}\). Since \(\varvec{x}_{i}\in L_{t_{i}}\) is arbitrary, \(\varvec{x}^{-}_{i}\) is also arbitrary, and vice versa. Obviously,
It follows from (2.14) that
This implies the average speed is larger than or equal to c. This completes the proof. \(\square \)
3 Stability of V-shaped traveling fronts in \(\mathbb {R}^{2}\)
This section establishes the stability result.
Lemma 3.1
For any \(M>0\), there exists a constant \(C>0\) such that
where \(V_{*}\) and \(v^{+}\) are given by Theorem 1.1 and (2.6), respectively. Moreover, we have
Proof
The assertions about \(v^{+}\) are very straightforward. Now we prove the assertions about \(V_{*}\). Since \( \partial _{x_{2}}V_{*}>0\) in \(\mathbb {R}^{2}\), it suffices to prove that for any sequence \(\{(x_{n},z_{n})\}_{n\ge 1}\subseteq \mathbb {R}^2\) satisfying \(|(x_{n},z_{n})|\rightarrow \infty \) and \(\left| z_{n}+m_{*}|x_{n}|\right| \le M\), there is
Now we prove this result by contradiction. Assume \(\lim _{n\rightarrow \infty }\partial _{x_{2}}V_{*}(x_{n}^{*},z_{n}^{*})=0\) for a certain sequence \(\{(x_{n}^{*},z_{n}^{*})\}_{n\ge 1}\). Since \(v^-<V<v^+\) in \(\mathbb {R}^{2}\) and notice that \(v^{-}\) is a subsolution, we have
where \(V_{\tau }\) is between \(V_{*}\) and \(v^{-}\). Under the condition \(\left| z_{n}^{*}+m_{*}|x_{n}^{*}|\right| \le M\) and \(|(x_{n}^{*},z_{n}^{*})|\rightarrow \infty \), there must be \(|x_{n}^{*}|\rightarrow \infty \), and thus,
Then, it follows that
which further implies that
Let \(n\rightarrow \infty \) in (3.1), and then, we have
which is a contradiction. Similarly,
and thus \(\partial _{x_{2}}V_{*}\rightarrow 0\) as \(\left| \frac{c}{s}(x_{2}+m_{*}|x_{1}|)\right| \rightarrow \infty \). This completes the proof. \(\square \)
Lemma 3.2
There exist positive constants \(\rho>0, \kappa >0\) and \(\delta \in \left( 0,\delta _{0}\right) \) such that
is a supersolution, where \(\xi \in \mathbb {R}\) is a constant and \(\delta _{0}\) is defined in (2.5).
Proof
Let \(\mathscr {\tilde{L}}[u]=u_{t}-(J*u-u)+s\partial _{x_{2}}u-f(u)\). Then, we have
where \(\tau \in (0,1)\). To prove the lemma, we argue as follows.
Case 1. \(|x_{2}+m_{*}|x_{1}||> R_{0}\) for some \(R_{0}>0\) large enough.
Without loss of generality, assume that \(R_{0}>0\) is large enough such that \(V_{*}< \delta _{0}\) or \(V_{*}>1-\delta _{0}\). Then, we have
It then follows by (2.5) that
Thus,
provided that \(\kappa <\kappa _{1}\).
Case 2. \(|x_{2}+m_{*}|x_{1}||\le R_{0}\).
In this case, it follows from Lemma 3.1 that \(\partial _{x_{2}}V_{*}\ge C:=C(R_{0})\). Thus,
provided that \(\rho >1+\frac{\max _{u\in [0,1+\delta _{0}]}|f'(u)|}{\kappa }\). Taking \(0<\kappa <\kappa _{1}\), \(\rho >1+\frac{\max _{u\in [0,1+\delta _{0}]}|f'(u)|}{\kappa }\) and combining the above two cases, we know \(\mathscr {\tilde{L}}[u^{+}]>0\) in \(\mathbb {R}^{2}\). This completes the proof. \(\square \)
The proof of the following lemma is very similar to that of Lemma 3.2, and we omit it here.
Lemma 3.3
There exist positive constants \(\rho>0, \kappa >0\) and \(\delta \in \left( 0,\delta _{0}\right) \) such that
is a supersolution, where \(\xi \in \mathbb {R}\) is a constant \(\delta _{0}\) is defined in (2.5).
Lemma 3.4
If \(u(\varvec{x},t;u_{0})\) is the solution of the Cauchy problem
where the initial function \(u_{0}\in C(\mathbb {R}^{2})\) satisfies \(u_{0}-v^{-}\in L^{1}(\mathbb {R}^{2})\) and
then for any fixed \(T>0\), we have
Proof
Let \(w(\varvec{x})=U\left( \frac{c}{s}(x_{2}+\varphi (x_{1}))\right) \). First we show that \(w-v^{-}\in L^{1}(\mathbb {R}^{2})\). In fact,
where \(\tau \in (0,1)\), \(\varsigma (\varvec{x})=\frac{c}{s}(x_{2}+\varphi (x_{1}))\) and \(\eta (\varvec{x})=\frac{c}{s}(x_{2}+m_{*}|x_{1}|).\) See (1.4) for \(A_{3}, \lambda _{3}\) and see (2.4) for \(\mu _{+}\). It follows that \(w-u_{0}\in L^{1}(\mathbb {R}^{2})\). It is also not difficult to prove that
Thus, it suffices to prove that
for any given \(T>0\). Let \(\Phi (\varvec{x},t):=u(\varvec{x},t;u_{0})-w(\varvec{x})\). Then, it satisfies
where \(\Phi _{\tau }=\tau u+(1-\tau )w \) with \(\tau \in (0,1)\). Let \(\hat{\Phi }\) be the solution of the following Cauchy problem
where \(M:=\max _{u\in [0,1]}|f'(u)|\). By the maximum principle, \(\hat{\Phi }\ge 0\) in \(\mathbb {R}^{2}\). By the comparison principle, it is easy to verify that
In the following, we estimate \(\hat{\Phi }(\varvec{x},t)\). Let \(\Psi (\varvec{x},t)=\hat{\Phi }(x_{1},x_{2}+st,t)\), then \(\Psi \) satisfies
The solution of (3.3) can be expressed as
where \(S(\varvec{x},t)=e^{-t}\delta _{0}(\varvec{x})+K_{t}(\varvec{x})\) is the fundamental solution of (3.3) with initial data \(\delta _{0}\), the Dirac measure at zero and \(K_{t}(\varvec{x})=\int \limits _{\mathbb {R}^{2}}e^{-t}(e^{\hat{J}(\varvec{y})t}-1)e^{i(\varvec{x})\cdot \varvec{y}}\mathrm{d}\varvec{y}\) with \(\hat{J}\) the Fourier transform of J. It is not difficult to verify that \(||S(\varvec{x},t)||_{L^{1}(\mathbb {R}^{2})}\le 3\). Then for any given \(T>0\),
For any \(\epsilon >0\) small enough, there exist \(R_{1}>0\) and \(R_{2}>0\) big enough such that
and
This implies that \(\Psi (\varvec{x},T)<\epsilon \) for \(x_{1}^{2}+x_{2}^{2}\ge R_{2}^{2}\) and thus \(\lim _{R\rightarrow +\infty }\sup _{x_{1}^{2}+x_{2}^{2}\ge R^{2}}\Psi (\varvec{x},T)=0\). Recall \(\hat{\Phi }(\varvec{x},t)=\Psi (x_{1},x_{2}-st,t)\) and we have \(\lim _{R\rightarrow +\infty }\sup _{x_{1}^{2}+x_{2}^{2}\ge R^{2}}\hat{\Phi }(\varvec{x},T)=0\). Then, the proof completes following (3.2). \(\square \)
Lemma 3.5
The solution \(u(\varvec{x},t;u_{0})\) of the Cauchy problem
depends continuously on the initial function \(u_{0}(\varvec{x})\). That is, if \(u_{1}(\varvec{x},t;u_{0,1})\) and \(u_{2}(\varvec{x},t;u_{0,2})\) are two solutions of (3.4) with initial values \(u_{0,1}\) and \(u_{0,2}\), respectively, then we have
for some A(t) depending only on t.
Proof
Let \(v(\varvec{x},t)=u(x_{1},x_{2}+st)\). Then \(v(\varvec{x},t)\) satisfies
The above problem is equivalent to the following integral equation
where \(\mu >0\) is a constant. Let \(w(\varvec{x},t):=v_{2}(\varvec{x},t)-v_{1}(\varvec{x},t)\), then it satisfies
It then follows that
By the Gronwall’s inequality, we have
where \(C_{1}=\mu +||f'||_{L^{\infty }(\mathbb {R}^{2})}\). Let \(A(t)=1+C_{1}te^{C_{1}t}\). Then, the proof is complete. \(\square \)
Now, define
Then, \(V^{*}\) satisfies
And by the comparison principle, there holds
Lemma 3.6
\(V_{*}(\varvec{x})\equiv V^{*}(\varvec{x})\) in \(\mathbb {R}^{2}\).
Proof
Assume on the contrary that \(V_{*}(\varvec{x})\not \equiv V^{*}(\varvec{x})\) in \(\mathbb {R}^{2}\). Then, they must be \(V_{*}(\varvec{x})<V^{*}(\varvec{x})\). By the aid of (2.7), we can find a \(\delta >0\) small enough and a proper \(\xi >0\) such that
Then, the comparison principle yields
Letting \(t\rightarrow \infty \) in the above inequality, we have
Define
Obviously \(\Lambda \ge 0\) and \(V^{*}(\varvec{x})\le V_{*}(x_{1},x_{2}+\Lambda )\). If \(\Lambda =0\), then the proof is done. Thus, we assume \(\Lambda >0\) to derive a contradiction. It follows from \(v^{-}(\varvec{x})<V_{*}(\varvec{x})<V^{*}(\varvec{x})<v^{+}(\varvec{x})\) that
which implies that there must be
By Lemma 3.1, there exists a constant \(R^{*}>0\) large enough such that
Define
Since
there exists a \(\sigma \in (0,\delta _{0})\) small enough such that
For \(\varvec{x}\in \mathbb {R}\backslash \Omega \), by (3.5) we have
To sum up, there is
Then by the comparison principle, we have
with \(\xi =\Lambda -2\rho \sigma \). Let \(t\rightarrow \infty \) in the above inequality, then we have
which contradicts to the definition of \(\Lambda \). Thus, \(\Lambda =0\) and the proof is complete. \(\square \)
Now we show that the curved fronts \(V_{*}\) are asymptotically stable under the condition that the initial perturbation is positive.
Proof of Theorem 1.2
Denote \(v(\varvec{x},t;v_{0})\) by \(v(\varvec{x},t)\) for simplicity. On the one hand, since \(v^{-}(\varvec{x})\le v_{0}(\varvec{x})\), by the comparison principle, we have
On the other hand, a similar argument as [13, proposition 2.5] can deduce that
for any \(\varvec{x}, \varvec{y}\in \mathbb {R}^{2}\) and \(|v_{t}|\le C\), where L and C are constants independent of \(\varvec{x},\varvec{y}\) and t. Thus, we have
Similarly, we have
(3.6) and (3.7) imply that it suffices to prove that for any \(\epsilon >0\), there exists \(T^{*}>0\) such that
Step 1. Let \(A^{*}=\sup _{\varvec{x}\in \mathbb {R}^{2}}\partial _{x_{2}}V_{*}(\varvec{x})\), and take \(\rho>0, \kappa >0\) as in Lemma 3.2. Then,
provided that \(\delta <\frac{\epsilon }{3\rho A^{*}}\). In other words,
Step 2. Fix \(\delta >0\) in step 1. By (3.6), for any \(T_{\delta }>0\), we have
Following from Lemma 3.4, there exists a \(R_{\delta }>0\) such that
Let \(\alpha >0\) be small enough such that
In other words, if \(\alpha \) is chosen to satisfy
then
Combining the inequalities (3.10) and (3.11), we have
Then, Lemma 3.3 and the comparison principle yield that
for \(t\ge 0\). Denote \(w^{t}_{+}:=w^{+}(\varvec{x},t)\) and applying the comparison principle again, we have
Since \(v(\varvec{x},t;v^{+})\) converges monotonically to \(V^{*}(\varvec{x})\) as \(t\rightarrow +\infty \), it follows from (3.8) that there exists a \(t_{1}>0\) such that
where \(v^{+,\delta }=v^{+}(x_{1},x_{2}+\rho \delta )\). On the other hand, Lemma 3.5 yields that
From the definition of \(w^{+}\), we know that there exists a \(T_{1}>0\) such that
Combining the above facts, we obtain that
for \(t\ge T_{1}\) and \(\varvec{x}\in \mathbb {R}^{2}\). It follows from (3.12) that
for \(t\ge T_{1}\) and \(\varvec{x}\in \mathbb {R}^{2}\). Take \(T^{*}=t_{1}+T_{1}+T_{\delta }\). By Lemma 3.6, we have
for \(t\ge T^{*}\) and \(\varvec{x}\in \mathbb {R}^{2}\). Combining the above inequality and (3.9), we obtain
This completes the proof. \(\square \)
References
Bates, P.W., Fife, P.C., Ren, X., Wang, X.: Traveling waves in a convolution model for phase transitions. Arch. Ration. Mech. Anal. 138, 105–136 (1997)
Bonnet, A., Hamel, F.: Existence of non-planar solutions of a simple model of premixed Bunsen flames. SIAM J. Math. Anal. 31, 80–118 (1999)
Brazhnik, P.K., Tyson, J.J.: On traveling wave solutions of Fisher’s equation in two spatial dimensions. SIAM J. Appl. Math. 60, 371–391 (1999)
Bu, Z.-H., Ma, L., Wang, Z.-C.: Conical traveling fronts of combustion equations in \(\mathbb{R}^3\). Appl. Math. Lett. 108, 106509 (2020)
Bu, Z.-H., Wang, Z.-C.: Global stability of V-shaped traveling fronts in combustion and degenerate monostable equations. Discrete Contin. Dyn. Syst. 38, 2251–2286 (2018)
Carr, J., Chmaj, A.: Uniqueness of travelling waves for nonlocal monostable equations. Proc. Am. Math. Soc. 132, 2433–2439 (2004)
Chan, H., Wei, J.: Traveling wave solutions for bistable fractional Allen–Cahn equations with a pyramidal front. J. Differ. Equ. 262, 4567–4609 (2017)
Chen, X.: Existence, uniqueness and asymptotic stability of travelling waves in nonlocal evolution equations. Adv. Differ. Equ. 2, 125–160 (1997)
Coville, J.: Equations de reaction diffusion non-locale. Mathématiques. Universitsé Pierre et Marie Curie-Paris VI, Franxcais (2003)
Coville, J., Dupaigne, L.: Travelling fronts in integrodifferential equations. C. R. Acad. Sci. Paris, Ser. I 337, 25–30 (2003)
Hamel, F., Nadirashvili, N.: Travelling fronts and entire solutions of the Fisher-KPP equation in \(\mathbb{R}^{N}\). Arch. Ration. Mech. Anal. 157, 91–163 (2001)
Hamel, F., Monneau, R., Roquejoffre, J.-M.: Existence and qualitative properties of multidimensional conical bistable fronts. Discrete Contin. Dyn. Syst. 13, 1069–1096 (2005)
Li, W.-T., Sun, Y.-J., Wang, Z.-C.: Entire solutions in the Fisher-KPP equation with nonlocal dispersal. Nonlinear Anal. Real World Appl. 11, 2302–2313 (2010)
Li, W.-T., Niu, H.-T., Wang, Z.-C.: Nonplanar traveling fronts for nonlocal dispersal equations with bistable nonlinearity. Differ. Integral Equ. 34, 265–294 (2021)
Ni, W.-M., Taniguchi, M.: Traveling fronts of pyramidal shapes in competition–diffusion systems. Netw. Heterog. Media 8, 379–395 (2013)
Ninomiya, H., Taniguchi, M.: Existence and global stability of traveling curved fronts in the Allen–Cahn equations. J. Differ. Equ. 213, 204–233 (2005)
Ninomiya, H., Taniguchi, M.: Global stability of traveling curved fronts in the Allen–Cahn equations. Discrete Contin. Dyn. Syst. 15, 819–832 (2006)
Niu, H.-T., Wang, Z.-C., Bu, Z.-H.: Curved fronts in the Belousov–Zhabotinskii reaction-diffusion systems in \(\mathbb{R}^{2}\). J. Differ. Equ. 264, 5758–5801 (2018)
Schumacher, K.: Travelling-front solutions for integro-differential equations. I. J. Reine Angew. Math. 316, 54–70 (1980)
Taniguchi, M.: Traveling fronts of pyramidal shapes in the Allen–Cahn equations. SIAM J. Math. Anal. 39, 319–344 (2007)
Taniguchi, M.: The uniqueness and asymptotic stability of pyramidal traveling fronts in the Allen–Cahn equations. J. Differ. Equ. 246, 2103–2130 (2009)
Taniguchi, M.: Multi-dimensional traveling fronts in bistable reaction–diffusion equations. Discrete Contin. Dyn. Syst. 32, 1011–1046 (2012)
Taniguchi, M.: An (N-1)-dimensional convex compact set gives an N-dimensional traveling front in the Allen–Cahn equation. SIAM J. Math. Anal. 47, 455–476 (2015)
Taniguchi, M.: Convex compact sets in \(\mathbb{R}^{N-1}\) give traveling fronts of cooperation–diffusion systems in \(\mathbb{R}^{N}\). J. Differ. Equ. 260, 4301–4338 (2016)
Wang, Z.-C.: Traveling curved fronts in monotone bistable systems. Discrete Contin. Dyn. Syst. 32, 2339–2374 (2012)
Wang, Z.-C., Bu, Z.-H.: Nonplanar traveling fronts in reaction–diffusion equations with combustion and degenerate Fisher-KPP nonlinearities. J. Differ. Equ. 260, 6405–6450 (2016)
Wang, Z.-C., Li, W.-T., Ruan, S.: Existence, uniqueness and stability of pyramidal traveling fronts in reaction–diffusion systems. Sci. China 59, 1869–1908 (2016)
Wang, Z.-C., Niu, H.-L., Ruan, S.: On the existence of axisymmetric traveling fronts in Lotka–Volterra competition–diffusion systems in \(\mathbb{R}^{3}\). Discrete Contin. Dyn. Syst. Ser. B 22, 1111–1144 (2017)
Zhang, G.-B., Li, W.-T., Wang, Z.-C.: Spreading speeds and traveling waves for nonlocal dispersal equations with degenerate monostable nonlinearity. J. Differ. Equ. 252, 5096–5124 (2012)
Acknowledgements
The work is supported by NNSF of China (11901330).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Niu, HT. Existence and stability of traveling curved fronts for nonlocal dispersal equations with bistable nonlinearity. Z. Angew. Math. Phys. 73, 90 (2022). https://doi.org/10.1007/s00033-022-01734-8
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00033-022-01734-8
Keywords
- Unbalanced bistable nonlinearity
- Nonlocal dispersal
- V-shaped traveling fronts
- Super- and subsolutions
- Stability