Loading [MathJax]/jax/output/SVG/jax.js

Multidimensional stability of pyramidal traveling fronts in degenerate Fisher-KPP monostable and combustion equations

  • Received: 01 April 2021 Revised: 01 July 2021 Published: 13 August 2021
  • Primary: 35K57, 35B35, 35C07; Secondary: 92D25

  • In this paper, multidimensional stability of pyramidal traveling fronts are studied to the reaction-diffusion equations with degenerate Fisher-KPP monostable and combustion nonlinearities. By constructing supersolutions and subsolutions coupled with the comparison principle, we firstly prove that under any initial perturbation (possibly large) decaying at space infinity, the three-dimensional pyramidal traveling fronts are asymptotically stable in weighted L spaces on Rn(n4). Secondly, we show that under general bounded perturbations (even very small), the pyramidal traveling fronts are not asymptotically stable by constructing a solution which oscillates permanently between two three-dimensional pyramidal traveling fronts on R4.

    Citation: Denghui Wu, Zhen-Hui Bu. Multidimensional stability of pyramidal traveling fronts in degenerate Fisher-KPP monostable and combustion equations[J]. Electronic Research Archive, 2021, 29(6): 3721-3740. doi: 10.3934/era.2021058

    Related Papers:

    [1] Denghui Wu, Zhen-Hui Bu . Multidimensional stability of pyramidal traveling fronts in degenerate Fisher-KPP monostable and combustion equations. Electronic Research Archive, 2021, 29(6): 3721-3740. doi: 10.3934/era.2021058
    [2] Cui-Ping Cheng, Ruo-Fan An . Global stability of traveling wave fronts in a two-dimensional lattice dynamical system with global interaction. Electronic Research Archive, 2021, 29(5): 3535-3550. doi: 10.3934/era.2021051
    [3] Jin Li, Yongling Cheng . Barycentric rational interpolation method for solving KPP equation. Electronic Research Archive, 2023, 31(5): 3014-3029. doi: 10.3934/era.2023152
    [4] Peng Gao, Pengyu Chen . Blowup and MLUH stability of time-space fractional reaction-diffusion equations. Electronic Research Archive, 2022, 30(9): 3351-3361. doi: 10.3934/era.2022170
    [5] Shao-Xia Qiao, Li-Jun Du . Propagation dynamics of nonlocal dispersal equations with inhomogeneous bistable nonlinearity. Electronic Research Archive, 2021, 29(3): 2269-2291. doi: 10.3934/era.2020116
    [6] J. Vanterler da C. Sousa, Kishor D. Kucche, E. Capelas de Oliveira . Stability of mild solutions of the fractional nonlinear abstract Cauchy problem. Electronic Research Archive, 2022, 30(1): 272-288. doi: 10.3934/era.2022015
    [7] Léo Girardin, Danielle Hilhorst . Spatial segregation limit of traveling wave solutions for a fully nonlinear strongly coupled competitive system. Electronic Research Archive, 2022, 30(5): 1748-1773. doi: 10.3934/era.2022088
    [8] Harish Bhatt . Second-order time integrators with the Fourier spectral method in application to multidimensional space-fractional FitzHugh-Nagumo model. Electronic Research Archive, 2023, 31(12): 7284-7306. doi: 10.3934/era.2023369
    [9] Yi Wei . The Riccati-Bernoulli subsidiary ordinary differential equation method to the coupled Higgs field equation. Electronic Research Archive, 2023, 31(11): 6790-6802. doi: 10.3934/era.2023342
    [10] Ibtissam Issa, Zayd Hajjej . Stabilization for a degenerate wave equation with drift and potential term with boundary fractional derivative control. Electronic Research Archive, 2024, 32(8): 4926-4953. doi: 10.3934/era.2024227
  • In this paper, multidimensional stability of pyramidal traveling fronts are studied to the reaction-diffusion equations with degenerate Fisher-KPP monostable and combustion nonlinearities. By constructing supersolutions and subsolutions coupled with the comparison principle, we firstly prove that under any initial perturbation (possibly large) decaying at space infinity, the three-dimensional pyramidal traveling fronts are asymptotically stable in weighted L spaces on Rn(n4). Secondly, we show that under general bounded perturbations (even very small), the pyramidal traveling fronts are not asymptotically stable by constructing a solution which oscillates permanently between two three-dimensional pyramidal traveling fronts on R4.



    In this paper, we investigate the large time behavior of solutions to the following Cauchy problem:

    {ut(t,x)=Δu(t,x)+f(u(t,x)),xRn,t>0,u(0,x)=u0(x),xRn, (1)

    where nN, ut=ut and Δ is the standard Laplace operator with respect to the space variables xRn. For some constants ς[0,1] and ι[0,1], the nonlinear reaction term fC1+ς([ι,1+ι],R) satisfies

    (H1): f(0)=f(1)=0,f(0)0,f(1)<0,f(u)0foru(0,1).

    Such equations arise in various phenomena in population dynamics, combustion and chemistry ecology (see [1]), where u typically stands for the concentration of a species or the temperature.

    In what follows, we shall study the multidimensional stability of three-dimensional pyramidal traveling fronts to Eq. (1) in Rn with n4. In order to motivate our study, let us recall some known results in the study of traveling fronts of Eq. (1). In R, traveling fronts are solutions taking the form

    u(t,x)=ϕf(p),p=xcft,

    where cf0 is the propagation speed and ϕf is the wave profile satisfying

    {ϕf(p)+cfϕf(p)+f(ϕf(p))=0,ϕf(p)<0,pR,ϕf(+)=0,ϕf()=1. (2)

    Such solution u(t,x)=ϕf(xcft) are called the planar traveling front since its level set is a hyperplane. Throughout the paper, we further assume that

    (H2): There exists ϕf(p)C2(R) with speed cf>0 satisfying (2) and

    limp+ϕf(p)ϕf(p)=Λ<Λ10,

    where Λ and Λ1 are two real roots of the equation μ2+cfμ+f(0)=0.

    The equation (1) with assumptions (H1)-(H2) is called degenerate Fisher-KPP monostable and combustion equation. In fact, it follows from [1,10] that the assumptions (H1)-(H2) hold with cf being the minimal wave speed and the unique wave speed of planar traveling front ϕf when the nonlinear reaction term f is of degenerate Fisher-KPP monostable type and combustion type, respectively. See [2,19] for more details.

    In Rn with n2, the function ϕf(zcft) is clearly still the solution of Eq. (1) with x=(x,y,z)Rn2×R×R. A very interesting question is to consider the asymptotic stability of one-dimensional traveling front ϕf(zcft) in n(2)-dimensional spaces. For this problem, one can refer to [9,11,20,21,22] and the references therein to Allen-Cahn equation. It is worth to mention that Matano et al. [14,13] investigated the asymptotic stability of one-dimensional traveling front under any initial spatial decaying perturbations by using sub-super solutions method combining with the comparison principle. Motivated by [14,13], Lv and Wang [12] and Bu and Wang [4] established the multidimensional stability of planar traveling fronts to Eq. (1) with Fisher-KPP nonlinearity, non-KPP monostable and combustion nonlinearity, respectively. He and Wu [8] using spectral method studied the stability of traveling front for degenerate Fisher type equations.

    However, due to the influence of curvature and spatial dimension, there are other types of traveling fronts in Rn with n2 which are called non-planar traveling fronts, since their level sets are not hyperplanes anymore. Readers can see for instance Bu and Wang [2,3], Hamel et. al. [7,6], Ninomiya and Taniguchi [15], Taniguchi [17,18] and Wang and Bu [19] for the existence and stability of two-dimensional V-shaped fronts, three-dimensional pyramidal fronts and multidimensional conical shaped fronts. Noting that it is also very interesting to investigate the multidimensional stability of nonplanar traveling fronts. See Sheng et. al. [16] and Cheng and Yuan [5] for the multidimensional stability of two-dimensional V-shaped fronts and three-dimensional pyramidal to Allen-Cahn equation under any spatially decaying initial perturbations, respectively. Recently, Bu and Wang [4] also established the stability of two-dimensional V-shaped fronts in Rn with n>2 to degenerate Fisher-KPP monostbale and combustion equations.

    In Rn with n3, we write x=(x,y,z,s) with xRn3 and (y,z,s)R×R×R. It follows from Wang and Bu [19] that the equation (1) exists a three-dimensional pyramidal fronts with the form u(t,y,z,s)=V(y,z,ς) under the assumptions (H1)-(H2) in R3, where ς=sct. For simplicity, we still write V(y,z,ς) as V(y,z,s). Let lN with l3 and {θj}1jl satisfy

    0θ1<θ2<<θl<2πandmax1jl(θj+1θj)<π,

    where θl+1:=θ1+2π. Let m=c2c2fcf and

    h(y,z)=max1jlhj(y,z)=max1jlm(ycosθj+zsinθj)for(y,z)R2,

    where c>cf. Then {(y,z,s)R3|s=h(y,z)} is a 3-dimensional pyramid. Let Γ denote the set of all edges of a pyramid.

    Theorem I (see Wang and Bu [19]) Assume that (H1)-(H2) hold. For any c>cf, Eq. (1) admits a traveling front of pyramidal shape satisfying

    Vyy+Vzz+Vss+cVs+f(V)=0, (3)
    limγ+sup(y,z,s)D(γ)|V(y,z,s)ϕf(cfc(sh(y,z)))|ϕβf(cfc(sh(y,z)))=0,β(Λ1Λ,1),

    where D(γ)={(y,z,s)R3|dist((y,z,s),Γ)>γ}. Moreover, one has sV(y,z,s)<0 for (y,z,s)R3 and

    ϕf(cfc(sh(y,z)))<V(y,z,s)<1,(y,z,s)R3.

    It is obvious that the three-dimensional pyramidal front in Theorem I is also the solution to Eq. (1) in Rn with n>3. The aim of this paper is to study the multidimensional stability of three-dimensional pyramidal fronts V(y,z,s) in Rn with n>3. Motivated by [14,4], we mainly use the super-sub solutions method combining with the comparison principle. However, since we are treating degenerate Fisher-KPP monostable and combustion equations in Rn with n4, many modifications and techniques are needed.

    In the following, we use the moving coordinate with speed c toward the s direction. Let ˜s=sct and u(t,x,y,z,s)=ϑ(t,x,y,z,˜s). For simplicity, we still denote ϑ(t,x,y,z,˜s) by ϑ(t,x,y,z,s). Then the Eq. (1) can be rewritten as

    {ϑtΔϑcϑsf(ϑ)=0,t>0,(x,y,z,s)Rn3×R3,ϑ(0,x,y,z,s)=ϑ0(x,y,z,s),(x,y,z,s)Rn3×R3. (4)

    In the sequel, the solution to Eq. (4) is written as ϑ(t,x,y,z,s;ϑ0). The main results in the present paper are as follows.

    Theorem 1.1. Assume that (H1)(H2) hold. Suppose that the initial value ϑ0(x,y,z,s) is of class C(Rn,[0,1]) with n>3 and satisfies

    limR+sup|x|+|y|+|z|+|s|R|ϑ0(x,y,z,s)V(y,z,s)|ϕβf(cfc(sh(y,z)))=0

    for some β(Λ1Λ,1). Then the solution ϑ(x,y,z,s;ϑ0) to Eq. (4) satisfies

    limt+sup(x,y,z,s)Rn|ϑ(t,x,y,z,s;ϑ0)V(y,z,s)|ϕβf(cfc(sh(y,z)))=0. (5)

    The above theorem shows that under the initial perturbations decaying as |x|+|y|+|z|+|s|+, the three-dimensional pyramidal traveling fronts are asymptotically stable in weighted L spaces on Rn(n4). In particular, when the initial perturbations further belong to L1 in a certain sense, the convergence rate for (5) is algebraic, see the following theorem for more detail.

    Theorem 1.2. Suppose that (H1)(H2) hold and the initial value ϑ0(x,y,z,s) to Eq. (4) satisfies

    V(y,z,sϑ0(x))ϑ0(x,y,z,s)V(y,z,sϑ+0(x)) (6)

    for some smooth functions ϑ0,ϑ+0L1(Rn3)L(Rn3) with n>3. Then for any β(Λ1Λ,1), the solution ϑ(t,x,y,z,s;ϑ0) to Eq. (4) satisfies

    sup(x,y,z,s)Rn|ϑ(t,x,y,z,s;ϑ0)V(y,z,s)|ϕβf(cfc(sh(y,z)))Ctn32,t>0, (7)

    where C>0 is a constant depending on β,f,||ϑ0||L1(Rn3),||ϑ0||L(Rn3),||ϑ+0||L1(Rn3) and ||ϑ+0||L(Rn3).

    If the initial perturbations in Theorem 1.2 keep the sign, then we can obtain that the convergence rate (7) is optimal in some sense.

    Proposition 1. Let ϑ0 be as in (6) and assume that either ϑ00,ϑ00 or ϑ+00,ϑ+00. Then for any β(Λ1Λ,1), there exist constants D1>0 and D2>0 such that

    D1(1+t)n32sup(x,y,z,s)Rn|ϑ(t,x,y,z,s;ϑ0)V(y,z,s)|ϕβf(cfc(sh(y,z)))D2tn32,t>0.

    Finally, by constructing a solution to Eq. (1) which oscillates permanently between two pyramidal traveling fronts, we show that the three-dimensional pyramidal traveling fronts are not asymptotically stable under general bounded perturbations (even very small) on R4.

    Theorem 1.3. Let n=4. Assume that (H1)(H2) hold. Then for any β(Λ1Λ,1) and ˉδ>0, there exists a bounded function ω(x)C(R) with ||ω||L(R)=ˉδ such that the solution u(t,x,y,z,s) to Eq. (1) with the initial value u0(x,y,z,s)=V(y,z,sω(x)) satisfies

    limm+sup|x|m!1,(y,z,s)R3|u(tm,x,y,z,s)V(y,z,sctm+(1)mˉδ)|ϕβf(cfc(sctmh(y,z)))=0,

    where tm=m(m!)24.

    Remark 1. From the perspective of dynamical systems, the above result yields that in the weighted Lloc(R4), the ω-limit set of the solution u to Eq. (1) contains at least two distinct points. And each of them is a translation of the same three-dimensional pyramidal traveling front.

    We organize this paper as follows. In Section 2, we give some preliminaries including the properties of the pyramidal traveling fronts, some known results on the curvature flow problem and a mollified pyramid. In Section 3, we prove that the three-dimensional pyramidal traveling fronts are asymptotically stable in Rn(n4) by constructing new types of supersolutions and subsolutions coupled with comparison principle. That is, we prove Theorems 1.1-1.2 and Proposition 1. In Section 4, we prove Theorem 1.3 which states the existence of solution to Eq. (1) which oscillates permanently with non-decaying amplitude.

    In this section, we state some known results which play an important role in the proving of the main results. Throughout the paper, let

    γ1:=suppR|ϕ(p)ϕ(p)|,γ2:=suppR|ϕ(p)ϕ(p)|,γ3:=supu[ι,1+ι]|f(u)|

    and fix ι(0,ι2) such that for any u(12ι,1+ι),

    32f(1)<f(u)<12f(1).

    We now recall some known results on the curvature flow problem. See [14] for more details. The mean curvature flow for a graphical surface w(t,x) on Rn3 is given by the following Cauchy problem:

    {wt1+|w|2=div(w1+|w|2),xRn3,t>0,w(0,x)=w0(x),xRn3. (8)

    Assume that on Rn3, the first and second derivatives of w with respect to x are bounded, then by direct calculation, there exists a constant k>0 large enough such that

    0=wt1+|w|2div(w1+|w|2)=wtΔw+n3i,j=1wxiwxjwxixj1+|w|2wtΔwk|w|2.

    It is clear that w(t,x) is a subsolution of the following Cauchy problem:

    {v+t=Δv++k|v+|2,xRn3,t>0,v+(0,x)=w0(x),xRn3.

    Taking the Cole-Hopf transformation w+(t,x)=exp(kv+(t,x)), we have

    {w+t=Δw+,xRn3,t>0,w+(0,x)=exp(kw0(x)),xRn3.

    Thus we can obtain that

    v+(t,x)=1klog|Rn3Γ(t,xη)exp(kw0(η))dη|, (9)

    where

    Γ(t,η)=1(4πt)n32exp(|η|24t).

    Therefore (9) gives an upper estimate for the solution w(t,x) to the Cauchy problem (8). Similarly, the lower estimate for w(t,x) can be given by the Cauchy problem

    {vt=Δvk|v|2,xRn3,t>0,v(0,x)=w0(x),xRn3.

    That is,

    v(t,x)=1klog|Rn3Γ(t,xη)exp(kw0(η))dη|. (10)

    Let k>0 be any constant and v±(t,x) be solutions to the following Cauchy problems

    {v±t=Δv±±k|v±|2,xRn3,t>0,v±(0,x)=w0(x),xRn3.

    The following lemma gives the large time behavior of v±(t,x), see Lemma 2.4 of [14].

    Lemma 2.1. If the initial value w0C(Rn3) is bounded and satisfies lim|x||w0(x)|=0, then the solutions v±(t,x) satisfy

    limtsupxRn3|v±(t,x)|=0,

    respectively. If we further assume that w0L1(Rn3), then

    supxRn3|v±(t,x)|1k||exp(kw0)1||L1(Rn3)tn32,t>0.

    Similar to the proof of Lemma 3.2 of [16] and Lemma 2.2 of [14], we can obtain the following key estimates about the three-dimensional pyramidal fronts and planar fronts, respectively.

    Lemma 2.2. Let V(y,z,s) be a pyramidal front to Eq. (3). Then there exists a positive constant ˜k1 (depending on f) such that

    ˜k1Vs(y,z,s)Vss(y,z,s)˜k1Vs(y,z,s),(y,z,s)R3. (11)

    Lemma 2.3. Let ϕf(p) be a planar front to Eq. (2). There exists a constant ˜k2>0 depending only on f such that

    ˜k2ϕf(p)ϕf(p)˜k2ϕf(p),pR.

    Let ˜k=max{˜k1,˜k2}. Thus we have

    {˜kVs(y,z,s)Vss(y,z,s)˜kVs(y,z,s),(y,z,s)R3,˜kϕf(p)ϕf(p)˜kϕf(p),pR. (12)

    The following lemma shows some properties on three-dimensional pyramidal traveling fronts V.

    Lemma 2.4. ([19,Lemmas 3.3 and 3.4]) Let V(y,z,s) be a pyramidal front to Eq. (3). One has

    limR+sup|sh(y,z)|RVs(y,z,s)ϕβf(cfc(sh(y,z)))=0,β(Λ1Λ,1), (13)

    and

    infδV(y,z,s)1δVs(y,z,s)<0foranyδ(0,ι). (14)

    Remark 2. Obviously, (13) implies that

    A1:=sup(y,z,s)R3Vs(y,z,s)ϕβf(cfc(sh(y,z)))<+,A2:=sup(y,z,s)R3,θ[0,1]ϕβf(cfc(s+θh(y,z)))ϕβf(cfc(sh(y,z)))<+.

    Finally, we show a mollified pyramid, which was constructed by Taniguchi [17]. Let ˜ρ(r)C[0,) satisfy the following properties:

    ˜ρ(r)>0,˜ρr(r)0forr0,˜ρ(r)=1ifr>0issmallenough,˜ρ(r)=erifr>0islargeenough,sayr>R0,2π0r˜ρ(r)dr=1.

    Clearly, the function ρ(y,z):=˜ρ(y2+z2) is of class C and R2ρ(y,z)dydz=1. Without loss of generality, suppose R0>1. For all non-negative integers i1 and i2 with 0i1+i23, we have

    |Di1yDi2zρ(y,z)|Mρ(y,z),(y,z)R2,

    where M>0 is a constant. Define a mollified pyramid {(y,z,s)R3s=φ(y,z)} as φ(y,z):=ρh associated with a pyramid {(y,z,s)R3s=h(y,z)}. That is,

    φ(y,z)=R2ρ(yy,zz)h(y,z)dydz=R2ρ(y,z)h(yy,zz)dydz. (15)

    Let

    S(y,z):=c1+|φ(y,z)|2cf, (16)

    where φ(y,z):=(φy(y,z),φz(y,z)) and |φ(y,z)|=φ2y(y,z)+φ2z(y,z). Then we have the following two lemmas, see [17,18].

    Lemma 2.5. Let φ and S be as in (15) and (16), respectively. For any fixed integers i10 and i20, one has

    sup(y,z)R2|Di1yDi2zφ(y,z)|<K1forsomeconstantK1>0,h(y,z)<φ(y,z)h(y,z)+2πm0r2˜ρ(r)dr,|φ(y,z)|<m,0<S(y,z)ccf,(y,z)R2

    and

    limλsup{S(y,z)|(y,z)R2,dist((y,z),Γ)λ}=0,limλsup{φ(y,z)h(y,z)|(y,z)R2,dist((y,z),Γ)λ}=0.

    Lemma 2.6. There exist two positive constants ν1 and ν2 such that

    0<ν1=inf(y,z)R2φ(y,z)h(y,z)S(y,z)sup(y,z)R2φ(y,z)h(y,z)S(y,z)=ν2<.

    In addition, for integers i10 and i20 with 2i1+i23, there exists a constant K2>0 such that

    sup(y,z)R2|Di1yDi2zφ(y,z)S(y,z)|<K2

    and

    |φyy(y,z)|,|φzz(y,z)|mM,(y,z)R2.

    In this section, we give the proof of asymptotic stability of three-dimensional pyramidal fronts V(y,z,s) in Rn with n4 under perturbation that decay at space infinity by constructing supersolutions and subsolutions coupled with comparison principle. That is, we prove Theorems 1.1-1.2 and Proposition 1. In the following, the symbols Δx and x denote the n3-dimensional Laplacian and gradient operators with respect to x, respectively. Let K(μ):=μ2+cfμ+f(0). Clearly, we have K(βΛ)<0 for any β(Λ1Λ,1). Take λ:=min{116K(βΛ),116f(1),1}.

    Lemma 3.1. Let ˜k>0 be defined as in (12). Then for β(Λ1Λ,1) and υ(0,1) with υ<min{c48cfΛmMK(βΛ),1}, there exist some constants δ0>0 and σ0>0 such that for any δ(0,δ0] and σσ0, and any function v+(t,x) satisfying

    v+t=Δxv++˜k|xv+|2,xRn3,t>0, (17)

    the function defined by

    V+(t,x,y,z,s)=V(y,z,sv+(t,x)σδ(1eλt))+δeλtϕβf(cfc(sv+(t,x)σδ(1eλt)φ(υy,υz)υ))

    is a supersolution to Eq. (4) in (0,+)×Rn.

    Proof. Let ξ(t,x,y,z,s):=cfc(sv+(t,x)σδ(1eλt)φ(υy,υz)υ), Y:=υy, Z:=υz and η(t,x,s):=sv+(t,x)σδ(1eλt). Using (3), (12) and (17), the direct calculation implies that

    H[V+]:=V+tΔV+cV+sf(V+)=Vηv+tσδλeλtVηδλeλtϕβf(ξ)cfcδβeλtϕβ1f(ξ)ϕf(ξ)v+tcfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)Vηη|xv+|2+VηΔxv+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|xv+|2c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)|xv+|2+cfcδβeλtϕβ1f(ξ)ϕf(ξ)Δxv+Vyyc2fc2β(β1)δeλtϕβ2f(ξ)ϕf(ξ)2φ2Y(Y,Z)c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)φ2Y(Y,Z)+cfcυβδeλtϕβ1f(ξ)ϕf(ξ)φYY(Y,Z)Vzzc2fc2β(β1)δeλtϕβ2f(ξ)ϕf(ξ)φ2Z(Y,Z)c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)φ2Z(Y,Z)+cfcυβδeλtϕβ1f(ξ)ϕf(ξ)φZZ(Y,Z)
    Vssβ(β1)δeλtϕβ2f(ξ)ϕf(ξ)2c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)c2fc2cVscfβδeλtϕβ1f(ξ)ϕf(ξ)f(V(y,z,η)+δeλtϕβf(ξ))=σδλeλtVηδλeλtϕβf(ξ)cfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)+Vη(v+t+Δxv+)Vηη|xv+|2+cfcδβeλtϕβ1f(ξ)[(v+t+Δxv+)ϕf(ξ)cfc|xv+|2ϕf(ξ)]c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|xv+|2c2fc2β(β1)δeλtϕβ2f(ξ)ϕf(ξ)2[1+|φ(Y,Z)|2]c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)[1+|φ(Y,Z)|2]+cfcυβδeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)cfβδeλtϕβ1f(ξ)ϕf(ξ)f(V(y,z,η)+δeλtϕβf(ξ))+f(V(y,z,η))=σδλeλtVηδλeλtϕβf(ξ)cfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)+(˜kVηVηη)|xv+|2+cfcδβeλtϕβ1f(ξ)[(˜kϕf(ξ)cfcϕf(ξ))|xv+|2]c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|xv+|2c2fc2β(β1)δeλtϕβ2f(ξ)ϕf(ξ)2[1+|φ(Y,Z)|2]c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)[1+|φ(Y,Z)|2]+cfcυβδeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)cfβδeλtϕβ1f(ξ)ϕf(ξ)f(V(y,z,η)+δeλtϕβf(ξ))+f(V(y,z,η))σδλeλtVηδλeλtϕβf(ξ)cfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)c2fc2β(β1)δeλtϕβ2f(ξ)ϕf(ξ)2[1+|φ(Y,Z)|2]c2fc2βδeλtϕβ1f(ξ)ϕf(ξ)[1+|φ(Y,Z)|2]+cfcυβδeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)cfβδeλtϕβ1f(ξ)ϕf(ξ)δeλtϕβf(ξ)f(V(y,z,η)+θδeλtϕβf(ξ)),

    where ξ=ξ(t,x,y,z,s), η=η(t,x,s) and θ=θ(t,x,y,z,s)(0,1).

    Since limp+ϕf(p)ϕf(p)=Λ, then there exists a constant R1>0 large enough such that

    32Λϕf(p)ϕf(p)12Λ,(βϕf(p)ϕf(p))2+cfβϕf(p)ϕf(p)+f(0)<12K(βΛ),p>R1. (18)

    By limp+ϕf(p)ϕf(p)=Λ, limp+ϕf(p)ϕf(p)=Λ2 and |φ(Y,Z)|<m, one has

    limp+c2fc2[(ϕf(p)ϕf(p))2ϕf(p)ϕf(p)](1+|φ(Y,Z)|2)=0

    uniformly in (Y,Z)R2. Thus there exists a constant R2>0 large enough such that

    |c2fc2[(ϕf(p)ϕf(p))2ϕf(p)ϕf(p)](1+|φ(Y,Z)|2)|<116K(βΛ), (19)

    for p>R2 and (Y,Z)R2. The assumption (H1) implies that there exists a constant K>0 such that

    |f(u1)f(u2)|K|u1u2|ς,u1,u2[ι,1+ι]. (20)

    Since |sh(y,z)|+ gives dist((y,z,s),Γ)+, then by Theorem I and the fact that h(y,z)<φ(υy,υz)υh(y,z)+2πm0r2˜ρ(r)drυ, there exists a constant R3>0 large enough such that for all δ<ι,

    |f(V(y,z,η)+θδeλtϕβf(ξ))f(0)|K|V(y,z,η)+θδeλtϕβf(ξ)|ς<116K(βΛ) (21)

    for any (x,y,z,s)Rn and t0 with ξ(t,x,y,z,s)>R3.

    Since limpϕf(p)=1, limpϕf(p)=0 and limpϕf(p)=0, then there exists a constant R4>0 large enough such that

    |ϕf(p)ϕf(p)|<c|f(1)|16cfmMand|βϕf(p)ϕf(p)|<|f(1)|8,p<R4. (22)

    Since |sh(y,z)|+ gives dist((y,z,s),Γ)+, then it follows from Theorem I and the fact that h(y,z)<φ(υy,υz)υh(y,z)+2πm0r2˜ρ(r)drυ that there exists a R5>0 large enough such that for any (x,y,z,s)Rn and t>0 with ξ(t,x,y,z,s)<R5,

    V(y,z,η(t,x,s))>1ι. (23)

    Let R:=max{R1,R2,R3,R4,R5} and ˉR:=R+2πm0r2˜ρ(r)drυ. It follows from (14) that there exists a constant β1>0 such that

    minˉRξ(t,x,y,z,s)ˉR(Vs(y,z,η(t,x,s)))>β1. (24)

    Let δ0:=ι. Take σ0>0 large enough such that

    σλβ1λγ22mMγ1γ3>0,σσ0. (25)

    Note that c2fc2(1+|φ(Y,Z)|2)1 for any (Y,Z)R2.

    Case 1. For (t,x,y,z,s)[0,+)×Rn3×R3 with ξ(t,x,y,z,s)>R. Using Vη>0, ϕf(ξ)>0, (18)–(21) with δ(0,δ0) and σσ0, we have

    H[V+]:=V+tΔV+cV+sf(V+)δeλtϕβf(ξ)[λ+c2fc2β((ϕf(ξ)ϕf(ξ))2ϕf(ξ)ϕf(ξ))(1+(φ(Y,Z))2)+β2(ϕf(ξ)ϕf(ξ))2[1c2fc2(1+(φ(Y,Z))2)]β2(ϕf(ξ)ϕf(ξ))2cfβϕf(ξ)ϕf(ξ)f(0)+cfcυβϕf(ξ)ϕf(ξ)Δφ(Y,Z)f(V(y,z,η)+θδeλtϕβf(ξ))+f(0)]δeλtϕβf(ξ)[λ+K(βΛ)1612K(βΛ)+116K(βΛ)+116K(βΛ)]0.

    Case 2. For (t,x,y,z,s)[0,+)×Rn3×R3 with ξ(t,x,y,z,s)<R, using Vη>0, ϕf(ξ)>0, (22) and (23) with δ(0,δ0) and σσ0, we have

    H[V+]:=V+tΔV+cV+sf(V+)δeλtϕβf(ξ)[λc2fc2βϕf(ξ)ϕf(ξ)(1+(φ(Y,Z))2)+cfcυβϕf(ξ)ϕf(ξ)Δφ(Y,Z)f(V(y,z,η)+θδeλtϕβf(ξ))]δeλtϕβf(ξ)[λ+f(1)8+f(1)8f(1)2]0.

    Case 3. For (t,x,y,z,s)[0,+)×Rn3×R3 with Rξ(t,x,y,z,s)R, using ϕf(ξ)>0, (24) and (25) with δ(0,δ0) and σσ0, we have

    H[V+]:=V+tΔV+cV+sf(V+)σδλeλtVηδλeλtϕβf(ξ)c2fc2βδeλtϕβf(ξ)ϕf(ξ)ϕf(ξ)[1+(φ(Y,Z))2]+cfcυβδeλtϕβf(ξ)ϕf(ξ)ϕf(ξ)Δφ(Y,Z)δeλtϕβf(ξ)f(V(y,z,η)+θδeλtϕβf(ξ))δeλt[σλβ1λsupξR|ϕf(ξ)||ϕf(ξ)|sup(Y,Z)R2|Δφ(Y,Z)|supξR|ϕf(ξ)||ϕf(ξ)|supu[ι,1+ι]|f(u)|]δeλt(σλβ1λγ22mMγ1γ3)0.

    By the above argument, we get H[V+]0 for (t,x,y,z,s)[0,+)×Rn3×R3. Namely, the function V+(t,x,y,z,s) is a supersolution to Eq. (4) in [0,+)×Rn.

    Lemma 3.2. Let ˜k>0 be defined as in (12). Then for β(Λ1Λ,1) and υ(0,1) with υ<min{cK(βΛ)48ΛcfmM,1}, there exist some constants δ1>0 and σ1>0 such that, for any δ(0,δ1] and σσ1, and any bounded functions v(t,x) satisfying

    vt=Δxv˜k|xv|2,xRn3,t>0, (26)

    the function defined by

    V(t,x,y,z,s)=V(y,z,sv(t,x)+σδ(1eλt))δeλtϕβf(cfc(s+v(t,x)+σδ(1eλt)φ(υy,υz)υ))

    is a subsolution to Eq. (4) in (0,+)×Rn.

    Proof. Let ξ(t,x,y,z,s)=s+v(t,x)+σδ(1eλt)φ(υy,υz)υ, η(t,x,s)=sv(t,x)+σδ(1eλt) and Y=vy,Z=vz. Using (3), (12) and (26), the direct calculation yields that

    H[V]=VtΔVcVsf(V)=Vηvt+σδλeλtVη+δλeλtϕβf(ξ)cfcδβeλtϕβ1f(ξ)ϕf(ξ)vtσδ2cfce2λtλβϕβ1f(ξ)ϕf(ξ)Vηη|xv|2+VηΔxv+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|xv|2+c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)|xv|2+cfcδβeλtϕβ1f(ξ)ϕf(ξ)ΔxvVyyVzz+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|φ(Y,Z)|2+c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)|φ(Y,Z)|2cfcδβυeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)Vηη+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2+c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)cVη+cfδβeλtϕβ1f(ξ)ϕf(ξ)f(V(y,z,s,η)δeλtϕβf(ξ))=σδλeλtVη+δλeλtϕβf(ξ)cfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)+Vη(vt+Δxv)Vηη|xv|2+cfcδβeλtϕβ1f(ξ)[(vt+Δxv)ϕf(ξ)+cfc|xv|2ϕf(ξ)]VyyVzzVηηcVηf(V)+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2[1+|φ(Y,Z)|2]
    +c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)[1+|φ(Y,Z)|2]+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2|xv|2cfcδβυeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)+cfδβeλtϕβ1f(ξ)ϕf(ξ)f(V(y,z,η)δeλtϕβf(ξ))+f(V(y,z,η))σδλeλtVη+δλeλtϕβf(ξ)cfcβσδ2λe2λtϕβ1f(ξ)ϕf(ξ)+c2fc2δβ(β1)eλtϕβ2f(ξ)ϕf(ξ)2[1+|φ(Y,Z)|2]+c2fc2δβeλtϕβ1f(ξ)ϕf(ξ)[1+|φ(Y,Z)|2]cfcδβυeλtϕβ1f(ξ)ϕf(ξ)Δφ(Y,Z)+cfδβeλtϕβ1f(ξ)ϕf(ξ)+δeλtϕβf(ξ)f(V(y,z,η)θδeλtϕβf(ξ)),

    where ξ=ξ(t,x,y,z,s), η=η(t,x,s) and θ=θ(t,x,y,z,s)(0,1).

    It follows from the boundedness of functions v(t,x) that there exists a constant ˉM>0 such that ||v||L((0,+)×Rn3)ˉM. And thus, Theorem I, (20) and h(y,z)φ(υy,υz)υh(y,z)+2πm+0r2˜ρ(r)drυ imply that there exists ˜R3>0 large enough such that for all δ<ι,

    |f(V(y,z,η)θδeλtϕβf(ξ))f(0)|K|V(y,z,η)θδeλtϕβf(ξ)|ς<116k(βΛ), (27)

    for (t,x,y,z,s)[0,+)×Rn3×R3 with ξ(t,x,y,z,s)>˜R3+2ˉM. Theorem I and the fact that h(y,z)φ(υy,υz)υh(y,z)+2πm+0r2˜ρ(r)drυ imply that there exists ˜R4>0 large enough such that

    V(y,z,η)>1ι, (28)

    for (t,x,y,z,s)[0,+)×Rn3×R3 with ξ(t,x,y,z,s)<˜R42ˉM.

    Let R1,R2 and R4 be defined as in (18), (19) and (22). Put ˆR:=max{R1,R2,R4,˜R3+2ˉM,˜R4+2ˉM} and ˆR:=ˆR+2πm+0r2˜ρ(r)drυ, then (14) yields that there exists a constant β1>0 such that

    minˆRξ(t,x,y,z,s)ˆRVη(y,z,η)β1>0. (29)

    Take σ1>0 large enough such that

    σλβ12+λ+γ2+2γ1mM+γ3<0,forσσ1. (30)

    Let δ1:=min{ι,β12γ1,k(βΛ)16σλγ1,|f(1)|8σλγ1} for σσ1.

    Similar to the proof of Lemma 3.1, we can get H[V]0 on (t,x,y,z,s)[0,+)×Rn3×R3. That is, the function V(t,x,y,z,s) is a subsolution of Eq. (4) on (t,x,y,z,s)[0,+)×Rn3×R3.

    Lemma 3.3. Assume that the initial value ϑ0C(Rn,[0,1]) with n>3 satisfies

    limR+sup|x|+|y|+|z|+|s|R|ϑ0(x,y,z,s)V(y,z,s)|ϕβf(cfc(sh(y,z)))=0,

    for some β(Λ1Λ,1). Then for any fixed T>0, we have

    limR+sup|x|+|y|+|z|+|s|R|ϑ(T,x,y,z,s;ϑ0)V(y,z,s)|ϕβf(cfc(sh(y,z)))=0.

    Proof. The proof of Lemma 3.3 is similar to that of Lemma 3.8 in [4], so we omit it here.

    We are now in the position to prove Theorem 1.1. break

    Proof of Theorem 1.1. We only show the lower estimate, as the upper estimate can be given in a similar way. We denote ϑ(t,x,y,z,s;ϑ0) by ϑ(t,x,y,z,s) for simplicity. Take constants ˜k>0 as in (12), δ:=min{δ0,δ1} and σmax{σ0,σ1,1}, where σ0, δ0 and σ1, δ1 are defined as in Lemmas 3.1 and 3.2, respectively. For any ε>0, set ˆε=min{log22γ1,ε8A1,ϵ22eγ1D/2υ} such that ˆεσ(0,δ), where A1 is defined as in Remark 2, D:=2πm+0r2˜ρ(r)dr and υ defined as in Lemmas 3.1 and 3.2. Since the initial value v0C(Rn,[0,1]) with n>3 satisfies

    limR+sup|x|+|y|+|z|+|s|R|ϑ0(x,y,z,s)V(y,z,s)|ϕβf(cfc(sh(y,z)))=0

    for some β(Λ1Λ,1), then the strong maximum principle yields that

    0<ϑ(t,x,y,z,s)<1,(x,y,z,s)Rn3×R3 and t>0.

    From Lemma 3.3, it follows that for any fixed T>0, there exists a constant R1>0 large enough that

    sup|x|+|y|+|z|+|s|R1|ϑ(T,x,y,z,s)V(y,z,s)|ϕβf(cfc(sh(y,z)))ˆεσ.

    Thus we can choose a continuous function w0(x)0 satisfying lim|x|w0(x)=0 and

    ϑ(T,x,y,z,s)V(y,z,sw0(x))ˆεσϕβf(cfc(s+w0(x)h(y,z)))

    for any (x,y,z,s)Rn3×R3, and hence,

    ϑ(T,x,y,z,s)V(y,z,sw0(x))ˆεσϕβf(cfc(s+w0(x)φ(υy,υz)υ))

    for any (x,y,z,s)Rn3×R3 from h(y,z)φ(y,z) and ϕf(p)>0, where υ is defined as in Lemma 3.2. Let v(t,x) be the solution of Cauchy problem

    {vt=Δxv˜k|v|2,xRn3,t>0,v(0,x)=w0(x),xRn3.

    Then it follows from Lemma 2.1 that there exists T1>0 large enough such that

    ˆεv(t,x)0forallxRn3andtT1.

    Therefore, the comparison principle together with the subsolution constructed in Lemma 3.2 yields

    ϑ(t,x,y,z,s)V(y,z,sv(tT,x)+ˆε(1eλ(tT)))ˆεσeλ(tT)ϕβf(cfc(s+v(tT,x)+ˆε(1eλ(tT))φ(υy,υz)υ))V(y,z,sv(tT,x)+ˆε(1eλ(tT)))ˆεσeλ(tT)ϕβf(cfc(s+v(tT,x)φ(υy,υz)υ))

    for tT+T1 and (x,y,z,s)Rn3×R3. Thus

    ϑ(t,x,y,z,s)V(y,z,s)ϕβf(cfc(sh(y,z)))V(y,z,sv(tT,x)+ˆε(1eλ(tT)))V(y,z,s)ϕβf(cfc(sh(y,z)))ˆεσeλ(tT)ϕβf(cfc(s+v(tT,x)φ(υy,υz)υ))ϕβf(cfc(sh(y,z))).

    Let v_(t,x):=v(tT,x)ˆε(1eλ(tT)). One has that 2ˆεv_(t,x)0 for xRn3 and tT+T1. γ1:=suppR|ϕf(p)ϕf(p)| yields that ϕf(p+˜p)eγ1p is increasing in ˜pR for each pR. Thus we have

    V(y,z,sv_(t,x))V(y,z,s)ϕβf(cfc(sh(y,z)))=Vs(y,z,sθv_(t,x))ϕβf(cfc(sh(y,z)))v_(t,x)=Vs(y,z,sθv_(t,x))ϕβf(cfc(sθv_(t,x)h(y,z)))ϕβf(cfc(sθv_(t,x)h(y,z)))ϕβf(cfc(sh(y,z)))v_(t,x)2ˆεsup(y,z,s)R3Vs(y,z,s)ϕβf(cfc(sh(y,z)))sup(t,x,y,z,s)Ωϕβf(cfc(sθv_(t,x)h(y,z)))ϕβf(cfc(sh(y,z)))2ˆεA1e2γ1ˆε,

    where θ(0,1) and Ω:=[T+T1,+)×Rn. From the fact that h(y,z)φ(y,z)h(y,z)+D for all (y,z)R2, it follows that

    ϕβf(cfc(s+v(tT,x)φ(υy,υz)υ))ϕβf(cfc(sh(y,z)))ϕβf(cfc(sˆεh(y,z)Dυ))ϕβf(cfc(sh(y,z)))eγ1D2υeγ1ˆε.

    Combining the above argument, we have

    ϑ(t,x,y,z,s)V(y,z,s)ϕβf(cfc(sh(y,z)))2ˆεA1e2γ1ˆεˆεeγ1D2υeγ1ˆε>ε2ε2>ε,

    for tT+T1 and (x,y,z,s)Rn3×R3. This completes the proof of Theorem 1.1.

    In this subsection, we give the proofs of Theorem 1.2 and Proposition 1 motivated by [14,4]. We firstly construct a pair of subsolution and supersolution to Eq. (4). Let v±(t,x) be the solutions to the following Cauchy problem:

    {v±t(t,x)=Δxv±(t,x)±˜k1|xv±(t,x)|2,xRn3,t>0,v±(0,x)=v±0(x),xRn3, (31)

    where ˜k1 is the positive constant defined as in Lemma 2.2.

    Lemma 3.4. Let V(y,z,s) be a pyramidal front to Eq. (3) and ϑ(t,x,y,z,s;ϑ0) be the solution to Eq. (4). Assume that the initial value ϑ0(x,y,z,s) satisfies

    V(y,z,sv0(x))ϑ0(x,y,z,s)V(y,z,sv+0(x)),(x,y,z,s)Rn3×R3.

    Then we have

    V(y,z,sv(t,x))ϑ(t,x,y,z,s)V(y,z,sv+(t,x)) (32)

    for all t0 and (x,y,z,s)Rn3×R3.

    Proof. Let w+(t,x,y,z,s)=V(y,z,sv+(t,x)). Now we show that the function w+(t,x,y,z,s) satisfies

    H[w+]:=w+tΔw+cw+sf(w+)0,

    which yields that w+(t,x,y,z,s) is a supersolution to Eq. (4).

    Using (3), Lemma 2.2 and (31), the direct calculation yields that

    H[w+]=v+tVsΣn3j=1[v+xjxjVs+(v+xj)2Vss]VyyVzzVsscVsf(V)=v+tVs+Δxv+Vs|xv+|Vss=|xv+|(˜k1VsVss)0.

    Similarly, we can show that the function V(y,z,sv(t,x)) is a subsolution to Eq. (4). Therefore it follows from the assumptions on initial value ϑ0(x,y,z,s) and the comparison principle that the inequalities (32) hold. break

    Proof of Theorem 1.2. Denote ϑ(t,x,y,z,s;ϑ0) by ϑ(t,x,y,z,s) for simplicity. For any t0 and (x,y,z,s)Rn2×R3, it follows from Lemma 3.4 that we have

    V(y,z,sv(t,x))ϑ(t,x,y,z,s)V(y,z,sv+(t,x))

    with v+0(x)=ϑ+0(x) and v0(x)=ϑ0(x) in (31). Thus for any β(Λ1Λ,1), we have

    ϑ(t,x,y,z,s)V(y,z,s)ϕβf(cfc(sh(y,z)))V(y,z,sv+(t,x))V(y,z,s)ϕβf(cfc(sh(y,z)))=Vs(y,z,sθv+(t,x))ϕβf(cfc(sh(y,z)))v+(t,x),

    where θ(0,1). Since the smooth functions ϑ+0L1(Rn3)L(Rn3) with n>3, then (9) implies that v+(t,x) are bounded uniformly in t0 and (x,y,z,s)Rn2×R3. From (15), it follows that there exists a positive constant D depending on β such that

    Vs(y,z,sθv+(t,x))ϕβf(cfc(sh(y,z)))=Vs(y,z,sθv+(t,x))ϕβf(cfc(sθv+(t,x)h(y,z)))ϕβf(cfc(sθv+(t,x)h(y,z)))ϕβf(cfc(sh(y,z)))sup(y,z,s)R3Vs(y,z,s)ϕβf(cfc(sh(y,z)))sup(t,x,y,z,s)[0,+)×Rnϕβf(cfc(sθv+(t,x)h(y,z)))ϕβf(cfc(sh(y,z)))D.

    Lemma 2.1 implies that there exists a constant D+>0 (depending on β, f, ||ϑ+0||L1 and ||ϑ+0||L) such that

    ϑ(t,x,y,z,s)V(y,z,s)ϕβf(cfc(sh(y,z)))DsupxRn3|ϑ+0(t,x)|D+tn32.

    In a similar way, we can obtain that there exists a constant D>0 (depending on β, f, ||ϑ0||L1 and ||ϑ0||L) such that

    ϑ(t,x,y,z,s)V(y,z,s)ϕβf(cfc(sh(y,z)))Dtn32.

    Let C:=max{D,D+}. We complete the proof.

    Proof of Proposition 1. From Theorem 1.2 and (32), it suffices to prove that the solutions v±(t,x) to Eq. (31) with v±0(x)=ϑ±0(x) satisfy v+(t,0)D2tn32 and v(t,0)D1(1+t)n32 for some constants Di>0,i=1,2. In fact, from the first inequality of (32), it follows that for t0,

    ϑ(t,0,0,0,0;ϑ0)V(0,0,0)ϕβf(0)V(0,0,v(t,0))V(0,0,0)ϕβf(0)mins[||v||L,0]|Vs(0,0,s)ϕβf(0)v(t,0)|D1(1+t)n32,

    where 0=(0,,0)n3. From (10), the explicit expression of the solution v(t,x) to (31) is given by

    v(t,x)=1˜k1log(Rn3Γ(t,xη)exp(˜k1ϑ0(η))dη).

    Since ϑ00 and ϑ00, then there exist a positive constant ϱ>0 and a open set Θ in Rn3 such that ϑ0ϱ in Θ. Therefore

    v(t,x)1˜k1log(1ΘΓ(t,xη)(1exp(˜k1ϱ))dη)1˜k1log(1|Θ|(1exp(˜k1ϱ))minηΘΓ(t,xη))|Θ|˜k1(1exp(˜k1ϱ))minηΘΓ(t,xη).

    And hence, v(t,0)D1(1+t)n32 for t0. Similarly, we can prove that v+(t,0)D2tn32 for t0. We complete the proof.

    In this section, we show Theorem 1.3. That is, we prove the existence of solution to Eq. (1) which oscillates permanently with non-decaying amplitude. To prove our main result, we need construct a sequence of subsolutions and supersolutions pushing the solution forth and back in the s-direction by combining Lemma 3.4 and the following auxiliary lemma.

    Lemma 4.1. [Lemmas 3.1 and 3.2 of [14]] Let ˜k1 be defined as in Lemma 2.2 and v±(t,x) be solutions to the Eq. (31) with n=4. Suppose that initial values v±0(x) are all bounded on R and satisfy

    {v+0(x)ˉδ,xR,v+0(x)ˉδ,|x|[m!+1,(m+1)!1]

    and

    {v0(x)ˉδ,xR,v0(x)ˉδ,|x|[m!+1,(m+1)!1]

    for some constant ˉδ>0 and some integer m2, respectively. Then there exists a constant B>0 which depends on ˉδ and ˜k1 such that

    sup|x|m!1v+(T,x)ˉδ+B|ζ|[0,2/m][m,+)eζ2dζ

    and

    sup|x|m!1v(T,x)ˉδB|ζ|[0,2/m][m,+)eζ2dζ,

    respectively, where T=m(m!)24.

    Proof of Theorem 1.3. Let

    Ωm:=[m!+1,(m+1)!1],ˆΩm:=[0,m!][(m+1)!,+).

    Let {v±j(t,x)}j=1,2, be two sequences of solutions to the Cauchy problem (31) with smooth initial values v±0,j(x) satisfying |v±0,j(x)|ˉδ in xR,

    v+0,j(x)={ˉδ,|x|Ω2j,ˉδ,|x|ˆΩ2jandv0,j(x)={ˉδ,|x|Ω2j+1,ˉδ,|x|ˆΩ2j+1,

    respectively. By the above definitions of v±0,j(x), we can choose a function ωC(R) such that

    v0,j(x)ω(x)v+0,j(x)forallj1andxR.

    Let ϑ(t,x,y,z,s) be the solution to Eq. (4) with initial value ϑ(0,x,y,z,s)=V(y,z,sω(x)). Then we have

    V(y,z,s+ˉδ)V(y,z,svj(t,x))ϑ(t,x,y,z,s)V(y,z,sv+j(t,x))V(y,z,sˉδ)

    from the definition of ˆϑ(x) and Lemma 3.4. Thus it follows from Lemma 4.1 that we have

    sup|x|(2j+1)!1ϑ(t2j+1,x,y,z,s)V(y,z,sˉδ)ϕβf(cfc(sh(y,z)))sup|x|(2j+1)!1V(y,z,svj(t2j+1,x))V(y,z,sˉδ)ϕβf(cfc(sh(y,z)))sup(y,z,s)R3|Vs(y,z,sˉδ+θ1B|ζ|[0,2/2j+1][2j+1,+)eζ2dζ)|ϕβf(cfc(sh(y,z)))×B|ζ|[0,2/2j+1][2j+1,+)eζ2dζ,

    and

    sup|x|(2j)!1ϑ(t2j,x,y,z,s)V(y,z,s+ˉδ)ϕβf(cfc(sh(y,z)))sup|x|(2j)!1V(y,z,sv+j(t2j,x))V(y,z,s+ˉδ)ϕβf(cfc(sh(y,z)))sup(y,z,s)R3|Vs(y,z,s+ˉδ+θ2B|ζ|[0,2/2j][2j,+)eζ2dζ)|ϕβf(cfc(sh(y,z)))×B|ζ|[0,2/2j][2j,+)eζ2dζ,

    where θ1,θ2(0,1), t2j=(2j)((2j)!)24 and t2j+1=(2j+1)((2j+1)!)24. By (13), the above two inequalities yield that

    limj+sup(y,z,s)R3sup|x|(2j+1)!1ϑ(t2j+1,x,y,z,s)V(y,z,sˉδ)ϕβf(cfc(sh(y,z)))=0

    and

    limj+sup(y,z,s)R3sup|x|(2j)!1ϑ(t2j,x,y,z,s)V(y,z,s+ˉδ)ϕβf(cfc(sh(y,z)))=0.

    Then the conclusion of Theorem 1.3 can be obtained from the above two limits. We complete the proof.



    [1] Multidimensional nonlinear diffusions arising in population genetics. Adv. Math. (1978) 30: 33-76.
    [2] Stability of pyramidal traveling fronts in the degenerate monostable and combustion equations I. Discrete Contin. Dyn. Syst. (2017) 37: 2395-2430.
    [3] Global stability of V-shaped traveling fronts in combustion and degenerate monostable equations. Discrete Contin. Dyn. Syst. (2018) 38: 2251-2286.
    [4] Z.-H. Bu and Z.-C. Wang, Multidimensional stability of traveling fronts in combustion and non-KPP monostable equations, Z. Angew. Math. Phys., 69 (2018), Paper No. 12, 27 pp. doi: 10.1007/s00033-017-0906-5
    [5] Multidimensional stability of disturbed pyramidal traveling fronts in the Allen-Cahn equation. Discrete Contin. Dyn. Syst. Ser. B (2015) 20: 1015-1029.
    [6] Solutions of semilinear elliptic equations in RN with conical-shaped level sets. Comm. Partial Differential Equations (2000) 25: 769-819.
    [7] Existence and qualitative properties of multidimensional conical bistable fronts. Discrete Contin. Dyn. Syst. (2005) 13: 1069-1096.
    [8] Spatial decay and stability of traveling fronts for degenerate Fisher type equations in cylinder. J. Differential Equations (2018) 265: 5066-5114.
    [9] Multidimensional stability of planar traveling waves. Trans. Amer. Math. Soc. (1997) 349: 257-269.
    [10] The evolution of reaction-diffusion waves in a class of scalar reaction-diffusion equations: algebraic decay rates. Phys. D (2002) 167: 153-182.
    [11] Multidimensional stability of traveling waves in a bistable reaction-diffusion equation II. Comm. Partial Differential Equations (1992) 17: 1901-1924.
    [12] Stability of planar waves in mono-stable reaction-diffusion equations. Proc. Amer. Math. Soc. (2011) 139: 3611-3621.
    [13] Large time behavior of disturbed planar fronts in the Allen-Cahn equation. J. Differential Equations (2011) 251: 3522-3557.
    [14] Stability of planar waves in the Allen-Cahn equation. Comm. Partial Differential Equations (2009) 34: 976-1002.
    [15] Existence and global stability of traveling curved fronts in the Allen-Cahn equations. J. Differential Equations (2005) 213: 204-233.
    [16] Multidimensional stability of V-shaped traveling fronts in the Allen-Cahn equation. Sci. China Math. (2013) 56: 1969-1982.
    [17] Traveling fronts of pyramidal shapes in the Allen-Cahn equations. SIAM J. Math. Anal. (2007) 39: 319-344.
    [18] The uniqueness and asymptotic stability of pyramidal traveling fronts in the Allen-Cahn equations. J. Differential Equations (2009) 246: 2103-2130.
    [19] Nonplanar traveling fronts in reaction-diffusion equations with combustion and degenerate Fisher-KPP nonlinearities. J. Differential Equations (2016) 260: 6405-6450.
    [20] Multidimensional stability of traveling waves in a bistable reaction-diffusion equation I. Comm. Partial Differential Equations (1992) 17: 1889-1899.
    [21] Multidimensional stability of traveling fronts in monostable reaction-diffusion equations with complex perturbations. Sci. China Math. (2014) 57: 353-366.
    [22] Stability of planar travelling waves for bistable reaction-diffusion equations in multiple dimensions. Appl. Anal. (2014) 93: 653-664.
  • This article has been cited by:

    1. 卓 陈, A Review of the Existence, Uniqueness, Stability, and Asymptotic Behavior of Traveling Wave Solutions in Reaction Diffusion Equations, 2024, 14, 2160-7583, 90, 10.12677/pm.2024.146230
  • Reader Comments
  • © 2021 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(2224) PDF downloads(134) Cited by(1)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog