Volume 2025, Issue 1 5781898
Research Article
Open Access

Asymptotic Lower Bound on the Spatial Analyticity Radius for Solutions of the Periodic Fifth Order KdV–BBM Equation

Tegegne Getachew

Corresponding Author

Tegegne Getachew

Department of Mathematics , Mekdela Amba University , Tulu Awulia , Amhara, Ethiopia , mkau.edu.et

Search for more papers by this author
First published: 27 April 2025
Academic Editor: Giovanni P. Galdi

Abstract

In this work, consideration is given to the initial value problem associated with the periodic fifth-order KdV–BBM equation. It is shown that the uniform radius of spatial analyticity σ(t) of solution at time t is bounded from below by ct−2/3 (for some c > 0), given initial data η0 that is analytic on the circle and has a uniform radius of spatial analyticity σ0. The proof of our main theorems is based on a contraction mapping argument, a method of approximate conservation law in a modified Gevrey spaces, Hölder’s inequality, Sobolev algebra, Cauchy–Schwartz inequality, and Sobolev embedding.

1. Introduction

We consider the initial value problem (IVP) for the fifth order Korteweg-de Vries (KdV)–Benjamin–Bona–Mahony (BBM) type equation on the circle ,
()
where the parameters γ1, γ2, δ1, δ2,  and γ are constants that satisfy certain constraints; see [13] for more details.

The IVPs for some dispersive equations have plenty of significant issues under the periodic setting compared to nonperiodic IVPs. The problem that makes a difference from the nonperiodic IVP is the presence of nontrivial resonance. In what follows, the fifth-order KdV–BBM equation (1) describes the unidirectional propagation of water waves, and this was introduced by Bona et al. [1] by using the second-order approximation in the two-way model, the so-called abcd-system derived in [4, 5].

If the parameter γ = 7/48, then sufficiently smooth solutions for the IVP (1) enjoys the energy conservation
()

Local well-posedness of the IVP (1) with data in the Sobolev spaces for s ≥ 1 was established by Carvajal et al. [2]. Moreover, for γ1, δ1 > 0 and γ = 7/48, the authors used the conserved energy quantity to prove global well-posedness of (1) for data in , s ≥ 2. Furthermore, the authors used the method of splitting argument to improve the global well-posedness result in for 1 ≤ s < 2. In addition to this, the authors also showed that the IVP (1) is ill-posed in the sense that the flow-map that takes initial data to solution can not be continuous for given data in . In line with this, the decay rate for (1) on the real line has been studied in [3, 6, 7].

As far as the author knows, the decay rate for (1) with initial data η0(x) that is analytic on the circle having a uniform radius of spatial analyticity σ0 > 0 has not yet been studied. Thus, in this work, we are interested in studying the property of spatial analyticity for the solution η(x, t) to the IVP (1), given initial data η0(x) that are real analytic in a strip of width 2σ0 around the x-axis of the complex plane. The question is then whether this property persists for all later times t, but with a possibly smaller and shrinking radius of analyticity σ(t) > 0, i.e., it is the solution u(x, t) of (1) analytic in Sσ(t for all t? For short times, it is shown that the radius of analyticity remains at least as large as the initial radius, i.e., one can take σ(t) = σ0. For large times on the other hand, we use the idea introduced in [8] (see also [9]) to show that σ(t) can decay no faster than t−2/3 as t. For this purpose, we need the modified Gevrey space , for σ > 0 and which is endowed with the norm
()
where 〈k2 = 1 + k2 and denotes the spatial Fourier transform of f given by
()
In fact, the space was introduced by Dufera et al. [10] from the Gevrey space by replacing the exponential weight eσ|k| with the hyperbolic weight cos h(σ|k|). Here, the Gevrey space is a Hilbert space of periodic functions endowed with the norm
()

It is clear that G0,s = H0,s = Hs, where denotes the L2-based Sobolev space of order s.

Since 1/2eσ|k| ≤ cos h(σ|k|) ≤ eσ|k|, the and -norms are equivalent in the sense that
()

The interest in the space is due to (6), and the Paley–Wiener Theorem [11], which states that for σ > 0 and s ≥ 0, a 2π-periodic function f belongs to if and only if f is the restriction to the real line of a function F, which is holomorphic in the strip and satisfies the bound .

Information about the domain of analyticity of a solution to a partial differential equation (PDE) can be used to gain a quantitative understanding of the structure of the equation, and to obtain insight into underlying physical processes. The idea of obtaining spatial analyticity was introduced by Kato and Masuda [12]. There are various studies concerning the properties of spatial analyticity of solutions for a large class of nonlinear partial differential equations. For instance, parabolic equations [13], Navier–Stoke equation [14], cubic Szegő equation [15], periodic generalized KdV equation [16,17], generalized Euler equation [18], second-order analytic nonlinear differential equations [19], semilinear Klein–Gordon equations [20] and the references therein.

Rapid progress has been made lately in obtaining an algebraic decay rate of the radius, i.e., σ(t) ~ tα for some α ≥ 1, to various nonlinear dispersive PDEs. For example, Himonas et al. in [21] studied the persistence of spatial analyticity for periodic solutions of the dispersion-generalized (g) KdV equation for β ≥ 2. For a class of analytic initial data with a uniform radius of analyticity σ0 > 0, the authors obtained an asymptotic lower bound σ(t) ≥ ctp on the uniform radius of analyticity σ(t) at time t, as t, where p = max{1, 4/β}. Moreover, Tesfahun [22] showed that the uniform radius of spatial analyticity σ(t) of solutions u(t) at time t to the KdV equation cannot decay faster than t−4/3 as t given initial data that is analytic with fixed radius σ0 > 0. Furthermore, in a paper [23], Himonas and Petronilho studied the IVP of the Benjamin–Bona–Mahony (BBM) equation with initial data that are analytic on the torus and have a uniform radius of analyticity σ0 > 0. In their work, they examined the evolution of the radius of spatial analyticity σ(t) of the solution u(t) at any future time t. The authors showed that the size of the radius of spatial analyticity persists for some time and after that, it evolves in such a way that its size at any time t is bounded below by ct−1 for some c > 0. A similar lower bound of spatial analyticity was obtained in [24, 25] for the Schrödinger equation with cubic nonlinearity on both the real line and the circle. The method used in these papers was first introduced by Selberg and Tesfahun [8] in the context of the 1d Dirac–Klein–Gordon equations, which is based on approximate conservation laws in a Gevrey space and Bourgan’s Fourier restriction method.

In an attempt to improve the decay rate obtained so far, the use of approximate conservation law in the modified Gevrey space can yield a decay rate of order t−1/(2α) for some 0 < α ≤ 1. For instance, in the paper [10], Dufera et al. obtained an asymptotic lower bound on the uniform radius of analyticity σ(t) of the solution u(·, t) of the Beam equation as t. Moreover, Esfahani and Tesfahun [26] showed persistence of spatial analyticity of solutions for the sixth-order Boussinesq equation with cubic nonlinearity, and they obtained a decay rate of order t−1/2 as t. The similar lower bound of spatial analyticity has been obtained in [6, 9, 27] for the fourth-order Schrödinger equation with cubic nonlinearity, fifth-order KdV–BBM equation and Kawahara equation on the real line. Furthermore, in the paper [28], Getachew shows that the radius of spatial analyticity σ(t) of the solution u(t) for the BBM equation on the circle does not decay faster than c|t|−2/3 (for some constant c > 0) as |t|⟶. Such a decay rate is obtained in the view of the inequality
()
which follows from an interpolation of .
We remark that the modified Gevrey spaces satisfy the embeddings
()
for any , 0 ≤ σ < σ and some constant C > 0. In particular, for σ = 0, and all , σ > 0, we have
()

As a consequence of property (9) and the existing well-posedness theory in (see [2]), we conclude that the IVP (1) with initial data η0 in for all σ0 > 0 and s ≥ 1 have a unique and global in time solution provide that γ1, δ1 > 0 and γ = 7/48.

The main result in this paper is as follows.

Theorem 1. Assume γ1, δ1 > 0. Let s ≥ 1, γ = 7/48, and for σ0 > 0. Then, the global solution η of (1) satisfies

()
with asymptotic lower bound
()
where c > 0 is a constant depending on the initial data norm .

Notation: throughout this paper, we use C to denote generic constant that may vary at each occurrence. For any positive quantities a and b, we use ab to indicate an estimate of the form aCb. Moreover, we write a ~ b if ab and ba. Furthermore, we denote by a+ the quantity a + ε for any ε > 0 and by Dα the differential operator given by .

2. Local Well-Posedness in Hσ,s

In this section, we prove the local well-posedness of the IVP (1) in the space Hσ,s for s ≥ 1 and σ > 0, using multilinear estimates in combination with a contraction mapping argument. In fact, we outline the argument in [3] that enables the authors to obtain the local well-posedness result for the IVP associated to the fifth oder KdV–BBM model on the real line in Gσ,s with s ≥ 1, σ > 0.

Taking the spatial Fourier transform of (1), we obtain
()
where
()
with
()
Defining the Fourier multipliers
()
In view of (12), the IVP (1) can be rewritten as in the following form:
()
where
()
Then, due to the Duhamel’s integral formula the integral equation for (16) and (17) is given by
()

Lemma 1 (see [23], Lemma 2.1, [2], Proposition 2.1.)Let with σ ≥ 0, and s ≥ 0. Then, we have

()
where ω(D) is the Fourier multiplier operator defined by with
()

Lemma 2. Let τ(D) be the operator given in (6), σ > 0, and s ≥ 0. Then, we have

()

Proof 1. Since γ1, δ1 > 0, there exists a constant C > 0 such that τ(k) ≤ Cω(k) for . Using this fact, the definition of Gσ,s-norm and the estimate in Lemma 1, we obtain

()

Lemma 3. Let ψ(D) be the operator given in (6), σ > 0, and s > 1/2. Then, we have

()

Proof 2. Note that

()
for all . Let
()

Now, using the definition of Gσ,s-norm, (24) and (25), triangle inequality, Parseval’s identity, the convolution (see [29], pp. 381),
()
From Lemma 1 and Sobolev algebra, we obtain
()

Lemma 4. Let ψ(D) be the operator given in (6), σ > 0, and s ≥ 1. Then, we have

()

Proof 3. Using Lemma 1 with s − 1 ≥ 0, we have

()

Combining the inequalities found in Lemmas 24, we obtain the following nonlinear estimate.

Lemma 5. Let F(η) be given as in (8). Then, for any s ≥ 1 and σ > 0, we have

()

Now, applying the contraction mapping argument to the integral equation (9) and using Lemma 5 yields the following local result.

Theorem 2. Let η0Hσ,s for s ≥ 1 and σ > 0. Then, there exists a unique solution

()
of the IVP (1) with existence time
()

Moreover,
()
Here, we use the notation
()

3. Approximate Conservation Law

Let us fix γ1, δ1 > 0 and γ = 7/48. By applying the operator cosh(σ|k|) to both sides of equation (1), the function v(x, t) = cos h(σ|k|)η(x, t) satisfies the equation
()
where
()
with
()
Define a modified energy functional associated with the function v by
()
For σ = 0, we have v = η. Since E(t) = E(0), we have the conservation
()
However, this fails to hold for σ > 0. Nevertheless, we will prove an approximate conservation law by establishing growth estimate. To do this, we differentiate the modified energy functional, and then using (35) and (36) and integration by parts, and obtain
()
Consequently, integration in time gives
()

Lemma 6. For all vH2, we have

()

Proof 4. Using (36) and (37), Plancherel’s Theorem and Cauchy–Schwartz inequality, we get

()

Then, estimate (42) reduces to proving the following estimates:
()
()
()
where α ∈ (0, 1] is any arbitrary constant.

To prove the estimates (44)–(46), we recall the following estimate from ([10], Lemma 3).

Lemma 7. Let p > 1 be any integer and such that |ξ1| ≤ |ξ2| ≤ ⋯≤|ξp| and . Let . Then,

()

From Lemma 7, one can deduce that
()
Then, for any α ∈ [0, 1], the interpolation between (47) and (48) yields
()
Proof of (44): by taking the Fourier transform, we write
()
where M(σ|ξ|) be given as in Lemma 7.
By symmetry, we may assume |k1| ≤ |k2|. Now, let
()
Then, for any choice of α ∈ (0, 1], applying (26) and (49) gives
()
Therefore, using Parseval’s identity, Hölder’s inequality, and Sobolev embedding, we obtain
()
Proof of (45): by taking the Fourier transform, we write
()
where M(σ|ξ|) be as in (49). Then, using (49) for α = 3/4, we get
()
By Parseval’s identity, Hölder’ inequality, and Sobolev embedding, we obtain
()
Proof of (46): similarly, by taking the Fourier transform, we write
()
where M(σ|k|) be given as in (49).
By symmetry, we may assume |k1| ≤ |k2| ≤ |k3|. Then, for any choice of α ∈ (0, 1], by using (26) and (49), we obtain
()
Therefore, using Parseval’s identity, Hölder’s inequality, and Sobolev embedding, we obtain
()

Theorem 3. (Approximate conservation law). Let η0Hσ,2 for all σ > 0, and η be the local-in-time solution to the IVP (1) that is constructed in Theorem 2. Then,

()

Proof 5. From (15), we have the bound

()
where T is as in (32).

On the other hand, we have
()

Then, plugging (61) and (62) into (42) and combining it with (41) yields the desired estimate (60).

4. Proof of Theorem 1

We closely follow the argument presented in [24, 30]. First, we consider the case s = 2. Then, the general case, s ≥ 1, will essentially reduce to s = 2 as shown in the next subsection.

4.1. The Case: s = 2

Suppose that for some σ0 > 0. From the local theory there is a unique solution
()
of (1) constructed in Theorem 2 with existence time T as in (32).
Since
()
we have
()
Now, following the argument in [24, 30], we can construct a solution on [0, T0] for arbitrarily large time T0 by applying the approximate conservation law in Theorem 3 so as to repeat the above local result on successive short time intervals of size T to reach T0 by adjusting the strip width parameter σ according to the size of T0. Doing so, we establish the bound
()
for σ satisfying
()
Thus, Eσ(t) < for t ∈ [0, T0], which in turn implies
()

4.2. The General Case: s ≥ 1

For any , we apply the Gevrey embedding inequality in (8) to get
()
From the local theory, there is a such that
()
and
()
where a > 0 is a constant depending on and σ0. Applying the Gevrey embedding inequality in (8), we now conclude that
()
and
()

This completes the proof of Theorem 1.

Conflicts of Interest

The author declares no conflicts of interest.

Funding

The author received no financial support for this manuscript.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

    The full text of this article hosted at iucr.org is unavailable due to technical difficulties.