Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Well-posedness of a highly nonlinear shallow water equation on the circle

Well-posedness of a highly nonlinear shallow water equation on the circle WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION ON THE CIRCLE NILAY DURUK MUTLUBAS Faculty of Engineering and Natural Sciences, Sabanci University, Turkey ANNA GEYER Delft Institute of Applied Mathematics, Faculty Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, The Netherlands. RONALD QUIRCHMAYR Department of Mathematics, KTH Royal Institute of Technology, Sweden. Abstract. We present a comprehensive introduction and overview of a recently derived model equation for waves of large amplitude in the context of shallow water waves and provide a literature review of all the available studies on this equation. Furthermore, we establish a novel result concerning the local well- posedness of the corresponding Cauchy problem for space-periodic solutions with initial data from the Sobolev space H on the circle for s > 3=2. Mathematics Subject Classi cation: Primary: 35Q35. Secondary: 35L30. Keywords: Shallow water equation, large amplitude waves, cubic nonlinearity, well-posedness. 1. Introduction and overview This paper addresses the local well-posedness of the model equation 3" u + u + uu (4u + 7u ) t x x xxx xxt 2 18 (1.1) 2 2 2 " " 2 3 = 2u u + uu 398uu u + 45u u + 154u x xx xxx x xx xxx 6 96 considered on the unit circle with initial data from the Sobolev space H for s > 3=2. Equation (1.1) was derived in [37] from the classical water wave problem for free surface gravity water waves over a at bottom, where the underlying incompressible ow is governed by Euler's equation. It serves as an asymptotic model for the horizontal velocity component of a unidirectional shallow water wave of large amplitude at a speci c depth. E-mail addresses: nilay.duruk@sabanciuniv.edu, a.geyer@tudelft.nl, ronaldq@kth.se. arXiv:2011.01290v1 [math.AP] 2 Nov 2020 2 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR Scalar model equations are e ective tools for studying the propagation of unidirec- tional water waves. In contrast to general solutions for the full governing equations, solutions of model equations are analytically (and numerically) tractable and provide accurate approximations for certain classes of waves, enabling an in-depth analysis of complex nonlinear wave phenomena. One of the most prominent equations in the class of shallow water wave models is the Korteweg-de Vries equation (KdV) 2" u + u + uu + u = 0; t x x xxx 3 6 here " and  are the dimensionless amplitude and shallowness parameter, see [9, 29, 28]. KdV was rst considered by Boussinesq [2] in 1877, before Korteweg and de Vries [32] derived it in 1895. It exhibits a balance between nonlinear and dispersive e ects enabling the formation of solitary waves, which where rst observed in a narrow canal by Russell [38]. Such waves preserve their shape while propagating at constant speed and can not be described by the linear dispersive water wave equation (formally obtained by setting " = 0 in the above equation), whose solutions decrease in height and disperse as they propagate. KdV is an accurate approximation for shallow water waves of small amplitude, i.e.   1 and " = O( ), which propagate in one direction. It forms a completely integrable bi-Hamiltonian system, and the existence of a Lax pair enables the application of nonlinear Fourier transform techniques to solve the Cauchy problem exactly for large classes of initial data and to determine the long-time behavior of the solution. Moreover, there exists a single global canonical coordinate system for space-periodic solutions (so-called Birkho coordinates) in which the time evolution becomes linear [30]. KdV is the simplest water wave model whose solitary waves are solitons (see the discussion in [13]), i.e. these localized waves preserve their shape after interacting with other waves of this type in conjunction with a phase shift. KdV has global solutions in time for a very large class of initial data, cf. [39], covering all physically relevant initial states (see the discussion in [4]). In particular KdV is not capable of describing breaking waves |waves that remain bounded whereas their slope becomes in nite in nite time. KdV belongs to a wider class of weakly nonlinear dispersive equations, the BBM equations [1] 2" u + u + uu +  ( u + u ) = 0; t x x xxx xxt with  0 (to avoid linear ill-posedness) and = + . Since this one-parameter family of equations does not account correctly for waves whose behavior is more nonlinear than dispersive, one is forced to consider more general regimes to include breaking waves, see [4]. A natural regime for describing larger waves is the shallow water regime for waves of moderate amplitude, i.e.   1 and " = O(). It contains the following two-parameter family of equations, which accurately approximate the water wave problem (cf. [10, 33]): 3" 2 2 u + u + uu +  ( u + u ) = " ( uu + u u ): t x x xxx xxt xxx x xx We refer to [33] for a comprehensive approach on the justi cation of KdV as an accurate shallow water approximation for the water wave problem. WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 3 The function u = uj describes the evolution of the horizontal velocity component of the ow eld at a xed dimensionless depth z 2 [0; 1]; the second parameter p 2 R can be chosen such that the coecients ; ; ;  satisfy 1 3 1 3 9 23 3 = p + ; = p + ; = p ;  = p ; 6 2 6 2 2 24 2 1 1 where  = (z ). It turns out that this family contains models that capture 2 3 wave breaking. The most prominent examples are the Camassa-Holm equation (CH) U + U + 3UU U = 2U U + UU ; t x x xxt x xx xxx and the Degasperis-Procesi equation (DP) U + U ^ + 4UU U = 3U U + UU : t x x xxt x xx xxx By means of a suitable scaling of U (provided that ;  ^ 6= 0) one can recover that both CH and DP belong to the before mentioned two-parameter family, cf. [28, 10]; they generalize KdV/BBM in the sense that they degenerate to KdV/BBM when " = O( ). CH was derived in [3] as an integrable shallow water wave model with peaked solitons; it was earlier derived in [21] as a novel bi-Hamiltonian equation. DP was rst considered in [14]. CH and DP share with KdV the rare property that they form completely integrable bi-Hamiltonian systems|in fact they are the only integrable equations of the before mentioned family, cf. [26, 27]. We refer to [11] and [8] for rigorous approaches on integrability of CH and DP, respectively. The classes of solutions of CH and DP are richer than the waves described by KdV: apart from capturing the phenomenon of wave breaking [5, 6, 7, 19], there is another fact that distinguishes CH and DP from KdV: they possess singular traveling wave solutions such as periodic and solitary waves with peaked and cusped crests, and their solitons are peaked [34, 35, 5, 12]. The model equation for the free surface  of moderate amplitude shallow water waves that corresponds to CH and DP|we denote it by SE|is given by 2 2 2 3" 3" +  +    + t x x x xxx xxt 2 8 12 12 3 2 3" 7" =   2  +  : x x xx xxx 16 24 It initially appeared in [28] in Johnson's re-derivation of CH directly from the governing equations, and was later studied in various papers. We refer to [10] for a rigorous justi cation of SE as an accurate shallow water approximation for the water wave problem (besides CH and DP), a local well-posedness result for initial data in H , s > 5=2, as well as wave breaking results. An improvement of local well-posedness in terms of lower Sobolev exponents can be found in [15]; we refer to [16] for an analogue well-posedness result on the periodic domain and wave breaking results for periodic solutions. Moreover, it was proved in [18] that the data-to-solution map is not uniformly continuous. SE has smooth solitary traveling wave solutions [23], which are orbitally stable [17], as well as smooth periodic waves and a rich collection of singular traveling waves [22]. Moreover, symmetric solutions of this equations must be traveling waves [24], adding evidence in support of the strong correlation between symmetry and traveling waves often observed in water wave theory. Equation (1.1) was derived to account for even larger waves than those captured by CH and DP (and SE) by appealing to the approach presented in [28], where 4 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR Johnson re-derived CH directly from the governing equations for gravity water waves by means of double asymptotic expansions in the wave parameters  and " of the velocity eld, the pressure and the free surface. This ansatz makes it possible to treat the amplitude and shallowness parameters independently over the course of the derivation. After determining suciently many terms in the asymptotic expansion, one may freely choose a speci c wave regime to discover the corresponding asymptotic models. Equation (1.1) lives in the regime which is determined by the relation " = O( ) and   1. This means that the amplitude parameter can| unlike in the CH regime|exceed the size of the shallowness parameter, which brings 2 2 in the additional highly nonlinear "  -terms when compared to CH and DP; for this reason we speak about a shallow water equation for waves of large amplitude. Equation (1.1) belongs to a two parameter family of shallow water equations of large amplitude: one of the free parameters determines the xed depth of the water, whereas the second parameter stems from the \BBM-trick" to include the u -term xxt in (1.1). The latter parameter was chosen such that the coecients of the quadratic third order terms in (1.1) satisfy the ratio 2 : 1 as in the case for CH, whereas the 2 2 particular choice for the dimensionless water depth entails that the "  -term can be written as an x-derivative, which structurally distinguishes (1.1) from all the other equations from the same family. Equation (1.1), as well as the entire two-parameter family describing the horizontal velocity component at any depth associated with (1.1), and the corresponding one- parameter family of equations for the evolution of the free surface elevation are locally well-posed on the real line in the classical H (R)-setting. This was shown in [37] with the help of Kato's semigroup approach for quasilinear evolution equations (see [31]), where also a blow-up criterion for solutions u of (1.1) in H (R) was established by means of the blow-up of ku k 1 . Since the appearance of this initial x L study, several papers by di erent authors addressed the local well-posedness of (1.1) in more general settings, re ned estimates for blow-up, proved persistence of certain asymptotic properties of the solution; furthermore weak traveling waves were fully classi ed. We sketch this progress in the following. In [40], the authors exploited the aforementioned structural properties of (1.1) to reformulate it as a nonlocal rst order equation to show local well-posedness of the corresponding Cauchy problem with initial data in H (R) for s > 3=2 (where 3=2 is the sharp lower bound) by applying Kato's theory. Furthermore, they proved an asymptotic persistence property for H (R) solutions|certain decay rates of the initial wave pro le are preserved under the time evolution. Even though the proofs leave much to be desired in terms of mathematical rigor and completeness of the arguments, see Remark 3.9, these results can be considered to be correct. The paper [25] classi es all (weak) traveling wave solutions of (1.1) in H (R) loc by means of a planar dynamical system approach. In particular it is shown that, in addition to smooth and singular traveling waves which are similar to those of CH and DP, equation (1.1) exhibits compactly supported traveling waves of depression with cusped or crested troughs and very peculiar waves having e.g. both peaked crests and troughs; such wave forms do not appear among the traveling waves of CH or DP. In [20] the local well-posedness for solutions of (1.1) on the real line was improved 3=2 to the Besov space setting B (R). The authors employed the earlier mentioned 2;1 nonlocal form of (1.1). In contrast to the previous well-posedness results they based WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 5 their proof on an iterative scheme, which was implemented by virtue of Littlewood- Paley decompositions. Moreover, they showed that if a solution u = u(x; t) blows up 3=2 T in B (R) with maximal life span T then ku (t)k dt = 1. Additionally x 1 2;1 0 L (R) they proved that certain real analytic initial data u = u (x) pass on analyticity 0 0 to the solution u (with respect to both the space and time variable) on some local time interval. The paper [41] considers local well-posedness of (1.1) in the Besov space B (R), p;q 1  p; q  1 and s > maxf1 + 1=p; 3=2g. In addition, the author presents a blow-up criterion for solutions in H (R), s > 3=2, in terms of the blow up of ku (t)k 1 L (R) as t approaches the maximal time of existence. Furthermore, asymptotic persistence estimates of solutions are established in L spaces for certain weight functions . The goal of this paper is to address the space-periodic Cauchy problem for (1.1). Our main result establishes local well-posedness for 1-periodic solutions in H (T), s > 3=2, cf. Theorem 2.1. This constitutes a new important aspect concerning the initial value problem of (1.1) which has not been considered in the literature so far: the local well-posedness results in [20, 37, 40, 41] have in common that they are formulated for decaying solutions on the real line and hence do not include periodic solutions. In particular the smooth periodic traveling wave solutions of (1.1), which were identi ed in [25], are captured by Theorem 2.1. Hence this result provides a rigorous foundation for the study of perturbations of these traveling waves opening the path for a stability analysis of smooth periodic traveling waves of (1.1). The proof of Theorem 2.1 relies|in contrast to the well-posedness results in [37]|on a reformulation of (1.1) as a nonlocal rst order equation (like in [20, 40, 41]), which 2 2 is possible due to the particular structure of the "  -coecients; it makes use of Kato's semigroup approach. 2. Local well-posedness result We address the local well-posedness of the Cauchy problem for (1.1) on the unit circle. That is, for a given initial data u : R ! R with u (x) = u (x + 1) for all 0 0 0 x 2 R we seek a unique solution u, which satis es > (1.1) is satis ed for x 2 R; t > 0 (2.1) u(0; x) = u (x) x 2 R u(t; x) = u(t; x + 1) x 2 R; t > 0 and depends continuously on the initial data u . By applying Kato's semigroup approach for quasilinear evolution equations [31] we show that (2.1) is locally well- s s posed in the Sobolev space H = H (T) for s > 3=2 with the usual norm kk ; per 2 0 2 T denotes the unit circle, i.e. T = R=Z. In particular, L = H , the L -norm per per coincides with kk which is induced from the standard inner product (;) in L . 0 0 per In full detail our main result reads as follows. Theorem 2.1. Let s > 3=2. For every u 2 H there is a maximal time of per s 1 s1 existence T 2 (0;1] and a unique solution u 2 C([0; T ); H )\C ((0; T ); H ) of per per the Cauchy problem (2.1). Moreover, the mapping which maps initial data u to unique local solutions u of s s (2.1) is continuous as a mapping from a neighborhood of u in H to C([0; T ]; H ), per per where [0; T ] denotes a suciently small but positive closed time interval. 6 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR 3. Proof of Theorem 2.1 In the following we provide a suitable adaptation of Kato's general theory [31] for our particular situation, which serves as the basis of our proof. Let us consider the abstract quasilinear initial value problem in a Hilbert space X : y = A(y)y + f (y) t > 0 (3.1) y(0) = y : Let Y be another Hilbert space which is continuously and densely embedded into X , and let S : Y ! X be a topological isomorphism. We assume the following: (I) For any given r > 0 it holds that for all y 2 B (0)  Y (the ball around the origin in Y with radius r) the linear operator A(y) : X ! X generates a strongly continuous semigroup T (t) in X which satis es ! t kT (t)k  e for all t 2 [0;1) y L(X) for a uniform constant ! > 0. (II) A maps Y into L(Y; X ); more precisely the domain D(A(y)) contains Y , and the restriction A(y)j belongs to L(Y; X ) for any u 2 Y . Furthermore, A is Lipschitz continuous in the sense that for all r > 0 there exists a constant C which only depends on r such that kA(y) A(z)k  C ky zk L(Y;X) 1 X for all y; z inside B (0)  Y . (III) For any y 2 Y there exists a bounded linear operator B(y) 2 L(X ) satisfying B(y) = SA(y)S A(y), and B : Y ! L(X ) is uniformly bounded on bounded sets in Y . Furthermore, for all r > 0 there exists a constant C which depends only on r such that kB(y) B(z)k  C ky zk L(X) 2 Y for all y; z 2 B (0)  Y . (IV) The map f : Y ! Y is locally X -Lipschitz continuous in the sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (y) f (z)k  C ky zk for all y; z 2 B (0)  Y; X 3 X r and locally Y -Lipschitz continuous in the sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (y) f (z)k  C ky zk for all y; z 2 B (0)  Y: Y 4 Y r Theorem 3.1 (Kato [31]). Let the assumptions (I){(IV) hold true. Then for any y 2 Y there exists a maximal time of existence T 2 (0;1] and a unique solution y 2 C([0; T ); Y )\C ((0; T ); X ) of the abstract Cauchy problem (3.1) in X . Moreover, the solution depends continuously on the initial data, i.e. the mapping which maps initial data y to unique local solutions y of (3.1) is continuous from a ~ ~ neighborhood of y in Y to C([0; T ]; Y ) for suciently small 0 < T  T . The rst step of our proof provides a reformulation of equation (1.1) into the form (3.1). To simplify the notation, let us consider the quasilinear evolution equation u u = u + u + uu t xxt 1 x 2 xxx 3 x (3.2) 2 3 + u u + uu + uu u + u u + u ; 1 x xx 2 xxx 1 x xx 2 xxx 3 x WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 7 with  > 0 and ; ; ; ; ; ; ; 2 R. Let  denote the rst order pseudo 1 2 3 1 2 1 2 3 2 1=2 di erential operator (1 @ ) . If the coecients , i = 1; 2; 3, satisfy = 2( + ); (3.3) 1 2 3 an application of  to (3.2) yields the abstract form u + A(u)u = f (u); (3.4) where the quasilinear part A(u) is given by 1 2 A(u) =  ( + u + u )@ ; (3.5) 2 2 2 x and the semilinear part f (u) satis es 2 3 2 2 1 2 2 2 3 2 2 f (u) =  @ + u + + u + u + u + ( 2 )uu : x 1 3 2 x x 2 2 3 2 (3.6) Remark 3.2. The above procedure requires  6= 0; if  is negative, the linear part of (3.2) becomes ill-posed, which explains the assumption that  is strictly positive. If (3.3) is satis ed, which is equivalent to the property that the cubic -terms can be written as an x-derivative, then (3.2) yields a scalar conservation law of the form u + (u) = 0. This is an immediate consequence of the reformulation t x (3.4) with (3.5) and (3.6), since the operators  and @ commute. Indeed, the commutation property of these operators can be seen from their representations 1 2 1 2 1 ^ ^ @ f = F (inf );  f = F (1 + n ) f ; x n n where F : C (T) ! S (Z) denotes the Fourier transform and (f ) is the sequence n n2Z of Fourier coecients associated with the Fourier series of f 2 C (T); here S (Z) denotes the Schwartz space of all rapidly decreasing sequences. The reformulation of (1.1) as a scalar conservation law opens the way to study global weak solutions of class C([0;1); H ) by introducing a distributional for- mulation of the corresponding Cauchy problem via multiplication of the scalar conservation law by suitable test functions (t; x) and integration over the time and space domain. This low-regularity class includes|unlike H with s > 3=2| functions with peaks or even cusps; we refer to [25] for the existence of weak traveling wave solutions for (1.1) exhibiting such singularities. Since (1.1) is of the form (3.2) and satis es (3.3) it admits the reformulation in (3.4). We may now normalize the coecients in this reformulation, since their particular values do not play a role in the subsequent analysis: they merely contribute to estimation constants, whose precise value is not of special interest. Therefore, we may henceforth consider the abstract operator equation u + A(u)u = f (u); (3.7) where A(u) = a(u)@ with a(u) = (1 + u + u ) (3.8) and 2 2 3 2 2 2 1=2 f (u) =  @ u + u + u + u + uu with  = (1 @ ) : (3.9) x x x s1 s As underlying function spaces we choose X := (H ;kk ) and Y := (H ;kk ) s1 s per per with s > 3=2, and consider the topological isomorphism S :=  : Y ! X between these spaces, which de nes an isometry, i.e. kuk = kuk for all u 2 H . For s1 s per 8 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR a given function g 2 H with r > 1=2 let us denote by M the corresponding per r r r r multiplication operator on H , i.e. M : H ! H ; w 7! gw. Since H , per per per per r > 1=2, is closed under multiplication, M is continuous. The subsequent lemmas show that the quasilinear part A and the semilinear part f given by (3.8) and (3.9), respectively, ful ll the assumptions (I){(IV) of Theorem 3.1. Lemma 3.3. For a 2 H the operator per 2 1 A := M @ = a(x)@ ; D(A) := fw 2 L : aw 2 H g a x x per per generates a strongly continuous semigroup T (t) in L which satis es per !t kT (t)k  e for all t  0 with ! = supja j: (3.10) 0 x x2T Proof. The idea of the proof is to reveal A as a bounded perturbation of a skew- adjoint operator. Then Stone's theorem and a standard perturbation result yield the assertion of the lemma. 2 2 Step 1. We rst observe that A : L  D(A) ! L is indeed a well-de ned per per linear operator, i.e. the action of A on an element w 2 D(A) is uniquely determined 2 2 and lies in L . To see this, we rst consider a general element w 2 L . Since the per per 1 1 1 usual product H  H ! H admits an extension to a continuous product per per per 1 1 1 H  H ! H , Leibniz' formula for general periodic distributions restricted per per per to this product yields that (aw) = a w + aw in H : x x x per For w 2 D(A) we have that both (aw) and a w lie in L , hence x x per Aw = (aw) a w = aw 2 L : x x x per Obviously H is contained in D(A), so A is densely de ned. per 1 2 Step 2. We show that C is a core for A in L , i.e. for every v 2 D(A) there per per 1 2 exists a sequence (v ) in C such that v ! v and Av ! Av in L . Let n n n n per per v 2 D(A). We may assume that supp(v)\ [0; 1] is contained in (0; 1), otherwise we can employ a suitable parametrization of T with a subordinated smooth partition of unity. In this way we can consider the restriction of v to (0; 1) and extend it to the whole real line by setting the extension identically 0 outside of (0; 1). The resulting function lies in L (R) and we denote it again by v. Now it suces to prove that there exists a sequence (v ) in C (R) with supp(v )  (0; 1) such that n n n v ! v and Av ! Av in L (R): n n Let  2 C (R) be a molli er with supp()  [0; 1] and  = 1, and consider v := v, where  is the convolution on R and  (x) := n(nx). Then v 2 C ((0; 1)) n n n for n  1 large enough and v ! v in L (R). We have to prove that (av ) ! (av) in L (R): (3.11) n x x By assumption and our considerations we have that both (av) and a v in L (R), x x thus also av 2 L (R), and hence (av ) ! av in L (R): (3.12) n x x WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 9 s1 1 Since a 2 H we may identify it as an L (R) element, thus the multiplication per with a constitutes a continuous operation in L (R) and we nd that a v ! a v in L (R): (3.13) x n x In view of the identity (av ) (av) = [a v a v] + [  (av ) av ] + [a(v )   (av )]; n x x x n x n x x n x n x which holds in L (R), together with (3.12) and (3.13) it suces to show that a(v )   (av ) ! 0 in L (R) (3.14) n x n x in order to infer (3.11). To this end, we consider the linear operators P v = a(v )   (av ); n  1: n n x n x 1 2 If v is regular enough, say v 2 H (R), then obviously P v ! 0 in L (R). In the following we show that the family fP g extends to a family of uniformly bounded n n1 linear operators in L (R). A well-known corollary of the uniform boundedness 1 2 principle and the density of H (R)  L (R) then yield that the sequence of operators (P ) converges pointwise to the trivial operator P  0 in L(L (R)); this implies n n1 in particular that (3.14) holds for our given v 2 D(A). We rst observe that every P extends to a continuous linear operator on L (R). Note that P v = a(( )  v) ( )  (av) +   (a v); n n x n x n x and for a general v 2 L (R) and arbitrary x 2 R we obtain that P v(x) = ( ) (y) a(x) a(x y) v(x y) dy +   (a v) (x) n n y n x = n  (ny) a(x) a(x y) v(x y) dy +   (a v) (x): y n x By recalling that supp()  [0; 1], we can write n  (ny) a(x) a(x y) v(x y) dy 1 h  i y y = n  (y) a(x) a x v x dy: n n Thus in view of the mean value theorem, we obtain the estimate n  (ny) a(x) a(x y) v(x y) dy supja j j (y)jjyj v x dy: x y x2R 0 10 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR 2 2 2 Let now C = sup ja j  (y)y dy. Then the Cauchy-Schwarz inequality and x2R y Fubini's theorem yield that n  (ny)(a(x) a(x y))v(x y) dy R 0 Z Z 1 h  i C v x dy dx R 0 Z Z 1 h  i = C v x dx dy 0 R = Ckvk : Moreover, we obtain by Young's inequality that k  (a v)k  ka vk  sup(ja j)kvk : n x 0 x 0 x 0 x2R We conclude that kP k 2  K n L(L ) per holds uniformly for all n  1 with K = ( C + sup ja j). It follows that C is a x2R per core for A in L . per : : Step 3. We show that A = A + M with domain D(A ) = D(A) is skew- 0 a 0 adjoint in L . Let A denote the adjoint of A with domain D(A ), which is per 0 0 well-de ned since A (and hence A ) is densely de ned. Fix w 2 D(A ), then 2 1 A w 2 L , and for ' 2 C  D(A ) we have that (A '; w) = ('; A w) . We 0 0 0 0 0 per per 0 consider the linear functional F : C ! R given by per F (') = (A '; w) = a' + a ' w dx = ('; A w) : w 0 0 x x 0 Obviously, F is continuous with respect to the L -norm. Integration by parts per yields that F (') coincides with the action of the regular periodic distribution 1 2 1 A w = (aw + a w) 2 L on the test function '. By density of C in 0 x x per per 2 1 L we infer that aw 2 H and A w = A w, i.e. w 2 D(A ), which implies 0 0 per per 0 A  A . For v 2 D(A ) let v :=   v. By integration by parts and the second step of 0 n n the proof we obtain for every w 2 D(A ) that (A w; v) = lim (A w; v ) = lim aw + a w v dx 0 0 0 n 0 x x n n!1 n!1 = lim a(v ) + a v w dx (3.15) n x x n n!1 2 = av + a v w dx = (w; A v) ; x x 0 0 in particular v 2 D(A ) and A  A . 0 0 Step 4. By the previous step the operator iA is self-adjoint and therefore Stone's theorem implies that A is the in nitesimal generator of a strongly continuous semigroup of contractions in L . A standard perturbation argument (see e.g. [36, per Thm. 3.1.1]) now yields that A is the generator of a strongly continuous semigroup T (s) in L , which satis es the estimate (3.10) by noting that a is a bounded per WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 11 function due to Sobolev's embedding theorem and therefore the operator norm of 2 2 M : L ! L can be estimated by sup ja (x)j. a x x per per x2T Let [;] denote the usual commutator of two operators. In the course of our analysis we will make use of commutator and product estimates taken from [33, Proposition B.10.(2)] and [31, Lemma A1], respectively. These results are stated for functions and distributions de ned on the real line (or R ); the corresponding assertions on the periodic domain T hold true analogously. We summarize these estimates formulated for one-dimensional periodic functions in the following lemma. Lemma 3.4. Let r > 1=2. (i) If 1=2 < t  r + 1, there exists a constant C > 0 such that r;t [ ; M ]h  C kgk khk g r;t r+1 t1 r+1 t1 for all g 2 H and h 2 H . per per (ii) If r < t  r, there exists a constant C > 0 such that r;t kfgk  C kfk kgk t r;t r t r t for all f 2 H and g 2 H . per per Lemma 3.5. For a 2 H the operator per s1 s ~ ~ : : A = M @ = a(x)@ ; D(A) = fw 2 H : aw 2 H g a x x per per s1 generates a strongly continuous semigroup T (t) in H which satis es per ! ~t kT (t)k  e for all t  0 with ! ~ = ka k : s1 x s1 Proof. Recall from Lemma 3.3 that the operator A = M @ with domain D(A) = a x 2 1 2 fw 2 L : aw 2 H g lies in G(1; !; L ) for ! = sup ja j=2. per per per x2T s1 1s 2 Let A :=  A . In the following we show that A A 2 L(L ). Since 1 1 per s1 1s s1 1s @ commutes with  and  , we may write A A = [ ; M ] @ . Let x 1 a x 2 1s 2 s2 w 2 L be arbitrary. Since  @ 2 L(L ; H ) we can directly apply Lemma per per per 3.4 (i) to infer that s1 1s k(A A)wk = k[ ; M ] @ wk 1 0 a x 0 1s Ckak k @ wk s x s2 Ckak kwk ; s 0 where the generic constant C solely depends on s. We just showed that A is a bounded perturbation of A, it therefore generates a strongly continuous semigroup (cf. [36, Thm. 3.1.1]); more speci cally, A 2 2  s1 s1 2 G(1; ! ; L ) for ! = ! + kA Ak 2 . Since  : H ! L is an per L(L ) per per per s1 (isometric) isomorphism, it follows from [36, Thm. 4.5.8] that H is A-admissible. per s1 That is, H is an invariant subspace of T (t), t  0, the strongly continuous per s1 semigroup generated by A, and the restriction of T (t) to H is strongly continuous per s1 with respect to kk . By [36, Thm. 4.5.5], the A-admissibility of H entails s1 per s1 ~ ~ that A, the part of A in H , generates the strongly continuous semigroup T (t), per s1 which equals the restriction of T (t) to H . The assertion of the Lemma follows per immediately.  12 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR A straight forward application of Lemma 3.5 shows that the operators A(u), u 2 H , which are de ned in (3.8), satisfy the assumption (I) of Kato's Theorem 3.1. per The following lemma establishes assumption (II). s s s1 Lemma 3.6. The operator A maps H into L(H ; H ). More precisely, the do- per per per s s s1 main D(A(u)) contains H , and the restriction A(u)j belongs to L(H ; H ) per per per per for any u 2 H . Furthermore, A is Lipschitz continuous in the sense that for all per r > 0 there exists a constant C which only depends on r such that kA(u) A(v)k s1  C ku vk (3.16) s 1 s1 L(H ;H ) per per for all u; v inside B (0)  H . per s s Proof. Recall that H is by de nition contained in D(A(u)) for arbitrary u 2 H per per s s1 (cf. Lemma 3.5). The restriction A(u)j belongs to L(H ; H ), since @ 2 H x per per per s s1 s1 L(H ; H ) and M 2 L(H ). a(u) per per per To see that the required estimate is satis ed, let u; v; w 2 H be arbitrary. per 2 2 k(A(u) A(v))wk = k(u v + u v )@ wk s1 x s1 2 2 C (ku vk +ku v k )k@ wk s1 s1 x s1 C (1 +kuk +kvk )kwk ku vk ; s1 s1 s s1 where C denotes a generic constant. This shows that for every r > 0 one can always nd a constant C such that (3.16) holds uniformly for all u; v 2 B (0)  H . 1 r per The two lemmas below verify the assumptions (III) and (IV), respectively; their proofs rely on Lemma 3.4. Lemma 3.7. For any u 2 H there exists a bounded linear operator B(u) 2 per s1 1 s s1 L(H ) satisfying B(u) = A(u) A(u), and B : H ! L(H ) is uniformly per per per bounded on bounded sets in H . Furthermore, for all r > 0 there exists a constant per C , which depends only on r, such that kB(u) B(v)k s1  C ku vk s (3.17) 2 H L(H ) per per for all u; v 2 B (0)  H . per s 1 Proof. Let u 2 H . Since @ commutes with  and  we obtain that per 1 1 B(u) = M @  M @ = [; M ] @ : x x x a(u) a(u) a(u) s1 Hence we can write  B(u) as s1 s 1 s1 B(u) = [ ; M ] @ + [M ;  ]@ : a(u) x a(u) x s1 s Let now w 2 H and u; v 2 H be arbitrary. In view of the above identity and per per Lemma 3.4 (i) we obtain the following estimate s1 B(u) B(v) w =  B(u) B(v) w s1 0 s 1 s1 [ ; M ] @ w + [ ; M ]@ w a(u)a(v) x a(u)a(v) x 0 0 Cka(u) a(v)k  @ w + @ w s x x s1 s2 2 2 C ku vk +ku v k kwk ; s s s1 where C is a generic constant independent of u; v and w. In particular, this shows s1 s that B(u) extends to a bounded linear operator on H for every u 2 H such that per per WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 13 s s1 s B : H ! L(H ) is uniformly bounded on bounded sets in H . Furthermore, per per per this estimation proves that there exists a constant C depending only on the radius of the ball B (0)  H such that (3.17) is satis ed for all u; v 2 B (0). r r per s s s1 Lemma 3.8. The map f : H ! H is locally H -Lipschitz continuous in the per per per sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (u) f (v)k  C ku vk for all u; v 2 B (0)  H ; (3.18) s1 3 s1 r per and locally H -Lipschitz continuous in the sense that for every r > 0 there exists a per constant C > 0, depending only on r, such that kf (u) f (v)k  C ku vk for all u; v 2 B (0)  H : s 4 s r per Proof. Recall that 2 2 3 2 2 f (u) =  @ u + u + u + u + uu : x x s s1 Let u 2 H , then u is contained in H , which is closed under multiplication per per s1 2 since s > 3=2, hence polynomials in u and u lie again in H . Since  @ maps x x per r r+1 H continuously to H for arbitrary r 2 R, it is clear that f is well-de ned per per s s and continuous as a mapping from H to H . An application of Lemma 3.4 (ii) per per yields the following estimation for arbitrary u; v 2 H : per 2 2 3 3 2 2 2 2 kf (u) f (v)k  C u v + u v + u v + u v + uu vv s1 x x x x s2 2 2 C 1 +ku + vk +ku + uv + v k ku vk s1 s1 s1 +ku + v k ku v k +ku k ku vk x x s1 x x s2 x s1 s1 +kv(u + v )k ku v k ; x x s1 x x s2 where C denotes a generic constant being independent of u and v. This shows that for every r > 0 there exists a constant C > 0 such that (3.18) holds for all u; v 2 B (0)  H . per Similarly, elementary estimates (which do not rely on the product inequality in Lemma 3.4) yield that f is locally H -Lipschitz continuous; we omit the details. per It is now an immediate consequence of Theorem 3.1 in conjunction with Lemmas 3.5, 3.6, 3.7 and 3.8, that the periodic Cauchy problem for equation (3.7) with A and f given by (3.8) and (3.9), respectively, is locally well-posed for initial data in H , s > 3=2, such that the unique local solution u with maximal life span per s 1 s1 T 2 (0;1] lies in C([0; T ); H ) \ C ((0; T ); H ) and depends continuously on per per the initial datum. As noted earlier, the particular values of the coecients in (3.8) and (3.9) do not play a role when it comes to local well-posedness. Therefore the periodic Cauchy problem for equation (1.1) is locally well-posed as well (in exactly the same sense), which nishes the proof of Theorem 2.1. Remark 3.9. Our proof of Theorem 2.1 can be directly adapted to the corre- sponding non-periodic initial value problem on the real line. This is not surprising, since Sobolev's embedding theorem, as well as product inequalities and commutator estimates possess analogous versions for the periodic and non-periodic case. Note that we carried over the crucial step in the proof of Lemma 3.3 to the corresponding non-periodic scenario by means of a smooth partition of unity; in particular one 14 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR immediately obtains an analogous non-periodic version of this lemma with the same proof. One delicate aspect of the proof is|apart from the application of the nontrivial commutator estimates in Lemma 3.4|the approximation technique in the second step in the proof of Lemma 3.3, which is inspired by the arguments in [6]. The domain D(A ) of the operator A in the proof of Lemma 3.3 contains H as 0 0 per a proper subspace, i.e. an element v 2 D(A ) does in general not lie in H . per Consequently the integration by parts formula is not directly applicable to the L per inner product (A w; v) in (3.15). 0 0 As noted in the introduction, the local well-posedness problem for (1.1) has been s1 earlier studied in [40] in the analogous framework on the real line, i.e. X = H (R) and Y = H (R). The authors chose basically the same approach, i.e. Kato's theory was applied to the nonlocal rst order reformulation of (1.1). A notable weakness of this paper is the presentation of the proofs: e.g. the integration by parts formula is applied under heavy abuse of notation in an unjusti ed way in the proof of [40, Lemma 3.1], the proof of [40, Lemma 3.3] is incomplete (only the trivial local H - s1 Lipschitz estimate is shown, the less trivial H -estimation is omitted), nontrivial commutator estimates are applied without providing any references|to mention just a few aws. Let us nally comment on some open questions regarding the shallow water wave equation (1.1). It is up to this point unknown whether or not a solution of (1.1) remains bounded in L on its whole interval of existence for the full space of initial data. This would give an answer to the important question whether or not wave- breaking is the only possible blow-up scenario (i.e. ku(t)k 1 remains bounded, L (R) whereas lim sup ku (t)k 1 = 1 whenever the maximal life span T is nite). x L (R) t%T Moreover, it is an open question if blow-up|or more speci cally wave-breaking|of solutions actually occurs for certain speci ed classes of initial data; neither it is known|apart from the traveling wave solutions established in [25]|whether there exist nontrivial initial wave pro les giving rise to global solutions. Acknowledgments The authors acknowledge the support of the Erwin Schr odinger International In- stitute for Mathematics and Physics (ESI) during the program \Mathematical Aspects of Physical Oceanography". R. Quirchmayr acknowledges the support of the European Research Council, Consolidator Grant No. 682537. References [1] T. B. Benjamin, J. L. Bona and J. J. Mahoney, Model equations for long waves in nonlinear dispersive systems, Phil. Trans. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 227, 47{78 (1972). [2] J. Boussinesq, Essai sur la theorie des eaux courantes, M emoires pr esent es par divers savants a l'Acad. des Sci. Inst. Nat. France, XXIII (1877). [3] R. Camassa and D. Holm, An integrable shallow water equation with peaked solitons, Phys. Rev. Lett. 71, 1661{1664 (1993). [4] A. Constantin, Nonlinear Water Waves with Applications to Wave-Current Interactions and Tsunamis, CBMS-NSF Regional Conference Series in Applied Mathematics 81, SIAM, Philadelphia (2011). [5] A. Constantin and J. Escher, Global Existence and Blow-up for a Shallow Water Equation, Ann. Scuola Norm. Sup. Pisa Cl. Sci. 26(4), 303{328 (1998). WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 15 [6] A. Constantin and J. Escher, Well-Posedness, Global Existence, and Blowup Phenomena for a Periodic Quasi-Linear Hyperbolic Equation, Comm. Pure Appl. Math. 51, 475{504 (1998). [7] A. Constantin and J. Escher, Wave breaking for nonlinear nonlocal shallow water equations, Acta Math. 181, 229{243 (1998). [8] A. Constantin, R. I. Ivanov and J. Lenells, Inverse scattering transform for the Degasperis- Procesi equation, Nonlinearity 23, 2559{2575 (2010). [9] A. Constantin and R. S. Johnson, On the Non-Dimensionalisation, Scaling and Resulting Interpretation of the Classical Governing Equations for Water Waves, J. Nonlinear Math. Phys. 15, 58{73 (2008). [10] A. Constantin and D. Lannes, The Hydrodynamical Relevance of the Camassa-Holm and Degasperis-Procesi Equations, Arch. Rational Mech. Anal. 192, 165{186 (2009). [11] A. Constantin and H. P. McKean, A shallow water equation on the circle, Comm. Pure Appl. Math. 52, 949{982 (1999). [12] A. Degasperis, D. D. Holm and A. N. W. Hone, A new integrable equation with peakon solutions, Theor. Math. Phys. 133, 1463{1474 (2002). [13] P. G. Drazin and R. S. Johnson, Solitons: an introduction, Cambridge Univ. Press, Cambridge, [14] A. Degasperis and M. Procesi, Asymptotic integrability In: Degasperis, A. and Gaeta, G. (eds.) Symmetry and Perturbation Theory, pp. 23{37. World Scienti c, Singapore (1999). [15] N. Duruk Mutluba s, On the Cauchy problem for a model equation for shallow water waves of moderate amplitude, Nonlinear Anal. Real World Appl. 14(5), 2022{2026 (2013). [16] N. Duruk Mutluba s, Local well-posedness and wave breaking results for periodic solutions of a shallow water equation for waves of moderate amplitude Nonlinear Anal. Theory Methods Appl. 97, 145{154 (2014). [17] N. Duruk Mutluba s and A. Geyer, Orbital stability of solitary waves of moderate amplitude in shallow water, J. Di erential Equations 255 (2), 254-263 (2013). [18] N. Duruk Mutluba s, A. Geyer and B. Matioc, Non-uniform continuity of the ow map for an evolution equation modeling shallow water waves of moderate amplitude, Nonlinear Anal. Real World Appl. 17(6), 322{331 (2014). [19] J. Escher, Y. Liu and D. Yin, Global weak solutions and blow-up structure for the Degasperis- Procesi equation J. Funct. Analysis 192, 457{485 (2006). [20] L. Fan and W. Yan, The Cauchy problem for shallow water waves of large amplitude in Besov space, J. Di er. Equ. 267(3), 1705{1730 (2019). [21] B. Fuchssteiner and A. S. Fokas, Symplectic structures, their B acklund transformations and hereditary symmetries, Physica D 4, 47{66 (1981). [22] A. Gasull and A. Geyer, Traveling surface waves of moderate amplitude in shallow water, Nonlinear Anal. Theory Methods Appl. 102, 105-119 (2014). [23] A. Geyer, Solitary traveling waves of moderate amplitude, J. Nonlinear Math. Phys. 19, 104{115 (2012). [24] A. Geyer, Symmetric waves are traveling waves for a shallow water equation modeling surface waves of moderate amplitude, J. Nonlinear Math. Phys. 22(4), 545{551 (2015). [25] A. Geyer and R. Quirchmayr, Traveling wave solutions of a highly nonlinear shallow water equation, Discrete Contin. Dyn. Syst. 38, 1567{1604 (2018). [26] R. I. Ivanov, On the Integrability of a Class of Nonlinear Dispersive Wave Equations J. Nonlinear Math. Phys. 12, 462{468 (2005). [27] R. I. Ivanov, Water waves and integrability, Philos. Trans. Roy. Soc. London A 365, 2267{2280 (2007). [28] R. S. Johnson, Camassa-Holm, Korteweg-de Vries and related models for water waves, J. Fluid Mech. 455, 63{82 (2002). [29] R. S. Johnson, A Modern Introduction to the Mathematical Theory of Water Waves Cambridge Univ. Press, Cambridge (1997). [30] T. Kappeler and J. P oschel, KdV & KAM, Ergeb. der Math. und ihrer Grenzgeb., Springer, Berlin-Heidelberg-New York, 2003. [31] T. Kato Quasi-linear equations of evolution, with applications to partial di erential equa- tions In: Spectral Theory and Di erential Equations, pp. 25-70. Springer Lecture Notes in Mathematics 448, Berlin (1975). [32] D. J. Korteweg and G. de Vries. On the change of form of long waves advancing in a rectangular canal and on a new type of long stationary waves. Phil. Mag. 39 422{443 (1895). 16 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR [33] D. Lannes, The Water Waves Problem: Mathematical Analysis and Asymptotics, American Math. Soc., Providence, RI (2013). [34] J. Lenells, Traveling wave solutions of the Camassa-Holm equation, J. Di er. Equ. 217, 393{430 (2005). [35] J. Lenells, Traveling wave solutions of the Degasperis-Procesi equation, J. Math. Anal. Appl. 306, 72{82 (2005). [36] A. Pazy, Semigroups of Linear Operators and Applications to Partial Di erential Equations Springer, New York (1983). [37] R. Quirchmayr, A new highly nonlinear shallow water wave equation, J. Evol. Equations 16, 539{567 (2016). [38] J. S. Russell, Experimental researches into the laws of certain hydrodynamical phenomena that accompany the motion of oating bodies, and have not previously been reduced into conformity with the known laws of the resistance of uids, Trans. R. Soc. Edinburgh 14, 47{109 (1839). [39] T. Tao, Low-regularity global solutions to nonlinear dispersive equations In: Surveys in Analysis and Operator Theory, pp. 19{48, Proc. Centre Math. Appl. Austral. Nat. Univ. 40 (2002). [40] S. Yang and T. Xu, Well-posedness and persistence property for a shallow water wave equation for waves of large amplitude, Appl. Anal. 98(5), 981{990 (2017). [41] S. Zhou, Well-posedness and wave breaking for a shallow water wave model with large amplitude, J. Evol. Equ., 20(1), 141{163 (2020). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Mathematics arXiv (Cornell University)

Well-posedness of a highly nonlinear shallow water equation on the circle

Mutlubas, Nilay Duruk; Geyer, Anna; Quirchmayr, Ronald
Mathematics , Volume 2020 (2011) – Nov 2, 2020
Download PDF
16 pages

Loading next page...
 
/lp/arxiv-cornell-university/well-posedness-of-a-highly-nonlinear-shallow-water-equation-on-the-mJ7KdRkqCt
ISSN
0362-546X
eISSN
ARCH-3343
DOI
10.1016/j.na.2020.111849
Publisher site
See Article on Publisher Site

Abstract

WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION ON THE CIRCLE NILAY DURUK MUTLUBAS Faculty of Engineering and Natural Sciences, Sabanci University, Turkey ANNA GEYER Delft Institute of Applied Mathematics, Faculty Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, The Netherlands. RONALD QUIRCHMAYR Department of Mathematics, KTH Royal Institute of Technology, Sweden. Abstract. We present a comprehensive introduction and overview of a recently derived model equation for waves of large amplitude in the context of shallow water waves and provide a literature review of all the available studies on this equation. Furthermore, we establish a novel result concerning the local well- posedness of the corresponding Cauchy problem for space-periodic solutions with initial data from the Sobolev space H on the circle for s > 3=2. Mathematics Subject Classi cation: Primary: 35Q35. Secondary: 35L30. Keywords: Shallow water equation, large amplitude waves, cubic nonlinearity, well-posedness. 1. Introduction and overview This paper addresses the local well-posedness of the model equation 3" u + u + uu (4u + 7u ) t x x xxx xxt 2 18 (1.1) 2 2 2 " " 2 3 = 2u u + uu 398uu u + 45u u + 154u x xx xxx x xx xxx 6 96 considered on the unit circle with initial data from the Sobolev space H for s > 3=2. Equation (1.1) was derived in [37] from the classical water wave problem for free surface gravity water waves over a at bottom, where the underlying incompressible ow is governed by Euler's equation. It serves as an asymptotic model for the horizontal velocity component of a unidirectional shallow water wave of large amplitude at a speci c depth. E-mail addresses: nilay.duruk@sabanciuniv.edu, a.geyer@tudelft.nl, ronaldq@kth.se. arXiv:2011.01290v1 [math.AP] 2 Nov 2020 2 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR Scalar model equations are e ective tools for studying the propagation of unidirec- tional water waves. In contrast to general solutions for the full governing equations, solutions of model equations are analytically (and numerically) tractable and provide accurate approximations for certain classes of waves, enabling an in-depth analysis of complex nonlinear wave phenomena. One of the most prominent equations in the class of shallow water wave models is the Korteweg-de Vries equation (KdV) 2" u + u + uu + u = 0; t x x xxx 3 6 here " and  are the dimensionless amplitude and shallowness parameter, see [9, 29, 28]. KdV was rst considered by Boussinesq [2] in 1877, before Korteweg and de Vries [32] derived it in 1895. It exhibits a balance between nonlinear and dispersive e ects enabling the formation of solitary waves, which where rst observed in a narrow canal by Russell [38]. Such waves preserve their shape while propagating at constant speed and can not be described by the linear dispersive water wave equation (formally obtained by setting " = 0 in the above equation), whose solutions decrease in height and disperse as they propagate. KdV is an accurate approximation for shallow water waves of small amplitude, i.e.   1 and " = O( ), which propagate in one direction. It forms a completely integrable bi-Hamiltonian system, and the existence of a Lax pair enables the application of nonlinear Fourier transform techniques to solve the Cauchy problem exactly for large classes of initial data and to determine the long-time behavior of the solution. Moreover, there exists a single global canonical coordinate system for space-periodic solutions (so-called Birkho coordinates) in which the time evolution becomes linear [30]. KdV is the simplest water wave model whose solitary waves are solitons (see the discussion in [13]), i.e. these localized waves preserve their shape after interacting with other waves of this type in conjunction with a phase shift. KdV has global solutions in time for a very large class of initial data, cf. [39], covering all physically relevant initial states (see the discussion in [4]). In particular KdV is not capable of describing breaking waves |waves that remain bounded whereas their slope becomes in nite in nite time. KdV belongs to a wider class of weakly nonlinear dispersive equations, the BBM equations [1] 2" u + u + uu +  ( u + u ) = 0; t x x xxx xxt with  0 (to avoid linear ill-posedness) and = + . Since this one-parameter family of equations does not account correctly for waves whose behavior is more nonlinear than dispersive, one is forced to consider more general regimes to include breaking waves, see [4]. A natural regime for describing larger waves is the shallow water regime for waves of moderate amplitude, i.e.   1 and " = O(). It contains the following two-parameter family of equations, which accurately approximate the water wave problem (cf. [10, 33]): 3" 2 2 u + u + uu +  ( u + u ) = " ( uu + u u ): t x x xxx xxt xxx x xx We refer to [33] for a comprehensive approach on the justi cation of KdV as an accurate shallow water approximation for the water wave problem. WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 3 The function u = uj describes the evolution of the horizontal velocity component of the ow eld at a xed dimensionless depth z 2 [0; 1]; the second parameter p 2 R can be chosen such that the coecients ; ; ;  satisfy 1 3 1 3 9 23 3 = p + ; = p + ; = p ;  = p ; 6 2 6 2 2 24 2 1 1 where  = (z ). It turns out that this family contains models that capture 2 3 wave breaking. The most prominent examples are the Camassa-Holm equation (CH) U + U + 3UU U = 2U U + UU ; t x x xxt x xx xxx and the Degasperis-Procesi equation (DP) U + U ^ + 4UU U = 3U U + UU : t x x xxt x xx xxx By means of a suitable scaling of U (provided that ;  ^ 6= 0) one can recover that both CH and DP belong to the before mentioned two-parameter family, cf. [28, 10]; they generalize KdV/BBM in the sense that they degenerate to KdV/BBM when " = O( ). CH was derived in [3] as an integrable shallow water wave model with peaked solitons; it was earlier derived in [21] as a novel bi-Hamiltonian equation. DP was rst considered in [14]. CH and DP share with KdV the rare property that they form completely integrable bi-Hamiltonian systems|in fact they are the only integrable equations of the before mentioned family, cf. [26, 27]. We refer to [11] and [8] for rigorous approaches on integrability of CH and DP, respectively. The classes of solutions of CH and DP are richer than the waves described by KdV: apart from capturing the phenomenon of wave breaking [5, 6, 7, 19], there is another fact that distinguishes CH and DP from KdV: they possess singular traveling wave solutions such as periodic and solitary waves with peaked and cusped crests, and their solitons are peaked [34, 35, 5, 12]. The model equation for the free surface  of moderate amplitude shallow water waves that corresponds to CH and DP|we denote it by SE|is given by 2 2 2 3" 3" +  +    + t x x x xxx xxt 2 8 12 12 3 2 3" 7" =   2  +  : x x xx xxx 16 24 It initially appeared in [28] in Johnson's re-derivation of CH directly from the governing equations, and was later studied in various papers. We refer to [10] for a rigorous justi cation of SE as an accurate shallow water approximation for the water wave problem (besides CH and DP), a local well-posedness result for initial data in H , s > 5=2, as well as wave breaking results. An improvement of local well-posedness in terms of lower Sobolev exponents can be found in [15]; we refer to [16] for an analogue well-posedness result on the periodic domain and wave breaking results for periodic solutions. Moreover, it was proved in [18] that the data-to-solution map is not uniformly continuous. SE has smooth solitary traveling wave solutions [23], which are orbitally stable [17], as well as smooth periodic waves and a rich collection of singular traveling waves [22]. Moreover, symmetric solutions of this equations must be traveling waves [24], adding evidence in support of the strong correlation between symmetry and traveling waves often observed in water wave theory. Equation (1.1) was derived to account for even larger waves than those captured by CH and DP (and SE) by appealing to the approach presented in [28], where 4 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR Johnson re-derived CH directly from the governing equations for gravity water waves by means of double asymptotic expansions in the wave parameters  and " of the velocity eld, the pressure and the free surface. This ansatz makes it possible to treat the amplitude and shallowness parameters independently over the course of the derivation. After determining suciently many terms in the asymptotic expansion, one may freely choose a speci c wave regime to discover the corresponding asymptotic models. Equation (1.1) lives in the regime which is determined by the relation " = O( ) and   1. This means that the amplitude parameter can| unlike in the CH regime|exceed the size of the shallowness parameter, which brings 2 2 in the additional highly nonlinear "  -terms when compared to CH and DP; for this reason we speak about a shallow water equation for waves of large amplitude. Equation (1.1) belongs to a two parameter family of shallow water equations of large amplitude: one of the free parameters determines the xed depth of the water, whereas the second parameter stems from the \BBM-trick" to include the u -term xxt in (1.1). The latter parameter was chosen such that the coecients of the quadratic third order terms in (1.1) satisfy the ratio 2 : 1 as in the case for CH, whereas the 2 2 particular choice for the dimensionless water depth entails that the "  -term can be written as an x-derivative, which structurally distinguishes (1.1) from all the other equations from the same family. Equation (1.1), as well as the entire two-parameter family describing the horizontal velocity component at any depth associated with (1.1), and the corresponding one- parameter family of equations for the evolution of the free surface elevation are locally well-posed on the real line in the classical H (R)-setting. This was shown in [37] with the help of Kato's semigroup approach for quasilinear evolution equations (see [31]), where also a blow-up criterion for solutions u of (1.1) in H (R) was established by means of the blow-up of ku k 1 . Since the appearance of this initial x L study, several papers by di erent authors addressed the local well-posedness of (1.1) in more general settings, re ned estimates for blow-up, proved persistence of certain asymptotic properties of the solution; furthermore weak traveling waves were fully classi ed. We sketch this progress in the following. In [40], the authors exploited the aforementioned structural properties of (1.1) to reformulate it as a nonlocal rst order equation to show local well-posedness of the corresponding Cauchy problem with initial data in H (R) for s > 3=2 (where 3=2 is the sharp lower bound) by applying Kato's theory. Furthermore, they proved an asymptotic persistence property for H (R) solutions|certain decay rates of the initial wave pro le are preserved under the time evolution. Even though the proofs leave much to be desired in terms of mathematical rigor and completeness of the arguments, see Remark 3.9, these results can be considered to be correct. The paper [25] classi es all (weak) traveling wave solutions of (1.1) in H (R) loc by means of a planar dynamical system approach. In particular it is shown that, in addition to smooth and singular traveling waves which are similar to those of CH and DP, equation (1.1) exhibits compactly supported traveling waves of depression with cusped or crested troughs and very peculiar waves having e.g. both peaked crests and troughs; such wave forms do not appear among the traveling waves of CH or DP. In [20] the local well-posedness for solutions of (1.1) on the real line was improved 3=2 to the Besov space setting B (R). The authors employed the earlier mentioned 2;1 nonlocal form of (1.1). In contrast to the previous well-posedness results they based WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 5 their proof on an iterative scheme, which was implemented by virtue of Littlewood- Paley decompositions. Moreover, they showed that if a solution u = u(x; t) blows up 3=2 T in B (R) with maximal life span T then ku (t)k dt = 1. Additionally x 1 2;1 0 L (R) they proved that certain real analytic initial data u = u (x) pass on analyticity 0 0 to the solution u (with respect to both the space and time variable) on some local time interval. The paper [41] considers local well-posedness of (1.1) in the Besov space B (R), p;q 1  p; q  1 and s > maxf1 + 1=p; 3=2g. In addition, the author presents a blow-up criterion for solutions in H (R), s > 3=2, in terms of the blow up of ku (t)k 1 L (R) as t approaches the maximal time of existence. Furthermore, asymptotic persistence estimates of solutions are established in L spaces for certain weight functions . The goal of this paper is to address the space-periodic Cauchy problem for (1.1). Our main result establishes local well-posedness for 1-periodic solutions in H (T), s > 3=2, cf. Theorem 2.1. This constitutes a new important aspect concerning the initial value problem of (1.1) which has not been considered in the literature so far: the local well-posedness results in [20, 37, 40, 41] have in common that they are formulated for decaying solutions on the real line and hence do not include periodic solutions. In particular the smooth periodic traveling wave solutions of (1.1), which were identi ed in [25], are captured by Theorem 2.1. Hence this result provides a rigorous foundation for the study of perturbations of these traveling waves opening the path for a stability analysis of smooth periodic traveling waves of (1.1). The proof of Theorem 2.1 relies|in contrast to the well-posedness results in [37]|on a reformulation of (1.1) as a nonlocal rst order equation (like in [20, 40, 41]), which 2 2 is possible due to the particular structure of the "  -coecients; it makes use of Kato's semigroup approach. 2. Local well-posedness result We address the local well-posedness of the Cauchy problem for (1.1) on the unit circle. That is, for a given initial data u : R ! R with u (x) = u (x + 1) for all 0 0 0 x 2 R we seek a unique solution u, which satis es > (1.1) is satis ed for x 2 R; t > 0 (2.1) u(0; x) = u (x) x 2 R u(t; x) = u(t; x + 1) x 2 R; t > 0 and depends continuously on the initial data u . By applying Kato's semigroup approach for quasilinear evolution equations [31] we show that (2.1) is locally well- s s posed in the Sobolev space H = H (T) for s > 3=2 with the usual norm kk ; per 2 0 2 T denotes the unit circle, i.e. T = R=Z. In particular, L = H , the L -norm per per coincides with kk which is induced from the standard inner product (;) in L . 0 0 per In full detail our main result reads as follows. Theorem 2.1. Let s > 3=2. For every u 2 H there is a maximal time of per s 1 s1 existence T 2 (0;1] and a unique solution u 2 C([0; T ); H )\C ((0; T ); H ) of per per the Cauchy problem (2.1). Moreover, the mapping which maps initial data u to unique local solutions u of s s (2.1) is continuous as a mapping from a neighborhood of u in H to C([0; T ]; H ), per per where [0; T ] denotes a suciently small but positive closed time interval. 6 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR 3. Proof of Theorem 2.1 In the following we provide a suitable adaptation of Kato's general theory [31] for our particular situation, which serves as the basis of our proof. Let us consider the abstract quasilinear initial value problem in a Hilbert space X : y = A(y)y + f (y) t > 0 (3.1) y(0) = y : Let Y be another Hilbert space which is continuously and densely embedded into X , and let S : Y ! X be a topological isomorphism. We assume the following: (I) For any given r > 0 it holds that for all y 2 B (0)  Y (the ball around the origin in Y with radius r) the linear operator A(y) : X ! X generates a strongly continuous semigroup T (t) in X which satis es ! t kT (t)k  e for all t 2 [0;1) y L(X) for a uniform constant ! > 0. (II) A maps Y into L(Y; X ); more precisely the domain D(A(y)) contains Y , and the restriction A(y)j belongs to L(Y; X ) for any u 2 Y . Furthermore, A is Lipschitz continuous in the sense that for all r > 0 there exists a constant C which only depends on r such that kA(y) A(z)k  C ky zk L(Y;X) 1 X for all y; z inside B (0)  Y . (III) For any y 2 Y there exists a bounded linear operator B(y) 2 L(X ) satisfying B(y) = SA(y)S A(y), and B : Y ! L(X ) is uniformly bounded on bounded sets in Y . Furthermore, for all r > 0 there exists a constant C which depends only on r such that kB(y) B(z)k  C ky zk L(X) 2 Y for all y; z 2 B (0)  Y . (IV) The map f : Y ! Y is locally X -Lipschitz continuous in the sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (y) f (z)k  C ky zk for all y; z 2 B (0)  Y; X 3 X r and locally Y -Lipschitz continuous in the sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (y) f (z)k  C ky zk for all y; z 2 B (0)  Y: Y 4 Y r Theorem 3.1 (Kato [31]). Let the assumptions (I){(IV) hold true. Then for any y 2 Y there exists a maximal time of existence T 2 (0;1] and a unique solution y 2 C([0; T ); Y )\C ((0; T ); X ) of the abstract Cauchy problem (3.1) in X . Moreover, the solution depends continuously on the initial data, i.e. the mapping which maps initial data y to unique local solutions y of (3.1) is continuous from a ~ ~ neighborhood of y in Y to C([0; T ]; Y ) for suciently small 0 < T  T . The rst step of our proof provides a reformulation of equation (1.1) into the form (3.1). To simplify the notation, let us consider the quasilinear evolution equation u u = u + u + uu t xxt 1 x 2 xxx 3 x (3.2) 2 3 + u u + uu + uu u + u u + u ; 1 x xx 2 xxx 1 x xx 2 xxx 3 x WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 7 with  > 0 and ; ; ; ; ; ; ; 2 R. Let  denote the rst order pseudo 1 2 3 1 2 1 2 3 2 1=2 di erential operator (1 @ ) . If the coecients , i = 1; 2; 3, satisfy = 2( + ); (3.3) 1 2 3 an application of  to (3.2) yields the abstract form u + A(u)u = f (u); (3.4) where the quasilinear part A(u) is given by 1 2 A(u) =  ( + u + u )@ ; (3.5) 2 2 2 x and the semilinear part f (u) satis es 2 3 2 2 1 2 2 2 3 2 2 f (u) =  @ + u + + u + u + u + ( 2 )uu : x 1 3 2 x x 2 2 3 2 (3.6) Remark 3.2. The above procedure requires  6= 0; if  is negative, the linear part of (3.2) becomes ill-posed, which explains the assumption that  is strictly positive. If (3.3) is satis ed, which is equivalent to the property that the cubic -terms can be written as an x-derivative, then (3.2) yields a scalar conservation law of the form u + (u) = 0. This is an immediate consequence of the reformulation t x (3.4) with (3.5) and (3.6), since the operators  and @ commute. Indeed, the commutation property of these operators can be seen from their representations 1 2 1 2 1 ^ ^ @ f = F (inf );  f = F (1 + n ) f ; x n n where F : C (T) ! S (Z) denotes the Fourier transform and (f ) is the sequence n n2Z of Fourier coecients associated with the Fourier series of f 2 C (T); here S (Z) denotes the Schwartz space of all rapidly decreasing sequences. The reformulation of (1.1) as a scalar conservation law opens the way to study global weak solutions of class C([0;1); H ) by introducing a distributional for- mulation of the corresponding Cauchy problem via multiplication of the scalar conservation law by suitable test functions (t; x) and integration over the time and space domain. This low-regularity class includes|unlike H with s > 3=2| functions with peaks or even cusps; we refer to [25] for the existence of weak traveling wave solutions for (1.1) exhibiting such singularities. Since (1.1) is of the form (3.2) and satis es (3.3) it admits the reformulation in (3.4). We may now normalize the coecients in this reformulation, since their particular values do not play a role in the subsequent analysis: they merely contribute to estimation constants, whose precise value is not of special interest. Therefore, we may henceforth consider the abstract operator equation u + A(u)u = f (u); (3.7) where A(u) = a(u)@ with a(u) = (1 + u + u ) (3.8) and 2 2 3 2 2 2 1=2 f (u) =  @ u + u + u + u + uu with  = (1 @ ) : (3.9) x x x s1 s As underlying function spaces we choose X := (H ;kk ) and Y := (H ;kk ) s1 s per per with s > 3=2, and consider the topological isomorphism S :=  : Y ! X between these spaces, which de nes an isometry, i.e. kuk = kuk for all u 2 H . For s1 s per 8 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR a given function g 2 H with r > 1=2 let us denote by M the corresponding per r r r r multiplication operator on H , i.e. M : H ! H ; w 7! gw. Since H , per per per per r > 1=2, is closed under multiplication, M is continuous. The subsequent lemmas show that the quasilinear part A and the semilinear part f given by (3.8) and (3.9), respectively, ful ll the assumptions (I){(IV) of Theorem 3.1. Lemma 3.3. For a 2 H the operator per 2 1 A := M @ = a(x)@ ; D(A) := fw 2 L : aw 2 H g a x x per per generates a strongly continuous semigroup T (t) in L which satis es per !t kT (t)k  e for all t  0 with ! = supja j: (3.10) 0 x x2T Proof. The idea of the proof is to reveal A as a bounded perturbation of a skew- adjoint operator. Then Stone's theorem and a standard perturbation result yield the assertion of the lemma. 2 2 Step 1. We rst observe that A : L  D(A) ! L is indeed a well-de ned per per linear operator, i.e. the action of A on an element w 2 D(A) is uniquely determined 2 2 and lies in L . To see this, we rst consider a general element w 2 L . Since the per per 1 1 1 usual product H  H ! H admits an extension to a continuous product per per per 1 1 1 H  H ! H , Leibniz' formula for general periodic distributions restricted per per per to this product yields that (aw) = a w + aw in H : x x x per For w 2 D(A) we have that both (aw) and a w lie in L , hence x x per Aw = (aw) a w = aw 2 L : x x x per Obviously H is contained in D(A), so A is densely de ned. per 1 2 Step 2. We show that C is a core for A in L , i.e. for every v 2 D(A) there per per 1 2 exists a sequence (v ) in C such that v ! v and Av ! Av in L . Let n n n n per per v 2 D(A). We may assume that supp(v)\ [0; 1] is contained in (0; 1), otherwise we can employ a suitable parametrization of T with a subordinated smooth partition of unity. In this way we can consider the restriction of v to (0; 1) and extend it to the whole real line by setting the extension identically 0 outside of (0; 1). The resulting function lies in L (R) and we denote it again by v. Now it suces to prove that there exists a sequence (v ) in C (R) with supp(v )  (0; 1) such that n n n v ! v and Av ! Av in L (R): n n Let  2 C (R) be a molli er with supp()  [0; 1] and  = 1, and consider v := v, where  is the convolution on R and  (x) := n(nx). Then v 2 C ((0; 1)) n n n for n  1 large enough and v ! v in L (R). We have to prove that (av ) ! (av) in L (R): (3.11) n x x By assumption and our considerations we have that both (av) and a v in L (R), x x thus also av 2 L (R), and hence (av ) ! av in L (R): (3.12) n x x WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 9 s1 1 Since a 2 H we may identify it as an L (R) element, thus the multiplication per with a constitutes a continuous operation in L (R) and we nd that a v ! a v in L (R): (3.13) x n x In view of the identity (av ) (av) = [a v a v] + [  (av ) av ] + [a(v )   (av )]; n x x x n x n x x n x n x which holds in L (R), together with (3.12) and (3.13) it suces to show that a(v )   (av ) ! 0 in L (R) (3.14) n x n x in order to infer (3.11). To this end, we consider the linear operators P v = a(v )   (av ); n  1: n n x n x 1 2 If v is regular enough, say v 2 H (R), then obviously P v ! 0 in L (R). In the following we show that the family fP g extends to a family of uniformly bounded n n1 linear operators in L (R). A well-known corollary of the uniform boundedness 1 2 principle and the density of H (R)  L (R) then yield that the sequence of operators (P ) converges pointwise to the trivial operator P  0 in L(L (R)); this implies n n1 in particular that (3.14) holds for our given v 2 D(A). We rst observe that every P extends to a continuous linear operator on L (R). Note that P v = a(( )  v) ( )  (av) +   (a v); n n x n x n x and for a general v 2 L (R) and arbitrary x 2 R we obtain that P v(x) = ( ) (y) a(x) a(x y) v(x y) dy +   (a v) (x) n n y n x = n  (ny) a(x) a(x y) v(x y) dy +   (a v) (x): y n x By recalling that supp()  [0; 1], we can write n  (ny) a(x) a(x y) v(x y) dy 1 h  i y y = n  (y) a(x) a x v x dy: n n Thus in view of the mean value theorem, we obtain the estimate n  (ny) a(x) a(x y) v(x y) dy supja j j (y)jjyj v x dy: x y x2R 0 10 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR 2 2 2 Let now C = sup ja j  (y)y dy. Then the Cauchy-Schwarz inequality and x2R y Fubini's theorem yield that n  (ny)(a(x) a(x y))v(x y) dy R 0 Z Z 1 h  i C v x dy dx R 0 Z Z 1 h  i = C v x dx dy 0 R = Ckvk : Moreover, we obtain by Young's inequality that k  (a v)k  ka vk  sup(ja j)kvk : n x 0 x 0 x 0 x2R We conclude that kP k 2  K n L(L ) per holds uniformly for all n  1 with K = ( C + sup ja j). It follows that C is a x2R per core for A in L . per : : Step 3. We show that A = A + M with domain D(A ) = D(A) is skew- 0 a 0 adjoint in L . Let A denote the adjoint of A with domain D(A ), which is per 0 0 well-de ned since A (and hence A ) is densely de ned. Fix w 2 D(A ), then 2 1 A w 2 L , and for ' 2 C  D(A ) we have that (A '; w) = ('; A w) . We 0 0 0 0 0 per per 0 consider the linear functional F : C ! R given by per F (') = (A '; w) = a' + a ' w dx = ('; A w) : w 0 0 x x 0 Obviously, F is continuous with respect to the L -norm. Integration by parts per yields that F (') coincides with the action of the regular periodic distribution 1 2 1 A w = (aw + a w) 2 L on the test function '. By density of C in 0 x x per per 2 1 L we infer that aw 2 H and A w = A w, i.e. w 2 D(A ), which implies 0 0 per per 0 A  A . For v 2 D(A ) let v :=   v. By integration by parts and the second step of 0 n n the proof we obtain for every w 2 D(A ) that (A w; v) = lim (A w; v ) = lim aw + a w v dx 0 0 0 n 0 x x n n!1 n!1 = lim a(v ) + a v w dx (3.15) n x x n n!1 2 = av + a v w dx = (w; A v) ; x x 0 0 in particular v 2 D(A ) and A  A . 0 0 Step 4. By the previous step the operator iA is self-adjoint and therefore Stone's theorem implies that A is the in nitesimal generator of a strongly continuous semigroup of contractions in L . A standard perturbation argument (see e.g. [36, per Thm. 3.1.1]) now yields that A is the generator of a strongly continuous semigroup T (s) in L , which satis es the estimate (3.10) by noting that a is a bounded per WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 11 function due to Sobolev's embedding theorem and therefore the operator norm of 2 2 M : L ! L can be estimated by sup ja (x)j. a x x per per x2T Let [;] denote the usual commutator of two operators. In the course of our analysis we will make use of commutator and product estimates taken from [33, Proposition B.10.(2)] and [31, Lemma A1], respectively. These results are stated for functions and distributions de ned on the real line (or R ); the corresponding assertions on the periodic domain T hold true analogously. We summarize these estimates formulated for one-dimensional periodic functions in the following lemma. Lemma 3.4. Let r > 1=2. (i) If 1=2 < t  r + 1, there exists a constant C > 0 such that r;t [ ; M ]h  C kgk khk g r;t r+1 t1 r+1 t1 for all g 2 H and h 2 H . per per (ii) If r < t  r, there exists a constant C > 0 such that r;t kfgk  C kfk kgk t r;t r t r t for all f 2 H and g 2 H . per per Lemma 3.5. For a 2 H the operator per s1 s ~ ~ : : A = M @ = a(x)@ ; D(A) = fw 2 H : aw 2 H g a x x per per s1 generates a strongly continuous semigroup T (t) in H which satis es per ! ~t kT (t)k  e for all t  0 with ! ~ = ka k : s1 x s1 Proof. Recall from Lemma 3.3 that the operator A = M @ with domain D(A) = a x 2 1 2 fw 2 L : aw 2 H g lies in G(1; !; L ) for ! = sup ja j=2. per per per x2T s1 1s 2 Let A :=  A . In the following we show that A A 2 L(L ). Since 1 1 per s1 1s s1 1s @ commutes with  and  , we may write A A = [ ; M ] @ . Let x 1 a x 2 1s 2 s2 w 2 L be arbitrary. Since  @ 2 L(L ; H ) we can directly apply Lemma per per per 3.4 (i) to infer that s1 1s k(A A)wk = k[ ; M ] @ wk 1 0 a x 0 1s Ckak k @ wk s x s2 Ckak kwk ; s 0 where the generic constant C solely depends on s. We just showed that A is a bounded perturbation of A, it therefore generates a strongly continuous semigroup (cf. [36, Thm. 3.1.1]); more speci cally, A 2 2  s1 s1 2 G(1; ! ; L ) for ! = ! + kA Ak 2 . Since  : H ! L is an per L(L ) per per per s1 (isometric) isomorphism, it follows from [36, Thm. 4.5.8] that H is A-admissible. per s1 That is, H is an invariant subspace of T (t), t  0, the strongly continuous per s1 semigroup generated by A, and the restriction of T (t) to H is strongly continuous per s1 with respect to kk . By [36, Thm. 4.5.5], the A-admissibility of H entails s1 per s1 ~ ~ that A, the part of A in H , generates the strongly continuous semigroup T (t), per s1 which equals the restriction of T (t) to H . The assertion of the Lemma follows per immediately.  12 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR A straight forward application of Lemma 3.5 shows that the operators A(u), u 2 H , which are de ned in (3.8), satisfy the assumption (I) of Kato's Theorem 3.1. per The following lemma establishes assumption (II). s s s1 Lemma 3.6. The operator A maps H into L(H ; H ). More precisely, the do- per per per s s s1 main D(A(u)) contains H , and the restriction A(u)j belongs to L(H ; H ) per per per per for any u 2 H . Furthermore, A is Lipschitz continuous in the sense that for all per r > 0 there exists a constant C which only depends on r such that kA(u) A(v)k s1  C ku vk (3.16) s 1 s1 L(H ;H ) per per for all u; v inside B (0)  H . per s s Proof. Recall that H is by de nition contained in D(A(u)) for arbitrary u 2 H per per s s1 (cf. Lemma 3.5). The restriction A(u)j belongs to L(H ; H ), since @ 2 H x per per per s s1 s1 L(H ; H ) and M 2 L(H ). a(u) per per per To see that the required estimate is satis ed, let u; v; w 2 H be arbitrary. per 2 2 k(A(u) A(v))wk = k(u v + u v )@ wk s1 x s1 2 2 C (ku vk +ku v k )k@ wk s1 s1 x s1 C (1 +kuk +kvk )kwk ku vk ; s1 s1 s s1 where C denotes a generic constant. This shows that for every r > 0 one can always nd a constant C such that (3.16) holds uniformly for all u; v 2 B (0)  H . 1 r per The two lemmas below verify the assumptions (III) and (IV), respectively; their proofs rely on Lemma 3.4. Lemma 3.7. For any u 2 H there exists a bounded linear operator B(u) 2 per s1 1 s s1 L(H ) satisfying B(u) = A(u) A(u), and B : H ! L(H ) is uniformly per per per bounded on bounded sets in H . Furthermore, for all r > 0 there exists a constant per C , which depends only on r, such that kB(u) B(v)k s1  C ku vk s (3.17) 2 H L(H ) per per for all u; v 2 B (0)  H . per s 1 Proof. Let u 2 H . Since @ commutes with  and  we obtain that per 1 1 B(u) = M @  M @ = [; M ] @ : x x x a(u) a(u) a(u) s1 Hence we can write  B(u) as s1 s 1 s1 B(u) = [ ; M ] @ + [M ;  ]@ : a(u) x a(u) x s1 s Let now w 2 H and u; v 2 H be arbitrary. In view of the above identity and per per Lemma 3.4 (i) we obtain the following estimate s1 B(u) B(v) w =  B(u) B(v) w s1 0 s 1 s1 [ ; M ] @ w + [ ; M ]@ w a(u)a(v) x a(u)a(v) x 0 0 Cka(u) a(v)k  @ w + @ w s x x s1 s2 2 2 C ku vk +ku v k kwk ; s s s1 where C is a generic constant independent of u; v and w. In particular, this shows s1 s that B(u) extends to a bounded linear operator on H for every u 2 H such that per per WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 13 s s1 s B : H ! L(H ) is uniformly bounded on bounded sets in H . Furthermore, per per per this estimation proves that there exists a constant C depending only on the radius of the ball B (0)  H such that (3.17) is satis ed for all u; v 2 B (0). r r per s s s1 Lemma 3.8. The map f : H ! H is locally H -Lipschitz continuous in the per per per sense that for every r > 0 there exists a constant C > 0, depending only on r, such that kf (u) f (v)k  C ku vk for all u; v 2 B (0)  H ; (3.18) s1 3 s1 r per and locally H -Lipschitz continuous in the sense that for every r > 0 there exists a per constant C > 0, depending only on r, such that kf (u) f (v)k  C ku vk for all u; v 2 B (0)  H : s 4 s r per Proof. Recall that 2 2 3 2 2 f (u) =  @ u + u + u + u + uu : x x s s1 Let u 2 H , then u is contained in H , which is closed under multiplication per per s1 2 since s > 3=2, hence polynomials in u and u lie again in H . Since  @ maps x x per r r+1 H continuously to H for arbitrary r 2 R, it is clear that f is well-de ned per per s s and continuous as a mapping from H to H . An application of Lemma 3.4 (ii) per per yields the following estimation for arbitrary u; v 2 H : per 2 2 3 3 2 2 2 2 kf (u) f (v)k  C u v + u v + u v + u v + uu vv s1 x x x x s2 2 2 C 1 +ku + vk +ku + uv + v k ku vk s1 s1 s1 +ku + v k ku v k +ku k ku vk x x s1 x x s2 x s1 s1 +kv(u + v )k ku v k ; x x s1 x x s2 where C denotes a generic constant being independent of u and v. This shows that for every r > 0 there exists a constant C > 0 such that (3.18) holds for all u; v 2 B (0)  H . per Similarly, elementary estimates (which do not rely on the product inequality in Lemma 3.4) yield that f is locally H -Lipschitz continuous; we omit the details. per It is now an immediate consequence of Theorem 3.1 in conjunction with Lemmas 3.5, 3.6, 3.7 and 3.8, that the periodic Cauchy problem for equation (3.7) with A and f given by (3.8) and (3.9), respectively, is locally well-posed for initial data in H , s > 3=2, such that the unique local solution u with maximal life span per s 1 s1 T 2 (0;1] lies in C([0; T ); H ) \ C ((0; T ); H ) and depends continuously on per per the initial datum. As noted earlier, the particular values of the coecients in (3.8) and (3.9) do not play a role when it comes to local well-posedness. Therefore the periodic Cauchy problem for equation (1.1) is locally well-posed as well (in exactly the same sense), which nishes the proof of Theorem 2.1. Remark 3.9. Our proof of Theorem 2.1 can be directly adapted to the corre- sponding non-periodic initial value problem on the real line. This is not surprising, since Sobolev's embedding theorem, as well as product inequalities and commutator estimates possess analogous versions for the periodic and non-periodic case. Note that we carried over the crucial step in the proof of Lemma 3.3 to the corresponding non-periodic scenario by means of a smooth partition of unity; in particular one 14 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR immediately obtains an analogous non-periodic version of this lemma with the same proof. One delicate aspect of the proof is|apart from the application of the nontrivial commutator estimates in Lemma 3.4|the approximation technique in the second step in the proof of Lemma 3.3, which is inspired by the arguments in [6]. The domain D(A ) of the operator A in the proof of Lemma 3.3 contains H as 0 0 per a proper subspace, i.e. an element v 2 D(A ) does in general not lie in H . per Consequently the integration by parts formula is not directly applicable to the L per inner product (A w; v) in (3.15). 0 0 As noted in the introduction, the local well-posedness problem for (1.1) has been s1 earlier studied in [40] in the analogous framework on the real line, i.e. X = H (R) and Y = H (R). The authors chose basically the same approach, i.e. Kato's theory was applied to the nonlocal rst order reformulation of (1.1). A notable weakness of this paper is the presentation of the proofs: e.g. the integration by parts formula is applied under heavy abuse of notation in an unjusti ed way in the proof of [40, Lemma 3.1], the proof of [40, Lemma 3.3] is incomplete (only the trivial local H - s1 Lipschitz estimate is shown, the less trivial H -estimation is omitted), nontrivial commutator estimates are applied without providing any references|to mention just a few aws. Let us nally comment on some open questions regarding the shallow water wave equation (1.1). It is up to this point unknown whether or not a solution of (1.1) remains bounded in L on its whole interval of existence for the full space of initial data. This would give an answer to the important question whether or not wave- breaking is the only possible blow-up scenario (i.e. ku(t)k 1 remains bounded, L (R) whereas lim sup ku (t)k 1 = 1 whenever the maximal life span T is nite). x L (R) t%T Moreover, it is an open question if blow-up|or more speci cally wave-breaking|of solutions actually occurs for certain speci ed classes of initial data; neither it is known|apart from the traveling wave solutions established in [25]|whether there exist nontrivial initial wave pro les giving rise to global solutions. Acknowledgments The authors acknowledge the support of the Erwin Schr odinger International In- stitute for Mathematics and Physics (ESI) during the program \Mathematical Aspects of Physical Oceanography". R. Quirchmayr acknowledges the support of the European Research Council, Consolidator Grant No. 682537. References [1] T. B. Benjamin, J. L. Bona and J. J. Mahoney, Model equations for long waves in nonlinear dispersive systems, Phil. Trans. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 227, 47{78 (1972). [2] J. Boussinesq, Essai sur la theorie des eaux courantes, M emoires pr esent es par divers savants a l'Acad. des Sci. Inst. Nat. France, XXIII (1877). [3] R. Camassa and D. Holm, An integrable shallow water equation with peaked solitons, Phys. Rev. Lett. 71, 1661{1664 (1993). [4] A. Constantin, Nonlinear Water Waves with Applications to Wave-Current Interactions and Tsunamis, CBMS-NSF Regional Conference Series in Applied Mathematics 81, SIAM, Philadelphia (2011). [5] A. Constantin and J. Escher, Global Existence and Blow-up for a Shallow Water Equation, Ann. Scuola Norm. Sup. Pisa Cl. Sci. 26(4), 303{328 (1998). WELL-POSEDNESS OF A HIGHLY NONLINEAR SHALLOW WATER EQUATION 15 [6] A. Constantin and J. Escher, Well-Posedness, Global Existence, and Blowup Phenomena for a Periodic Quasi-Linear Hyperbolic Equation, Comm. Pure Appl. Math. 51, 475{504 (1998). [7] A. Constantin and J. Escher, Wave breaking for nonlinear nonlocal shallow water equations, Acta Math. 181, 229{243 (1998). [8] A. Constantin, R. I. Ivanov and J. Lenells, Inverse scattering transform for the Degasperis- Procesi equation, Nonlinearity 23, 2559{2575 (2010). [9] A. Constantin and R. S. Johnson, On the Non-Dimensionalisation, Scaling and Resulting Interpretation of the Classical Governing Equations for Water Waves, J. Nonlinear Math. Phys. 15, 58{73 (2008). [10] A. Constantin and D. Lannes, The Hydrodynamical Relevance of the Camassa-Holm and Degasperis-Procesi Equations, Arch. Rational Mech. Anal. 192, 165{186 (2009). [11] A. Constantin and H. P. McKean, A shallow water equation on the circle, Comm. Pure Appl. Math. 52, 949{982 (1999). [12] A. Degasperis, D. D. Holm and A. N. W. Hone, A new integrable equation with peakon solutions, Theor. Math. Phys. 133, 1463{1474 (2002). [13] P. G. Drazin and R. S. Johnson, Solitons: an introduction, Cambridge Univ. Press, Cambridge, [14] A. Degasperis and M. Procesi, Asymptotic integrability In: Degasperis, A. and Gaeta, G. (eds.) Symmetry and Perturbation Theory, pp. 23{37. World Scienti c, Singapore (1999). [15] N. Duruk Mutluba s, On the Cauchy problem for a model equation for shallow water waves of moderate amplitude, Nonlinear Anal. Real World Appl. 14(5), 2022{2026 (2013). [16] N. Duruk Mutluba s, Local well-posedness and wave breaking results for periodic solutions of a shallow water equation for waves of moderate amplitude Nonlinear Anal. Theory Methods Appl. 97, 145{154 (2014). [17] N. Duruk Mutluba s and A. Geyer, Orbital stability of solitary waves of moderate amplitude in shallow water, J. Di erential Equations 255 (2), 254-263 (2013). [18] N. Duruk Mutluba s, A. Geyer and B. Matioc, Non-uniform continuity of the ow map for an evolution equation modeling shallow water waves of moderate amplitude, Nonlinear Anal. Real World Appl. 17(6), 322{331 (2014). [19] J. Escher, Y. Liu and D. Yin, Global weak solutions and blow-up structure for the Degasperis- Procesi equation J. Funct. Analysis 192, 457{485 (2006). [20] L. Fan and W. Yan, The Cauchy problem for shallow water waves of large amplitude in Besov space, J. Di er. Equ. 267(3), 1705{1730 (2019). [21] B. Fuchssteiner and A. S. Fokas, Symplectic structures, their B acklund transformations and hereditary symmetries, Physica D 4, 47{66 (1981). [22] A. Gasull and A. Geyer, Traveling surface waves of moderate amplitude in shallow water, Nonlinear Anal. Theory Methods Appl. 102, 105-119 (2014). [23] A. Geyer, Solitary traveling waves of moderate amplitude, J. Nonlinear Math. Phys. 19, 104{115 (2012). [24] A. Geyer, Symmetric waves are traveling waves for a shallow water equation modeling surface waves of moderate amplitude, J. Nonlinear Math. Phys. 22(4), 545{551 (2015). [25] A. Geyer and R. Quirchmayr, Traveling wave solutions of a highly nonlinear shallow water equation, Discrete Contin. Dyn. Syst. 38, 1567{1604 (2018). [26] R. I. Ivanov, On the Integrability of a Class of Nonlinear Dispersive Wave Equations J. Nonlinear Math. Phys. 12, 462{468 (2005). [27] R. I. Ivanov, Water waves and integrability, Philos. Trans. Roy. Soc. London A 365, 2267{2280 (2007). [28] R. S. Johnson, Camassa-Holm, Korteweg-de Vries and related models for water waves, J. Fluid Mech. 455, 63{82 (2002). [29] R. S. Johnson, A Modern Introduction to the Mathematical Theory of Water Waves Cambridge Univ. Press, Cambridge (1997). [30] T. Kappeler and J. P oschel, KdV & KAM, Ergeb. der Math. und ihrer Grenzgeb., Springer, Berlin-Heidelberg-New York, 2003. [31] T. Kato Quasi-linear equations of evolution, with applications to partial di erential equa- tions In: Spectral Theory and Di erential Equations, pp. 25-70. Springer Lecture Notes in Mathematics 448, Berlin (1975). [32] D. J. Korteweg and G. de Vries. On the change of form of long waves advancing in a rectangular canal and on a new type of long stationary waves. Phil. Mag. 39 422{443 (1895). 16 N. DURUK MUTLUBAS, A. GEYER, AND R. QUIRCHMAYR [33] D. Lannes, The Water Waves Problem: Mathematical Analysis and Asymptotics, American Math. Soc., Providence, RI (2013). [34] J. Lenells, Traveling wave solutions of the Camassa-Holm equation, J. Di er. Equ. 217, 393{430 (2005). [35] J. Lenells, Traveling wave solutions of the Degasperis-Procesi equation, J. Math. Anal. Appl. 306, 72{82 (2005). [36] A. Pazy, Semigroups of Linear Operators and Applications to Partial Di erential Equations Springer, New York (1983). [37] R. Quirchmayr, A new highly nonlinear shallow water wave equation, J. Evol. Equations 16, 539{567 (2016). [38] J. S. Russell, Experimental researches into the laws of certain hydrodynamical phenomena that accompany the motion of oating bodies, and have not previously been reduced into conformity with the known laws of the resistance of uids, Trans. R. Soc. Edinburgh 14, 47{109 (1839). [39] T. Tao, Low-regularity global solutions to nonlinear dispersive equations In: Surveys in Analysis and Operator Theory, pp. 19{48, Proc. Centre Math. Appl. Austral. Nat. Univ. 40 (2002). [40] S. Yang and T. Xu, Well-posedness and persistence property for a shallow water wave equation for waves of large amplitude, Appl. Anal. 98(5), 981{990 (2017). [41] S. Zhou, Well-posedness and wave breaking for a shallow water wave model with large amplitude, J. Evol. Equ., 20(1), 141{163 (2020).

Journal

Mathematics – arXiv (Cornell University)

Published: Nov 2, 2020

References

  • Model equations for long waves in nonlinear dispersive systems
    Benjamin
  • Essai sur la theorie des eaux courantes
    Boussinesq
  • An integrable shallow water equation with peaked solitons
    Camassa
  • Nonlinear water waves with applications to wave-current interactions and tsunamis
    Constantin
  • Global existence and blow-up for a shallow water equation
    Constantin
  • Global existence and blowup phenomena for a periodic quasi-linear hyperbolic equation
    Constantin
  • Wave breaking for nonlinear nonlocal shallow water equations
    Constantin
  • Inverse scattering transform for the degasperis-procesi equation
    Constantin
  • On the non-dimensionalisation, scaling and resulting interpretation of the classical governing equations for water waves
    Constantin
  • The hydrodynamical relevance of the Camassa–Holm and Degasperis-Procesi equations
    Constantin
  • A shallow water equation on the circle
    Constantin
  • A new integrable equation with peakon solutions
    Degasperis
  • Asymptotic integrability
    Degasperis