ebook img

Interaction of modulated gravity water waves of finite depth PDF

0.42 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Interaction of modulated gravity water waves of finite depth

INTERACTION OF MODULATED GRAVITY WATER WAVES OF FINITE DEPTH 6 1 IOANNISGIANNOULIS 0 2 Abstract. We consider the capillary-gravity water-waves problem of finite n depth withaflat bottom of one ortwo horizontal dimensions. We derive the a modulation equations of leading and next-to-leading order in the hyperbolic J scalingforthreeweaklyamplitude-modulatedplane-wavesolutionsofthelin- 1 earized problem in the absence of quadratic and cubic resonances. We fully 3 justifythederivedsystemofmacroscopicequationsinthecaseofpuregravity waves, i.e. in the case of zero surface tension, employing the stability of the ] P water-wavesproblemonthetime-scaleO(1/ǫ)obtainedbyAlvarez-Samaniego andLannes. A . h t a 1. Introduction m Asignificantpartintheresearchonwaterwavesisbasedonthestudyofasymp- [ totic limits derived from the original water-waves problem, which is considered as 1 providingthe complete descriptionof their behavior. The benefit of this method is v that these reduced models have, on the one hand, a clearer structure and, on the 5 5 other hand, they highlight a particular qualitative feature of the wave evolution. 2 Depending on the aspect of the nature of the water waves one is interested in, 0 one has to employ the relevant asymptotic scaling, obtaining in each case a differ- 0 ent macroscopic limit. Considering that from the outset the original water-waves . 2 problem has some fundamental characteristics, the result is a plethora of different 0 equations which are presumed to describe approximatively the behavior of water 6 waves in different situations and regarding different aspects. While the choice of 1 the relevant asymptotic scaling requires a thorough understanding of the initial : v model, from an analytical point of view the crucial question is the justification of i X the derived model, i.e. the proof that its solutions indeed are approximations of r solutions to the original problem. a Concerning the initial set-up of the water waves problem one could distinguish roughly between (a) gravity or capillary-gravitywaves (the latter ones taking into account together with gravity also the surface tension as driving forces for the evolution of the waves), (b) finite- or infinite-depth water, and (c) two- or three- dimensional space, where in the former case one considers a vertical plane in the water domain that contains the (dominant) direction of evolution, assuming that the waves are (nearly) constant in the direction normal to the plane. Of course, this is only a very rough classification(e.g. in the case of finite depth one can con- sider shallow or deep water, flat bottoms or bottoms with some (smooth or rough) topography, or even moving bottoms etc.), but it seems to be the prevalent one in the mathematical-analytical literature, where the fundamental question concerns Date:February2,2016. 2010 Mathematics Subject Classification. Primary76B15;76B45, 35B35,35L03,35L05. 1 2 IOANNISGIANNOULIS the well-posednessofthe water-wavesproblemandinparticularthe existence time of its solutions. In the case of gravity water waves in two dimensions and for infinite depth, first local well-posedness results were obtained by Nalimov [41] in 1974 for small Sobolev initial data, and by Shinbrot [48] and Kano and Nishida [32] for analytic initial data. The method of Nalimov was employed to prove local well-posedness in the case of finite depth by Yosihara [56] and by Craig [11], who obtained also first rigorous justification results of the Korteweg-de Vries (KdV) and Boussinesq approximations. However, in the case of infinite depth, the crucial breakthrough was made by the work of S. Wu, who presented localwell-posedness results for the two-andthree-dimensionalcaseswithoutsmallnessassumptionsontheinitialdata in[52],[53],whichsheextendedtoalmostglobalandglobalexistence,respectively, in [54], [55]. Independently, global existence in three dimensions was shown by Germain, Masmoudi, and Shatah in [20]. Inthecaseoffinitedepththefirstgeneralwell-posednessresultforgravitywaves inthree dimensions wasobtainedby Lannes[36] in2005. This resultwasextended in[2],wherethecharacteristicdimensionlessparametersoftheoriginalwater-waves problem have been worked out in order to provide a stable fundament for the derivation and justification of various asymptotic limits. The results obtained, are presented in more detail in the survey [37], to which we take explicit reference in the presentpaper. Concerningthe well-posedness ofcapillary-gravitywater-waves, we mention exemplarily only the more recent selection [5, 45, 24, 8, 47, 10, 4, 40, 1, 21, 38], to which we refer for more details on the various results obtained, their development and their extensions. As mentioned above, for each original water-wave problem with its own char- acteristics, different asymptotic limits can be obtained. The main distinction of the derivedmodels is with respect to the shallowness parameter µ=H2/L2 of the 0 originalequations,whereH >0 is the waterdepthandL=1 is the characteristic 0 horizontal length-scale. For µ 1 we speak of shallow water, while for µ 1 and ≪ ≈ µ > 1 of deep water (with the limiting case of infinite depth as µ ). This → ∞ classificationis not arbitrary. Indeed, the maindifference inthe behaviorof water- waves in these two cases is that for increasing water depth the rˆole of dispersion becomes more dominant, see, e.g., [37, 1.3]. Of course within each of these two § mainclassesofmodels,finerdistinctionscanbe,andindeedare,made. Sinceinthe present article we consider the deep (though finite) water case µ>1, we refrain to mentionanyofthevariousshallowwatermodels,butrefertothesurvey[37],which seems to give a complete accountof the ”state-of-the-art”in 2013 concerning their derivation and justification. However, we would like to mention the justification of the celebrated KdV-equation in the two-dimensional case in [44, 45] after some first results in [11, 33], and with improvements in [7, 25, 26]. We will address modulation equations further below, after presenting in the following the capillary-gravitywater waves equations. The capillary-gravity water waves problem of finite depth 0 < √µ < with ∞ a flat bottom extending over all of Rd, d = 1,2, can be written in the following non-dimensionalized form, due to Zakharov [57], Craig, C. Sulem, P.-L. Sulem [14, 15, 49], and Alvarez-Samaniego, Lannes [2, 37]: (1.1) ∂ U + (U)=0, U =(ζ,ψ)T, =( 1 , 2 )T, t Nǫ,σ Nǫ,σ Nǫ,σ Nǫ,σ INTERACTION OF MODULATED GRAVITY WATER WAVES OF FINITE DEPTH 3 where (1.2) 1 (U)= [ǫζ]ψ, Nǫ,σ −G ζ ǫ ǫ( [ǫζ]ψ+ǫ ζ ψ)2 (1.3) 2 (U)=ζ 1 ∇ + ψ 2 G ∇ ·∇ . Nǫ,σ − Bo∇· 1+ǫ2 ζ 2 2|∇ | − 2 1+ǫ2 ζ 2 (cid:16) |∇ | (cid:17) |∇ | Here, the unknown functionspof time t [0,T), T > 0, and space X Rd are the surface elevation ζ : [0,T) Rd R ∈and the trace ψ : [0,T) Rd∈ R of the × → × → velocity potential of the fluid at the surface. The scaling parameter 0 < ǫ 6 1 is the steepness of the wave, i.e., the ratio of the amplitude of the surface elevation above the still water level ζ =0 to the characteristic horizontal length L=1. The secondtermin (1.3) correspondsto the surface tension,which is essentially the mean curvature of the surface scaled by the (inverse) Bond number 1 = σ , Bo ρg where σ,ρ,g are the (dimensionless) coefficients of the surface tension, the fluid density and the gravityacceleration,respectively. When σ =0, this term is absent and we speak of gravity water waves. The most important term in the above formulation is the Dirichlet-Neumann operator (1.4) [ǫζ]ψ =∂zΦ(,ǫζ) (ǫζ) Φ(,ǫζ)= 1+ (ǫζ)2∂nΦ(,ǫζ) G · −∇ ·∇ · |∇ | · where the velocity potential of the fluid Φ solves thpe boundary value problem for the Laplace equation ∆X,zΦ=0, √µ6z 6ǫζ, (1.5) − (Φ(,ǫζ)=ψ, ∂zΦ(, √µ)=0 · · − in the fluid domain at time t>0, Ω = (X,z) Rd+1 : √µ6z 6ǫζ(t,X) , ǫ,t { ∈ − } with Dirichlet data ψ at the surface and Neumann boundary data at the bottom. With n in (1.4) being the upwardunit normalvector atthe surface ǫζ, we see that the Dirichlet-Neumann operator [ǫζ] relates the Dirichlet data ψ to the normal G derivative of the potential Φ at the surface, thus justifying its name. In particular, the first equation of the system (1.1), (1.6) ∂ ζ [ǫζ]ψ =0, t −G codifies the physical assumption that fluid particles at the surface stay there for all times. We note also that the Dirichlet-Neumann operator is linear in ψ but nonlinear in ζ. The second equation of the gravity water-wavesproblem (1.1) (with σ =0), ǫ ǫ( [ǫζ]ψ+ǫ ζ ψ)2 (1.7) ∂ ψ+ζ+ ψ 2 G ∇ ·∇ =0, t 2|∇ | − 2 1+ǫ2 ζ 2 |∇ | originates from the Euler equation for the fluid velocity (ǫΦ) of an inviscid, X,z ∇ homogeneous, incompressible, and irrotational fluid in Ω under the influence of ǫ,t gravity and with constant external (atmospheric) pressure at the surface. Inte- grating the Euler equation over the space variables (X,z) and evaluating it at the surface z =ǫζ, one gets 1 (1.8) ∂ (ǫΦ)(,ǫζ)+ (ǫΦ)(,ǫζ)2+ǫζ =0. t X,z · 2|∇ · | 4 IOANNISGIANNOULIS By use of the chain rule on ψ = Φ(,ǫζ), one obtains with (1.4) and (1.6) that · (1.8) is equivalent to (1.7). Thus, determining U = (ζ,ψ) via the system (1.6), (1.7), we have all required data to solve (1.5) for Φ (under reasonable regularity assumptionsonU andundertheconditionthattheflowisatrestas (X,z) ). | |→∞ From the Euler equation we can then determine also the pressure of the fluid. It was Zakharov who noted in [57] that the knowledge of U = (ζ,ψ) is sufficient for solvingthewater-wavesprobleminthisway,whiletheuseoftheDirichlet-Neumann operator (1.4) in the formulation of the system (1.6), (1.7) is mainly due to Craig, C. Sulem and P.-L. Sulem in [14, 15]. The non-dimensionalized version (1.1) of the water-waves problem, that we use forthe dispersive,deep watercaserelevantin this article,is relyingona moregen- eral one, derived by Alvarez-Samaniego and Lannes first in [2] and then presented in more detail in [37], which works out all characteristic parameters of the water wave problem. This is particularly useful for a systematic and analytically reliable derivationofallpossibleasymptoticlimitsonemaybeinterestedin. Sincetheulti- mategoalofthepresentarticleisthejustification(seeSection4)ofthemodulation equationsformallyderivedinSection3, andsinceourjustificationresult(Theorem 4.2) followsdirectly fromthe stability propertyof the water-wavesproblemas pre- sentedbyLannesin[37](seehereTheorem4.1),wechosetostudythewater-waves problem from the beginning in the form (1.1). This is also the reason for the (at a first glance unusual) notation of the water-depth by √µ. For a full derivation of the water-wavesproblemin the form (1.1), and an extended and detailed overview of its recent analytical state of the art, we refer the reader to [37]. The water-wavesproblem (1.1) has the linearization around (ζ,ψ)=(0,0) ∂ ζ [0]ψ =0, t (1.9) −G (∂tψ+ζ− B1o∆ζ =0, with [0]ψ = ∂ Φ(,0), where Φ solves (1.5) with Dirichlet data Φ(,0) = ψ at z G · · the surface ζ = 0. Considering the Fourier transform of (1.5) with respect to the horizontal variables X Rd, we obtain for each ξ Rd a second-order ODE for ∈ ∈ Φ(ξ, )intheverticalvariablez withboundaryvaluesatz = √µandz =0,which · − can be solved uniquely, yielding ∂ Φ(ξ,0)=g (ξ)ψ(ξ), g (ξ)= ξ tanh(√µξ ), ξ Rd. z 0 0 | | | | ∈ Thus, in the Fourier-multiplier notation d b (1.10) f\(D)u(ξ)=f(ξ)u(ξ), ξ Rd, with D = i , ∈ −∇ we obtain b [0]ψ = D tanh(√µD )ψ. G | | | | Moreover,we obtain that (1.9) allows for plane wave solutions of the form ζ (1.11) ei(ξX ωt)+c.c., ξ Rd, ω R, ζ,ψ C ψ · − ∈ ∈ ∈ (cid:18) (cid:19) (with c.c. denoting the complex conjugate of the preceding term(s)), provided the dispersion relation iω (1.12) ζ = ψ and ω2 =ω2(ξ), 1+ 1 ξ 2 Bo| | INTERACTION OF MODULATED GRAVITY WATER WAVES OF FINITE DEPTH 5 is satisfied, with the dispersion function (1.13) ω(ξ)= (1+ 1 ξ 2)g (ξ), g (ξ)= ξ tanh(√µξ ), ξ Rd. Bo| | 0 0 | | | | ∈ q Inthecaseoflinearsystemsonecanconstructmorecomplicatedsolutions(wave packets) by superposition of plane waves via Fourier transformation. The analog to this in nonlinear systems is the consideration of modulated plane waves. In the most simple case of amplitude modulation we replace the constants ζ,ψ C in ∈ (1.11) by slowly varying functions ζ(t,X ) (1.14) ′ ′ ei(ξX ωt)+c.c., ζ,ψ :[0, ) Rd C, ψ(t′,X′) · − ∞ × → (cid:18) (cid:19) where t =ǫt, X =ǫX with 0<ǫ61 are new, macroscopictime- and space-vari- ′ ′ ables. The question then is whether the nonlinear system allows approximatively forsolutionsofthisform. Bythiswemeansolutionswhichmaintain theaboveform at least locally with respect to the macroscopic time t 6 T, or, equivalently, for ′ t6T/ǫ. Typically, by inserting the two-scale ansatz (1.14) into the nonlinear sys- tem, one obtains formally the necessary conditions, viz. the modulation equations, whichthe macroscopicfunctionsζ,ψ havetosatisfy. Themodulationequationsre- veal some qualitative, macroscopic feature in the behavior of the nonlinear system under investigation, which depends of course strongly on the macroscopic scaling usedfor the modulation. This approachhasbeen usedwidely inthe physicslitera- tureforallsortsofdispersivesystems. Indeed,oneoftheoldestfieldsofapplication havebeen waterwaves,asis exemplifiedprominently inthe workofWhitham[51], to which we refer for a methodical exposition of the ideas behind modulation from the physical point of view. Innonlineardispersivesystems the centralmodulationequationisthe nonlinear Schr¨odingerequation(nlS),sinceitcapturestheinterplaybetweennonlinearityand dispersiongoverningthe deformation of the envelopes of the wave packets (see e.g. [49] for an overview). For this, the right two-scale ansatz is not (1.14) but rather ζ(t ,X ) (1.15) ′′ ′′ ei(ξX ωt)+c.c., ζ,ψ :[0, ) Rd C, ψ(t′′,X′′) · − ∞ × → (cid:18) (cid:19) witht =ǫt, X =X ωt,where ω isthe groupvelocityofthe wavepacket. ′′ ′ ′′ ′ ′ −∇ ∇ Inserting this ansatz (with a correspondingpolarizationconditionfor ζ) into (1.1), one obtains the nlS equation, which for two-dimensional gravity waves takes the form 1 ∂t′′ψ−i2ω′′∂x′′2ψ+ic|ψ|2ψ =0 with c R depending on ξ,ω and ∂ , ∂ denoting differentiation with respect to ∈ t′′ x′′ t , x . It was derived by Zakharov [57] for infinite depth and by Hasimoto and ′′ ′′ Ono [30] for finite depth. In the three-dimensional case of finite depth instead of the nlS one obtains for the scaling (1.15) the Davey-Stewartson (DS) system [16] (see[49]foradetaileddiscussionofitsproperties). However,ininfinitedepthagain the nlS is obtained as the modulation equation for the scaling (1.15). Concerning the justification of these modulation equations, this has been achievedfor the two- dimensionalgravitywater-wavesproblembyTotzandWu[50]inthecaseofinfinite depth and by Du¨ll, Schneider, and Wayne [18] in the case of finite depth. In the three-dimensional capillary-gravity case there exist consistency results for the nlS 6 IOANNISGIANNOULIS equation[12]andfortheDSsystem[13],whereconsistencymeansthattheamount by which the approximate solution fails to satisfy the original problem (i.e. the residual)tends to zeroin the asymptotic limit with respect to some relevantnorm. In the present article we use the hyperbolic scaling (1.14) and consider three modulatedplanewavesofthatform. Weareinterestedinthemodulationequations thatgovernthe macroscopicdynamics ofthese wavesnotonly inleading orderbut also with respect to their macroscopic corrections of order O(ǫ). For the sake of clarity, we first present our exact assumptions, and discuss them afterwards. We make the two-scale ansatz for approximate solutions of (1.1) ζ ζ ζ ζ (1.16) U = a = 0 +ǫ 1 +ǫ2 2 a ψ ψ ψ ψ a 0 1 2 (cid:18) (cid:19) (cid:18) (cid:19) (cid:18) (cid:19) (cid:18) (cid:19) with ζ ζ ζ 0 = 0j e +c.c.+ 00 , ψ ψ j ψ 0 0j 00 (cid:18) (cid:19) j (cid:18) (cid:19) (cid:18) (cid:19) X ζ ζ ζ ζ 1 = 1j e + 1ji e +c.c.+ 10 , ψ ψ j ψ ji ψ 1 1j 1ji 10 (cid:18) (cid:19) j (cid:18) (cid:19) ji (cid:18) (cid:19) (cid:18) (cid:19) X X ζ ζ ζ ζ ζ 2 = 2j e + 2ji e + 2jik e +c.c.+ 20 , ψ ψ j ψ ji ψ jik ψ 2 2j 2ji 2jik 20 (cid:18) (cid:19) j (cid:18) (cid:19) ji (cid:18) (cid:19) jik (cid:18) (cid:19) (cid:18) (cid:19) X X X defined according to the following notations and assumptions: Notation 1.1. (1) All functions ζ ,ψ on the right of (ζ ,ψ )T, n = 0,1,2, ... ... n n are complex-valued and depend only on the macroscopic time- and space-variables 06t =ǫt6T, X =ǫX =ǫ(x,y) Rd (d=1,2 and X =x R if d=1), where ′ ′ ∈ ∈ 0 < ǫ 6 1. Differentiation with respect to t and X is denoted by ∂ and . The ′ ′ t′ ∇′ abbreviation c.c. denotes the complex conjugate of all preceding terms. (2) We introduce the index-sets J = 1,2,3 , { } I = (1,1), (2,2), (3,3), (1, 2), (1, 3), (2, 3) (1,2),(1,3),(2,3) =I , < { ± ± ± }⊃{ } K = (1,1,1), (2,2,2), (3,3,3), { (1,1, 2), (1,1, 3), (2,2, 3),(2,2, 1), (3,3, 1), (3,3, 2), ± ± ± ± ± ± (1,2,3), (1,2, 3), (1,3, 2), (2,3, 1) . − − − } We denote summation over these index-sets by := , := , := . Xj Xj∈J Xji (jX,i)∈I Xjik (j,iX,k)∈K (3a) The functions e for j J, e for (j,i) I and e for (j,i,k) K are j ji jik ± ∈ ∈ ∈ defined through e j(t,X)= e±i(ξj·X−ωjt), eji = ejei, ejik = ejeiek, ± INTERACTION OF MODULATED GRAVITY WATER WAVES OF FINITE DEPTH 7 where the wave-vectors ξ Rd 0 and the frequencies ω =ω(ξ )>0 satisfy for j j j ∈ \{ } each j J the dispersion relation ω2 =ω2(ξ ) with the dispersion function (1.13). ∈ j j We assume that the plane waves e , j J, are mutually different, i.e. j ∈ (ξ ,ω )=(ξ ,ω ) j,i J, j =i. j j i i 6 ∀ ∈ 6 Occasionally, we will refer to e , e and e as the first-, second- and third-order j ji jik harmonics, respectively, and to 1= e0 as the zeroth-order harmonic. (3b) In analogy to the index-notation for the harmonics, we use the abbreviations ξ = ξ , ξ =ξ +ξ , ξ =ξ +ξ +ξ , j j ji j i jik j i k ± ± ω = ω , ω =ω +ω , ω =ω +ω +ω , j j ji j i jik j i k ± ± and b =1+ 1 ξ 2, b =1+ 1 ξ 2, b =1+ 1 ξ 2, j Bo| j| ji Bo| ji| jik Bo| jik| g =g (ξ ), g =g (ξ ), g =g (ξ ). j 0 j ji 0 ji jik 0 jik Finally, we denote g = g (ξ ), g = g (ξ ), H = 1 (ξ ) , j′ ∇ 0 j j′i ∇ 0 ji j 2∇′·Hg0 j ∇′ where (ξ) is the Hessian matrix of the function g at ξ Rd, see (1.13). Hg0 0 ∈ (Note, in particular, g0(0,0)=0, ∇g0(0,0)=(0,0) and Hg0(0,0)=2√µI.) In this notation the following identities hold true: (1.17) 2ω ω =b g + 1 2ξ g , where ω = ω(ξ ), j∇ j j j′ Bo j j ∇ j ∇ j and b 1 ω (1.18) (ξ ) ψ = j H ψ (b ω )2ψ + 1 j∆ψ , ∇′·Hω j ∇′ 0j ω j 0j − ω j∇Bo j ·∇′ 0j Bo b ′ 0j j j j where 1 ω (1.19) ω = ω 1 2 jξ . ∇Bo j b ∇ j − Bo b j j j (cid:16) (cid:17) (3c) The plane waves e , j J, satisfy the non-resonance conditions j ∈ ω2 =ω2(ξ )=b g (j,i) I ji 6 ji ji ji ∀ ∈ and ω2 =ω2(ξ )=b g (j,i,k) K. jik 6 jik jik jik ∀ ∈ (4) For u Hs(Rd), s R, see (4.3), of the form ∈ ∈ k u(X)= u˜i(X′)eiξi·X i=1 X we use the notations u˜ = k u˜ and | |Hs i=1| i|Hs k P k u′(X)= iξiu˜i(ǫX)eiξi·X, u′′(X)= ( ξi 2)u˜i(ǫX)eiξi·X, −| | i=1 i=1 X X such that (with ∆ = ) ′ ′ ′ ∇ ·∇ (1.20) u=u +ǫ u and ∆u=u +2ǫ u +ǫ2∆u. ′ ′ ′′ ′ ′ ′ ∇ ∇ ∇ · 8 IOANNISGIANNOULIS Themotivationforthespecialformoftheansatz(1.16)isthatwewanttoinclude inourformalexpansionofthecapillary-gravitywater-wavesequation(1.1)thecase ofquadraticinteractionoftwomodulatedplanewaves. Byquadraticinteractionwe mean the situation where two such waves generate by multiplication a third plane wave through the (quadratic) resonance of their phases, e.g. e e = e . In this 1 2 3 case one has to consider from the outset all three involved modulated plane waves in order to obtain a closed system of macroscopic equations, and the interaction is manifested macroscopicallyin leading order,which means that an expansionup to O(ǫ)-terms in (1.16) would be sufficient. However, as will be explained below, in the case of pure gravity waves, which is our main focus in the present paper, no such quadratic resonances arise. Such resonancesexistonlyifsurfacetensionisincludedintheoriginalwater-wavesequa- tion, see [46] for the two-dimensional case (d = 1). Then, naturally, the question arises, whether even in this quadratically non-resonant case any macroscopic cou- pling can be detected in the next-to-leading-order correction of the leading order amplitudes or in the non-oscillating mean field generated by the waves. Wanting to perform the (unsurprisingly, very cumbersome) formal expansion of the water waves equation for three modulated pulses only once, we chose the ansatz (1.16), which is usefull in both cases (quadratically resonant and non-resonant) and for waves with or without surface tension. As expected, indeed also in the quadratically non-resonantcase, the interaction of modulated waves can be traced in the second-order macroscopic system. More precisely,weobtaininSection3thatinorderfortheapproximationU of (1.16)to a satisfyformallythe waterwavesequation(1.1)uptoresidualtermsoforderO(ǫ3), i.e. (1.21) ∂ U + (U )=ǫ3(r1,r2)T, t a Nǫ,σ a 2 2 the macroscopic modulation equations ∂ ψ + ω ψ =0, t′ 0j ∇ j ·∇′ 0j (3.17) ∂∂tt′′ψ2ψ1j00+−∇√ωµj∆·∇′ψ′0ψ01j==PEjj(cid:16)(gj2−|ξj|2)∂t′+2ωbjjξj ·∇′(cid:17)|ψ0j|2, with j J = 1,2,3 and ∈ { } 1 b (3.9) Ej = i2∇′·Hω(ξj)∇′ψ0j −iψ0j 2ωj (gj2−|ξj|2)∂t′+ξj ·∇′ ψ00+E˜j j (cid:16) (cid:17) have to be satisfied, where E˜ consists of cubic products of the leading-order am- j plitudes ψ , see (3.10). The other macroscopic functions appearing in U can be 0j a determined via ψ ,ψ ,ψ or can be chosen arbitrarily. 0j 00 1j Ofcourse,moregenerallyonecouldconsideralsoanarbitrarynumberofN N, ∈ N >3, modulated plane waves,as was done for other physical settings, e.g. in [23, 22]. Inorderto keepthe presentationmoresimple andexplicit, we chose,however, to consider here only three pulses. Note that the results derived in the present paper can be used in order to obtain the (non-resonant) macroscopic dynamics up to next-to-leading order for two pulses or even for a single pulse, by equating the macroscopiccoefficientsofthe superfluouswavesto zero,cf. fortwo pulses e.g.[29] and for a single pulse [6] (see also Remark 3.3) and [49, 11.1], [37, 8.2.5]. § § INTERACTION OF MODULATED GRAVITY WATER WAVES OF FINITE DEPTH 9 Theansatz(1.16)consistsofallfirst-andhigher-orderharmonictermsexpected to arise up to order ǫ2 due to the nonlinear nature of the water wavesproblem, in- cludingthenon-oscillatingtermsforthemeanfieldwhicharisefromtheinteraction of a plane wave with its complex conjugate. Note here, that the index sets I and K of Notation 1.1(2) represent all possibly different second- and third-order har- monics. In this context, we point out that actually we are interested in the weakly nonlinear approximation of solutions to the water-wavesproblem (1.1) with ǫ=1, that is to say in an approximation of the form ǫU with U as in (1.16). However, a a the expansion of the water waves equation in Section 2 and the derivation of the modulationequationsinSection 3areperformedfor the equationinthe form(1.1) and the ansatz U in (1.16). a The reason why the ansatz U includes also terms of order ǫ2, although we are a interested only in an approximation U of up to next-to-leading-order terms of a,1 order ǫ (i.e. consisting only of the first two terms of U ), is that the determining a equationsforthefunctionsζ ,ψ ariseatorderO(ǫ2),cf.Corollary2.5. Moreover, 1j 1j for the justification of the approximation U over time scales of order O(1/ǫ) we a,1 need first to consider an approximation that satisfies (1.1) formally up to residual terms of order O(ǫ3), see Section 4 for the details, and [34] for a more general presentation of this approach. The two non-resonance conditions of Notation 1.1(3c) imply that none of the higher-order harmonic terms are plane waves (or non-oscillating). The first set of conditionsisessentialforthe formofthemodulationequations,see(3.17),yielding thatinleadingorderthe macroscopicamplitudes ψ arejust transportedwiththe 0j group velocity ω of the wave packet, without any macroscopic interaction. In j ∇ the opposite case, when quadratic resonances appear, one obtains the ’three-wave- interaction equations’, a coupled system of three semilinear transport equations for ψ containing for each j J quadratic products of the other two amplitudes, 0j ∈ according to the existing resonances, see [46]. As stated above, for pure gravity water waves this quadratic non-resonance condition is always satisfied. This is known since the 1960s, see [42], while the existence of quadratic resonances in the caseofcapillary-gravitywavesofinfinite depthwas firstprovenin[39]. For amore generaldiscussionofresonancesofwaterwaveswereferto[28,46]andthereferences given therein. For the sake of completeness we give in the following Remark 1.1 a shortanalyticalproofofthenon-existenceofquadraticresonancesforgravitywater waves of finite depth. Remark 1.1. Assume there are ξ ,ξ ,ξ +ξ Rd 0 , such that 1 2 1 2 ∈ \{ } (ω(ξ ) ω(ξ ))2 =ω2(ξ +ξ ) with ω(ξ)= ξ tanh(√µξ ). 1 2 1 2 ± | | | | q Sinceω is aradialfunction, takingsquare-rootsandpossiblyconsideringthe oppo- site of some wave-vector ξ and relabeling, the equation on the left can always be j written in the form ω(ξ )+ω(ξ )=ω(ξ +ξ ), ξ ,ξ Rd 0 . 1 2 1 2 1 2 ∈ \{ } 1 Mutiplying by µ4 >0 and setting ξ ξ ξ 2 1 2 x=√µξ >0, λ= | | >0, c= · [ 1,1], 1 | | ξ ξ ξ ∈ − 1 1 2 | | | || | 10 IOANNISGIANNOULIS solving this equation for ξ ,ξ Rd 0 is equivalent to finding roots (x,λ,c) 1 2 ∈ \{ } ∈ (0, ) (0, ) [ 1,1] of the function ∞ × ∞ × − r (x,λ,c)= (1+λ2+2cλ)tanh(x 1+λ2+2cλ) λtanh(xλ) √tanhx. 0 − − q But r (x,λ,c)6r (x,λ,1) and p p 0 0 g(x(1+λ)) g(xλ) r (x,λ,1)= − √tanhx, where g(y)= ytanhy, 0 √x − p with r (x,0,1)=0 and 0 d r (x,λ,1)=√x(g (x(1+λ)) g (xλ))<0 λ>0, 0 ′ ′ dλ − ∀ the latter due to the strict decreasing of tanhy+y(1 tanh2y) g′(y)= − , y >0. 2√ytanhy Hence, we conclude r (x,λ,c)<0 for all (x,λ,c) (0, ) (0, ) [ 1,1]. (cid:3) 0 ∈ ∞ × ∞ × − While the first (quadratic) non-resonance condition of Notation 1.1(3c) is es- sential for the form of the derived modulation equations, the second (cubic) non- resonance condition is much less so. Indeed, if the first non-resonance condition holds true, cubic resonances do not change the form of the modulation equations, but merely contribute additional cubic products of the leading order amplitudes ψ to the source term of one of the equations for the next-to-leading order ampli- 0j tudes ψ , if a third-order harmonic is equal to one of the three considered plane 1j waves. (Note, that in general if plane waves are generated through resonant inter- action one always has to include the generated wave in the original ansatz, here (1.16),inordertoobtainaclosedsystemofmodulationequations.) Forthe sakeof simplicity we do not consider these cases explicitly here, but prefer to impose the cubic non-resonance condition of Notation 1.1(3c) instead. Nevertheless, since the same justification result holds true for these modified modulation equations, the presentpapercoverscompletelythe justificationofthemodulationequationsupto next-to-leading order for three weakly amplitude-modulated gravity water waves, provided possibly existing cubic resonances generate only one of the three plane waves considered. We close this introduction by outlining the structure of the article and com- menting on its main results. In the following Section 2, after inserting the ansatz (1.16) for the approximation U into the water waves equation (1.1), we expand a with respect to the steepness parameter 0 < ǫ 1 up to residual terms of formal ≪ order O(ǫ3), for which we give estimates in the Sobolev norms used for the justifi- cationofthemodulationequations(3.17)inSection4. Thepreciseformulasforthe more involved,though structurally simple, macroscopic coefficients are given in an Appendix. Thus, as a byproduct, we provide a complete formal explicit expansion includingalltermsoforderε2 for threeamplitude modulatedplanewaveswiththe hyperbolic scaling t = ǫt, X = ǫX for the capillary-gravity water waves problem ′ ′ of finite depth, that can be used independently, also for only one or two waves. Then, in Section 3, we derive the necessary conditions on the macroscopic co- efficients of U in order for the latter to satisfy (1.1) up to the residual terms of a

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.