![Arxiv:1801.06086V1 [Nlin.PS] 18 Jan 2018 Late Time Dynamics, in the Absence of Dissipation](http://data.docslib.org/img/a900220b21f9f1363536b0a801566da4-1.webp)
Phononic Rogue Waves
E. G. Charalampidis, J. Lee and P. G. Kevrekidis Department of Mathematics and Statistics, University of Massachusetts Amherst MA 01003-9305, USA
C. Chong∗ Department of Mathematics, Bowdoin College, Brunswick, ME 04011, USA (Dated: January 19, 2018) We present a theoretical study of extreme events occurring in phononic lattices. In particular, we focus on the formation of rogue or freak waves, which are characterized by their localization in both spatial and temporal domains. We consider two examples. The first one is the prototypical nonlinear mass-spring system in the form of a homogeneous Fermi-Pasta-Ulam-Tsingou (FPUT) lattice with a polynomial potential. By deriving an approximation based on the nonlinear Schr¨odinger(NLS) equation, we are able to initialize the FPUT model using a suitably transformed Peregrine soliton solution of the NLS, obtaining dynamics that resembles a rogue wave on the FPUT lattice. We also show that Gaussian initial data can lead to dynamics featuring rogue wave for sufficiently wide Gaussians. The second example is a diatomic granular crystal exhibiting rogue wave like dynamics, which we also obtain through an NLS reduction and numerical simulations. The granular crystal (a chain of particles that interact elastically) is a widely studied system that lends itself to experimental studies. This study serves to illustrate the potential of such dynamical lattices towards the experimental observation of acoustic rogue waves.
I. INTRODUCTION
Extreme wave events, such as freak or rogue waves, are waves that seem to appear out of nowhere, and then vanish without a trace [1–3]. The term rogue wave was first coined to describe an ocean wave that has an amplitude greater than twice the significant wave height [1]. Based on the classical description of waves that assumes a Rayleigh distribution of wave heights, a rogue wave should be an extremely rare event [1]. The measurement of an ocean rogue wave (the Draupner wave) in 1995 initiated an intense interest in the subject of extreme events. It has been found that ocean rogue waves occur more regularly than the statistical description predicts [1], and a number of alternative mechanisms for the formation of rogue waves has been produced [1]. One such approach is through the derivation of simple modulation equations such as the nonlinear Schr¨odinger (NLS) equation from the underlying equations of motion [4]. The Peregrine soliton solution of the focusing NLS equation sits atop a finite background, and is localized in both space and time [5]. The maximum amplitude of the Peregrine soliton is three times greater than the background upon which it sits, and is therefore a prominent rogue wave candidate. Such structures have been studied in various media, including nonlinear optics [6–9], mode-locked lasers [11], superfluid helium [12], hydrodynamics [13–15], Faraday surface ripples [16], parametrically-driven capillary waves [17], plasmas [18], ultra-cold gases [19] and electrical transmission lines [20]. A unifying theme of these varied physical settings of rogue waves is the relevance of the NLS setting as an approximate model equation. Rogue waves in discrete systems are far less studied. One example of such a study concerns rogue waves in the integrable Ablowitz-Ladik lattice [21], which is known to have an exact solution that has similar properties as the NLS Peregrine soliton. At the level of granular systems, the pioneering work of [22] was the first one to recognize the potential of such systems for unusually large (rogue) fluctuations in arXiv:1801.06086v1 [nlin.PS] 18 Jan 2018 late time dynamics, in the absence of dissipation. The present study concerns a different discrete system, namely phononic lattices, which are systems that manipulate pressure waves (as opposed to photonic latices in which light waves are manipulated). Arguably, one of the most prototypical phononic lattice is the Fermi-Pasta-Ulam-Tsingou (FPUT) lattice, which describes a one-dimensional system of masses coupled through weakly nonlinear springs [23]. While the amount of research efforts in the direction of the FPUT lattice is immense (see the book [24], but also the recent review [25]), rogue waves in FPUT lattices have not been reported on, to the best of our knowledge. In the small amplitude limit, the NLS equation is once again a valid modulation equation, suggesting that Peregrine-soliton-type dynamics are possible in phononic lattices. To demonstrate that a phononic rogue wave could in principle be observed experimentally, we conduct a study in the case of an one-dimensional chain of beads interacting through Hertzian contacts, i.e. granular crystals. Over the
∗ [email protected] 2 2
3 12 3 (a) (b) Temporal Profile (x=0) (c) 2.5 10 Spatial Profile (t=0) 2.5
2 8 2 ) ) 2 k 1.5 k 1.5 ( 6 ( |A| ω ω 1 4 1
0.5 2 0.5
0 0 0 0 0.5 1 1.5 2 2.5 3 -5 0 5 0 0.5 1 1.5 2 2.5 3 k X,T k
FIG.FIG. 1. (a) 1. Dispersion(a) Dispersion relation relation for for the the monomer monomer FPUT FPUT lattice lattice with withK2K=2 1,= which 1, which consists consists of an of acoustic an acoustic branch branch only. (b) only. The (b) The spatialspatial profile profile of the of the Peregrine Peregrine soliton soliton at at tt == 0 and the the temporal temporal profile profile at x at=x 0.= (c) 0. Dispersion (c) Dispersion relation relation for the for dimer the FPUT dimer FPUT latticelattice with withK2 K=32 =/2and 3/2 and⇢ =0ρ =. 08,.8, which which consists consists of of an an acoustic acoustic branch branch (lower (lower branch) branch) and opticaland optical branch branch (upper (upper branch). branch). dynamiclast two responses. decades, granular At the crystals same time, have receivedit is possible considerable to easily attention, access as and is now arrange summarized the media in a wide in a range wide of range of reviews [26–32]. Granular crystals are remarkably tunable, which permits one to access weakly or strongly nonlinear configurationsdynamic responses. (homogeneous, At the periodic, same time, chains it is possible with impurities, to easily accesschains and with arrange local resonators, the media in disordered a wide range chains, of etc.). Theseconfigurations aspects make (homogeneous, the study of periodic, granular chains crystals with fascinating impurities, chains from both withfundamental local resonators, and disordered applied perspectives chains, and [30]. Themany remainder others). of These the paper aspects is make structured the study as follows. of granular In Sec. crystals II, we fascinating examine afrom homogeneous both fundamental FPUT and lattice. applied We derive a focusingperspectives NLS [31]. equation, which describes the modulation of small amplitude and rapidly oscillating plane waves in time andThe space. remainder The of Peregrine the paper soliton is structured of the as NLS follows. equation In Sec. is II, used we examine to initialize a homogeneous the FPUT FPUT system lattice. which We leads derive to rogue- likea wave focusing dynamics. NLS equation, The prediction which describes based on the the modulation NLS approximation of small amplitude coincides andrapidly with the oscillating numerical plane simulations waves in of the FPUTtime lattice and space. up until The Peregrine the formation soliton of of the NLS large equation amplitude is used wave. to initialize While the the FPUT NLS system approximation which leads sees to rogue- a decreasing andlike “vanishing” wave dynamics. of the The large prediction amplitude based wave on back the NLS towards approximation the background, coincides the with presence the numerical of a modulational simulations of instabilitythe causesFPUT the lattice formation up until of outward the formation propagating of the large waves amplitude from the wave. center While of the the lattice. NLS approximation We also explore sees a generalized decreasing pulse and “vanishing” of the large amplitude wave back towards the background, the presence of a modulational instability like initial data (in the form of Gaussian wave packets), which can lead to wide variety of behavior including soliton causes the formation of outward propagating waves from the center of the lattice. We also explore generalized pulse dynamics,like initial breathing data (in dynamics, the form of and Gaussian rogue wave wave packets), dynamics. which The can leadamplitude to wide of variety the initial of behavior value including “selects” soliton the type of observeddynamics, dynamics. breathing In Sec.dynamics, III we and consider rogue wave a diatomic dynamics. granular The amplitude crystal lattice. of the initial Using conditions a focusing “selects” NLS equation the type derived as anof envelope observed dynamics. approximation In Sec. of III this we diatomic consider a chain, diatomic we granular once again crystal use lattice. the Peregrine Using a focusing soliton solution NLS equation of the NLS equationderived to as initialize an envelope the lattice approximation dynamics. of We this find diatomic qualitatively chain, we similar once again behavior use the to that Peregrine reported soliton for thesolution homogeneous of FPUTthe lattice NLS equation for all massto initialize ratios the tested. lattice A dynamics. noteworthy We finding find qualitatively is that the similar sensitivity behavior to to boundary that reported e↵ects for depends the on the chosenhomogeneous mass FPUT ratio. latticeSec. IV for draws all mass conclusions ratios tested. and A discusses noteworthy future finding directions. is that the sensitivity to boundary effects appears to depend on the chosen mass ratio. Sec. IV draws conclusions and discusses future directions.
II. HOMOGENEOUS FERMI-PASTA-ULAM-TSINGOU LATTICES II. HOMOGENEOUS FERMI-PASTA-ULAM-TSINGOU LATTICES
A.A. Theoretical Theoretical Set-up Set-up
TheThe prototypical prototypical Fermi-Pasta-Ulam-Tsingou Fermi-Pasta-Ulam-Tsingou (FPUT) (FPUT) lattice lattice has has the the form form
u¨n = V 0(un+1 un) V 0(un un 1) (1) u¨n = V 0(un+1 un) V 0(un −un 1) (1) − − − with with 2 3 V 0(x) = K2x + K3x + K4x , (2) 2 3 V 0(x)=K2x + K3x + K4x , (2) where n I, with I a countable index set, and un = un(t) R is the displacement of the nth particle from equilibrium ∈ ∈ position at time t. Equation (1) with I = Z has the Hamiltonian where n I,withI a countable index set, and un = un(t) R is the displacement of the n-th particle from equilibrium 2 2 position at time t. Equation (1) with I = Z has the1 Hamiltonian H = u˙ 2 + V (u u ). 2 n n+1 − n n X∈Z 1 2 H = u˙ + V (un+1 un). The linear problem (i.e. when K = K = 0) is solved2 byn 3 4 n X2N i(kn+ωt) un(t) = e The linear problem (i.e. when K3 = K4 = 0) is solved by
i(kn+!t) un(t)=e for all k [0, ⇡], where ! and k are related through the dispersion relation, 2 2 2 !(k) =4K2 sin (k/2), 3 for all k [0, π], where ω and k are related through the dispersion relation, ∈ 2 2 ω(k) = 4K2 sin (k/2), such that the cutoff point of the acoustic band is 2√K2, see Fig. 1(a). Motivated by prior works on rogue waves where the Peregrine soliton is used to describe the formation of such structures, we first derive the NLS equation from Eq. (1). When deriving the NLS equation as a modulation equation, one uses the multiple scale ansatz
u (t) ψ (t) := ε (B(X,T ) + [A(X,T )E + c.c.]) ,E = ei(k0n+ω0t),X = ε(n + ct),T = ε2t, (3) n ≈ n where ε 1 is a small parameter, effectively parametrizing the solution amplitude (and also its inverse width). Directly substituting this ansatz into Eq. (1) and equating the various orders of ε leads to the dispersion relation 2 ω = ω(k ), at (ε) the group velocity relation c = ω0(k ), at (ε ) and the nonlinear Schr¨odingerequation 0 0 O 0 O i∂ A(X,T ) + ν ∂2 A(X,T ) + ν A(X,T ) A(X,T ) 2 = 0, (4) T 2 X 3 | | 3 at (ε ), where ν2 = ω00(k0)/2 > 0 and ν3 is a lengthy wavenumber-dependent expression. Full details of the derivationO of the NLS equation− starting from Eq. (1), including the higher-order terms of the ansatz, can be found e.g. in [33–35]. Since we seek standing wave solutions, we choose the wavenumber to be at the edge of the acoustic band k0 = π, such that the group velocity vanishes, ω0 = 2√K2, and 4 ν = (3K K 4K2) = b. 3 k0=π 2 4 3 | K2√K2 −
Since ν2 > 0, the NLS equation (4) will be focusing if b > 0. For our numerical computations, we consider the case example of K2 = K4 = 1 and K3 = 1/√2 such that ν2 = 1/4 and b = 4. The equation for B(X,T ) is defined in terms of A(X,T ), 4K (1 cos(k)) ∂ B(X,T ) = 3 − A(X,T ) 2. (5) X (ω (k))2 (ω (0))2 | | 0 − 0
B. Peregrine Initial Data
The focusing NLS equation has the one-parameter family of Peregrine soliton solutions [5] given by
P0 4(1 + 2P0 iT ) A(X,T ) = 1 eiP0T , (6) 2 2 2 2 ν3 − 1 + P0X + 4P T r ν2 0 ! where P0 > 0 is an arbitrary parameter. This solution is localized in space and time and has a maximum (located at (x, t) = (0, 0)) that is three times greater than its background, which are the features we desire to describe a rogue wave, see Fig. 1(b). Using the Peregrine soliton for the envelope function and a wave number k0 = π, K2 = K4 = 1 and K3 = 1/√2 we arrive at the following approximation √η 4(1 + 2iη t) u (t) = 1 ei(πn+(2+η)t) + c.c. + εB(εn, ε2t) , η = P ε2, (7) n 2 − 1 + 8ηn2 + 4(ηt)2 0 where B is defined in Eq. (5). It will be convenient to represent the solution in the strain variable formulation, that is, yn = un+1 un since the term B in the ansatz, which introduces a linear slope, will vanish. The parameter 2 − η = P0ε > 0 selects the background amplitude (since yn(0) 2√η as n ) and the frequency of oscillation | | → → ±∞ 2 + η, which lies above the cutoff of the acoustic band ω0 = 2. To test the validity of the multiscale analysis, we perform numerical simulations of the FPUT model Eq. (1) using Eq. (7) as initial data. For instance, see Fig. 2 for a simulation with ε = 0.02, X [ 40, 40] and T [ 5, 5]. In this simulation, our initial time is t = 5ε2, such that t = 0 should correspond to a∈ peak− at the middle∈ node− n = 0. The simulations are sensitive to the boundary− conditions since the background is non-zero (we employ boundary conditions that are periodic in the strain). Therefore, we take a larger spatial domain to reduce the influence of the boundary, since we are mainly concerned with the core of the solution. For times before the rogue wave appears (i.e. t < 0) the FPUT dynamics is predicted by the NLS dynamics (compare Fig. 2(a) and 2(b)). After the formation of the rogue wave, i.e., for t > 0, the FPUT dynamics departs from the NLS prediction. In the FPUT case, the large amplitude portion of the wave breaks into smaller, but still large relative to the background, waves. We believe that the emergence of these waves stemming from the Peregrine soliton core is a byproduct of the modulational instability of the NLS background as transcribed into the FPUT lattice and as seeded by the large amplitude perturbation induced by the wave structure. 4
5 0.12 5 0.12 (a) (b) 0.1 0.1
0.08 0.08 t
2 0 0.06 0 0.06 T ϵ
0.04 0.04
0.02 0.02
-5 -5 -20 -10 0 10 20 -20 -10 0 10 20 ϵn X
FIG.FIG. 2. 2. (a) (a) Simulation Simulation of of Eq. Eq. (1) (1) with with ε"==0 0..02,02, XX [[ 4040,,40]40] and and TT [[ 55,,5]5] that that is is initialized initialized with with Eq. Eq. (3) (3) with with A givengiven by ∈2 − ∈2 − Eq.Eq. (6) (6) with with PP00 == 1. 1. Color Color intensity intensity corresponds corresponds to to the the strain strain yynn((tt)) .. Notice Notice that that the the space-time space-time evolution evolution here here and and in in the the | | 2 figuresfigures that that follow follow is is given given in in terms terms of of the the rescaled rescaled variables variables εn"n|andand ε"| 2tt forfor space space and and time, time, respectively. respectively. (b) (b) Corresponding Corresponding NLSNLS prediction prediction 2 2ε"AA((X,TX, T)).. Note Note that that the the background background amplitude amplitude of of yynn((tt00)) isis the the same same as as 2 2ε" AA((X,TX, T00)) .. || || || || || ||
Note: I changed the definition of theC. strain Gaussian so that Initial the Data above expression is correct The parameter 2 ⌘ = P0" > 0 selects the background amplitude (since yn(0) 2p⌘ as n ) and the frequency of oscillation | | ! ! ±1 2+It⌘ has, which been lies shown above through the cuto the↵ rigorousof the acoustic work of band [36]! that0 = Peregrine-like 2. structures are a generic by-product of the so-calledTo test gradient the validity catastrophe of the multiscale phenomenon analysis, that the we (focusing)perform numerical NLS is subject simulations to for of localized the FPUT initial model data Eq. in the (1) semi-using classicalEq. (7) as limit. initial This data. featureIn the has ledcode, also I to solve very in clean the recent strain observations formulation. of Peregrine But solving solitons in in displacements optical systems then [37]. Also,converting at a numerical to strain level, will systematic give the explorations same result of (I Gaussian checked initial this data already), have led as to along rogue-like as the waves BC inis theconsistent. focusing NLSFor equation the flow for of sufficiently the text, broad it seemed Gaussians more [19]. natural When sufficiently to refer to broad the (so displacement as to be rescalable formulation. to the semi-classical So (7) is regime),the correct the waves ref. evolvingFor instance, through see the Fig. equations 2 for a of simulation motion focus with their" =0 energy.02, X to the[ 40 center, 40] andin a PeregrineT [ 5, 5]. structure. In this Evensimulation, more remarkably, our initial time such is initialt = data5"2, subsequentlysuch that t = lead 0 should to the correspond formation to of a an peak2 array at of the essentially middle2 node identical n = 0. (up The to smallsimulations corrections) are sensitive Peregrine-like to the boundary structures, conditions arising at since the the poles background of the so-called is non-zero tritronqu´eesolution (we employ boundary of the conditions Pain´ev´eI equation.that are periodic On the otherin the hand, strain). if the Therefore, Gaussian we is take sufficiently a larger narrow, spatial then domain a solution to reduce more the akin influence to a soliton of the forms;boundary, see thesince top we panel are mainly of Fig. concerned 3 for a few with examples. the core Here, of the we solution. investigate For iftimes a similar before phenomenology the rogue wave is appears possible (i.e. in thet< FPUT0)the lattice.FPUT dynamics More specifically, is predicted we consider by the NLS initial dynamics data for (compare the envelope Fig. function 2(a) andA( 2(b)).X,T ) Afterof the the form formation of the rogue wave, i.e., for t>0, the FPUT dynamics departs from the NLS prediction. In the FPUT case, the large amplitude X2 portion of the wave breaks into two smaller,A(X,T but= 0) still := largeA (X relative) = exp to the− background,. waves. We conjecture that the(8) G 4σ2 discrepancy between theory and numerics is due to a combination of boundary e↵ects, but also the fact that the NLS Inapproximation Fig. 3 results becomes for simulations less accurate for the for parameter larger amplitude values σ waves.20.1, 10.5, 2.5, 1.3 for ε = 0.1 and ε = 0.05 are shown. Note the strong resemblance to the NLS prediction, however,∈ after{ the main peak} forms, there is noticeable distortion between the NLS prediction and the actual FPUT dynamics, just as the case in the Peregrine example in the previous subsection. In this simulation the tails are decayingC. Gaussian to zero, and Initial thus Data any potential boundary effects should be minimal. These findings confirm once again the genericity of the gradient catastrophe scenario of [36], although presumably theIt non-integrability has recently been of shown the present through lattice the rigorous distorts work the of “Christmas-tree” [35] that Peregrine-like pattern structures of the subsequent are a generic Peregrines by-product in comparisonof the so-called to the gradient NLS paradigm. catastrophe Nevertheless, phenomenon the that pattern the (focusing) is still clearly NLS is discernible subject to and for progressively localized initial reverts data to in breathingthe semi-classical and ultimately limit. toThis solitonic feature solutions has led as alsoσ decreases to very clean (i.e., recent along the observations horizontal of direction). Peregrine On solitons the other in optical hand, thesystems trend [36]. of decreasing Also, at aε numerical(along the level, vertical systematic direction) explorations makes the of patterns Gaussian appear initial data more have and led more to “NLS-like” rogue-like waves as is expectedin the focusing by the NLS increased equation accuracy for su ofciently the NLS broad approximation Gaussians [19]. in the The limit low ofamplitude small ε. and broad waves when evolving through the equations of motion focus their energy to the center in a Peregrine structure. Even more remarkably, such initial data lead to the formation of an array of essentially identical (up to small corrections) Peregrine-like structures, arising at the poles of the so-calledIII. tritronqu´ee DIATOMIC solution GRANULAR of the Pain´ev´eI CRYSTAL equation. If the Gaussian is narrower, then a solution more akin to a soliton forms, see the top panel of Fig. 3 for a few examples. Here, we investigate if a similar phenomenology is possible in the FPUT lattice.A. More Theoretical specifically, Set-up we consider initial data for the envelope function A(X, T) of the form
We now turn our attention to another variant of the FPUT model that considersX2 a so-called Hertzian contact [26, 27] for the nonlinearity rather than the polynomialA(X, T = nonlinearity 0) = AG(X)=exp considered in Eq.. (1). Such a nonlinearity is relevant(9) in 4 2 the description of granular crystals when only considering forces due✓ to elastic◆ compression between the particles. In thisIn Fig. case, 3 the results model for equations simulations are for the parameter values 20.1, 10.5, 2.5, 1.3 for " =0.1 and " =0.05 are shown. Note the strong resemblance to the NLS prediction, however,2 { after the main} peak forms, there is noticeable An p An+1 p distortion between the NLSu¨n prediction= [δ0,n and+ u then 1 actualun]+ FPUT dynamics,[δ0,n+1 + justun asun the+1]+ case, in the Peregrine example(9) mn − − − mn − 55
30 30 30 30 1.2 1.2 0.9 0.5 25 25 25 25 1 0.8 1 0.7 0.4 20 20 20 20 0.8 0.8 0.6 0.3
t t t 0.5 t 2 15 2 15 0.6 2 15 2 15 ϵ 0.6 ϵ ϵ ϵ 0.4 0.2 10 0.4 10 0.4 10 0.3 10 0.2 0.1 5 0.2 5 0.2 5 5 (a) (b) (c) 0.1 (d) 0 0 0 0 0 -20 -10 0 10 20 -20 -10 0 10 20 -20 -10 0 10 20 -20 -10 0 10 20 ϵn ϵn ϵn ϵn
30 30 30 30 0.8 0.8 0.45 0.7 0.7 0.4 25 25 25 25 0.2 0.6 0.6 0.35 20 20 20 0.3 20 0.5 0.5 0.15
t t t 0.25 t
2 15 0.4 2 15 0.4 2 15 2 15 ϵ ϵ ϵ ϵ 0.2 0.1 0.3 0.3 10 10 10 0.15 10 0.2 0.2 0.1 5 5 5 5 0.05 0.1 0.1 (e) (f) (g) 0.05 (h) 0 0 0 0 0 -20 -10 0 10 20 -20 -10 0 10 20 -20 -10 0 10 20 -20 -10 0 10 20 ϵn 44 ϵn 44 ϵn 44 ϵn 44
5 0.12 530 5 0.12 0.12 530 5 0.12 0.12 530 5 0.12 0.12 530 0.12 (a) (b) (a) 4(b)4 (a) 4(b)4 (a) 2 4(b)4 44 0.1 0.1 0.1 0.1 3 0.1 0.1 0.1 0.1 1 5 0.12 5 5 0.123 0.12 5 5 0.12 0.12 5 5 0.12 0.12 5 0.12 (a) 0.08 20 (b) (a) 0.08 0.08 20 (b) (a) 0.08 0.08 20 (b) (a) 0.08 1.50.08 20 (b) 0.08 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
t t t 2 t
2 0 0.06 0 2 0 0.06 0.06 0 2 0 0.06 0.06 0 2 0 0.06 0.06 0 0.06 T T T T
ϵ ϵ 2 ϵ ϵ 0.08 0.08 0.08 0.08 0.08 0.081 0.08 0.080.5 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 t t 10 t 10 t 10 2 0 0.06 100 2 0 0.06 0.06 0 2 0 0.06 0.06 0 2 0 0.06 0.06 0 0.06 T T T T ϵ ϵ 1 ϵ 1 ϵ 0.02 0.02 0.02 0.50.02 0.04 0.02 0.04 0.04 0.02 0.04 0.04 0.02 0.04 0.04 0.02 0.04
-5 -5 (i)-5 -5 (j)-5 -5 (k)-5 -5 (l) -20 -10 0 10 20 0.02 -200 -10-20 0 -10 10 0 20 10 200.02 0.02 -200 -10-20 0 -10 10 0 20 10 200.020 0.02 -200 -10-20 0 -10 10 0 20 10 200.020 0.02 -200 -10 0 10 20 0.020 -5 ϵn -5-30-5 -15X 0ϵn 15 30 -5-30-5 -15X 0ϵn 15 30 -5-30-5 -15X 0ϵn 15 30 -5-30 -15X 0 15 30 -20 -10 0 10 20 -20-20 -10 -10 0 0 10 10 20 20 -20-20 -10 -10 0 0 10 10 20 20 -20-20 -10 -10 0 0 10 10 20 20 -20 -10 0 10 20 ϵn X ϵn X ϵn X ϵn X FIG.FIG. 2. 2. (a) (a) Simulation Simulation of of Eq. Eq. (1) (1) with with""=0=0.02,.02,XX [[ 4040,,40]40]FIG. andFIG. andT 2.T 2. (a) (a)[[ Simulation55 Simulation,,5]5] that that is is of initializedof initialized Eq. Eq. (1) (1) with with with" Eq." Eq.=0=0 (3). (3)02,.02, with withXX AA[[ givengiven4040,,40]40] by byFIG. andFIG. andT 2.T 2. (a) (a)[[ Simulation55 Simulation,,5]5] that that is is of initializedof initialized Eq. Eq. (1) (1) with with with" Eq." Eq.=0=0 (3). (3)02,.02, with withXX AA[[ givengiven4040,,40]40] by byFIG. andFIG. andTT 2. 2. (a)[ (a)[ 5 Simulation5, Simulation,5]5] that that is is ofinitialized initializedof Eq. Eq. (1) (1) with with" Eq. Eq."=0=0 (3) (3).02,.02, with withXX AA[[givengiven4040,,40]40] by by and andTT [[ 55,,5]5] that that is is initialized initialized with with Eq. Eq. (3) (3) with withAAgivengiven by by 22 22 22 22 22 22 22 22 Eq.Eq. (6) (6) with withPP00== 1. 1. Color Color intensity intensity corresponds corresponds to to the the strain strainEq.Eq.yyn (6)n( (6)(tt)) with. with. Notice NoticePP00== that that 1. 1.Color theColor the space-time space-time intensity intensity evolutioncorresponds evolutioncorresponds here here to to and theand the instrain instrain the theEq.Eq.yyn (6)n( (6)(tt)) with. with. Notice NoticePP00== that that 1. 1.Color theColor the space-time space-time intensity intensity evolutioncorresponds evolutioncorresponds here here to to and theand the instrain instrain the theEq.Eq.yynn (6)( (6)(tt)). with. with Notice NoticePP00= that= that 1. 1. theColor the Color space-time space-time intensity intensity evolution correspondsevolution corresponds here here to to and and the the in strainin strain the the yynn((tt)).. Notice Notice that that the the space-time space-time evolution evolution here here and and in in the the FIG.FIG. 2. 2. (a) (a) Simulation Simulation of of Eq. Eq. (1) (1) with with""=0=0.02,.02,XX [[ 4040FIG.,FIG.,40]40]|| and2. and 2. (a)| (a)|2T2T Simulation Simulation[[ 55,,5]5] that that of of Eq.is isEq. initialized initialized (1) (1) with with with" with"=0=0 Eq.. Eq.02,.02, (3)X (3)X with with[[ 4040AAFIG.,,FIG.40]given40]given|| and and2. 2. by by|(a)|2 (a)T2T Simulation Simulation[[ 55,,5]5] that that of of is Eq. is Eq. initialized initialized (1) (1) with with with" with"=0=0 Eq.. Eq.02,.02, (3) (3)XX with with[[ 4040AAFIG.,FIG.,40]givengiven40]|| and2. and 2. by by|(a)|2 (a)T2T Simulation Simulation[[ 55,,5]5] that that of of Eq.is isEq. initialized initialized (1) (1) with with with" with"=0=0 Eq.. Eq.02,.02, (3)X (3)X with with[[ 4040AA,,40]given40]given|| and and by by||2T2T [[ 55,,5]5] that that is is initialized initialized with with Eq. Eq. (3) (3) with withAAgivengiven by by figuresfigures that that follow follow is is given given in in terms terms of of the the rescaled rescaled variables variables22 "figures"nfiguresnandand that" that" tt2for2for follow follow space space is is and given and given time, time, in in terms terms respectively. respectively. of of the the rescaled rescaled (b) (b) Corresponding Corresponding variables variables22 "figures"nfiguresnandand that" that" tt2for2for follow follow space space is is and given and given time, time, in in terms terms respectively. respectively. of of the the rescaled rescaled (b) (b) Corresponding Corresponding variables variables22 ""figuresnfiguresnandand" that" thatttfor2for2 follow follow space space is is and andgiven given time, time, in in terms termsrespectively. respectively. of of the the rescaled rescaled (b) (b) Corresponding Corresponding variables variables22 ""nnandand"" tt2for2for space space and and time, time, respectively. respectively. (b) (b) Corresponding Corresponding Eq.Eq. (6) (6) with withPP00== 1. 1. Color Color intensity intensity corresponds corresponds to to the the strainEq. strainEq. (6) (6)yy withn withn((tt)).P.P Notice0 Notice0== 1. 1. thatColor that Color the the intensity intensity space-time space-time corresponds corresponds evolution evolution to to here here the the andstrainEq. andstrainEq. in(6) in (6)y the theyn withn with((tt))..PP Notice0 Notice0== 1. 1. that thatColor Color the the intensity intensity space-time space-time corresponds corresponds evolution evolution to to here here the the and strainEq.and strainEq. in(6) in (6) they they withn withn((tt)).P.P Notice0 Notice0== 1. 1. thatColor that Color the the intensity intensity space-time space-time corresponds corresponds evolution evolution to to here here the the andstrain andstrain in iny the theynn((tt)).. Notice Notice that that the the space-time space-time evolution evolution here here and and in in the the NLSNLS prediction prediction 2 2""AA((X,X, T T)).. Note Note that that the the background background amplitude amplitudeNLSNLS of ofprediction| prediction| yynn((t|t0|02)2) isis 2 2" the" theAA( same(X, sameX, T T)) as. as. Note 2 Note 2""AA( that(X, thatX, T T0 the0 the)).. background background amplitude amplitudeNLSNLS of ofprediction| prediction| yynn((t|t0|02)2) isis 2 2" the" theAA( same(X, sameX, T T)) as. as. Note 2 Note 2""AA( that(X, thatX, T T0 the0 the)).. background background amplitude amplitudeNLSNLS of of prediction| prediction|yynn((tt0|0)|2)2 isis 2 2the" the"AA( same(X, sameX, T T) as) as.. Note 2 Note2""AA(( thatX,X, that T T0 the0) the).. background background amplitude amplitude of of|| yynn((t|t0|02)2) isis the the same same as as 2 2""AA((X,X, T T00)).. figuresfigures that that follow follow|| is is given given|| in in terms terms of of the the rescaled rescaled variables variablesfiguresfigures""FIG.nn thatand| thatand| " follow" followt||t 3.forfor|| is spaceis spaceSimulation given given and and|| in in time,terms time, terms|| respectively. respectively.of of the| the| of rescaled rescaled Eq. (b) (b) variables variables Corresponding Corresponding (1)figuresfigures" that"nnand that|and that| "" follow isfollowt||tforfor initialized|| spaceis spaceis given given and and|| in in time, termstime,| terms| respectively. respectively.with of of| the| the rescaled rescaled Eq. (b) (b) variables variablesCorresponding Corresponding (3)figuresfigures with""nn thatand| that|and" follow"A follow|t|tforforgiven|| is spaceis space given given and and|| in in by time,terms| time, terms| Eq. respectively. respectively.of of| the| the (9). rescaled rescaled (b) (b)" variables variables Corresponding Corresponding=0"."n1(toppanels)andnand|and| "" t||tforfor space space and and time, time,|| respectively. respectively.|| " =0 (b) (b).05 Corresponding Corresponding NLSNLS prediction prediction 2 2""AA((X,X, T T)).. Note Note that that the the background background amplitude amplitudeNLSNLSFIG. prediction prediction of of yy 3.nn((t 2t0 20")")AisAis( Simulation(X, theX, the T T same) same).. Note Note as as 2 that 2" that"AA(( theX, theX,of TT background0 background0)). Eq.. (1) amplitude amplitudeNLSNLS that prediction prediction of of yyn isn((tt0 20 2)" initialized)"AisAis(( theX, theX, TT same) same).. Note Note as as 2 2 that" that"A withA((X, theX, the T T0 background0 background)). Eq.. (3) amplitude amplitudeNLSNLS with prediction prediction of of yyAnn((t 2t0 20")given")AisAis((X, theX, the T T same) same)..by Note Note as as 2 thatEq. 2" that"AA(( theX, theX, (8)TT background0 background0)).. with amplitude amplitude thefollowing of of yynn((tt00)) isis the the same samewidth as as 2 2""AA parameters((X,X, T T00)).. || || (bottom|| || panels).|| || In all|| panels X || [ 40|||| , 40]|| and|| T || [ 5, 5]. The|| values|| of|| the|| width|| parameter | | are|| (from left|| to|| right) theNote: correct I changed ref. For the instance, definition see Fig.of the 2 for strain a simulation sothe thatNote:σ with correct= the I20" changedabove=0. ref.1.02, for expressionXFor the panels instance,[ definition40, 40] is seeand correct (a,e,i), Fig.ofT the 2[The for5 strain, 5]. a parameterσ simulation In= so thisthe 10thatNote: with. correct5 the Ifor" changedabove=0 ref. panels.02, expressionXFor the instance,[ definition40 (b,f,j),, 40] is seeand correct Fig.ofT σ the 2[The= for5 strain, 5]. a2 parameter simulation.5 In so thisforthe thatNote:panels with correct the I" above=0changed ref..02, (c,g,k) expressionXFor the[ instance, definition40 and, 40] is and seecorrectσ Fig.ofT = the 2[The for15 strain,.5].3 aparameter simulation forIn sothis panelsthat with the" above=0 (d,h,l)..02, expressionX [ In40, 40] panelsis and correctT [The (a)-(d)5, 5]. parameter In this 2 2 2 2 2 2 22 2 2 2 2 2 2 2 2 2 theNote:simulation,⌘ = correctP0" I changed> our0 ref. selects initialFor the thetime instance, background definition is t = see5" amplitude, Fig.of such the 2 that for strain (sincet a= simulation 0the shouldyNote: sosimulation,n⌘(0)= thatcorrectP correspond0 with" I2 thep changed> our⌘020 ref."as selectsabove=0 initialn. to1.02, a,For the10 thepeaktime expressionX instance,) background. definition5 isandat,[t the2= the40. middle5, see540] frequency," is1 amplitude, Fig.ofand such. correct3 node the 2T that. offorn strainEGC: oscillation(since=[tThe a= 0.5 simulation, 0 The5]. parametershouldthey soNote:simulation,n⌘(0) In= that correctNote thisP correspond0 with" I2 thep> changed our⌘0" ref.as selectsabove=0 initialthatn to.02, aForthepeak time expressionX instance,) backgroundthedefinition isandat[t the= the40 middlebottom,5see40] frequency" is amplitude,Fig.and suchof correct node theT2 that of forn strainoscillation (since=[tThe panela= 0.5 simulation, 0 The5]. parametershouldtheyNote: sonsimulation,⌘(0) In= thatcorrect thisP correspondcorresponds0 with" I2p the changed>⌘ our0 ref."as selectsabove=0 initialn to. a02,For thepeak thetime expressionX instance,) backgrounddefinition andat is [t the= theto40 middle,see5 frequency40]" isthe amplitude, Fig.ofand such correct node the 2T that respectiveof forn strainoscillation= (since[tThe a 0.=5 simulation, 0The5]. parametershouldy son(0) In that this correspond with2 thep numerical⌘"as above=0n to.02, a peak expressionX) andat[ the the40simulations middle, 40] frequency is and correct nodeT ofn oscillation=[The 0.5, The5]. parameter In this 2 2 the2 perturbation2 parameter2 2 value is2 ε = 0.1.2 In panels2 2 (e)-(h) the2 perturbation2 parameter2 2 value is ε = 0.05.2 Panels (i)-(l)2 simulation,⌘ = P0" > our0 selects initial thetime background is t = 5" amplitude, such that (sincet =simulation,|⌘ 0= shouldynP|(0)0!" correspond> our20p selects initial⌘ as!n to± thetime1 a backgroundpeak is t)= andat the5 the" amplitudemiddle, suchfrequency that node (sincet ofn=| oscillation=simulation, 0⌘ 0.should=yn| The(0)P!0" correspond> our20p selects⌘ initialas!n to± thetime1 apeak background is )t = andat the5 the" middle amplitude, frequency such thatnode (sincet ofn=| oscillation=simulation,⌘ 0 0.= shouldyn| TheP(0)!0" correspond> our20p selects initial⌘ as! n± to thetime1 a backgroundpeak is t)= andat the5 the" amplitudemiddle, suchfrequency that node (sincet ofn=| oscillation= 0 0.shouldyn| The(0)! correspond2p⌘ as!n to±1 a peak) andat the the middle frequency node ofn oscillation= 0. The simulations2+⌘, which are lies sensitive above the to the cuto boundary↵ of the conditionsacoustic band since!0 thesimulations2+=| 2.background⌘,2 which| !{ are lies sensitiveis non-zeroabove! the to± (we the cuto1 employboundary↵ of the boundary conditionsacoustic} conditions band since!0 thesimulations2+=| 2.background⌘, which| ! are lies sensitiveis non-zeroabove! the to± (we the cuto1 employboundary↵ of the boundary conditionsacoustic conditions band since!0 thesimulations=2+| 2.background⌘, which| ! are lies is sensitive non-zero above! the to (we± the cuto1 employ boundary↵ of the boundary conditionsacoustic conditions band since!0 the=| 2.background| ! is non-zero! ± (we1 employ boundary conditions simulations2+thatTo⌘, are whichtest periodic the are lies validity sensitive abovein the of strain). the to the the cuto multiscale boundary Therefore,↵ of the analysis, conditionsacoustic we take we aband perform larger sincesimulations2+that!correspond0 theTo spatialof⌘ numerical=, are whichtest2. background theperiodic domain the are lies simulations validity sensitive NLS abovein to is the reduce non-zeroto of strain). the to the ofequation thethe cuto multiscale boundary(weFPUT influenceTherefore,↵ respectiveof employ the model analysis, of conditionsacoustic we boundary the Eq.with take boundary, we (1) aband performnumerical largerusing since conditionssimulations2+that Gaussian!0 theTo spatial numerical⌘= are,test2.background which periodic domain the are liessimulations validitysimulations sensitive in above to is initialthe reduce non-zero of strain). the tothe of the the cutomultiscale (we boundaryFPUT influenceTherefore, data↵ of employ of the model analysis,the of conditionsacoustic we (forboundary the Eq. take boundary,NLS we (1) a performband the largerusing since conditionssimulations2+that equation!0 thespatialTo numerical⌘ same=, are whichtest2. background periodic domain the are liessimulations validity sensitivevalueswith abovein to is the reduce non-zero of strain). the to theof Gaussian the the cuto multiscaleof boundary(weFPUT influence Therefore,↵ of employ)ofRef.[19]. the model analysis, of conditionsacoustic we theboundaryinitial Eq. take boundary, we (1) aband perform using larger since conditions data,!0 the spatial numerical= 2.background see domain simulations e.g. to is reduce non-zero Ref. of the (weFPUT influence [19]. employ model of boundary the Eq. boundary, (1) using conditions thatsinceEq.To are (7) testwe periodic asare the initial mainly validity in data. concerned the ofIn strain). the the multiscale with code, Therefore, the I core solve analysis, of we the in take the solution. we a strainperformthat largersinceEq.To For are (7)formulation. testwe spatial timesnumerical periodic asare the initial mainly before domainvalidity in data.simulations the concerned Butthe toof rogueIn strain). solving thereduce the wave multiscale with of code, Therefore,the in appears the displacementsFPUT influence I core solve analysis, (i.e. of wemodel the int< of take the solution. thewe0)the Eq. then a strainperform boundary,that larger (1)sinceEq.To For using are (7)formulation. wetest spatial timesnumerical periodicas are the initial mainly before domain validity indata.simulations the concerned But the to rogueofIn strain). solving reducethe the wave multiscalewith of code, the Therefore,in appears the displacementsFPUT influence I core solve analysis, (i.e. of model we the int< of take the solution. the we0)the Eq. then straina performboundary,that larger(1)sinceEq.To For using are formulation. (7) testwe spatialtimes numerical periodic asare the initial before mainly domainvalidity in data.simulations the But concerned the roguetoofIn strain). solving thereduce the wave multiscale with of code, Therefore,thein appears the displacementsFPUT influence I core solve analysis, (i.e. of wemodel the int< of take the solution. thewe0)the Eq. then a strainperform boundary, larger (1) For using formulation. spatial timesnumerical before domain simulations the But to rogue solving reduce wave of the in appears displacementsFPUT influence (i.e. modelt< of the0)the Eq. then boundary, (1) using sinceEq. (7) we asare initial mainly data. concernedIn the with code, the I core solve of the in the solution.sinceEq. strain (7) we For asare formulation. times initial mainly before data. concerned theIn But therogue solving with code, wave the in I appears core solve displacements of the in (i.e. the solution.t
wn 1 vn 1, wn vn 1 | − − | | − | reduce to the dimer FPUT lattice,
ρv¨n = V 0(wn vn) V 0(vn wn 1), (14) − − − − w¨ = V 0(v w ) V 0(w v ), (15) n n+1 − n − n − n where
2 3 3 3 3 V 0(x) = K x + K x + K x ,K = ,K = ,K = . 2 3 4 2 2 3 −8 4 −48 T 0 0 T i(kn+ωt) The linearized Eqs. (14) and (15) (i.e. where K3 = K4 = 0) have solutions of the form (vn, wn) = (v , w ) e , where k and ω are related through the dispersion relation
2 2 1 1 4 2 k ω(k) = K2 1 + 1 + sin , (16) ± ρ ± ρ − ρ 2 s where the minus and plus signs correspond to the acoustic and optical bands, respectively, of the dispersion relation, see Fig. 1(c). At the wavenumber k = π the upper cutoff frequency of the acoustic band is ω (π) = √2K2 and − the lower cutoff frequency of the optical band is ω+(π) = 2K2/ρ. Since ρ < 1 there is a band gap of size
2K2 ω+(π) ω (π) = (1 √ρ). In order to find a rogue wave,p we proceed in the same way as in the previous − − ρ − section. Namely, weq derive a focusing NLS equation from Eqs. (14) and (15) in order to obtain an approximation that has the Peregrine soliton as the envelope function. This approximation should describe a rogue wave of the dimer granular crystal for small amplitudes. We will make use of numerical simulations to test the role of the nonlinearity stemming from the Hertzian contact. To derive the NLS equation, we use the following ansatz [35],
vn(t) = ε (B(X,T ) + [A(X,T )E(n, t; 0, ω+(π)) + c.c.]) , (17)
wn(t) = εB(X,T ), (18) where
i(k0n+ω0t) 2 E(n, t; k0, ω0) = e ,X = εn, T = ε t. Here, we have already selected the plane wave at the bottom of the optical band to be modulated by the envelope function A, since the notation is less cumbersome than in the general wavenumber case. Substitution of this ansatz into Eqs. (14) and (15) leads to the focusing NLS equation at order ε3 2 ω00 (π) K ω (π) i∂ A(X,T ) + ν ∂2 A(X,T ) + ν A(X,T ) A(X,T ) 2 = 0, ν = + , ν = 2 + (3K K 4K ). (19) T 2 X 3 | | 2 − 2 3 2 2 4 − 3
Note that, since ν3 < 0 and ω+00 (π)/2 < 0, both ν2 and ν3 are negative such that Eq. (19) is focusing. The function B(X,T ) is defined in terms of− A(X,T ) via
4K3 2 ∂X B(X,T ) = A(X,T ) . − K2 | |
Since ν2 and ν3 are negative, the Peregrine soliton for Eq. (19) is the same as of Eq. (6) but with the appropriate sign changes:
P0 4(1 2P0 iT ) iP T A(X,T ) = 1 e− 0 . (20) 2 − 2 2 2 − ν3 − 1 P0X + 4P T r − ν2 0 ! 2 Substituting this expression into Eq. (17) leads to a plane wave that oscillates with temporal frequency ω+(π) ε P0 (and hence lies within the band gap of the spectrum, since ε 1) that is modulated by a Peregrine soliton. − 87
3 0.6 3 0.8 0.8 3 1.5 0.8 1.5 1.5 0.6 0.6 1.5 0.4 0.6 0 0 0 0.4 0 0.4 0.4 0.2 -1.5 -1.5 -1.5 0.2 0.2 -1.5 0.2 (a) (b) (c) (d) -3 0 -3 0 -3 0 -3 0 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10
3 3 1 3 3 0.9 1 1.2 1.5 1.5 1.5 0.75 0.75 1.5 0.6 0.8 0 0 0 0.5 0.5 0 0.3 0.4 -1.5 -1.5 0.25 -1.5 0.25 -1.5 (e) (f) (g) (h) -3 0 -3 0 -3 0 0 -10 -5 0 5 10 -12 -8 -4 0 4 8 12 -16 -8 0 8 16 -50 -25 0 25 50
FIG. 4. Same as Fig. 4, but plotted with a higher resolution. In particular, the spatiotemporal evolution of the strain variable FIG.for values 4. Simulation of the mass of ratio Eq. (9) parameter that is initialized⇢ of (from with left to Eq. right) (17)⇢ and=0. Eq.1, 0.2 (18), 0.3and0 with A.4andgiven⇢ by=0 Eq..5, 0 (20).6, 0. with7and0P0 .=9isdepicted 1, ε = 0.5, Xat the[ 40 top, 40] and and bottomT [ panels,5, 5]. Colorrespectively. intensity corresponds to the strain yn . The values of the mass ratio parameter are (a) ρ =∈ 0.−1, (b) ρ = 0.2, (c)∈ −ρ = 0.3, (d) ρ = 0.4, (e) ρ = 0.5, (f) ρ = 0.6, (g) ρ| =| 0.7, (h) ρ = 0.9,
B.ACKNOWLEDGEMENTS Peregrine Initial data
WeThis conduct material a number is based of upon simulations work supported of the fully by nonlinear the National dimer Science crystal Foundation model Eq. under (9) using Grant the No. ansatz DMS-1615037. in Eqs. (17) andPGK (18) gratefully for various acknowledges mass ratios. support The from results the are US-AFOSR summarized under in Fig. FA9550-17-1-0114. 4. For small values of the mass ratio ρ, the dynamics are similar to the monomer FPU chain studied above. There is the appearance of a large amplitude peak, seemingly out of nowhere, but then rather than disappearing “without a trace”, as the NLS Peregrine soliton predicts, the large amplitude portion of the wave breaks into smaller, but still large relative to the background, waves (compare Fig. 4 and Fig. 2). The same feature persists for larger mass ratios, however, the secondary pulses become broader. This[1] isE. part Pelinovsky of the and manifestation C. Kharif (eds.), of theExtreme modulational Ocean Waves instability(Springer, ofthe NY, corresponding 2008). background. For sufficiently large[2] C. mass Kharif, ratios E.ρ Pelinovsky,, more waves and A.seem Slunyaev, to emergeRogue as Waves a result in the of the Ocean instability(Springer, and NY, the 2009). time scale of their interaction appears[3] A. R.to Osborne,be shorter.Nonlinear Ocean Waves and the Inverse Scattering Transform (Academic Press, Amsterdam, 2010). We[4] D. have H. Peregrine, also observed J. Austral. a substantial Math. Soc. sensitivity B 25,16(1983). to boundary conditions and a rapid propagation of the resulting excitations,[5] M. J. Ablowitz, reflecting M. from D. Kruskal, the boundary and J. F. towards Ladik, SIAM the coreJ. Appl. of the Math., Peregrine36,428-437(1979). structure. It is relevant to note here that[6] forD. R. uncompressed Solli, C. Ropers, granular P. Koonath, crystals, and solitary B. Jalali, waves Nature are450 found,1054(2007). to exist at special mass ratios (the so-called anti- resonances),[7] B. Kibler and et al., severe Nature wave Phys. attenuation6,790(2010). occurs at other special mass ratios (the so-called resonances), [40–43]. It would[8] B. be Kibler particularly et al., Sci. interestsing Rep. 2,463(2012). to explore whether such phenomena have an analogue in the case of precompressed [9] J. M. Dudley, F. Dias, M. Erkintalo, and G. Genty, Nat. Photon. 8,755(2014). diatomic[10] B. Frisquet granular et crystalsal., Sci. Rep. and6 whether,20785(2016). they have any implications towards the formation of the Peregrine solitons. Future[11] C. studies Lecaplain, concerning Ph. Grelu, resonances J. M. Soto-Crespo, and anti-resonances and N. Akhmediev, of precompressed Phys. Rev. Lett. diatomic108,233901(2012). granular chains would therefore not[12] onlyA. N. be Ganshin, interesting V. B.in theirEfimov, own G. right, V. Kolmakov, but might L. P.also Mezhov-Deglin, help explain theand observed P. V. E. McClintock, deviations from Phys. granular Rev. Lett. crystal101, dynamics065303 and (2008). the NLS predictions. [13] A. Chabchoub, N. P. Ho↵mann, and N. Akhmediev, Phys. Rev. Lett. 106,204502(2011). [14] A. Chabchoub, N. Ho↵mann, M. Onorato, and N. Akhmediev, Phys. Rev. X 2,011015(2012). [15] A. Chabchoub and M. Fink,IV. Phys. DISCUSSION Rev. Lett. 112,124101(2014). AND FUTURE DIRECTIONS [16] H. Xia, T. Maimbourg, H. Punzmann, and M. Shats, Phys. Rev. Lett. 109,114502(2012). [17] M. Shats, H. Punzmann, and H. Xia, Phys. Rev. Lett. 104,104503(2010). [18]In theH. Bailung, present S. study, K. Sharma, we have and Y. definitively Nakamura, illustrated Phys. Rev. the Lett. potential107,255005(2011). of phononic lattices to support rogue wave structures.[19] E. G. Charalampidis, Our preliminary J. Cuevas-Maraver, considerations focused D. J. Frantzeskakis, on the FPUT and lattice P. G. Kevrekidis, as a prototypicalRogue waves example in ultracold wherebosonic rogue waves seas, couldarXiv:1609.01798. be excited by using the Peregrine soliton solution of the derived NLS equation as initial data. For sufficiently wide[20] Gaussians,Y. Shen, P. we G. also Kevrekidis, found rogue-wave G. P. Veldes, patterns D. J. Frantzeskakis, in line with the D. universality DiMarzio, X. of Lan, the andgradient V. Radisic, catastrophe Phys. mechanismRev. E 95, suggested032223 by (2017). [36]. However, for Peregrine and Gaussian initial data, the formation of the large amplitude structures led eventually[21] N. Akhmediev to deviations and A. in Ankiewicz, the FPUT Phys. dynamics Rev. E from4,046603(2011). the expected predictions of the NLS approximation. While part [22] E. Fermi, J. Pasta, and S. Ulam. Studies in nonlinear problems, I. Los Alamos report, LA 1940, 1955. of[23] theG. observed Gallavotti, discrepanciesThe Fermi–Pasta–Ulam may be attributed Problem: to A boundary Status Report effects,(Springer-Verlag, the predominant Berlin, reason Germany, for this 2008). phenomenology is[24] theP. presence G. Kevrekidis. of theNonlinear modulational waves instability in lattices: for Past, the present, background future. on IMA top J of Appl which Math the76 Peregrine,389-423(2011). structure is formed. We[25] alsoV. F.considered Nesterenko, a diatomicDynamics granular of Heterogeneous crystal to Materials demonstrate(Springer-Verlag, that rogue New wave York, dynamics NY, 2001). is possible in a system that is[26] highlyS. Sen, accessible J. Hong, in J. experiments Bang, E. Avalos, in a and space-time R. Doney, resolved Phys. Rep. way462 [30].,21(2008). A key challenge in that regard concerns the large scales considered in this paper (where the NLS approximation is valid) leading to large lattices. However, it 8 may be interesting to try relevant ideas in smaller lattices; some studies have considered lattices as large as N = 81 nodes [44], or even N = 188 nodes in [45]. While this paper establishes important first steps for the realization of phononic rogue waves, future theoretical studies should consider further steps in some of these directions; another important one involves the suitable initialization with Peregrine-like initial data, as these lattices permit considerable control e.g. over driving the boundaries, but are less amenable to a distributed initialization over the entire chain. Such topics are presently under consideration and will be reported in future publications.
ACKNOWLEDGEMENTS
PGK gratefully acknowledges discussions with S. Sen at an early stage of this work. This material is based upon work supported by the National Science Foundation under Grant No. DMS-1615037. PGK gratefully acknowledges support from the US-AFOSR under FA9550-17-1-0114.
[1] E. Pelinovsky and C. Kharif (eds.), Extreme Ocean Waves (Springer, NY, 2008). [2] C. Kharif, E. Pelinovsky, and A. Slunyaev, Rogue Waves in the Ocean (Springer, NY, 2009). [3] A. R. Osborne, Nonlinear Ocean Waves and the Inverse Scattering Transform (Academic Press, Amsterdam, 2010). [4] Catherine Sulem and Pierre-Louis Sulem, The Nonlinear Schr¨odinger Equation: Self-Focusing and Wave Collapse (Springer, NY, 1999) [5] D. H. Peregrine, J. Austral. Math. Soc. B 25, 16 (1983). [6] D. R. Solli, C. Ropers, P. Koonath, and B. Jalali, Nature 450, 1054 (2007). [7] B. Kibler et al., Nature Phys. 6, 790 (2010). [8] B. Kibler et al., Sci. Rep. 2, 463 (2012). [9] J. M. Dudley, F. Dias, M. Erkintalo, and G. Genty, Nat. Photon. 8, 755 (2014). [10] B. Frisquet et al., Sci. Rep. 6, 20785 (2016). [11] C. Lecaplain, Ph. Grelu, J. M. Soto-Crespo, and N. Akhmediev, Phys. Rev. Lett. 108, 233901 (2012). [12] A. N. Ganshin, V. B. Efimov, G. V. Kolmakov, L. P. Mezhov-Deglin, and P. V. E. McClintock, Phys. Rev. Lett. 101, 065303 (2008). [13] A. Chabchoub, N. P. Hoffmann, and N. Akhmediev, Phys. Rev. Lett. 106, 204502 (2011). [14] A. Chabchoub, N. Hoffmann, M. Onorato, and N. Akhmediev, Phys. Rev. X 2, 011015 (2012). [15] A. Chabchoub and M. Fink, Phys. Rev. Lett. 112, 124101 (2014). [16] H. Xia, T. Maimbourg, H. Punzmann, and M. Shats, Phys. Rev. Lett. 109, 114502 (2012). [17] M. Shats, H. Punzmann, and H. Xia, Phys. Rev. Lett. 104, 104503 (2010). [18] H. Bailung, S. K. Sharma, and Y. Nakamura, Phys. Rev. Lett. 107, 255005 (2011). [19] E. G. Charalampidis, J. Cuevas-Maraver, D. J. Frantzeskakis, and P. G. Kevrekidis, Rom. Rep. Phys. 70, 504 (2018). [20] Y. Shen, P. G. Kevrekidis, G. P. Veldes, D. J. Frantzeskakis, D. DiMarzio, X. Lan, and V. Radisic, Phys. Rev. E 95, 032223 (2017). [21] N. Akhmediev and A. Ankiewicz, Phys. Rev. E 83, 046603 (2011). [22] D. Han, M. Westley, and S. Sen, Phys. Rev. E 90, 032904 (2014). [23] E. Fermi, J. Pasta, and S. Ulam. Studies in nonlinear problems, I. Los Alamos report, LA 1940, 1955. [24] G. Gallavotti, The Fermi–Pasta–Ulam Problem: A Status Report (Springer-Verlag, Berlin, Germany, 2008). [25] P. G. Kevrekidis. Nonlinear waves in lattices: Past, present, future. IMA J Appl Math 76, 389-423 (2011). [26] V. F. Nesterenko, Dynamics of Heterogeneous Materials (Springer-Verlag, New York, NY, 2001). [27] S. Sen, J. Hong, J. Bang, E. Avalos, and R. Doney, Phys. Rep. 462, 21 (2008). [28] G. Theocharis, N. Boechler, and C. Daraio, Nonlinear phononic periodic structures and granular crystals, in Acoustic Metamaterials and Phononic Crystals, Springer-Verlag, Berlin, Germany, 2013, 217–251. [29] A. F. Vakakis, Analytical methodologies for nonlinear periodic media, in Wave Propagation in Linear and Nonlinear Periodic Media (International Center for Mechanical Sciences (CISM) Courses and Lectures), Springer-Verlag, Berlin, Germany, 2012, 257. [30] C. Chong, M. A. Porter, P. G. Kevrekidis, and C. Daraio, J. Phys.: Condens. Matter, 29, 413003 (2017). [31] M. A. Porter, P. G. Kevrekidis, and C. Daraio, Physics Today 68, 44 (2015). [32] Y. Starosvetsky, K. Jayaprakash, M. A. Hasan, and A. Vakakis, Dynamics and Acoustics of Ordered Granular Media, World Scientific, Singapore, 2017. [33] G. Schneider, Appl. Anal. 89, 1523 (2010). [34] G. Huang, Z.-P. Shi, and Z. Xu, Phys. Rev. B 47, 14561 (1993). [35] G. Huang and B. Hu, Phys. Rev. B 57, 5746 (1998). [36] M. Bertola and A. Tovbis, Comm. Pure Appl. Math. 66, 678 (2013). [37] A. Tikan, C. Billet, G. El, A. Tovbis, M. Bertola, T. Sylvestre, F. Gustave, S. Randoux, G. Genty, P. Suret, and J.M. Dudley Phys. Rev. Lett. 119, 033901 (2017). 9
[38] C. Chong, P. G. Kevrekidis, G. Theocharis, and C. Daraio, Phys. Rev. E 87, 042202 (2013). [39] C. Chong, F. Li, J. Yang, M. O. Williams, I. G. Kevrekidis, P. G. Kevrekidis, and C. Daraio, Phys. Rev. E 89, 032924 (2014). [40] E. Kim, R. Chaunsali, H. Xu, J. Jaworski, J. Yang, P. G. Kevrekidis, and A. F. Vakakis, Phys. Rev. E 92, 062201 (2015). [41] Y. Zhang, D. Pozharskiy, D. M. McFarland, P. G. Kevrekidis, I. G. Kevrekidis, and A. F. Vakakis, Exp. Mech. 57, 505 (2017). [42] K. R. Jayaprakash, Y. Starosvetsky, and A. F. Vakakis, Phys. Rev. E 83, 036606 (2011). [43] K. R. Jayaprakash, Y. Starosvetsky, A. F. Vakakis, and O. V. Gendelman, J. Nonlinear Sci. 23, 363 (2013). [44] N. Boechler, G. Theocharis, S. Job, P. G. Kevrekidis, M. A. Porter, and C. Daraio, Phys. Rev. Lett. 104, 244302 (2010). [45] R. Carretero-Gonz´alez,D. Khatri, M.A. Porter, P.G. Kevrekidis, and C. Daraio, Phys. Rev. Lett. 102, 024102 (2009).