A mechanical analog of quantum bradyons and tachyons

Aur´elienDrezet1, Pierre Jamet1, Donatien Bertschy1, Arnaud Ralko1, C´edricPoulain1,2 1Univ. Grenoble Alpes, CNRS, Grenoble INP, Institut Neel, F-38000 Grenoble, France and 2Univ. Grenoble Alpes, CEA, LETI, F-38000 Grenoble, France∗

We present a mechanical analog of a quantum wave-particle duality: a vibrating string threaded through a freely moving bead or ‘masslet’. For small string amplitudes, the particle movement is governed by a set of non-linear dynamical equations that couple the wave field to the masslet dynamics. Under specific conditions, the particle achieves a regime of transparency in which the field and the particle’s dynamics appear decoupled. In that special case, the particle conserves its momentum and a guiding wave obeying a Klein-Gordon equation, with real or imaginary mass, emerges. Similar to the double-solution theory of de Broglie, this guiding wave is locked in phase with a modulating group-wave co-moving with the particle. Interestingly, both subsonic and supersonic particles can fall into a quantum regime as with the slower-than-light bradyons and hypothetical, faster-than-light tachyons of particle physics.

PACS numbers: *43.40.Le,03.65.-w,42.15.Dp

I. INTRODUCTION tive interpretations, however deterministic, of quantum mechanics. The foundations of Quantum mechanics (QM) mainly In this paper, we focus on a type of mechanical ana- rely on the pioneering work of Louis de Broglie elabo- log closer in spirit to the original double solution of de rated during his PhD [1] and for which he received the Broglie, i.e., transverse waves on which a small bead of Nobel prize. The key assumption of de Broglie’s intuitive mass mp and of stiffness kp is submitted to the string impulse and moves without friction. This ‘masslet’ acts approach is that any mass mp of matter acts like a clock 2 as a particle somewhat similar to a free-to-move defect of pulsation ωp such that its mass energy mpc balances or impedance jump; the density and elasticity are locally its vibrational energy ~ωp where c is the light velocity altered at the particle location. Its dynamical behavior is and ~ the Planck constant. Then, relying on special rela- tivity to estimate how this moving clock would appear for governed by usual momentum transfer at the impedance an immobile observer, de Broglie showed that the clock jump, giving rise to reflection, transmission or absorption must be in phase with a superluminal phase wave, the of the incoming wave. The masslet being free to move, guiding wave, giving birth to the celebrated but still mys- these mechanisms are accompanied with the radiation terious wave-particle duality. After de Broglie discovered force phenomenon common to acoustic and electromag- this phase wave, he proposed in late 1926 a mechanical netic fields. This drives the particle in the whole field analog [2–4], ‘the double solution theory’. This analog – incident and scattered – and eventually leads to mem- was subsequently followed by some hydrodynamical and ory effects. In the spirit of de Broglie’s assumptions, the sliding masslet can be viewed as a a moving clock of pul- very interesting analogs e.g. the Madelung approach [5]. p In the same series of works [3, 6], de Broglie also intro- sation ωp = kp/mp in its rest frame. duced the pilot-wave interpretation nowadays known as In the past, quite similar models were first proposed de Broglie-Bohm – or ‘Bohmian’ mechanics – after its by Rayleigh and Helmholtz [13–15] to study the vibra- rediscovery in 1952 by D. Bohm [7, 8]. While the quest tions of a loaded string. More recently, Boudaoud et for a classical analog of QM is legitimate, it is far from al. studied the self-adaptation of free-to-move beads on being an easy task and de Broglie failed in extending his a string submitted to acoustic noise [16] in the context earlier results. Quite recently, the interest on this subject of soap film and pattern formation. A few years ago,

arXiv:2002.08147v1 [quant-ph] 19 Feb 2020 was renewed by the pioneering work initiated by Couder Borghesi showed relying on a relativistic framework, that and Fort on bouncing droplets [9–11] in which one or such a system yields a wave particle duality governed by several droplets hit a vertically shaken bath and gener- the also relativistic Klein-Gordon equation [17]. Here, ate a surface wave. Among other works (see [12] for a we restrict ourselves to the non-relativistic Newtonian review), these droplets, sometimes referred to as walkers, framework and unravel the emergence of a transparency were shown to not only mimic a wave particle duality regime in which the particle-string interaction, and thus at macroscale, but also to reproduce most, if not all, of the radiation force, vanish. As we show, this regime, is QM features. This is not that surprising because these reminiscent of de Broglie’s double-solution [3] and leads models share some features with the double solution and to a Schr¨odinger equation for the phase wave associated pilot wave theories [3, 7], known to be possible alterna- to the particle. Very interestingly, two classes of trans- parent particles are possible candidates: (i) the class of subsonic particles that travel uniformly along the string with a velocity smaller than the speed of sound on the ∗Electronic address: [email protected] string and (ii) the class of supersonic (i. e., faster-than- 2

accelerated vertically and horizontally by the local string acceleration and subsequently moves along it. Since the N(t) particle has locally a density or elasticity which are dif- ferent from the string, its inertia will in turn stretch the z (t) p string and act as a source for the field u(t, x), generating scattered waves. Thereby, the particle acts as a clock moved by and within the wave field that it has formerly u(t,x) produced. In this work, we only study non-relativistic movements, i.e. we assume that the velocity of particle and wave are

x p(t) much smaller than the celerity of light defined by the ve- locity c of the elastic waves along the string. This allows FIG. 1: Sketch of the string-mass mechanical analog. A us to define two distinct regimes associated to the Mach bead of coordinates (xp, zp) slides without friction on a string number Ma = |vp,x/c| where vp,x is the horizontal ve- with local transverse amplitude u(t, x). A vertical spring acts locity of the particle. Ma < 1 is the subsonic regime, on the bead as as a vertical restoring force and features an while Ma > 1 is the supersonic one. As we show in the internal clock in the de Broglie double solution of quantum following, the subsonic regime is physically equivalent to mechanics. the bradyonic regime as defined in particle physics for slower-than-light quanta, while the supersonic regime is sound) particles. These two families of dynamical mo- similar to the tachyonic one with faster than light quan- tions are hereafter referred to as bradyons and tachyons, tum particles. The mechanical analogy with special rel- respectively in reference to their quantum counterparts ativity is remarkable and we demonstrate that the ap- in particle physics (where the speed of sound must be re- plication of an equivalent Lorentz transformation (where placed by the speed of light). The paper is structured as the sound velocity replaces the velocity of light) is crucial follows: In Section. II, we describe the system and derive for a quantum description of mechanical analogs. the general dynamical equations. We discuss analytical solutions and present the numerical approach employed for studying the dynamics. In Section. III, we discuss in B. Deriving the equations detail the transparency regimes by focusing on the two quantum analogs of bradyons and tachyons. Concluding Since our system possesses several degrees of freedom remarks are addressed in Section. IV. coupled by a holonomic condition, the Lagrangian for- malism is well suited to derive the equations of motion. Obviously, the total action I of the system can be split II. THE STRING-MASSLET SYSTEM in three parts as

A. General description Z ZZ Z I = dtLp + dtdxLs + dtLint. (1) Our system consists of an elastic string of linear den- sity λ stretched along x by tension T . The string oscil- where Lp and Ls are respectively the Lagrangian of the lates in the transverse z-direction and the vibration is particle and the Lagrangian density of the string (indi- characterized by the field u(t, x). The effect of gravity is cated by the cursive letter) in the absence of the holo- nomic coupling constraint Lint.. We focus first on the neglected throughout the paper. A ring-shaped mass mp expression of the Lagrangian Lp of the particle in our – or bead – is threaded on this string at x = xp and can slide without friction. system. Since we impose an attraction towards the string Importantly, in order to make this particle behave as a baseline, a potential clock, the masslet is placed in an additional elastic force 1 field corresponding to a spring of stiffness k . Thus, it V (z ) = m ω2z2 (2) p p 2 p p p exerts an elastic force towards the non-deformed axis of the string (z = 0) (see illustration in Fig. 1). is introduced, with ωp a pulsation or equivalently kp = Since the mass cannot escape the string, its verti- 2 mpωp a stiffness applied to the particle solely. The par- cal position zp must satisfy the holonomic condition ticle Lagrangian then takes the form zp(t) = u(t, xp(t)) at any time t. Furthermore, the ab- sence of any dissipation (and in particular of friction be- 1 L (t, z , v ) = m v2 − V (z ) (3) tween the particle and the string) imposes that the reac- p p p 2 p p p tion force N(t) from the string to the particle is locally normal to the string at the particle’s location (in the where we have introduced ~vp = vp,x~ux + vp,z~uz with dxp(t) dzp(t) limit of small vibration amplitudes i.e the reaction force vp,x(t) = dt and vp,z(t) = dt the longitudinal is mainly vertical). When the string vibrates, the mass is and transverse particle velocity respectively. Without the 3 holonomic constraint, the variational Lagrange principle point-out that Eqs. 10a, 10b, and 10c can easily be inter- R δ[ dtLp] = 0 leads to the usual Newtonian equations preted using a Newtonian language. The Lagrange multi- plier N(t) is the vertical reaction force acting on the slid- 2 mpx¨p(t) = 0, mp[¨zp(t) + ωpzp(t)] = 0. (4) ing mass while −∂xu(t, x)|x=xp(t) N(t) is the horizontal component of this reaction force (i.e., in the linear limit of Similarly, L is obtained by taking the continuum limit s small wave-amplitude we have ∂ u(t, x)| = tan θ ' of a chain of springs with constant stiffness and consid- x x=xp(t) sin θ where θ is the angle between the reaction force N ering small amplitude vibrations (i.e., neglecting nonlin- and the z vertical direction). We emphasize that Eq. 10c earities): ˜ 2 2 can be rewritten as λ(t, x)∂t u(t, x) = T ∂xu(t, x) = 0 1 1 with λ˜(t, x) = λ + mδ(x − x (t)). In other words the L (u, ∂ u, ∂ u) = λ(∂ u(t, x))2 − T (∂ u(t, x))2 (5) p s t x 2 t 2 x mass m can also be interpreted as a local translatable defect in the linear density λ. This could have an im- In this equation, λ and T are respectively the linear den- pact for physical interpretations. Moreover, the previ- sity and the tension of the string. We also define our local ous dynamics is completed by an analysis of energy and speed of sound c = pT/λ of the transverse waves propa- momentum conservation in the coupled system (i.e., of gating along the string. Once again, for L = 0, one can int. prime integrals of motion). From the full action and La- obtain from the variational principle δ[R dt R dxL ] = 0 s grangian Eq. 1 (or alternatively from Eqs. 10a, 10b, and the wave-equation: 10c) we deduce after lengthy but straightforward calcu- 2 2 lations the local energy-momentum conservation laws for λ∂t u(t, x) − T ∂xu(t, x) = u(t, x) = 0 (6) the field coupled to the mass: where  is the usual linear d’Alembert operator. The specific nature of the holonomic constraint is ide- ∂tε(t, x) + ∂xSx(t, x) = −Nδ(x − xp(t)∂tu(t, x)| ally grasped by the coupling Lagrangian: x=xp(t) ∂tgx(t, x) + ∂xTxx(t, x) = Nδ(x − xp(t)∂xu(t, x)|x=xp(t) Lint.(t, xp, zp,N) = N · [zp − u(t, xp)], (7) (12) or equivalently by a Lagrangian density Lint. (such that R where ε = 1 T [ 1 (∂ u)2 + (∂ u)2] and g = − T ∂ u∂ u Lint. = dxLint.): 2 c2 t x x c2 t x are respectively the u−field energy and linear (pseudo) Lint.(u, xp, zp,N) = N · [zp − u(t, x)]δ(x − xp), (8) momentum density of the acoustic field along the string. 2 Similarly Sx = c gx and Txx = ε are respectively the en- where N(t) is a Lagrange multiplier defining an addi- ergy and (pseudo) momentum density flow along the x di- tional variable in the variational problem. Variations rection (Txx is the constraint tensor of the field which has with respect to N lead automatically to the holonomic only one component here). Moreover, combining Eq. 12 constraint: with Eqs. 10a, 10b, and 10c leads to

zp = u(t, xp). (9) d Z 1 dxε(t, x) + m (x ˙ )2 Taking into account the action of the mass on the string, dt 2 p p the wave dynamics follows a new equation obtained from  R R 1  2 2 2 the variational principle δ[ dt dx(Ls + Lint.)] = 0, + mp (z ˙p) + ωp(zp) = 0 (13a) whereas symmetrically the modified dynamic of the slid- 2 ing mass is obtained from δ[R dt(L + L )] = 0. Alto- d Z  p int. dxg (t, x) + m x˙ = 0 (13b) gether this leads to the following set of coupled equations: dt x p p

which shows that the total energy and linear momentum mpx¨p(t) = −∂xu(t, x)|x=x (t) N(t), (10a) p of the system are conserved.  2  N(t) = mp z¨p(t) + ωpzp(t) , (10b) We stress that despite its non-relativistic nature our N(t) model differs from the one proposed in [17] which consid- u(t, x) = − δ(x − x (t)) (10c)  T p ered a Klein-Gordon wave equation from the start, i.e., Ω2 [ + m ]u(t, x) = − N(t) δ(x−x (t)). In [17] it was shown with the holonomic condition Eq. 9. Interestingly, after  c2 T p that such a Klein-Gordon equation with a source term elimination of N(t) Eqs. 10a and 10b can equivalently be can be used to generate a mechanical analog of wave- rewritten as particle duality. Here we show that it is not necessary  2  to introduce such a complication. An inhomogeneous x¨p(t) = −∂xu(t, x)|x=xp(t) z¨p(t) + ωpzp(t) (11) d’Alembert equation, i.e., as in Eq. 10c, is already suffi- Remarkably, this equation is independent of the mass mp cient to reproduce wave-particle duality. As we show in a fact which is reminiscent of acoustic analogs of gravi- the following this is in complete agreement with original tational forces in hydrodynamical systems [19]. We also ideas developed by de Broglie before 1927 [18]. 4

III. THE TRANSPARENCY REGIME is required to bring a stationary solution. Actually, a ‘first-order’ Poincar´e-Lorentz transformation of the form: 0 0 vp A. The bradyonic or subsonic regime: x = x − vpt, t = t − c2 x is already sufficient to ensure wave-particle duality the invariance of the free wave equation. This is be- vp cause the term − c2 x results from a time synchronization Let us now focus on a particular class of motion for procedure for clocks located at different points of iner- 0 which the coupling force N(t) between the field and the tial frames R and R as already proposed by Poincar´ein particle cancels: 1900 [20]. Moreover, the d’Alembert operator  being in- 0 variant under the usual Lorentz transformation  =  , 2 mp[¨zp(t) + ωpzp(t)] = N(t) = 0 (transparency) (14) we make use of the following coordinate change:  v  yielding a purely oscillatory motion x0 = γ (x − v t) , t0 = γ t − p x , (19) p p p c2

zp(t) = A cos (ωpt + ϕ) (15) q 2 −1 vp with γp = 1 − c2 so that in the Lorenz-Poincar´e with A and ϕ real constants. Eqs. 10a and 10c yield: group in dimension 1+1, the field u(t, x) appears as a scalar invariant field, i.e., u(t, x) = u0(t0, x0). We point x¨p(t) = 0, u(t, x) = 0 (16) out that the variable t0 and x0 have no direct physical which results in an inertial movement for the particle: meaning here since we are working in the context of New- tonian dynamics where the time is absolute. However,

xp(t) = vpt + xp,i (17) Eq. 19 is used as a mathematical tool for finding the so- lutions of the d’Alembert equation under the restriction with xp(0) = xp,i and vp(t) = vp two real constants. Ma < 1 (in this context we emphasize that Voigt [21] In this configuration the different degrees of freedom already used the Lorentz transformation as a mathemat- R 1 2 are decoupled and in particular dxε(t, x), 2 mp(x ˙ p) , ical tool in optics but devoid of physical interpretation). 1  2 2 2 R 2 mp (z ˙p) + ωp(zp) , dxgx(t, x) and mpx˙ p are con- Searching for a standing solution in the Lorentzian co- stant of motions. We emphasize that the oscillatory moving frame R0 that ensures a ‘time-space’ separation 0 0 0 0 0 motion zp(t) is reminiscent of de Broglie’s original idea u (t , x ) = F (t )G(x ), one gets: [1, 18] of associating a local clock to any particle. Here, we have a more detailed mechanical model for which we 1 d2F d2G see that the uniform motion given by Eq. 17 is dynami- G − F = 0 (20) c2 dt02 dx02 cally linked to the harmonic oscillation defined by Eq. 15. Concerning the string field u(t, x), several solutions which turns into a set of equation: are a priori available. Indeed the general solution of d2F d2G ω02 u(t, x) = 0 reads u(t, x) = f(t − x/c) + g(t + x/c) cor- 02  02 + ω F = 0, 02 + 2 G = 0 (21) responding to two arbitrary pulses f(t) and g(t) propa- dt dx c gating respectively along the +x and −x direction. Con- with ω0 a complex constant to be determined. For the sidering for example the case g = 0 and using the holo- case of interest when ω0 is real, F and G are harmonic so vp xp,i 0 0 0 nomic condition Eq. 9 one gets f(t(1 − c ) − c ) = that we obtain an ‘amplitude modulated’ field u (t , x ): A cos (ω t + ϕ), i.e., p ω0  u0(t0, x0) = B cos (ω0t0 + η) cos x0 + ξ = u(t, x)  ω  x   c f(t) = A cos p t + p,i + ϕ . (18) 1 − vp c h  v  i ω0  c = B cos ω0γ t − p x + η cos γ (x − v t) + ξ p c2 c p p This motion would correspond to a mass ‘surfing’ on a ωp (22) monochromatic propagative wave with pulsation vp . 1− c However, while Eq. 18 is interesting in itself we are not B, η and ξ being three real constants. They can be deter- going here to further develop this approach. Instead, we mined by using the holonomic condition expressed for the now follow the physical intuitions of de Broglie and seek case of the uniform motion: zp(t) = u(t, x = vpt + xp,i). solutions for the homogeneous equation u = 0 such that It follows that: 0 in a co-moving inertial frame R translating at velocity  ω0 ω0γ x v  v with the particle, the field u would appear as sta- u(t, x = vt + x ) = B cos t + η − p p,i p p p,i γ c2 tionary. Interestingly, a Galilean transformation of the p 0 0 ω0  coordinates (t = t, x = x−vpt) which lets the time flow × cos γ x + ξ unchanged fails to bring a standing solution. This coor- c p p,i dinate change, although legitimate here for a Newtonian approach (small Mach number) is however too crude to which, by identification with z0(t) given by Eq. 15 yields: allow for clock synchronization at a distance. A coordi- ω0γ x v ϕ = η − p p,i p + 2πn (with n ∈ ), (23) nate transformation that does not let the time unaltered c2 Z 5

ω0  A = B cos γ x + ξ , (24) which is the guidance formula introduced by de Broglie c p p,i in his pilot-wave interpretation [3, 6] and which leads ∂xSbrad. to vp = Q in the limit Ma  1 in agreement and above all the relation: mp with Bohmian mechanics [6, 7]. Therefore, altogether ω0 ωp = . (25) we recover de Broglie’s assumptions casting the double γp solution theory [3] with a ψ-wave also called the guiding To give a meaningful interpretation of this condi- wave (solution of the Klein-Gordon equation) and the u- tion, we can write: u0 = u in Eq. 22 in the form wave (solution of the homogeneous d’Alembert equation) u = B cos S cos Φ with associated with the particle’s movement. Importantly, in brad. brad. our approach (and in contradistinction to [17]) we start 0 0 Sbrad. = ω t + η = ωt − kx + η from the d’Alembert equation and not from the Klein- ω0 ω Gordon equation. Still, we are able to obtain a guiding- Φ = x0 + ξ = (x − v t) + ξ (26) brad. c c p wave ψ which is solution of Eq. 28 i.e., the Klein-Gordon equation. In other words the mass term of the Klein- 0 2 with ω = ω γp, k = ωvp/c . Besides, the following dis- Gordon equation has been generated from the u−field persion relation applies itself. This agrees with a model already presented by 2 02 de Broglie in 1925 [18], i.e., two years before the model ω 2 ω 2 − k = 2 (27) named traditionally the ‘double solution’[3] and based on c c the Klein-Gordon equation. between the pulsation ω and the wave-vector k. The The condition (25) on the frequency can also be written quantity Sbrad. plays the role of a phase for a plane (carry- in terms of phase and becomes a phase-locking condition, iSbrad. ing) wave, ψ(t, x) = e solution of the Klein-Gordon since for x = xp(t) one gets: equation: ω0   02 0 0 1 2 2 ω Sbrad. = ω t + η = t + ϕ = ωpt + ϕ (32) ∂ − ∂ ψ(t, x) = − ψ(t, x). (28) γp c2 t x c2 which expresses the phase-locking of the particle’s clock The phase velocity v is defined by the condition ph. (with pulsation ω ) to that of the wave (with pulsation dS = 0, which implies p brad. ω or ω0). Following de Broglie [18, 22], we can rewrite ω c2 the total wave u as a sum of waves (by means of the vph. = = > c (29) trigonometric identity: 2 cos F cos G = cos (F + G) + k vp cos (F − G)) to give a physically meaningful interpreta- in accordance with the formulas obtained by Louis de tion Broglie in his PhD manuscript [2]. Besides, Φbrad. in Eq. 26 defines an envelop (i.e., a group) velocity which B   x0   identifies with the particle’s velocity vp (setting dΦbrad. = u0(t0, x0) = cos ω0 t0 + + η + ξ 0 we get dx/dt = vp). Furthermore, we deduce the 2 c Rayleigh formula v = dω = v which was also obtained B   x0   gr. dk p + cos ω0 t0 − + η − ξ (33) by de Broglie. This is clearly reminiscent of Hamilton 2 c dHp formula = vp if we identify the Hamiltonian or en- dPp 2 or equivalently ergy function Hp := mpc γp and the linear momentum Pp := mpγpvp with respectively Qω and Qk where Q is B h v x i u(t, x) = cos ω0γ (1 − p )(t + ) + η + ξ a constant having the dimension of an action. The sim- 2 p c c ilarity between Q and ~ is clear and is reenforced if we B h 0 vp x i write Eq. 28 in the limit M  1 as + cos ω γp(1 + )(t − ) + η − ξ (34) a 2 c c 2 Q 2 2 It is another way to express u as the sum of two iQ∂tΨ ' − ∂xψ(t, x) + mpc ψ(t, x) (30) 2mp counter propagating waves u(t, x) = u+(t, x) + u−(t, x) B which is identical to Schr¨odinger’sequation after the sub- with (i) u− = 2 cos [ω−t + ω−x/c + η + ξ] a wave with a low-frequency Doppler shift ω = ω(1 − vp ) stitution Q → ~ (similarly in the Klein-Gordon Eq. 28 we − c 02 2 2 propagating along the −x direction, and (ii) ω mpc can replace c2 by Q2 in agreement with standard text- B u+ = 2 cos [ω+t − ω+x/c + η − ξ] a wave with a books). Moreover, we can also write Hp = −Q∂tSbrad. vp high-frequency Doppler shift ω+ = ω(1 + c ) propa- and Pp = Q∂xSbrad. which are reminiscent of Hamilton- gating along the +x direction. Experimentally, this Jacobi equations with QS playing the role of an action decomposition will allow us to generate the resulting and therefore we have modulated u wave appearing in Eq. 22 as the sum of 2 ∂xSbrad. two plane waves. vp = −c (31) ∂tSbrad. 6

(a) tion u(t, x) = 0 and this simplifies the interpretation of Eq. 34 as the sum of two plane waves whereas de Broglie had to involve a complicated sum of advanced and re- tarded waves emitted by the particle itself [18, 22]. More generally, it is clear that the equations of mo- tion presented here are strongly non-linear, and except for a few specific regimes as the ones discussed in the present paper, no analytical solutions are usually reach- able. In order to further investigate the physics of our model, we have employed a numerical scheme based on (b) standard finite differences through a Runge-Kutta of or- der 4 (see for example [24]). Since the mutual interac- tion between the mass and the string only enters as an external source term for the each other, we have imple- mented separately their differential equations and solve the full system self consistently. We first start by fixing initial compatible configurations for both the string and the mass. During the time resolution process, an update of the string is performed, accounting the presence of the mass. Once this space-time update is achieved, we pro- ceed to the update of the mass now accounting back to FIG. 2: The transparent regime for a bradyonic (i.e., sub- the new state of the string. And we continue so on and sonic) particle plotted from the analytical solutions described so forth until desired time periods. The convergence is in the paper. An equivalent animation obtained from the nu- ensured by respecting the von Neumann stability crite- merical code of the full set of equation is available in the sup- rion. Despite the above mentioned non-linearities in the plemental material [23]. For this example, we have vp/c = 0.1, complete set of equations, the algorithm appears to be xp,i = 0.1 for a string length of L = 1 and c = 1 (in well stable and reproduces very precisely the analytical natural units). We have chosen a ψ−wave temporal period solutions discussed here. (transparency). This validates T = 2π = 10 and η = ξ = 0. The continuous thick gray phase ω our approach and indicates that the coupling between the line is the total field u(x, t) which can be decomposed into mass and the string does not seem to be that crucial in the phase-wave ψ (continuous blue line) and the group wave (dashed gray line). The whole field is shown at two different the convergence. For this first work, we have employed instants (a) t = 0 and (b) t = 0.08 clearly demonstrating the our algorithm in order to complement with the dynam- subsonic (or bradyonic) regime of the particle. The particle’s ics the discussed analytical solutions. In particular, we initial condition is set so that it is at a maximum of the group have recorded a movie [23] showing the behavior of the phase Φbrad. (see Eq. 26) and is clearly locked to it along time. mass in the transparency regime. The question of other The red arrows in panel (a) and (b) compare the position of possible emerging exotic regimes in our coupled system the particle at time t = 0 and t = 0.08 whereas the blue arrow is very interesting though, and deserves a proper study. show the position and displacement of a phase maximum at We dedicate a more systematic numerical analysis to a the two instants t = 0 and t = 0.08. future work. The calculations above deserve some comments in re- The analytical solution is plotted in Fig. 2 for this lation with de Broglie’s picture. Indeed, it is noteworthy that the pulsation ωp remains unchanged regardless of transparency regime (the normalized parameters are in- 0 dicated in the caption). Going back to Eq. 26 we have the particle’s speed and is in particular different from ω 2π (which is velocity dependent). It is slightly different from for the phase-wave a temporal period Tphase = ω and a 2π c2 de Broglie’s results and is a consequence of our mixed spatial period λphase = which have to be compared ω vp approach, combining a Lorenz transformation (by means with the temporal and spatial period of the group wave of c the speed of sound) in a Newtonian framework for 2π c 2π Tgroup = ≥ Tphase, λgroup = c ≤ λphase (as illus- ω vp ω which relativistic dynamics do not apply. More precisely, trated in Fig. 2 for λphase/λgroup = 10. Importantly, we in de Broglie’s approach which is fully relativistic (i.e., have here the well-known de Broglie formula [1]: with the sound celerity replaced everywhere by the light velocity) the particle’s internal clock with pulsation ωp is 2πQ defined in the rest-frame R0, not in the laboratory frame Pp = (35) λphase R. The phase-locking condition reads now: 0 0 It is interesting to note that in de Broglie approach 0 0 ω ω γpxp,ivp 0 Sbrad. = ω t +η = t+η − 2 = ωpt +ϕ (36) the wave field was a solution of a non-homogeneous equa- γp c tion including a source term for the particle u(t, x) ∝  0 ω and we have thus ω = ωp = which differs from δ(x − xp(t)) (i.e., like in Eq. 10c). Here, instead the γp transparency condition N(t) = 0 imposes the field equa- Eq. 25 by a prefactor γp associated with a relativistic 7 time-dilation (we have also η = ϕ since ϕ is now defined (a) in the rest frame[17, 18]). In the non-relativistic limit vp where c  1 and γp ' 1 de Broglie’s theory reduces 0 to ω ' ω = ωp. This is identical in our model to the regime Ma  1 (where c is now the sound velocity) so that the difference between the two approaches vanishes for sufficiently slow particle motions. It is interesting to remark that in the Ma  1 regime the Hamiltonian 2 Hp := mpc γp which was formally introduced reduces to 2 1 2 Hp ' mpc + 2 mpvp which, up to an additive constant, is the translational kinetic energy associated with the parti- (b) cle motion along x (similarly Pp ' mpvp which identifies with the translational linear momentum of the particle). 2 Since c is here the sound velocity, mpc can not physi- cally be identified with a rest energy which is a relativistic Einsteinian concept. This once again stresses the similar- ities and differences between our mechanical analog and de Broglie’s own approach. For similar reasons we have no right to identify the integration factor Q discussed previously with the Planck constant ~. This could only hold in de Broglie’s model for a genuine quantum parti- cle. Still, the mechanical analogy works fine for Ma  1 FIG. 3: The transparent regime for a tachyonic (i.e., super- and could be actually extended to the relativistic regime sonic) particle of velocity wp/c = 10, and initial coordinate by implementing a covariant, i.e. Einsteinian, mechani- Xp,i = 0 (in natural units). The two panels correspond to cal model. Here, we nevertheless stick to the Newtonian observation times (a) t = 0 and (b) t = 0.08. The other pa- framework which is closer to the experimental realization rameters of the wave and string are unchanged with respect 2π 2π c to Fig. 2. In particular, we have Tphase = = = 1 (see and is already sufficient to grasp the essential features of Ω ω vp de Broglie’s mechanical model. text). Like the subsonic case, the particle is locked to the group wave cos Φtach. (dashed gray line) and is clearly faster than the velocity of the phase-wave cos Stach. on the string (blue continuous line). The red and blue arrows compare the B. The Tachyonic or supersonic regime displacement of the particle and phase wave between t = 0 and t = 0.08. So far, we have discussed the dynamics of a particle moving on the string with a velocity vp smaller than that of the waves on this string. Since this motion is this purpose, we see from Eq. 19 that if we set x0 = 0, 0 completely governed by the interaction with the field u, we get x = vpt and therefore the ct axis corresponds we do not expect the particle to spontaneously cross the to the trajectory of a particle with velocity vp < c (this sound barrier without the use of an external force. How- is of course the definition of a comoving rest frame for 0 ever, one could choose initial conditions such that vp > c. such a ‘bradyonic’ motion). If we now set t = 0, we c2 We would then find a whole new supersonic regime, which get x = t = wpt which corresponds to a case where vp we can investigate. As explained in the introduction we the x0 axis identifies with the trajectory of a tachyonic choose to refer to such a particle as a ‘tachyon’ by analogy c2 particle with velocity wp = > c. This is the first hint with the eponymous hypothetical particles introduced in vp special relativity, which are a class of solution for the dy- that the dynamics of tachyons is completely symmetrical namics associated with faster than light motions [25–28] with that of ‘normal’ bradyonic particles. Indeed, we can rewrite Eq. 19 as (symmetrically the case studied previously with vp < c is referred to as ‘bradyonic’ motion [25–28]). Indeed, this 1 w  velocity c appears in the Lorentz transformation that we 0 0 p t = Γp(wpt − x), x = cΓp 2 x − t , (37) use for the field, as in the Lorentz transformation of spe- c c cial relativity, and appears as an asymptotic limit for the 2 wp −1/2 vp velocity vp in both cases. However one should once again with Γp = ( c2 −1) = γp c and where we see, by com- not confuse those two velocities: it is a fundamental limit paring with the original Lorentz transformation, that the within the special relativity framework, while it is not in roles of space and time have been reversed in the tachy- the context of a particle sliding on a string. Supersonic onic case, compared to the bradyonic one. Furthermore, particles (analogous to supersonic aircrafts in a fluid) are like for Eq. 19 the physical meaning of Eq. 37 is not im- indeed possible solutions. Thus, it is completely reason- mediate and here we use it mainly as a mathematical able to study these ‘acoustic’ tachyons, which are phys- tool for guessing at an interesting solution of the wave ically viable solutions of the dynamical equations. For equation. More precisely, let us consider once again the 8

wp stationary field K = Ω c2 which becomes

 0  2 02 0 0 0 0 0 ω 0 Ω Ω u (t , x ) = B cos (ω t + η) cos x + ξ (38) − K2 = − (43) c c2 c2 defined with the variables t0 and x0. From the analysis and the Klein-Gordon equation for the wave Ψ = eiStach. made before we can use this field to match the motion of (with Stach. = Kx − Ωt + ξ) c2 a tachyonic particle with velocity wp = . For this we vp 02 use in Eq. 38 a wave with the same pulsation ω0 as in the Ω Ψ = + Ψ (44) badryonic case. However, as we will see below it implies c2 that we use a different spring with pulsation Ω 6= ω . 02 p p where the signs in front of Ω have been reversed com- Hence, using Eq. 37 we get in the laboratory frame: c2 ω02 pared with c2 in Eq. 27 and Eq. 28. This is specific of  ω0  tachyonic motions where the pulsation or mass can en- u(t, x) = B cos − Γ (x − w t) + η c p p visioned as purely imaginary Ω˜ = iΩ0 (for instance we Ω˜ h 0  wp  i have Ω = r [25, 26, 28]). Eq. 44 thus equivalently × cos −ω Γp t − x + ξ . (39) w2 2 1− p c c2 Ω˜ 2 We find that, once again, the roles of two quantities have reads Ψ = − c2 Ψ which is the usual form for the Klein- been swapped. We have now Gordon equation but now with a purely imaginary mass. Once more, we point out that our supersonic particle is ω0 Φ = ω0t0 + η = − Γ (x − w t) + η only superficially looking as a tachyon. Indeed, genuine tach. c p p relativistic tachyons would induce reluctant causality and ω0  w  thermodynamic problems [25, 26, 28, 29] which are often S = x0 + ξ = −ω0Γ t − p x + ξ (40) tach. c p c2 considered as fatal objections to their mere existence. Here the analogy with tachyon is not complete. For in- Here, the carrying phase wave and the envelope have stance observe that we can in analogy with the subsonic been swapped: the supersonic phase wave that we had regime define the Hamiltonian and linear momentum as in the bradyonic case is now traveling alongside the par- 2 Hp := mpc Γp = QΩ and Pp := mpwpΓp = QK. Phys- ticle; the group wave is now slower than the particle and ically this corresponds to a tachyonic particle of imagi- has become the phase wave (i.e., Stach. = Φbrad. and 0 2 0 2 nary mass imp = iQΩ /c = iQω /c while the physical Sbrad. = Φtach.). The field itself remains unchanged, ‘Newtonian’ mass is of course mp. These expressions are and only the roles of its two components with respect in general clearly different from the usual kinetic energy to the particle have been changed. Thus, all the results and momentum of a Newtonian particle. that were obtained in the case of a particle with veloc- To illustrate the tachyon dynamics we show in Fig. 3 ity vp can be used for a tachyonic particle of velocity 2 an example for this transparency regime (the normalized wp = c /vp if we keep in mind the symmetrical nature parameters are indicated in the caption). Here we have of this supersonic regime. More precisely, considering a 2π for the phase wave the temporal period Tphase = Ω = uniform motion xp(t) = wpt + Xp,i with Xp,i a constant 2π c 2π c2 2π = 1 and a spatial period λphase = = c = and using the holonomic condition Eq. 9 together with ω vp Ω wp ω the oscillatory z (t) motion of Eq. 15 we get by identifi- 0.1 which have to be compared with the temporal and p 2π c 2π spatial period of the group wave Tgroup = = = cation with Eq. 39: Ω wp ω 2π 2π c 0 0.1 ≤ Tphase, λgroup = Ω c = ω v = 1 ≥ λphase (as illus- ω ΓpXp,iwp p ϕ = −ξ − 2 + 2πp (with p ∈ Z), trated in Fig. 3 for λphase/λgroup = 0.1. Clearly the role c of phase and group waves have been swapped compared ω0  A = B cos ΓpXp,i − η , to the bradyonic case. This stresses the strong symmetry c existing between our bradyonic and tachyonic particle. 0 ω0 c Remarkably, it means that a same u−wave can carry ω vp Ωp = = . (41) several bradyons of velocity vp associated with a spring of Γp γp c2 pulsation ωp and tachyons of velocity wp = v but with Comparing the frequency-locking condition with the re- p a spring of pulsation Ωp (compare Figs. 2 and 3). The sult obtained in Eq. 25 we deduce the constraint different particles could actually move together on the c same string since the transparency condition N(t) = 0 Ωp = ωp (42) ensures that the particles do not interact with the wave vp and completely ignore each other. Therefore, we have and consequently Ωp < ωp. here the possibility to generate collective excitations ac- There are, however, a few noteworthy differences in tually reminiscent of bosons surfing on a given wave. some of the equations, mainly the dispersion relation be- Consider for instance the case of n particles (bradyons or 0 tween the wave pulsation Ω = Ω Γp and wave vector tachyons) coherently driven by a carrying u−wave. The 9 situation shows some similarity with a Fock state |ni in to figure out and visualize clearly a possible mechanical quantum mechanics for a collective excitation (e.g., pho- interpretation of the wave particle duality and especially tons or phonons) with n energy quanta. We could imag- of the phase-harmony. It gives a deeper understanding of ine a statistical ensemble of such strings carrying a vari- the possible cause for the clock synchronization on which ous number of quanta n = 0, 1, ..., +∞ and subsequently leans de Broglies theory. develop a particle bosonic statistics with partition func- Interestingly our approach which was developed for nε ω0 Pn=+∞ − K T both the ‘bradyonic’ (i. e., subsonic) and ‘tachyonic’ (i. e., tion Z = e B (KB is the Boltzmann constant n=0 supersonic) regime is a priori not limited to the uniform and T the temperature of the ensemble). The energy εω0 motions. Indeed, the set of equations 10a, 10b, and 10c is the Hamiltonian Hp associated with the bradyons or tachyons characterized by the frequency ω0. We naturally can easily be extended in order to include the effects of obtain Planck’s law or more generally and realistically a a more complex potential V (t, xp, zp). For example, by bosonic statistics if the number of particle is finite and adding a longitudinal potential V (t, xp) Eq. 10a becomes fixed. Clearly, this issue is very interesting for developing m x¨ (t) = −∂ u(t, x)| N(t) − ∂ V (t, x)| mechanical analogs of quantum statistics and will deserve p p x x=xp(t) x x=xp(t) further studies in connection with equilibrium and non- (45) equilibrium thermodynamics. whereas Eqs. 10b, and 10c are let unaffected. The model could thus in principle be applied to regimes involving IV. CONCLUSIONS confining potentials which are of particular interest for further studies on mechanical analogs of QM. In that con- We have developed a simple non-relativistic mechan- text, the development of the numerical method employed ical analog of de Broglie’s wave-particle duality. In the for generating the movie (see [23]) will bring further in- model considered, an oscillating bead or particle mim- sight into the complex dynamic of the system (especially ics an internal quantum clock in phase with a transverse when the ‘transparency’ condition N(t) = 0 is no longer acoustic wave field u(t, x). Despite its non relativistic valid). Finally, while the basis of our mechanical model nature the model is based on a Lorentz transformation is particularly simple, the complexity of the various dy- where the sound celerity c of waves on the string replaces namics (with N = 0 and N 6= 0) open interesting and the light velocity. Therefore the model offers strong sim- fundamental questions. We think in particular of the ilarities with de Broglie’s approach which was based on unexplored tachyonic regime as well as the highly likely special relativity. De Broglie named this phase-locking chaotic regimes that can also be addressed experimen- feature the ‘phase-harmony’ [3] since it emphasizes the tally. For all these reasons, we hope that this work will 2 stimulate further studies exploiting the potential of de fundamental role of the quantum relation mpc = ~ωp. Like with de Broglie’s theory the acoustic u−field here Broglie’s mechanical interpretation of quantum physics. generates a phase-wave Ψ acting as a guiding field for the particle in pure analogy with Bohmian interpreta- tion of QM. In other words, in the presented model the Acknowledgments u-field unifies the particle and the wave in a insepara- ble structure as summarized by the holonomic condition The authors wish to thank J. Duplat for helpful dis- zp = u(t, xp). Thus, one can speak about ‘wave monism’ cussions and comments. Authors also acknowledge the and in that limited sense our analogy deciphers the mean- continuous support of Neel institute during this project. ing of wave-particle duality. Therefore, our work enables C. Poulain thanks the CEA-Leti for its support.

[1] L. de Broglie, Ann. Phys. 10, 22-128 (1925). [10] Y. Couder and E. Fort, Phys. Rev. Lett. 97, 154101 [2] L. de Broglie, Nature 118, 441-442 (1926). (2006). [3] L. de Broglie, J. Phys. Radium 8 225-241 (1927) trans- [11] D. M. Harris, J. Moukhtar, E. Fort, Y. couder and lated in: de L. Broglie and L. Brillouin Selected papers J. W. M Bush, Phys. Rev. E 88, 011001(R). on wave mechanics (Blackie and Son, Glasgow, 1928). [12] J. W. M. Bush J, Phys. Today 68 47-53 (2015). [4] L. de Broglie, Ann. Fond. de Broglie 12 1-23 (1987). [13] Lord Rayleigh, The theory of sounds (University of [5] E. Madelung, Z. Phys. 40, 322326 (1927). Michigan library, 1894) reprinted by Dover publications [6] G. Bacciagaluppi A. Valentini, Quantum theory at the 1945. crossroads: Reconsidering the 1927 Solvay Conference [14] H. V. Helmholtz,Vorlesungen ¨uber theoretische physik, (Cambridge Univ. Press, Cambridge, 2009). Vol 3 (Ambrosuis Barth, Leipzig, 1908). [7] D. Bohm,Phys. Rev. 85, 166179 (1952). [15] P. M. Morse and K. U. Ingard, Theoretical Acoustics [8] D. Bohm and B. J. Hiley, The undivided Universe (Rout- (Princeton University Press, Princeton, 1992). ledge, London, 1993). [16] A. Boudaoud, Y. Couder and M. Ben Amar, Eur. J. Phys. [9] Y. Couder S. Proti`ere,E. Fort, Nature 437, 208 (2005). B 9, 159-165 (1999). 10

[17] C. Borghesi, Ann. Fond. de Broglie 42, 161-196 (2017). entific press, Singapore, 2016). [18] L. de Broglie, C. R. Acad. Sci. 180, 498-500 (1925). [25] J. P. Terletsky, J. Phys. Radium 21, 681-684 (1960). [19] P. Y. Gires, J. Duplat, A. Drezet, C. Poulain, EPL 127, [26] O. M. P. Bilaniuk, V. K. Deshpande and E. C. G. Sudar- 34002 (2019). shan, Am. J. Phys. 30, 718-723 (1962). [20] H. Poincar´e, La th´eorie de Lorentz et le principe de [27] G. Feinberg, Phys. Rev. 159, 1089-1105 (1967). r´eactionin Recueil des travaux offerts par les auteurs `a [28] J. P. Terletsky, J. Phys. Radium 23, 910-920 (1962). H. A. Lorentz, pp. 252-278 (Nijhoff, The Hague, 1900). [29] J. S. Bell, ‘La nouvelle cuisine’ in Between science and [21] W. Voigt Voigt, G¨ottingerNachrichten 8, 177238 (1887). technology, North-Holland Delta Series , edited by A. [22] L. de Broglie, Ondes et mouvements (Gauthier Villars, Sarlemijn and P.Kroes, pp. 97-115 (Elsevier, Amster- Paris 1926). dam, 1990). [23] Follow this link to see the attached movie XXXXX. [24] J. Loustau, Numerical differential equations (World sci-