A synchrotron-like pumped ring resonator for water waves

©Isis Vivanco,1 Alexander Egli,1 Bruce Cartwright,2

Juan F. Marı́n ,3, ∗ and Leonardo Gordillo 1, †

1
Departamento de Fı́sica, Universidad de Santiago de Chile,
Av. Victor Jara 3493, Estación Central, Santiago, Chile
2
arXiv:2406.00264v1 [physics.flu-dyn] 1 Jun 2024

The University of Newcastle, Callaghan, New South Wales 2308, Australia

Pacific Engineering Systems International,
277–279 Broadway, Glebe, New South Wales 2037, Australia
3
Departamento de Fı́sica, Facultad de Ciencias Naturales,
Matemática y del Medio Ambiente,
Universidad Tecnológica Metropolitana,
Las Palmeras 3360, Ñuñoa 780-0003, Santiago, Chile.

Abstract
The wave-like behaviour of matter in quantum physics has spurred insightful analogies between
the dynamics of particles and waves in classical systems. In this study, drawing inspiration from
synchrotrons that resonate to accelerate ions along a closed path, we introduce the concept of a
synchrowave: a waveguide designed to generate and sustain travelling water waves within a closed
annular channel. In analogy to unavoidable energy losses in conventional particle accelerators
due to electromagnetic radiation and inelastic collisions, the system displays undesired water-
wave dampening, which we address through the synchronised action of underwater wavemakers.
Our analogies extend the resonance mechanisms of synchrotrons to generate gravity waves in
closed waveguides efficiently. A proof-of-concept experiment at a laboratory scale demonstrates
the unique capability of this technique to build up anomalously large travelling waves displaying
a flat response in the long-wave limit. Besides quantifying the performance of wave generation,
our findings offer a framework for both industrial and computational applications, opening up
unexplored possibilities in hydraulics, coastal science and engineering. In a broader context, our
experimental apparatus and methods highlight the versatility of a simple yet powerful concept: a
closed-path continuous-energy-pumping scheme to effectively harvest prominent resonant responses
within wave-supporting systems displaying weak dissipation.

1
I. INTRODUCTION

The concept of matter waves is a core idea introduced during the early stages of quantum
physics. Some of the most intriguing properties of quantum systems useful in modern
applications are rooted in wavelike phenomena. Quantum-like effects that are distinctive
of wave dynamics, such as interference and tunnelling phenomena [1, 2], confinement in
quantum corrals [3], quantised responses [4], bound-state generation [5], pilot-wave dynamics
[6], quantisation of orbits [7], have been scrutinised in different fields of classical physics that
support waves, ranging from electromagnetism to acoustics. However, there is no better
playground to explore and visualise such analogies than fluid mechanics [8–12]. This stands
even though energy dissipation in fluids is orders of magnitude larger than at a quantum
scale. Indeed, water-wave-decay inhibition [13, 14] is possible and has been achieved by
using experimental configurations that compensate energy losses through moving boundaries
[13, 15], leading to inviscid-like water-wave behaviour. Similar mechanisms have been helpful
in studying energy transmission in tsunami events [16], internal waves in stratified fluids
induced by tides[17], and nature-inspired locomotion strategies in viscous flows [18].

Energy compensation in weakly dissipative systems often implies the existence of reso-
nances, i.e. abnormal amplifications of the system response at given frequencies. Under cer-
tain conditions, a resonance can be engineered to devise valuable applications. For instance,
synchrotrons are particle accelerators operating under the resonance and synchronisation
principles. Their use as a radiation source is widely spread for applications in fundamental
research, applied physics, and the medical industry[19]. In a synchrotron, ions emitted from
a particle source accelerate while restrained to travel along a closed path at constant speed.
The curved trajectory is achieved through the action of a static magnetic field. However,
ions on a curved path lose energy through electromagnetic radiation. Ions also suffer inelas-
tic scattering with other particles within the ring, which are present even under ultra-high
vacuum conditions [20]. Such energy losses are compensated through the action of an alter-
nating electric field applied at every segment, or building block, of the closed path. When
properly synchronised with the electrical field in each building block, the synchrotron be-
comes resonant, and the ions describe sustained propagation in a closed path at constant
speed under the action of the Lorentz force.

Inspired by the motion of charged particles in a synchrotron, here we propose and study

2
a synchrotron of water waves, or synchrowave, a device that builds up gravity waves on
a channel and sustains its propagation along a closed circular path at a constant speed.
Inspired by the segmented alternating electric fields that drive the particles in conventional
synchrotrons, a synchrowave sustains travelling waves by pumping energy at discrete build-
ing blocks through synchronous periodic actuation of servomotors under a soft bottom.
Such periodic motion uses the recently proposed wave generation technique coined as pedal
wavemaking [15]. We study the linear response of surface gravity waves by analysing the
hydrodynamic equations for viscous fluids under appropriate boundary conditions. We show
that the synchrowave displays resonance, wave synchronisation, and filtering-like behaviour
as a function of the wavelength, frequency, and viscosity. To round off, we designed and set
up a tabletop synchrowave with 64 building blocks to test our hypotheses. We demonstrate
that under resonant conditions, a small amplitude motion of the servomotors sustains large
amplitude waves at the surface. Our experimental results agree quantitatively with our
theoretical findings and match Smoothed-Particle Hydrodynamics numerical simulations as
well.

THE SYNCHROTRON-LIKE WATER WAVEGUIDE

Our proof of concept of a synchrowave designed to sustain travelling water waves with a
programmed wavelength is shown in Fig. 1(a). The water is confined in a 125-cm-diameter
annular waveguide or channel, 3.6-cm wide, made of laser-cut acrylic cylindrical walls and
3d-printed synthetic-polymer (PLA) and plastic-ABS parts, as shown in Fig. 1(b). A soft
elastomer cast seal bed on the base is pressed to ensure a waterproof and snug fit, as shown
in Fig. 1(b).
The synchrowave is assembled from 64 building blocks, each including a servomotor fixed
to the soft-bed section of the channel. Each servomotor has an individual transmission
system with a connecting rod and a cam for converting servomotor rotational motion to
a vertical push. The transmission system also comprises linear bearings and a piston, as
shown in Fig. 1(c). The purpose of the servomotor array is to pump energy constructively
to the waves as they travel, thus compensating for their losses. Servomotors are wired to
power supplies and connected to a central microcontroller that establishes a daisy-chain
communication protocol for independent control. We placed the channel between a high-

3
FIG. 1. The Synchrowave: a synchrotron water waveguide machine. (a) Three-
dimensional scheme of the synchrowave. The annular waveguide channel for gravity waves consists
of two annular walls and a soft bed on a set of 64 servomotors uniformly spanned along the bottom
for multipoint action. The free-surface motion is recorded using a high-speed camera and an LED
panel. (b) The zoom-in view highlights the servomotors’ arrangement below the soft bed. (c) A
zoom-in on the base of each unit shows every component involved in motion transmission, which
allows independent control. (d) Snapshot of a sampling window on the experimental system under
the action of the moving water bed.

speed camera and an LED panel, as shown in Fig. 1(a), to record the dynamics of the free
surface in a window (see Fig. 1(a,d)).
For each experimental run, we start with the fluid at rest and slowly increase the servo-
motors’ oscillatory motion amplitude at both fixed input frequency and wavelength. Once
the fluid’s surface wave amplitude has reached a stable value, we use a boundary-detection
image algorithm to measure the fluid displacement, and hence, the wave amplitude η0 . The
details of the experimental protocol and image processing are described in Methods. Figure 2
summarises our experimental characterisation of the synchrowave response to the bottom
periodic motion using an eight-servomotor wavelength, corresponding to λ ≈ 49.1 cm. Fig-
ure 2(a) shows the system gain in amplitude G, defined as the ratio of the wave amplitude η0
to the bottom displacement ζ, as a function of the dimensionless input frequency ω ≡ ω/ω0 ,
i.e. the ratio of the forcing frequency ω to the natural frequency ω0 , which depends on λ
√
via the dispersion relation of water waves on uniform depth, ω0 (k) = gk tanh kh, Here,

4
 (a) c = 46.7 cm/s

c = 44.2 cm/s

t c = 47.9 cm/s
t
101 c = 38.0 cm/s
x t

t x
c = 63.8 cm/s
G ≡ η0 /ζ

c = 24.6 cm/s
x

x
t t

100 η0

x
x
ζ

resonance
(out-of-phase synchronization)

(b) π 


1  sinuous mode 
π
φ

2
varicose mode (fast)
(slow)
0
0.5 1 1.5
ω ≡ ω/ω0

FIG. 2. Resonant response of gravity waves in a synchrotron water waveguide: Ex-

perimental results in a water channel of depth h = 1.3 cm for a bottom actuation of wavelength
λ/R = π/8 (λ ≃ 49.1 cm) at an amplitude a = 1 mm. Servomotors are synchronised to oscillate at
different frequencies, keeping the traveling-phase scheme. (a) Gain (wave-to-bottom displacement
ratio) of the system as a function of the dimensionless forcing frequency. Theoretical prediction
(dash-dotted blue line) and experimental measurements (red triangles) exhibit a resonance peak.
Spatiotemporal diagrams at different forcing frequencies are shown in the insets, evidencing the
emergence of travelling gravity waves on the free surface around the resonant condition. The speed
c of the wave is shown at the top of each snapshot. (b) Phase between the wave and the bottom
displacement as a function of the driving frequency of the servomotors.

5
k = 2π/λ and h = 1.3 cm is the water depth. The data shows that the travelling gravity
wave gradually builds up as the frequency of servomotors increases from f = 0.5 Hz up to
f = 0.875 Hz, where the amplitude of the gravity wave reaches its overall maximum value.
The maximal gain is placed at ω = ωres , close to the natural frequency condition ω = 1.

Synchrowave-generated waves have the same frequency of oscillations as the bottom mo-
tion. Since the amplitude of the servomotors’ oscillations a is small compared to the wave-
length (λ/a ≃ 491), the system is linear and responds at the same input frequency. Yet,
different synchronisation regimes exist, as evidenced by the phase shift between the servo-
motors’ motion and the generated gravity wave. Figure 2(b) shows the phase shift of the
wave as a function of the servomotors’ frequency. We observe the emergence of a slow mode
for frequencies far below the resonance condition, ω ≪ ωres , where the free surface and the
soft bottom slowly oscillate in anti-phase, thus generating a varicose fluid layer (see left
inset in Fig. 2(b)). When increasing the frequency, the system approaches the resonance
condition. At resonance, ω = ωres , the surface-wave and servomotor displacement become
out-of-phase (see bottom inset in Fig. 2(a)). Far above the resonance condition, ω ≫ ωres ,
we observed the emergence of a fast mode, in which the free surface and the servomotors
oscillate in phase. In this latter case, the synchronisation establishes a sinuous fluid layer in
which the water wave replicates the bottom displacement (see right inset in Fig. 2(b)).

After having depicted our experiment and summarised its main outcomes, we draw a
parallel between conventional synchrotrons and synchrowaves, emphasizing their operat-
ing mechanisms’ resemblance and differences (see Table I). Further details are provided in
Methods.

TA-
BLE 1: Analogies between synchrotrons and synchrowaves: Parallel between the con-
stituent parts and the physical processes in charged-particle synchrotrons and water-wave syn-

6
chrowaves.

SYNCHROWAVE THEORETICAL MODELLING

To provide further insights into the dynamic response of synchrowaves under general
synchronisation schemes, we model the synchrowave as a fluid layer of uniform depth h in
the x − z plane and kinematic viscosity ν, under the action of gravity, g = −gk. The
bed on which fluid lies is at z = −h and is deformable, i.e., it is allowed to displace and
push the liquid above. The model neglects the effect of the curvature of the waveguide
of the synchrowave, as it is negligible compared to the wavenumber of the water waves.
A description and quantitative justification for the accounted simplification is provided in
Methods.
A vertical travelling-wave displacement introduced by the servomotors at the base can
thus be described by the complex field zb , given by

zb (x, t) = ζ exp[i (kx − ωt)], (1)

where ζ is the complex amplitude of the forced vertical motion, and the phase propagates
in the positive x-direction (hereon the streamwise direction), with prescribed wavenumber k
and angular frequency ω. We use the dimensionless wavenumber k ≡ kh and frequency ω ≡
ω/ω0 . Likewise, the amplitude of the surface wave, the vertical motion of the servomotors
at the bottom, and the space coordinates in dimensionless form are η 0 = η0 /h, ζ = ζ/h,
z = z/h, and x = x/h, respectively. Solving the linearised system (see Methods), we obtain
the gain G, i.e. the ratio of the free-surface-wave complex amplitude η 0 to bed vertical-
oscillation amplitude ζ, in terms of the parameters of the system, given by

1 + iΘ cosh αk − iΘ cosh k
G(ω, k) =
  h    i .
cosh αk cosh k 1 − ω12 1 − tanhc αk + sinhc αk + k sinhc2 k(α+1)
2 2 k(α−1)
+ sinhc
tanhc k sinhc k 2 2
(2)
p p
Here, Θ ≡ 2ν/ gh3 is the dimensionless viscosity of the fluid, and α ≡ 1 − iω/νk 2 is
a dimensionless complex parameter. The functions sinhc(x) = (sinh x)/x and tanhc(x) =
(tanh x)/x are the cardinal hyperbolic sine and tangent, respectively [21]. Figure 3(a) shows
that the gain derived from equation (2), exhibits a resonance peak for ω = ω res ≃ 1, i.e,

7
 G
(a) 102 (c) 102 102

high frequencies pass

s viscosity
G

100 ncie
freque dependent
low a s s 100
ot p

λ ≡ λ/h
do n
10−2 101
(b) π
ω res /ω 0 10−2
1
π
φ

2
ω=1
0
100 −1 10−4
10−1 100 101 10 100 101
ω ω

FIG. 3. Theoretical frequency-response curves: Prediction for a 40 m-deep channel filled with
a fluid of viscosity ν = 1800 cSt (Θ = 4.55 × 10−4 ). (a) The gain of the synchrowave for λ = 10,
√
exhibiting a resonant peak near the frequency ω0 = gk tanh kh. The system behaves as a filter
of low frequencies. (b) The phase ϕ = − arg(Y ), revealing synchronisation at frequency ranges
above the resonance condition (indicated by diamonds). (c) Gain of the system as a function of
the wavelength and the frequency of the bottom motion.

for ω ≃ ω0 . At such precise synchronous conditions, the motion is optimally transferred

from the servomotors to the surface of the fluid, generating the largest amplitude wave
able to circulate indefinitely in the channel as long as the action of the servomotors is
sustained. This mechanism has a complete analogue in conventional synchrotrons, where
beam losses are compensated through alternating electric fields synchronously applied with
a precise (resonant) frequency through radio-frequency cavities. These points of input of
energy must be conveniently placed along the circular path of the curved beam. Indeed,
the spatial separation between such cavities and the phase difference between their emitted
fields induce an effective space dependence and wavelength of the alternating electric field
along the loop. The analogy between this mechanism in conventional synchrotrons and the
”field” generated by the motion of our array of servomotors thus becomes evident.

Figure 3(a) also shows that the synchrowave filters slow frequencies as there is a persistent
gain decay under ω res , while in the fast frequency spectrum limit, the gain reaches a plateau.
The system’s temporal asymptotic response, far from the resonance, is, hence, that of a high-
pass filter, a behaviour reported for impulsively forced water waves in the context of tsunami

8
 (a) 102
(b) 103 (d) 1.00
ω res
101 ω0
102 0.99
100 100 101 102
G

G
λ
10−1 (e) 1.00
101
ω res
ω 0 0.99
10−2

10−3 100 0.98

0.8 0.9 1 1.1 0.8 0.9 1 1.1 10−5 10−4 10−3 10−2
ω ω Θ
(c)
102
Gmax

101

100
100 101 102
λ

FIG. 4. Performance of the synchrowave: Synchrowave response as a function of viscosity and

wavelength following our theoretical modelling. Gain as a function of the dimensionless frequency
for different values of: (a) the dimensionless wavelength λ = [1, 20] at fixed dimensionless viscosity
Θ = 4.5 × 10−4 ; and (b), the dimensionless viscosity Θ = 4.5 × [10−7 , 10−3 ] at fixed wavelength
λ = 10 (c) Maximum gain in the synchrowave, Gmax , vs. wavelength for Θ = 7.15 × 10−6 . (d )
Frequency of the resonant mode as a function of the dimensionless wavelength λ and (e) the
dimensionless viscosity Θ.

generation [16, 22, 23]. Likewise, the theoretical model can predict the synchronisation
modes as a function of frequency. The Bode phase diagram shown in Fig. 3(b) shows the
phase ϕ = − arg(η 0 ) of the gravity wave as a function of the input frequency ω. It shows
clearly that the wave synchronises in two different modes: Far below the resonant frequency,
the fluid layer propagates through the waveguide in a varicose mode, while far above, it
follows the bottom in a sinuous mode.
To contrast our theoretical results, we appended in Fig. 1 our predictions for the gain and
the phase (dash-dotted lines), for easy comparison with the synchrowave experimental data.
Given that our theoretical model does not consider viscous effects due to the friction with the
lateral walls, we used fluid viscosity as a parameter to be fitted, keeping the remaining ones
matched with our experimental conditions. The agreement between theory and experiments
is remarkable in both the gain and the phase response.

9
 The model provides a mechanical characterisation of the system under general conditions.
Figure 3(c) summarises the synchrowave gain as a function of both the input frequency and
wavelength, evidencing how synchrowaves filter short waves. For a long wavelength, the gain
remains almost constant [see top left plateau in Fig. 3(c)] reaching saturation around λ ≃ 10,
and plumbing when the wavelength decreases further. Thus for a random distribution of
wavelengths and frequencies far enough from the resonance, only those components with high
frequencies and long wavelengths will build up. This property may find relevant applications
in the recreation of oceanic waves in engineered tanks, given that ocean waves due to storms,
earthquakes, and tides are long [24].
Besides displaying the high-pass filter and resonant behaviour for any forcing wavelength,
the synchrowave has an overall optimal wavelength, as revealed by Figure 4(a). Figure 4(b)
shows the intuitive but still noteworthy feature that the resonance peak becomes prominent
and sharper as the viscosity decreases. At fixed viscosity, the response also displays a
maximum gain Gmax for each input frequency. Gmax displays an overall high-gain low-
pass filter behaviour when plotted against λ, as depicted in Fig. 4(c). Likewise, Fig. 4(d )
shows that the resonance frequency ω res ∼ 1 in a two-orders-of-magnitude-span wavelength
interval. Finally, we also found in our theoretical expressions that the properties of the
waves are very robust to viscosity effects. Figure 4(e) shows that ω res ∼ 1 inside a wide
range of small viscosity values, Θ ∈ [10−6 , 10−2 ] for an arbitrary wavelength λ = 10.

NUMERICAL SIMULATIONS

To test our results, we performed numerical simulations on a fluid layer with periodic
boundary conditions along the streamwise direction using the Smoothed Particle Hydrody-
namics (SPH) method [25, 26]. All our simulations consider a fluid of density ρ0 = 1 g/cm3
in a domain of depth h = 1 cm and length L = 20 cm, under the action of a moving bottom.
The details of the numerical methods are given in Methods. We study the SPH system
response to servomotors in an input frequency span ω ∈ [0.1, 10] and fixed λ = 10.
Figure 5 presents the results from SPH simulations. In Fig. 5(a), we confirm that the
synchrotron water waveguide exhibits typical resonance features, where the gain reaches
the value Gmax = 3.1143 at ω = ω res . Moreover, SPH simulations confirm the high-pass
filtering behaviour of the system: the gain strongly decays for decreasing values of ω below

10
 (a)
100
G

10−1

10−2

(b) π
theory
SPH simulations
1
π
φ

100 101
ω

FIG. 5. Gain and phase of the synchrotron water waveguide: Results for λ = 10. Compar-
ison between SPH simulations (red diamonds) and analytical results using equation (2) (dashed
blue line). Theoretical curves corresponds to Θ = 0.0958 (ν = 1.5 cSt).

ω res . In contrast, if the frequency is increased above the resonance, the gain asymptotically
decays towards unity. Given that our SPH simulations have been purposedly made highly
viscous, the resonant peak is relatively broad, and the resonant frequency is slightly below
the value corresponding to the inviscid case, ω = 1. Such behaviour was previously observed
in Fig. 4(e) in our model. In Fig. 5(a), we show with a dashed blue line the predicted curve
following equation (2) for the gain corresponding to Θ = 0.0958, which is in remarkable
agreement with SPH simulations.

Finally, Fig. 5(b) shows the phase as a function of frequency obtained from SPH simu-
lations, along with the corresponding theoretical curve for Θ = 0.0958. We introduced a
shift ∆ϕ = 0.21 in the theoretical curve of the phase to consider numerical effects due to
the artificially introduced compressibility of fluid particles in our SPH formulation [27]. We
observe that the resonance condition occurs slightly below ω = 1, consistent with our previ-
ous observations. SPH simulations confirm that the frequency of surface waves synchronises
with the servomotors at fast frequencies.

11
 CONCLUSIONS

In this work, we emulated the resonance phenomenon that occurs in a synchrotron of

charged particles into a new synchrotron of gravity water waves. Instead of a system where
an oscillating electromagnetic field is induced at discrete magnetic cavities distributed along
a circular path for accelerating charged particles, we designed a ring-shaped water channel
made of discrete mechanical actuators, or building blocks, each one including a servomo-
tor that pushes water from the bottom of the channel to excite and sustain gravity-wave
trains at a constant speed. We have built a lab-sized synchrowave using 64 servomotors
uniformly distributed along the bottom of a 125-cm-diameter annular waveguide. We anal-
ysed the response of travelling waves as a function of the synchronisation properties of the
servomotors. We studied and characterised the resonance of gravity graves generated at
the free surface of the viscous fluid using experimental data measured in the synchrowave
and in simplified theoretical and numerical models. For our theoretical study, we analysed
the linear Navier-Stokes equations in an infinite two-dimensional fluid layer of finite depth,
providing a closed-form expression for the system gain as a function of fluid and bottom-
motion parameters. The system exhibits a resonance peak and behaves like a high-pass
filter for frequencies and a long-pass filter for wavelength. The resulting resonant peak can
be sharp for small viscosity and display remarkable amplification factors. Our theoretical
model accurately captures the observed experimental frequency response of the synchrowave
and shows that one can always tune the frequency at which the system resonates for a given
wavelength. Our main result is that synchrowaves can generate large amplitude long waves
efficiently, with orders-of-magnitude amplifications for ordinary conditions.

Fluid parameters naturally affect the resonance: Increasing viscosity value decreases the
system gain, widens the resonance peak and introduces a slight resonant-frequency shift.
Accordingly, we have also shown an optimal wavelength for a given combination of viscosity
and frequency at which the gain attains the maximum possible value. Thus, our wave-
generation setup is efficient and can accurately mimic wave dynamics. Since our results
can be used to design water waves with an on-demand wavelength, even in viscous fluids in
narrow water channels, we test our hypotheses numerically using a basic formulation of SPH,
which can be a highly viscous method without modern correction techniques. We performed
a complete numerical characterisation of the resonance, showing excellent agreement with

12
the theoretical predictions even at large viscosity.
Our theoretical modelling within the framework of dimensionless variables can be scaled
to describe physical systems significantly larger than our tabletop synchrowave. A scaled-up
water synchrowave should be able to display even sharper resonances with colossal gains.
The results of this work suggest promising applications in the study of oceanic phenomena
involving long gravity waves at the surface of the sea, such as ordinary tides, trans-tidal
waves, and other waves generated by storms and earthquakes. In addition, our results may
shed light on the open question of how relatively small-amplitude perturbations generate
large-amplitude surface wave events, such as super-tsunamis appearing after small-intensity
earthquakes and rogue waves appearing without warning in otherwise benign conditions.
It is still shocking how the simple, even obvious, scheme of pumping bits of energy along
a closed path to build up amplitude after every turn can end in huge outcomes, especially
when the energy losses are weak. The synchrowave, as well as the synchrotron, exploit
this simple idea. One could wonder if such a scheme exists unprompted in nature. The
Drake Passage, which separates South America’s southern tip and Antarctica’s northern
tip, creates the only water belt across the globe with no relevant landmasses within. In
this ocean region extending from the latitudes between parallels 50◦ and 60◦ south, circular
unidirectional currents and tides find enough room to create free-surface resonant waves.
For centuries, sailors have known this region for being almost impossible to sail due to their
unexpected surges, rogue waves and unpredictable weather [28]. A 19th-century sailor saying
may provide an emotional yet accurate account of drifting on a planetary-scale synchrowave:
Beyond 40 degrees south, there is no law. Beyond 50 degrees south, there is no God.

II. METHODS

THE SYNCHROWAVE: A SYNCHROTRON-BASED DESIGN

Our experimental development is inspired by a fluid mechanics analogue of synchrotrons of

charged particles, already summarised in Table I. In conventional synchrotrons, beams of acceler-
ated particles are bent between dipole magnets. The trajectory of particles is deflected under the
Lorentz force, and the storage ring is achieved by arranging a series of such bending magnets with
appropriate separation around a closed loop. In our synchrowave, the curved walls of the annu-

13
lar channel naturally deflect the generated waves, closing their path after a turn in a waveguide.
Curvature in synchrotrons and synchrowave introduces an effective loss. Indeed, the synchrotron’s
charged particles under the Lorentz force emit radiation as their direction of propagation changes,
thus leading to a slow energy decay. In complete analogy, the walls on our synchrotron water
waveguide also lead to wave losses in gravity waves due to viscous boundary layers and meniscus,
leading to effective friction between sloshing waves and the curved walls. Moreover, particles in
conventional synchrotrons undergo unavoidable inelastic collisions with other particles in the ac-
celerator, even under the best high-vacuum conditions achieved to date [20]. The analogous loss
in our synchrowave machine is dampening due to viscosity in the bulk, which accounts for internal
inelastic processes within the fluid due to adjacent-fluid layers’ relative motion. The 64 building
blocks distributed along the channel with servomotors compensate for the loss mechanisms. The
oscillating mechanical waves introduced by the array of servomotors are equivalent to the oscillat-
ing electromagnetic fields generated between the magnets conveniently distributed in conventional
synchrotrons. The critical point of synchrotrons and our synchrowave is the choice of a synchroni-
sation scheme for the oscillating field to enable an optimal transfer of input energy to sustain the
circular orbits at resonance.

EXPERIMENTAL DEVELOPMENT

The synchrowave is a 3.6-cm-wide channel along a 125-cm-diameter circumference enclosed by

two 30.0-cm-high concentric walls. All pieces and parts of the synchrowave were designed using
SolidWorks©, and manufactured using 3D printing and laser cutting. The setup is placed on a
structure built with T-slotted aluminium profiles that allow the proper levelling of the channel, as
shown in Fig. 6.

The soft elastomer base that rests over the acrylic base, as can be seen in Figs. 1(b) and 1(d),
is made of a polyaddition polymer made of room-temperature polymerising silicone from Esprit
Composite (Shore hardness scale OO). This material can reproduce fine details through moulding.
It is also resistant to ageing and moisture. The base of the synchrowave comprises two 3-mm-wide
ring-shaped acrylic layers, on which the pistons of the servomotors fit in such a way that popper
motion is guaranteed.

We used 64 coreless Dynamixel AX – 18A servomotors, manufactured by Robotis, connected

14
in series and fixed to the acrylic base channel by rubber restraints, and fed by eight power supplies
of 12V and 29A regularly distributed along the loop. Each servomotor can process one bit of data
sequentially through a single communication channel composed of a receptor and a transmitter.
Each servomotor has a range of movement between 0 and 300◦ , with an approximate resolution of
0.29◦ . Each unit of the structure independently converts the servomotor’s orbital motion to the
piston’s vertical motion. The units contain steel shafts, laser-cut acrylic and 3D-printed PLA/ABS
parts.
Servomotors are controlled through the microcontroller OpenCM 9.04 using an open-source code
written in Arduino IDE. This code assigns an ID number i between 0 and 63 to each servomotor
according to its position along the synchrowave. The servomotors’ rotational motion is responsible
for generating the pistons’ vertical displacement. The vertical position of i-th motor as a function
of time and space is defined as
 
t 2π
zi = A sin 2πf − i , (3)
1000 λn
where A is the bottom movement amplitude in millimetres, f is the rotation frequency in Hertz, t
is time in milliseconds, i is the servomotor ID, and λn is the number of servomotors that complete
one spatial period. To satisfy the periodic boundary conditions, λn must be a power of 2 with
2 ≤ λn ≤ 64. If λn = 2, then 32 wavelengths fit in the synchrowave, while for λn = 64, one single
long wavelength fits in the whole domain.
Each servomotor receives the maximum angular position as input, which can be mapped to a
unique piston vertical position. These parameters are related after the system geometry shown in
Fig. 6: When the motor is positioned in its initial position, the connecting rod has a 45◦ angle with
the horizontal. Then, the angular position of the i-th motor is related to its piston position zi by
 2
zi + r2 − b2 180◦

θi = arccos + 45◦ . (4)
2rzi π
The code we implemented allows real-time control of the frequency, wavelength, and amplitude.
The amplitude is limited by the dimensions of the system of motion transfer, and the maximal
frequency is limited by the servomotors’ response time. There is also a constraint for the maximal
torque, but all the measurements are below this limit.
To characterise the experimental gain of water waves in the synchrowave, we recorded the
motion of the free surface by locating a high-speed camera and a backlit LED panel in a section
of the channel, as shown in Fig. 1(a). With a wavelength of 49.09 cm and 1.3 cm water height, we

15
 (a) (b)
Initial
position
xi b

r Equilibrium
position

FIG. 6. Experimental development of the synchrowave: (a) Photograph of the synchrowave

tabletop setup. (b) Transmission system to transform angular motion to vertical at each building
block. The pivoting point is r = 5-mm apart from the centre of rotation of the servomotor, and
the connecting rod is b = 17−mm long.

performed a frequency sweep between 0.5 and 1.3 Hz, recording 1024×520-pixel-resolution videos
for each frequency (1 pixel corresponds to 0.0624 mm). The sampling frequency is adjusted to
fully resolve the bottom motion by recording 256 frames in 4 periods. In each video frame, we
used an image detection algorithm to find the position of the free surface and bottom. The data is
conveniently plotted in spatiotemporal diagrams for each frequency, showing the free surface motion
as a function of time and space, as shown in Fig. 2(a). In addition to the bottom displacement
data, this is used to derive the experimental amplitudes, η0 and ζ, defined as the peak-to-peak
displacement of the wave and the bottom. The experimental gain G for each frequency is simply
defined as
η0
G(ω) = . (5)
ζ

The experimental phase can also be obtained straightforwardly by measuring the time shift between
the instants at which the wave and bottom displacement reach their maximal value throughout a
cycle.

CALCULATION OF THE GAIN OF THE SYNCHROWAVE

The synchrowave fluid mechanics is governed by the Navier-Stokes (NS) equations for incom-
pressible fluids,
1
∂t u + (u · ∇)u = − ∇P + ν∇2 u + g, (6a)
ρ

∇ · u = 0, (6b)

16
where u = ui + wk is the velocity field, P is the pressure, ν is the kinematic viscosity, g = −gk
is the acceleration of gravity, and ρ is the density of the fluid (water). We shall assume that the
system is invariant along the y-direction, i.e. perpendicular to the walls.
The input drive of the system is assumed to follow the pedal-wavemaker driving[15], which
assumes an orbital periodic motion at each point of the water bed with a phase that travels at a
constant speed. This approximation for our synchrotron will be justified later. Thus, the vertical
motion of the bottom is given by equation (1) satisfying the no-slip boundary condition at the
bottom,
∂t zb = w |z=−h , (7)

whereas the horizontal periodic displacements of pedal wavemakers, which can be described by the
complex field xb , is given by
xb (x, t) = iχ exp[i (kx − ωt)], (8)

where χ is the amplitude of the horizontal component of the pedalling motion. The horizontal
input drive must also comply with the no-slip boundary condition at the bottom,

∂t xb = u |z=−h . (9)

The linear response under study, i.e. the output of the system of equations (6), to the input drive
of equation (1) is a surface gravity wave, denoted as η, which in the linear regime has the same
frequency ω and wavenumber k as the bottom motion,

η(x, t) = η0 exp[i (kx − ωt)], (10)

where η0 is a complex amplitude. At the top of the fluid domain, the velocity field must comply
with linearised boundary conditions for a free surface, namely

∂t η = w |z=η , (11a)

P
2ν∂z w − = 0, (11b)
ρ z=η

∂x w + ∂z u |z=η = 0. (11c)

where equation (11a) follows from the kinematic boundary condition, and equation (11b) and (11c)
from normal and tangential stress balance. Following Vivanco et al.[15], we solve the linearised
version of equations (6) using the Helmholtz decomposition [29],

u = ∇ϕ + ∇ × (ψj) , (12)

17
where ϕ is the velocity potential and ψ, the Stokes stream function. After replacing the Helmholtz
decomposition in the governing equations, the velocity potential ϕ must satisfy,

∇2 ϕ = 0, (13a)

P
∂t ϕ + + gz = 0, (13b)
ρ

while the diffusion equation governs the stream function ψ,

∂t ψ = ν∇2 ψ. (14)

Accordingly, the exact solutions of the velocity potential and the stream function are of the form
[15],
ω
ϕ(x, z, t) = [A cosh(kz) + B sinh(kz)] exp[i(kx − ωt)], (15a)
k

ω
ψ(x, z, t) = [C cosh(mz) + D sinh(mz)] exp[i(kx − ωt)], (15b)
k
p
where m = k 2 − iω/ν is a complex wavenumber that accounts for viscous effects.

Hereon, we use the dimensionless wavenumber k = kh and m = mh, and the dimensionless
√
frequency ω = ω/ω0 , where ω0 = gk tanh kh is the dispersion relation of gravity waves. Likewise,
the amplitude of the surface wave, the horizontal bottom motion, the vertical bottom motion, the
vertical coordinate, and the horizontal coordinate in dimensionless form are η 0 = η0 /h, χ = χ/h,
ζ = ζ/h, z = z/h, and x = x/h, respectively.

Replacing the solutions of equation (15) the boundary conditions of equations (11), we ob-
tain the following linear system of equations for the dimensionless vector (A, B, C, D)T ≡
(A, B, C, D)T /h,
    
3/2 3/2
 ωΘβk i 1 −iαk Θ A 0
    
 0 i β 0  B   0 
  =  , (16)
    

 i cosh k −i sinh k −α sinh m α cosh m   C  χ
    
    
−i sinh k i cosh k cosh m − sinh m D ζ
p
with α ≡ m/k, β ≡ 1 − i/θ, and Θ ≡ 2ν/ gh3 is the dimensionless viscosity of the fluid. After
substitution of the solution to equation (16) into the kinematic boundary condition (11a), one

18
obtains

2
 
(1) (0) (0)
 
iαθ αθ ω 2 ΓB − ΓB − ΓS P
A= (0)
ζ cosh k +
2
, (17a)
ΓB (θ − i) αθ ω 2 (ΓS + θΓB )
(1 + iθ)P
B= , (17b)
θ(ΓS + θΓB )
P
C=D=    , (17c)
ΓS + θ cosh m α sinh k + (1 + ασ) cosh k − θ sinh m cosh k + (α + σ) sinh k

3/2 (0)
with θ = Θk /ω, σ = ω 2 (θ−i)2 , and P = ζ[θ cosh k−(θ−i) cosh m]. The constants ΓS and ΓS are
(0)
related to the angular frequency, depth, wavenumbers and viscosity through ΓS = (i−θ)αθω 2 and
2
 
(0)
ΓS = (i − θ) 1 + αθ ω 2 , whereas the remaining constants are given by ΓB = α sinh k cosh m −
(1) (0) (1)
cosh k sinh m, ΓB = α sinh k sinh m − cosh k cosh m, and ΓB = ΓB − ΓB + σ(α cosh k cosh m −
sinh k sinh m).

The velocity potential and the stream function can be built by replacing the expressions in equa-
tion (17) into equation (15). The complete velocity field can be computed through equation (12).
Finally, the kinematic boundary condition (11a) implies that η 0 = C + iB, which can be written
as η 0 = Gx χ + Gz ζ, where Gx and Gz are the gain due to the horizontal and vertical component
of the pedal-wavemaker motion, respectively. Finally, one obtains

 
iΘ 1
Gx (ω, k) = sinh αk − sinh k , (18a)
Ξ(ω, k) α


1 + iΘ cosh αk − iΘ cosh k
Gz (ω, k) = , (18b)
Ξ(ω, k)

with the function Ξ(ω, k) defined as

2 k(α − 1)
       
1 tanhc αk sinhc αk 2 2 k(α + 1)
Ξ(ω, k) ≡ cosh αk cosh k 1 − 2 1− + +k sinhc + sinhc .
ω tanhc k sinhc k 2 2
(19)
A graphical inspection of the plots of the expressions above for a fixed wavelength shows that
Gz /Gx ≫ 1 for all ω, and thus, the vertical motion of the pedal wavemakers is the main contributor
of the total gain by far. Therefore, the theory of pedal wavemakers provides a good framework for
modelling the vertical input drive of the synchrotron servomotors. Accordingly, the total gain G
is well approximated by Gz , and thus by the entire expression in equation (2).

19
 CURVATURE EFFECTS ON GRAVITY WAVES

To get an insight into the effects of the wall curvature on the dynamics of gravity waves in our
synchrotron, we study the three-dimensional free-surface equations in cylindrical coordinates. Let
R1 and R2 be the radius of the inner and outer walls of the synchrotron, respectively. In Cartesian
coordinates, the dynamic and kinematic boundary condition now reads as follows,

1h i
∂t ϕ + (∂x ϕ)2 + (∂y ϕ)2 + (∂z ϕ)2 + gη = 0, (20a)
2 z=η

∂t η + ∂x ϕ∂x η + ∂y ϕ∂x η − ∂z ϕ|z=η = 0. (20b)

For simplicity, we omit the bottom motion in this calculation since we are interested in the curvature
effects on the surface waves via their dispersion relation. Thus, at the bottom of the channel, we
have
∂z ϕ|z=−h = 0, (21)

whereas at the walls of the synchrotron, we have

x∂x ϕ + y∂y ϕ||x2 +y2 =R2 = 0. (22a)

1

x∂x ϕ + y∂y ϕ||x2 +y2 =R2 = 0. (22b)

2

Equations (20) and the incompressibility condition (13a) written in the cylindrical coordinates
(r, φ, z) reads as follows,
"  2 #
1 2 1 2
∂t ϕ + (∂r ϕ) + ∂φ ϕ + (∂z ϕ) + gη = 0, (23a)
2 r
z=η
1
∂t η + ∂r ϕ∂r η + 2 ∂φ ϕ∂φ η − ∂z ϕ = 0, (23b)
r z=η
1 1
∂r (r∂r ϕ) + 2 ∂φφ ϕ + ∂zz ϕ = 0, (23c)
r r

whereas the boundary conditions at the curved walls, equations (22), gives ∂r ϕ = 0 at r = R1 and
r = R2 . Linearising equations (23a) and (23b), we obtain

∂t ϕ + gη|z=0 = 0, (24a)

∂t η − ∂z ϕ|z=0 = 0. (24b)

Differentiating equation (24a) with respect to time, we find upon substitution of equation (24b)

∂tt ϕ + g∂z ϕ|z=0 = 0, (25)

20
with solutions of the form

cosh [k (z + h)]
ϕ (r, φ, z, t) = exp [i (mφ − ωt)] R (r) , (26)
cosh kh

where ω 2 = gk tanh kh is the classical water-wave dispersion relation. On the other hand, the
function R(r) is given by the solutions of the Bessel differential equation,

d2 f df
ξ2 + ξ 2 − m2 f = 0,


2
+ξ (27)
dξ dξ

where ξ ≡ kr is a dimensionless variable such that R(r) = f (ξ). Thus, R (r) = αm Jm (kr) +
βm Ym (kr), where Jm (ξ) and Ym (ξ) are the Bessel functions of the first and second kind, respec-
tively. The boundary conditions at the curved walls require
 
dJm dYm
k αm + βm = 0, for ξ = kR1 and ξ = kR2 . (28)
dξ dξ

This condition is met if

ξ=kR2
dYm
= 0. (29)
dJm ξ=kR1

Defining the mean radius of the synchrotron and the water-channel half-width of the as R =
(R1 + R2 ) /2 and e = (R2 − R1 ) /2, respectively, and performing the Taylor expansion of equa-
tion (29) for λ ≫ e (i.e., ke ∼ 0), we obtain
 
d dYm
+ O k 2 e2 = 0.


ke (30)
dξ dJm ξ=kR

Developing the derivatives and using equation (27) leads to

m2 dJm −2
  
2
+ O k 2 e2 = 0.


1− 2 (31)
πξ ξ dξ
ξ=kR

Thus, we have kR = m + (1/3!)((ke)2 )/m + O k 4 e4 = 0, from which follows that

1 e2
 
kR
= 1+ + ... . (32)
m 6 R2

The later expression shows that in an annular waveguide, the classical water-wave dispersion re-
lation ω 2 = gk tanh kh remains valid, but for a modified wavenumber k, that at leading order is
the azimuthal wavenumber multiplied by the waveguide mean radius, mR. Notice that this value
corresponds to the perimetral wavenumber (the wavelength along the arclength is λp = 2πR/m).
The correction terms are powers of the channel width-to-diameter ratio, and the leading one is
of quadratic order. For our setup, the corrections are of the order of 10−4 or higher, and thus
curvature effects are negligible.

21
 NUMERICAL SPH FORMULATION

Our numerical setup for SPH simulations is schematised in Fig. 7(a). A fluid layer of density
ρ0 = 1 gr/cm3 and depth h = 1 cm is initially at rest in a fluid domain of length L = 20 cm, that
neglects curvature effects. In the SPH method, the density, pressure, and velocity field variables
and gradients are obtained numerically from a Lagrangian formulation of the Navier-Stokes equa-
tions using a combination of two approximation techniques: the kernel approximation and the
particle approximation. SPH provides a powerful method for the simulation of fluids under high
deformations, such as free surface waves, shock waves, high vorticity flows, and spraying, all very
common phenomenons in the study of coastal and ocean dynamics [30–33]. The SPH method can
be highly dissipative without modern amending techniques compared to other computational fluid
dynamics methods [34]. This combination of features poses SPH as an appropriate framework for
testing the robustness of the resonance phenomena in our synchrowave regarding viscosity. Here,
we briefly overview the SPH formulation used in this work, which was based on the Open Software
PySPH [35]. For further details, Liu and Liu give a comprehensive and detailed presentation of
the SPH method in Ref. [36], and Monaghan in Refs. [37, 38].

The SPH method for numerically solving the hydrodynamic equations is based on two main
approximations: the kernel approximation and the particle approximation. In the kernel approxi-
mation, the continuity and momentum equations are multiplied by a kernel function W = W (|r −
r′ |; hw ) and integrated along r′ in the fluid domain, leading to the so-called weighted integrals[36].
The kernel W is a smooth, normalised, and monotonically decaying function of the distance |r−r′ |,
typically with a bell-like shape. Formally, it is assumed that W (|r − r′ |; hw ) → δ(|r − r′ |) as
hw → 0, where δ is the Dirac delta function[38]. Moreover, kernels are usually chosen to vanish
for |r − r′ | > hw , i.e. they have a compact support domain of radius hw around r′ . This first
approximation step gives the field variables and gradients as smoothed, local space-averaged values
around each fluid particle placed at r′ , where only close-enough fluid particles contribute to the
space averaging. The radius hw of the compact support domain is known as the smoothing length.
We used a cubic-spline kernel with a smoothing length of hw = 0.13 cm in our SPH simulations.

The formal space discretisation of the fluid domain is performed using particle approximation,
and weighted integrals are estimated as discrete sums on a set of fluid particles. We used N = 2343
fluid particles and 852 solid particles simulating the soft elastomer bottom, all with diameter

22
 (a) t = 0T u (cm/s) (d) η (cm)
3 3
2
1
0 5.4

u (cm/s)
(b) t = 15 T 2
3 5.3

t (s)
2
1
0 5.2
1
u (cm/s)
(c) t = 15.125 T
3 5.1
2
1
0 5 0
0 5 10 15
x (cm)

FIG. 7. Numerical protocol for simulating the synchrotron water waveguide. (a) Nu-
merical setup used in SPH simulations. Waves are generated by the vertical motion of the bed
(black solid particles) in a two-dimensional fluid layer (turquoise fluid particles) of depth h = 1
and length L = 20. (b) The numerical simulation begins with a gradual increase in the amplitude
of the bottom motion. At some t > 0, one observes a regular travelling wave at the free surface.
(c) Surface detection algorithm performed on a fully developed gravity wave with λ = 10. (d )
Spatiotemporal diagram obtained from the concatenation of consecutively detected free surfaces.
The depicted case corresponds to ω = 1.01, i.e. close to resonance.

D = 0.1 cm. In this pure SPH formulation, the soft elastic bottom is represented by discrete
solid particles. We use three layers of solid SPH particles for better performance of the bottom
boundary and to avoid the percolation of fluid particles during the bottom motion. The boundary
condition at the moving bottom is formulated in terms of a generalised wall boundary condition
[39], which is based on a local force balance between the bottom solid particles and the fluid
particles to prevent wall penetration. We implemented the weakly-compressible SPH formulation
[27] for better computational performance in the simulation of the free-surface flow, keeping the
density fluctuations small (∼ 1%) [37]. The resulting discrete SPH formulation for the continuity

23
and momentum equations for the i-th particle is
dρi X
= mj uij · ∇i Wij (hw ), (33a)
dt
j̸=i
!
dui X Pi Pj
=− mj + 2 ∇i Wij (hw ) + Πi + gi , (33b)
dt ρ2i ρj
j̸=i

where Wij (hw ) = W (rij ; hw ), rij = ri − rj , uij = ui − uj , and ∇i denotes a gradient with respect
to the coordinates of the i-th particle. The artificial viscous acceleration Πi is a numerical term
originally introduced to control nonphysical instabilities appearing in problems with shock waves
[35–38]. We use the SPH formulation of the entropically damped artificial compressibility term
[35, 40],
X mj rij · ∇i Wij (hw )
Πi := 2αhw C uij , (34)
ρi + ρj (|rij |2 + ϵh2w )
j̸=i

where α = 0.2, C = 442.94 cm/s, and ϵ = 0.01 is a parameter introduced to avoid singularities for
vanishing |rij |. The pressure at the i-th particle is specified by its density via the usual equation
of state for weakly compressible fluids, namely
 γ 
P0 ρi
Pi = −1 , (35)
γ ρ0
where P0 and ρ0 are the reference pressure and density, respectively. For fluids, it is common to
use γ = 7. Finally, time integration of equations (33) is achieved using standard time-integration
methods for ordinary differential equations. This work used a second-order predictor-corrector
integrator with time-step ∆t = 1 × 10−5 s. We have chosen P0 in equation (35) large enough to
keep the density fluctuations around 1% [37].
Every numerical simulation starts with a gradual increase of the amplitude of the vertical
oscillations at the bottom from zero to a fixed maximum value, A = A/h = 0.042. As this happens,
the system responds with a sustained build-up of a smooth surface wave without neither defects or
fronts travelling throughout the numerical domain, as depicted in Fig. 7(b). Once gravity waves
form, the surface profiles are first extracted from a boundary-detection algorithm applied to the set
of SPH fluid particles [see Fig. 7(c)]. The boundary detection is performed at time steps given by
∆T = T /8, where T is the period of the bottom motion. Our measurements account for an integer
number of oscillations of the servomotors. A typical spatiotemporal diagram of the waves as they
travel streamwise is depicted in Fig. 7(d ). As the output of SPH numerical is Lagrangian, i.e.
a set of particles located throughout the domain, each carrying its hydrodynamic quantities, the
measurement of the wave response requires post-processing. After applying a boundary-detection

24
algorithm to the set of SPH fluid particles to detect the free surface at each time step, we generate a
spatiotemporal diagram collecting all the detected boundaries in time, as shown in Fig. 7(d ). Then,
we compute the two-dimensional Fast Fourier Transform in space and time of such spatiotemporal
diagram and find its maximum, which we denote as Fmax (ηenvl ). Following equation (1), the
bottom moves as ηb (x, t) = A cos(kx − ωt) ∝ Re [zb (x, t)], i.e., a wave force travelling towards the
streamwise direction. We compute similarly the maximum of the fast Fourier transform of this
pre-programmed spatiotemporal dynamics corresponding to the servomotors, Fmax (ηb ). Finally,
we obtain the gain and the phase of the system with G = |Fmax (ηenvl )| / |Fmax (ηb )| and ϕ =
− arg [Fmax (ηenvl )/Fmax (ηb )] + ∆ϕ0 , respectively, where ∆ϕ0 is the accumulated phase shift from
the beginning of the simulation to the first boundary detection.

ACKNOWLEDGMENTS

The authors thank Enrique Cerda, Claudio Falcón, Christophe Josserand and Thomas
Séon for fruitful discussions on the experimental setup and its reach. The authors are
also grateful to Bas Dijkhuis, Narek Halsdorfer, Jeremy Riffo, Lucas Martı́nez, Giuliana
Álvarez, for their technical help in building the first synchrowave. J.F.M. and A.E. acknowl-
edge the financial support of Agencia Nacional de Investigación y Desarrollo (ANID—Chile)
through the grant FONDECYT Postdoctorado No. 3200499. B.C. thanks the Australian
Research Council Linkage Project Number LP190101283. L.G. and I.V. thank the FONDE-
CYT/Iniciación grant No. 11170700 and No. 1221103.

DATA AVAILABILITY STATEMENT

The data that support the findings of this work are available from the corresponding
author upon reasonable request.

AUTHOR CONTRIBUTIONS STATEMENT

Conceptualisation, BC, LG and JFM; formal analysis, IV, JFM, and LG; experimental
investigation, IV and LG; methodology, BC, JFM and LG; software, JFM and AE; super-
vision, JFM and LG; validation, JFM and LG; writing–original draft, JFM; writing–review

25
and editing, IV and LG. All authors have read and agreed to the published version of the
manuscript.

COMPETING INTERESTS

The authors declare no competing interests.

∗ [email protected]
† [email protected]
[1] Th. Martin and R. Landauer, “Time delay of evanescent electromagnetic waves and the anal-
ogy to particle tunneling,” Phys. Rev. A 45, 2611–2617 (1992).
[2] Konstantinos Papatryfonos, Mélanie Ruelle, Corentin Bourdiol, André Nachbin, John W M
Bush, and Matthieu Labousse, “Hydrodynamic superradiance in wave-mediated cooperative
tunneling,” Communications Physics 5, 142 (2022).
[3] Pedro J Sáenz, Tudor Cristea-Platon, and John W M Bush, “Statistical projection effects in
a hydrodynamic pilot-wave system,” Nature Physics 14, 315–319 (2018).
[4] Stéphane Perrard, Matthieu Labousse, Marc Miskin, Emmanuel Fort, and Yves Couder,
“Self-organization into quantized eigenstates of a classical wave-driven particle,” Nature Com-
munications 5, 3219 (2014).
[5] Chia Wei Hsu, Bo Zhen, A Douglas Stone, John D Joannopoulos, and Marin Soljačić, “Bound
states in the continuum,” Nature Reviews Materials 1, 16048 (2016).
[6] John W.M. Bush, “Pilot-wave hydrodynamics,” Annual Review of Fluid Mechanics 47, 269–
292 (2015).
[7] Emmanuel Fort, Antonin Eddi, Arezki Boudaoud, Julien Moukhtar, and Yves Couder, “Path-
memory induced quantization of classical orbits,” Proceedings of the National Academy of
Sciences 107, 17515–17520 (2010).
[8] Y Couder, S Protière, E Fort, and A Boudaoud, “Walking and orbiting droplets,” Nature
437, 208 (2005).
[9] Yves Couder and Emmanuel Fort, “Single-particle diffraction and interference at a macro-
scopic scale,” Phys. Rev. Lett. 97, 154101 (2006).

26
[10] A. Eddi, E. Fort, F. Moisy, and Y. Couder, “Unpredictable tunneling of a classical wave-
particle association,” Phys. Rev. Lett. 102, 240401 (2009).
[11] A. Eddi, J. Moukhtar, S. Perrard, E. Fort, and Y. Couder, “Level splitting at macroscopic
scale,” Phys. Rev. Lett. 108, 264503 (2012).
[12] M. Labousse, A. U. Oza, S. Perrard, and J. W. M. Bush, “Pilot-wave dynamics in a harmonic
potential: Quantization and stability of circular orbits,” Phys. Rev. E 93, 033122 (2016).
[13] P. J. Cobelli, V. Pagneux, A. Maurel, and P. Petitjeans, “Experimental observation of trapped
modes in a water wave channel,” Europhysics Letters 88, 20006 (2009).
[14] E. Monsalve, A. Maurel, P. Petitjeans, and V. Pagneux, “Perfect absorption of water waves
by linear or nonlinear critical coupling,” Applied Physics Letters 114, 013901 (2019).
[15] Isis Vivanco, Bruce Cartwright, A. Ledesma Araujo, Leonardo Gordillo, and Juan F. Marin,
“Generation of gravity waves by pedal-wavemakers,” Fluids 6, 222 (2021).
[16] Timothée Jamin, Leonardo Gordillo, Gerardo Ruiz-Chavarrı́a, Michael Berhanu, and Eric
Falcon, “Experiments on generation of surface waves by an underwater moving bottom,”
Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 471,
20150069 (2015).
[17] P Maurer, S J Ghaemsaidi, S Joubaud, T Peacock, and P Odier, “An axisymmetric inertia-
gravity wave generator,” Experiments in Fluids 58, 143 (2017).
[18] Anupam Pandey, Zih-Yin Chen, Jisoo Yuk, Yuming Sun, Chris Roh, Daisuke Takagi, Sungyon
Lee, and Sunghwan Jung, “Optimal free-surface pumping by an undulating carpet,” Nature
Communications 14, 7735 (2023).
[19] Jens Als-Nielsen and Des McMorrow, Elements of modern X-ray physics (John Wiley & Sons,
2011).
[20] V. Baglin, P. Chiggiato, and R. Kersevan, “Vacuum challenges at the beam energy frontier,”
in The Future of the Large Hadron Collider (2024) Chap. 29, pp. 423–426,.
[21] Javier Sánchez-Reyes, “The hyperbolic sine cardinal and the catenary,” The College Mathe-
matics Journal 43, 285–290 (2012).
[22] K. Kajiura, “The leading wave of a tsunami,” B. Earthq. Res. I. Tokyo 41, 535–571 (1963).
[23] Joseph L. Hammack, “A note on tsunamis: their generation and propagation in an ocean of
uniform depth,” Journal of Fluid Mechanics 60, 769–799 (1973).
[24] Walter Munk, “Origin and generation of waves,” Coast. Eng. Proc. 1, 1 (1950).

27
[25] Leon B Lucy, “A numerical approach to the testing of the fission hypothesis,” The Astronom-
ical Journal 82, 1013–1024 (1977).
[26] R. A. Gingold and J. J. Monaghan, “Smoothed particle hydrodynamics: theory and applica-
tion to non-spherical stars,” Monthly Notices of the Royal Astronomical Society 181, 375–389
(1977).
[27] Jason P. Hughes and David I. Graham, “Comparison of incompressible and weakly-
compressible SPH models for free-surface water flows,” Journal of Hydraulic Research 48,
105–117 (2010).
[28] E. Shackleton, South: The story of the 1914-1917 expedition (William Heinemann Ltd., 1919).
[29] H. L. Lamb, Hydrodynamics (Cambridge University Press, 1932).
[30] Leonardo Di G. Sigalotti, Hender López, Arnaldo Donoso, Eloy Sira, and Jaime Klapp, “A
shock-capturing SPH scheme based on adaptive kernel estimation,” Journal of Computational
Physics 212, 124 – 149 (2006).
[31] Leonardo Di G. Sigalotti, Hender López, and Leonardo Trujillo, “An adaptive SPH method
for strong shocks,” Journal of Computational Physics 228, 5888 – 5907 (2009).
[32] P.N. Sun, A. Colagrossi, S. Marrone, M. Antuono, and A.M. Zhang, “Multi-resolution delta-
plus-sph with tensile instability control: Towards high reynolds number flows,” Computer
Physics Communications 224, 63 – 80 (2018).
[33] Balachander Gnanasekaran, Gui-Rong Liu, Yao Fu, Guangyu Wang, Weilong Niu, and Tao
Lin, “A smoothed particle hydrodynamics (sph) procedure for simulating cold spray process
- a study using particles,” Surface and Coatings Technology 377, 124812 (2019).
[34] L. McCue, L. Alford, W. Belknap, G. Bulian, L. Delorme, A. Francescutto, and A. Vakakis,
“An overview of the minisymposium on extreme ship dynamics,” Proceedings of the 2005
SIAM Conference on Applications of Dynamical Systems 43, 55–61 (2006).
[35] Prabhu Ramachandran, Aditya Bhosale, Kunal Puri, Pawan Negi, Abhinav Muta, A. Dinesh,
Dileep Menon, Rahul Govind, Suraj Sanka, Amal S. Sebastian, Ananyo Sen, Rohan Kaushik,
Anshuman Kumar, Vikas Kurapati, Mrinalgouda Patil, Deep Tavker, Pankaj Pandey, Chan-
drashekhar Kaushik, Arkopal Dutt, and Arpit Agarwal, “Pysph: A python-based framework
for smoothed particle hydrodynamics,” ACM Trans. Math. Softw. 47 (2021), 10.1145/3460773.
[36] M. B. Liu and G. R. Liu, “Smoothed Particle Hydrodynamics (SPH): an overview and recent
developments,” Archives of Computational Methods in Engineering 17, 25–76 (2010).

28
[37] J J Monaghan, “Smoothed particle hydrodynamics,” Reports on Progress in Physics 68, 1703–
1759 (2005).
[38] J.J. Monaghan, “Smoothed particle hydrodynamics and its diverse applications,” Annual Re-
view of Fluid Mechanics 44, 323–346 (2012).
[39] S. Adami, X.Y. Hu, and N.A. Adams, “A generalized wall boundary condition for smoothed
particle hydrodynamics,” Journal of Computational Physics 231, 7057–7075 (2012).
[40] Prabhu Ramachandran and Kunal Puri, “Entropically damped artificial compressibility for
SPH,” Computers & Fluids 179, 579–594 (2019).

29