Arbitrary mechanical memory encoding via nonlinear waves in bistable metamaterials

Audrey A. Watkins; Giovanni Bordiga; Katia Bertoldi; Mingxing Mu; Vincent Tournat

REVIEW 4 major objections 3 minor 33 references

Arbitrary mechanical memory encoding via nonlinear waves in bistable metamaterials

T0 review · 4 major / 3 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single shaped boundary pulse can set any state of a chain of bistable units that ignore each other when still.

desk verdict A genuinely new writing mechanism for bistable metamaterials, but the 'arbitrary encoding' claim runs ahead of the experiments. read the letter →

arxiv 2508.20321 v1 pith:L4UOIGD4 submitted 2025-08-27 physics.app-ph

Audrey A. Watkins , Giovanni Bordiga , Mingxing Mu , Vincent Tournat , Katia Bertoldi This is my paper · ORCID

classification physics.app-ph

keywords mechanicalmetamaterialsbistableelementsnonlinearwavesmemorymass-in-massunitcellswave-drivenactuationtransitiongraphstateencoding

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a chain of bistable mass-in-mass units, whose inner and outer masses show no static coupling, can nevertheless be reprogrammed from one end by sending a single nonlinear wave through it. The wave accelerates the outer masses, and the resulting inertial forces push selected inner masses over their energy barriers, so a shaped boundary pulse can set any combination of units into either of their two stable states. The authors support the claim with experiments on a three-cell chain and with numerical integration of a lumped model, and they show that redistributing mass between inner and outer components enlarges the regions where nearby inputs still write the same state. If correct, the idea turns a static, uncoupled bistable array into a remotely addressable mechanical memory with $2^N$ possible states, written by one input rather than by touching each unit.

What carries the argument

The central object is the mass-in-mass unit cell: an outer mass $m_{\rm out}$ connected by linear springs to its neighbors and by a bistable von Mises truss to an inner mass $m_{\rm in}$. The argument runs through the two-degree-of-freedom equations of motion in Eq. (1), which combine linear coupling, viscous damping fitted to experiment, Coulomb friction on the outer mass, and the truss potential $E(\delta_n)$. The key feature of that potential is its bistable energy landscape with minima at $\delta = 0$ and $\delta = 2l_0\tan\theta_0$; the truss switches when dynamic inertial forces exceed the barrier. The machinery works because static decoupling and dynamic coupling coexist: at rest the truss configuration does not affect neighboring cells, but during a wave the inner masses are accelerated independently, so each cell can be tipped selectively. The model is used to sweep the amplitude-frequency plane for every initial state, to define a sensitivity score $S$ from the largest contiguous pixel areas of each final state, and to guide the mass optimization.

What would settle it

Take a three-cell chain with the optimized masses (outer 34 g, inner 23 g), initialize it in state $(0,0,0)$, and apply a symmetric sine pulse located in one of the predicted large contiguous regions, repeating many times. If the final state varies across repeats, or if a pulse chosen from such a region lands in a state different from the model's prediction, the claimed robustness and accessibility of all transitions fails.

Watch

Extended reading notes

Core claim

The central discovery is that statically decoupled bistable elements become individually writable through inertial coupling during wave propagation. In each unit cell an outer mass and an inner mass are connected by a bistable von Mises truss; the truss switches only when the relative displacement $\delta_n$ is driven past its energy barrier. Because the inner mass is not coupled to neighboring cells, quasi-static loading leaves it untouched, but a dynamic boundary pulse accelerates the outer masses and transmits inertial forces to the inner masses, enabling selective switching. The final state is an extremely sensitive function of pulse amplitude and frequency: three nearly identical pulses landed the same chain in three different states. The authors find that reducing both masses to 34 g outer and 23 g inner enlarges the contiguous regions in the amplitude-frequency plane, and the resulting transition graph shows all eight states of a three-cell system reachable from every starting state. They further show experimentally that when the exact measured waveform is used as input, the model reproduces transitions that a best-fit sine pulse misses.

Load-bearing premise

The load-bearing premise is that the equations of motion fitted to one built sample remain accurate enough across untested pulse shapes and mass ratios, so the heat maps and transition graph are trustworthy guides to what a real chain will do.

Editorial extensions

If this is right

A memory array built from statically decoupled bistable units can be written remotely: one shaped pulse at the boundary sets the whole configuration, eliminating the need to actuate each element locally.
Mass distribution inside each unit cell is a design knob for robustness; moving to smaller masses increased the largest contiguous parameter regions from $S = 3899$ to $S = 5095$ pixels squared, making the written state less sensitive to input jitter.
For the optimized three-cell chain, all eight bit states are dynamically reachable from all eight starting states, so the array can be cycled through arbitrary memory contents without resetting.
The sensitivity heat maps provide a practical target-selection procedure: looking up a pulse in the amplitude-frequency plane tells the user which input should write a desired state.
Because the same wave mechanism is argued to extend to longer chains and to two-dimensional tessellations, the scheme is offered as a route to mechanical memories with more than three bits.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The directed transition graph behaves like a finite automaton: each pulse is a transition arrow between states, so composing pulses could implement Boolean logic or counters inside the same device, a consequence the paper does not develop.
Distorted input waveforms are treated as experimental error, but the fact that the full measured waveform changes the final state means waveform shape is an extra encoding channel; writing could intentionally use asymmetric or harmonically rich pulses to reach states that symmetric sine pulses miss.
The optimization scan varied only the two masses; varying truss stiffness, rest angle, coupling stiffness, and damping should produce larger or more robust regions, and failing to find such regions for a given bit would reveal intrinsic limits of the writing scheme.
Reliability in practice may require per-sample calibration: fabrication imperfections slightly alter truss parameters, so a pulse chosen from a nominal heat map might write a different state in one particular physical chain.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The thing to know about this paper is that the core architecture is new and worth taking seriously: a mass-in-mass bistable unit that is statically decoupled yet dynamically switchable, so a single boundary pulse can flip internal units remotely. That is a real step beyond local actuation or predefined couplings. The experiments on a 3-unit chain show repeatable writing of specific states, and the mass-tuning optimization reduces sensitivity in a sensible, well-explained way. The authors also deserve credit for reporting the discrepancies honestly—they show where the model misses, and they recover eight 'missing' transitions by feeding the actual measured waveform into the simulation. That is good scientific practice, not a cover-up.

The soft spot is the gap between the title and the evidence. The word 'arbitrary' and the conclusion's 'single boundary pulse is sufficient to set a desired state' go beyond what is demonstrated. Out of 56 directed state pairs, 32 predicted transitions were not verified experimentally; of those, 8 are recovered with the real input, leaving about 24 unverified edges. It is possible those are narrow regions of parameter space rather than fundamental absences, but the paper does not show that. The transition maps in Figs. 2d,e and 3b,c come from a model whose damping coefficients and friction are fitted to experimental data, so those maps are calibrated predictions, not independent validations. The model is standard and matches experiments for specific inputs, so the concept is plausible, but the 'all states accessible' claim rests on simulation extrapolation.

A minor but real issue: the data availability line in the preprint has the URL as 'github.com/bertoldi-collab/mass in mass' with a space, which will not resolve. That should be fixed before submission.

My take: this deserves a serious referee, but the authors should be pushed to either verify more of the missing transitions or soften the language. A claim of 'controlled writing of many states' or 'arbitrary in principle, with a subset verified' would be accurate and still interesting. As it stands, the overclaim is the main thing separating this from a clean acceptance.

Referee Report

4 major / 3 minor

Summary. The paper proposes a one-dimensional metamaterial made of three statically decoupled but dynamically coupled mass-in-mass bistable units, and claims that shaped boundary pulses can selectively switch internal units, enabling remote, programmable, and 'arbitrary' mechanical memory encoding. The authors develop a lumped-parameter model (Eq. (1)) with damping and friction coefficients fitted to experiments, use it to compute final-state maps over frequency-amplitude space (Figs. 2d,e), optimize the inner and outer masses to reduce input sensitivity (Fig. 3), and experimentally test a subset of the predicted transitions on a redesigned sample (Fig. 4). The manuscript reports that many predicted transitions were not experimentally verified, and that using the actual experimental waveform recovers only a minority of these missing transitions.

Significance. If the central claim is supported, the work would be a valuable advance: it would show that a single boundary pulse can write arbitrary bit patterns in a chain of bistable elements whose static states are otherwise decoupled, avoiding labor-intensive local actuation. The study combines clean experiments with a standard and physically reasonable model, and it is a strength that the authors openly report discrepancies and provide a reproducible data repository. However, the significance is currently tempered by the gap between the model-generated transition graph and the experimentally verified transitions: the central 'arbitrary' claim rests on a calibrated model whose predictions are only partially confirmed.

major comments (4)

[Experimental results, Fig. 4d] The claim that 'all eight states are dynamically accessible, regardless of the metamaterial's initial configuration' (Fig. 3c) is not fully supported by the experiments. The paper reports that 32 of the numerically predicted directed transitions (out of 56 possible nonzero transitions for eight states) were not verified experimentally, and only eight of these were recovered when using the actual experimental input waveform. This leaves at least 24 predicted transitions without experimental confirmation. Because the 'arbitrary encoding' claim depends on the completeness of this transition graph, the manuscript should either provide experimental verification for the remaining transitions or explicitly restrict the claim to the verified subset and justify the extrapolation with a quantified uncertainty analysis.
[Model and validation, Eq. (1), Fig. S3, Fig. S4] The damping coefficients cout and cin and the kinetic friction coefficient µk are fitted to experimental decay data (Fig. S3), and model validation is shown for a limited set of input waveforms (Fig. S4). The parameter-space maps in Figs. 2d,e and 3b,c and the mass optimization then extrapolate this calibrated model across the full frequency-amplitude range and to modified mass values. This is a fit-then-predict sequence rather than independent validation, so the predictive capability of the model in untested regions is an assumption. The authors should provide out-of-sample validation, for example by testing the model on mass configurations or waveforms not used in the calibration, or by clearly stating the limits of the calibrated model's predictive range.
[Conclusions, last paragraph] The statement that 'the results readily extend to larger arrays containing more memory bits' is asserted rather than demonstrated; all experiments and simulations are for N=3. The mechanism relies on wave propagation, multiple reflections, and nonlinear interactions, so finite-size and length-dependent effects could alter the accessibility of transitions. Please provide supporting evidence or limit the claim to the demonstrated system size.
[Results, Fig. 3c and Eq. (3)] The sensitivity metric S in Eq. (3) measures the summed contiguous pixel areas for transitions, but it is not by itself a measure of whether every transition can be robustly addressed. The transition graph in Fig. 3c contains edges with very light colors, corresponding to very small contiguous parameter regions, and the manuscript does not demonstrate that such high-sensitivity transitions can be reliably targeted in practice. The claim that the optimized metamaterial is 'significantly improved robustness' should be quantified against the experimental success rate for all edges, not only those with large contiguous areas.

minor comments (3)

[Eq. (3)] The summation limits in Eq. (3) are written as '2n−1', but they should presumably be '2^N−1' for the eight-state system; please correct the notation and define n versus N consistently.
[Data availability] The data repository link appears as 'github.com/bertoldi-collab/mass in mass'; please provide the exact, machine-readable URL without spaces.
[Fig. 2 and Fig. 3 captions] The figures would be easier to interpret if the color maps included a legend identifying each final state by its binary representation, since the colors alone are not immediately intuitive.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: fitted damping coefficients are used for independent final-state predictions, and the central mechanism is experimentally demonstrated.

full rationale

The paper's derivation chain is not circular. Eq. (1) is a Newtonian equation of motion with the bistable potential Eq. (2) derived from the von Mises truss geometry; the only fitted parameters are the viscous damping coefficients cout and cin and the Coulomb friction coefficient µk, explicitly fit to experimental decay data (Fig. S3). The paper then uses this calibrated model to predict final states across frequency–amplitude sweeps (Figs. 2d,e and 3b,c). Those final-state maps are not the same data to which the coefficients were fitted, and the manuscript independently checks the model against experimental transition outcomes (Figs. 4a,c,d). The fact that 32 numerically predicted transitions were not experimentally verified is an evidentiary and extrapolation concern, not a circular reduction: no predicted outcome is an identity or a renamed fit. Self-citations (e.g., refs [11,15,28]) are contextual background and are not load-bearing; no uniqueness theorem or ansatz is imported from the authors' prior work. The statement that all eight states are dynamically accessible is a direct simulation result of Eq. (1), not a consequence of fitting those transitions. Therefore the circularity score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on a fitted, deterministic model of the bistable units and on an extrapolation from 3 units to larger systems. No new physical entities are introduced.

free parameters (4)

cout = 0.22 kg/s
Viscous damping coefficient on the outer mass, fitted to experimental decay data (Fig. S3).
cin = 0.1 kg/s
Viscous damping coefficient on the inner mass, fitted to experimental decay data (Fig. S3).
mu_k = 0.0675
Kinetic friction coefficient between outer mass and platform; value presumably from independent friction measurement or fit, but not stated as measured.
beta = 100 m^-1 s
Smoothing constant in the tanh friction model; chosen for numerical convenience, not physics.

assumptions (4)

domain assumption The von Mises truss elastic energy E(δ) in Eq. (2) exactly represents the potential of each bistable element.
Standard truss kinematics; assumes ideal geometry and no plastic deformation or fatigue.
domain assumption The system dynamics are fully captured by lumped masses, linear springs, viscous damping, and Coulomb friction (Eq. 1).
Neglects higher-order modes, contact nonlinearities, and variability in fabrication beyond the stated tolerances.
domain assumption The final state is a deterministic function of the input pulse.
Assumes no stochasticity from thermal noise, friction fluctuations, or external vibrations during the transient.
ad hoc to paper The behavior observed for N=3 units extends to larger arrays and 2D tessellations.
Stated in the conclusion without demonstration; the sensitivity and control properties may scale differently with N.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Arbitrary mechanical memory encoding via nonlinear waves in bistable metamaterials." pith.science (2026). https://pith.science/paper/L4UOIGD4

@misc{pith2026250820321,
  author       = {Pith},
  title        = {Pith review of: Arbitrary mechanical memory encoding via nonlinear waves in bistable metamaterials},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L4UOIGD4}},
  note         = {Machine review of arXiv:2508.20321}
}

read the original abstract

Mechanical metamaterials composed of bistable elements have recently emerged as promising platforms for mechanical memory. Traditional approaches to writing information in these systems typically rely on localized actuation or predefined coupling schemes, which are often labor-intensive or lack adaptability. In this work, we introduce a one-dimensional metamaterial consisting of mass-in-mass bistable units that are statically decoupled yet dynamically switchable, allowing arbitrary mechanical information to be encoded through nonlinear waves applied at the boundary of the system. Through a combination of experiments and simulations, we demonstrate that tailored input signals can selectively trigger state transitions deep within the structure, enabling remote and programmable bit writing. This approach opens a new avenue for mechanical memory, harnessing the robustness of bistable elements and the tunability of nonlinear wave-driven actuation.

Figures

Figures reproduced from arXiv: 2508.20321 by the authors.

**Figure 1.** Bistable mass-in-mass metamaterial. a) Photo and b) schematic of the 3-units bistable mass-in-mass metamaterial. c) Schematic of the unit cell. d) Experimentally characterized bistable elastic energy for the three von Mises trusses. e) Displacement input A (a bipolar pulse of amplitude A =8.8 mm and nominal frequency f=9 Hz) delivered by the shaker. f) Time and spatio-temporal response triggered by input A. g) Final… view at source ↗

**Figure 1.** First, as illustrated by un(t) in Fig. 1f, a pulse with A = 8.8 mm and f = 9 Hz initiates a wave that propagates through the structure, reflects multiple times, and eventually dissipates. Second, when the accelerations of the outer masses are sufficiently large, the resulting inertial forces on the inner masses can overcome their energy barriers, causing them to switch states. This is evident from the evolution of… view at source ↗

**Figure 2.** Effect of small perturbations in the input signal. a) 3 slightly different input signals, and b) spatio-temporal plots of the corresponding responses, with c) photos of the 3 different final states reached. d)-e) Numerically predicted final states as a function of input frequency and amplitude for a sinusoidal input reached from d) α0 = (0, 0, 0) and e) the seven other states. cell as moutu¨n + k(2un − un+1 − un−1)+… view at source ↗

Figures from the paper (2 more)

**Figure 3.** Figure 3: Mass-dependent response space. a) Sensitivity function to input perturbations as defined by Eq. (3) as a function of mout and min, with both configurations of masses used in this article. b) Color-coded final states reached from all possible initial configurations in r…

**Figure 4.** Figure 4: b(i) is best approximated by a symmetric sinusoidal pulse with (A, f) = (10.9 mm, 8 Hz). When excited with such a sinusoidal input, the model predicts a final state of (0,1,1), in agreement with the experiment (Fig. 4c(i)). In contrast, the input in Fig. 4b(ii) is best…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 28 canonical work pages

[1]

Bertoldi, V

K. Bertoldi, V. Vitelli, J. Christensen, and M. Van Hecke, Nature Reviews Materials 2, 1 (2017)

work page 2017
[2]

P. Jiao, J. Mueller, J. R. Raney, X. Zheng, and A. H. Alavi, Nature communications 14, 6004 (2023)

work page 2023
[3]

Barchiesi, M

E. Barchiesi, M. Spagnuolo, and L. Placidi, Mathematics and Mechanics of Solids 24, 212 (2019)

work page 2019
[4]

S. Shan, S. H. Kang, J. R. Raney, P. Wang, L. Fang, F. Candido, J. A. Lewis, and K. Bertoldi, Adv. Mater 27, 4296 (2015)

work page 2015
[5]

P. A. Deymier, Acoustic metamaterials and phononic crystals, Vol. 173 (Springer science & Business media, 2013)

work page 2013
[6]

Dorn and D

C. Dorn and D. M. Kochmann, Extreme Mechanics Let- ters 65, 102091 (2023)

work page 2023
[7]

Celli and S

P. Celli and S. Gonella, Applied Physics Letters 106, 091905 (2015)

work page 2015
[8]

K. K. Dudek, M. Kadic, C. Coulais, and K. Bertoldi, Shape morphing metamaterials (2025), arXiv:2501.14804 [physics.app-ph]

arXiv 2025

Show all 33 references

[9]

L. Jin, A. E. Forte, B. Deng, A. Rafsanjani, and K. Bertoldi, Advanced Materials 32, 2001863 (2020)

2020
[10]

X. Ni, H. Luan, J.-T. Kim, S. I. Rogge, Y. Bai, J. W. Kwak, S. Liu, D. S. Yang, S. Li, S. Li, Z. Li, Y. Zhang, C. Wu, X. Ni, Y. Huang, H. Wang, and J. A. Rogers, Nature Communications 13, 5576 (2022)

2022
[11]

J. R. Raney, N. Nadkarni, C. Daraio, D. M. Kochmann, J. A. Lewis, and K. Bertoldi, Proceedings of the National Academy of Sciences 113, 9722 (2016)

2016
[12]

L. Jin, R. Khajehtourian, J. Mueller, A. Rafsanjani, V. Tournat, K. Bertoldi, and D. M. Kochmann, Pro- ceedings of the National Academy of Sciences 117, 2319 (2020)

2020
[13]

J. L. Silverberg, A. A. Evans, L. McLeod, R. C. Hayward, T. Hull, C. D. Santangelo, and I. Cohen, science 345, 647 (2014)

2014
[14]

O. R. Bilal, A. Foehr, and C. Daraio, Proceedings of the National Academy of Sciences 114, 4603 (2017)

2017
[15]

T. Chen, M. Pauly, and P. M. Reis, Nature 589, 386 (2021)

2021
[16]

L. S. Novelino, Q. Ze, S. Wu, G. H. Paulino, and R. Zhao, Proceedings of the National Academy of Sciences 117, 24096 (2020)

2020
[17]

Haghpanah, L

B. Haghpanah, L. Salari-Sharif, P. Pourrajab, J. Hopkins, and L. Valdevit, Adv. Mater 28, 7915 (2016)

2016
[18]

J. E. Pechac and M. J. Frazier, Applied Physics Letters 122 (2023)

2023
[19]

Treml, A

B. Treml, A. Gillman, P. Buskohl, and R. Vaia, Pro- ceedings of the National Academy of Sciences 115, 6916 (2018)

2018
[20]

Z. Liu, H. Fang, J. Xu, and K.-W. Wang, Advanced Sci- ence 10, 2305146 (2023)

2023
[21]

Masana, S

R. Masana, S. Khazaaleh, H. Alhussein, R. Crespo, and M. Daqaq, Applied Physics Letters 117 (2020)

2020
[22]

Jules, A

T. Jules, A. Reid, K. E. Daniels, M. Mungan, and F. Lechenault, Physical Review Research 4, 013128 (2022)

2022
[23]

Yasuda, T

H. Yasuda, T. Tachi, M. Lee, and J. Yang, Nature com- munications 8, 962 (2017)

2017
[24]

T. Mei, Z. Meng, K. Zhao, and C. Q. Chen, Nature com- munications 12, 7234 (2021)

2021
[25]

Jiang, L

Y. Jiang, L. M. Korpas, and J. R. Raney, Nature commu- nications 10, 128 (2019)

2019
[26]

L. J. Kwakernaak and M. van Hecke, Physical Review Letters 130, 268204 (2023)

2023
[27]

Yasuda, P

H. Yasuda, P. R. Buskohl, A. Gillman, T. D. Murphey, S. Stepney, R. A. Vaia, and J. R. Raney, Nature 598, 39 (2021)

2021
[28]

Gutierrez-Prieto, C

E. Gutierrez-Prieto, C. M. Meulblok, M. van Hecke, and P. M. Reis, Dynamic driving enables independent control of material bits for targeted memory (2025), arXiv:2508.16257 [cond-mat.soft]

2025 arXiv
[29]

R. C. Merkle, Nanotechnology 4, 114 (1993)

1993
[30]

R. C. Merkle, R. A. Freitas, T. Hogg, T. E. Moore, M. S. Moses, and J. Ryley, Journal of Mechanisms and Robotics 10, 061006 (2018)

2018
[31]

Mei and C

T. Mei and C. Q. Chen, Nature Communications 14, 5204 (2023)

2023
[32]

J. Liu, M. Teunisse, G. Korovin, I. R. Vermaire, L. Jin, H. Bense, and M. van Hecke, Proceedings of the National Academy of Sciences of the United States of America121, e2308414121 (2024)

2024
[33]

Nadkarni, A

N. Nadkarni, A. F. Arrieta, C. Chong, D. M. Kochmann, and C. Daraio, Phys. Rev. Lett. 116, 244501 (2016)

2016

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Bertoldi, V

K. Bertoldi, V. Vitelli, J. Christensen, and M. Van Hecke, Nature Reviews Materials 2, 1 (2017)

work page 2017

[2] [2]

P. Jiao, J. Mueller, J. R. Raney, X. Zheng, and A. H. Alavi, Nature communications 14, 6004 (2023)

work page 2023

[3] [3]

Barchiesi, M

E. Barchiesi, M. Spagnuolo, and L. Placidi, Mathematics and Mechanics of Solids 24, 212 (2019)

work page 2019

[4] [4]

S. Shan, S. H. Kang, J. R. Raney, P. Wang, L. Fang, F. Candido, J. A. Lewis, and K. Bertoldi, Adv. Mater 27, 4296 (2015)

work page 2015

[5] [5]

P. A. Deymier, Acoustic metamaterials and phononic crystals, Vol. 173 (Springer science & Business media, 2013)

work page 2013

[6] [6]

Dorn and D

C. Dorn and D. M. Kochmann, Extreme Mechanics Let- ters 65, 102091 (2023)

work page 2023

[7] [7]

Celli and S

P. Celli and S. Gonella, Applied Physics Letters 106, 091905 (2015)

work page 2015

[8] [8]

K. K. Dudek, M. Kadic, C. Coulais, and K. Bertoldi, Shape morphing metamaterials (2025), arXiv:2501.14804 [physics.app-ph]

arXiv 2025