Active Elastic Solid Model (AES) Dynamics

• AES is a minimal active-matter framework that couples elastic forces with active agent dynamics through a nonlinear feedback loop.
• It employs linear spring networks and dynamical matrix analysis to capture selective mode actuation and collective ordering in various configurations.
• AES formulations range from discrete models to continuum and variational approaches, offering practical insights into active solid behavior and morphogenesis.

Searching arXiv for recent and foundational papers on the Active Elastic Solid model. I’m checking arXiv for the AES literature and related active-solid papers. The Active Elastic Solid model (AES) denotes a class of active-matter models in which force-generating units are elastically coupled, so that active forcing deforms a solid network and the resulting deformation feeds back on the active degrees of freedom. In its original formulation, polar active agents act on the nodes of a two-dimensional elastic lattice, and the ensuing displacement field nonlinearly reorients the active agents; later formulations recast the same logic in rigid-limit zero-mode manifolds, continuum active sheets, variational statics, and pressure-stabilized shells (Baconnier et al., 2021, Hernández-López et al., 2023). Across these variants, AES serves as a minimal framework for selective collective actuation, rigid body motion, shape change, and polar-ordering dynamics in active solids.

1. Microscopic definition and elasto-active feedback

In the minimal AES construction, each lattice node $i$ carries a displacement ${\bf u}_i$ from a reference configuration and a polarity unit vector $\hat{\bf n}_i$. Nodes are elastically coupled by linear springs with stiffnesses $k_{ij}$, while each node exerts a fixed active force $F_0\hat{\bf n}_i$ on the lattice. In the overdamped, no-noise limit, the dimensional dynamics are

$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$

After nondimensionalizing lengths by the elastic deflection $l_e=F_0/k$, times by $\zeta/k$, and introducing the elasto-active feedback parameter $\pi=l_e/l_a$, with ${\bf u}_i$0 the self-alignment length, the equations become

${\bf u}_i$1

where ${\bf u}_i$2 is the dynamical matrix of the reference lattice and its eigenvalues ${\bf u}_i$3 are the squared normal-mode frequencies (Baconnier et al., 2021).

The defining feature of AES is the nonlinear feedback loop between activity and elasticity. The polarity field drives the solid through the active push ${\bf u}_i$4, while the instantaneous velocity or displacement rotates the polarity through the transverse projector ${\bf u}_i$5. This closes a mechanosensitive loop without requiring explicit Vicsek-type alignment, external control circuitry, or inertial dynamics. In this sense, AES is a mechanically mediated active-ordering model rather than a purely kinematic flocking model.

2. Instability, global bifurcation, and mode selectivity

A central result of the original AES analysis is that the onset of collective actuation is controlled by the elastic spectrum but is not reducible to simple excitation of the softest mode. Linearization about a static polarity configuration shows that the first fixed point to lose stability does so at

${\bf u}_i$6

where ${\bf u}_i$7 is the smallest eigenvalue of ${\bf u}_i$8. More generally, no fixed point remains stable for

${\bf u}_i$9

with $\hat{\bf n}_i$0 measuring the local orthogonality of modes $\hat{\bf n}_i$1 and $\hat{\bf n}_i$2 (Baconnier et al., 2021).

The transition is explicitly described as a global bifurcation rather than a local Hopf bifurcation. In the one-particle toy model, below $\hat{\bf n}_i$3 there is an entire ring of neutrally stable fixed points of radius $\hat{\bf n}_i$4; above threshold, these fixed points destabilize simultaneously and a stable limit cycle of radius $\hat{\bf n}_i$5 appears, with oscillation frequency

$\hat{\bf n}_i$6

This result is important because it rules out the common identification of AES oscillations with a standard single-fixed-point Hopf scenario.

Mode selection in the full lattice dynamics is highly specific. When the dynamics are projected onto the normal-mode basis, numerical and experimental results show that above threshold almost all power condenses into exactly two degenerate modes $\hat{\bf n}_i$7. The selected pair maximizes the instability bound

$\hat{\bf n}_i$8

so spatial extension and local orthogonality matter as much as eigenfrequency. Consequently, the actuated modes are not necessarily the lowest-energy ones: in the triangular cluster with $\hat{\bf n}_i$9, the selected pair is the two lowest-energy modes, whereas in the kagome cluster with $k_{ij}$0, the selected pair is modes $k_{ij}$1 and $k_{ij}$2. In the same study, numerics reproduced $k_{ij}$3 and $k_{ij}$4, and experiments with “hexbug” microrobots showed a frozen-disordered phase, a heterogeneous coexistence regime, and then full collective actuation.

3. Rigid-limit reduction, zero modes, and effective Landau theory

A later AES formulation addresses active solids that host zero-energy deformation modes and are subject to noise. In that setting, the system is composed of $k_{ij}$5 active units at the nodes of an elastic network, each with a two-dimensional position $k_{ij}$6, a polarization angle $k_{ij}$7, and unit vector $k_{ij}$8. When active forces are sufficiently weak that no spring is appreciably stretched, one imposes bond-length constraints exactly and projects the motion onto the manifold of zero modes $k_{ij}$9. The rigid-limit condition is

$F_0\hat{\bf n}_i$0

so no harmonic $F_0\hat{\bf n}_i$1 modes are excited (Hernández-López et al., 2023).

Within this limit, the elastic force on node $F_0\hat{\bf n}_i$2 is expressed purely through zero-mode projections,

$F_0\hat{\bf n}_i$3

and the orientation dynamics can be written in terms of an effective angle potential

$F_0\hat{\bf n}_i$4

For $F_0\hat{\bf n}_i$5, the angular dynamics are fast relative to the slow motion along the zero-mode manifold, which motivates an adiabatic approximation. The fast variables equilibrate under

$F_0\hat{\bf n}_i$6

and one introduces mode order parameters

$F_0\hat{\bf n}_i$7

In the large-$F_0\hat{\bf n}_i$8 limit, the joint distribution of $F_0\hat{\bf n}_i$9 acquires a Landau large-deviation form,

$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$0

with

$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$1

$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$2

The minima of this effective free energy select which zero mode is spontaneously actuated. The theory predicts single-mode actuation for translation, pure rotation, or pure auxetic contraction, all with a transition at $$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$3, and it also resolves two-mode competition. In a free lattice with translation plus rotation, the global minimum is always translation, while rotation is metastable because of a finite-$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$4 barrier. In a rotation-plus-auxetic network, the free-energy landscape has up to four local minima and yields two transitions as $$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$5 is lowered: first auxetic actuation at $$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$6, then rotation at much smaller $$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$7. This establishes a statistical-mechanical route from microscopic force alignment to spontaneous rigid-body motion and shape-changing mechanisms.

4. Continuum AES and turbulent polar-ordering dynamics

The AES framework also admits a continuum description. Lie, Simonsen and Dommersnes formulated a minimal model of a self-propelled elastic sheet in which a two-dimensional network of beads connected by linear springs is represented by an elastic displacement field $$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$8, strain tensor

$$\zeta\,\dot{\bf u}_i = F_0\,\hat{\bf n}_i - \sum_j k_{ij}({\bf u}_i-{\bf u}_j),$$9

and elastic energy density

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$0

For an isotropic linear solid in two dimensions,

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$1

The active and frictional terms enter through a substrate-drag stress $$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$2, with $$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$3, and an active internal stress

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$4

where $$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$5 is a unit-vector polarization field. The overdamped force balance is

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$6

while the discrete polarity dynamics align each $$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$7 with the net elastic force $$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$8 through

$$\dot{\hat{\bf n}}_i = \frac{1}{\tau}\,P_\perp(\hat{\bf n}_i)[{\bf u}_i] = \frac{1}{\tau}\,(\hat{\bf n}_i\times{\bf u}_i)\times\hat{\bf n}_i.$$9

In rescaled units with length $l_e=F_0/k$0 and time $l_e=F_0/k$1, one sets $l_e=F_0/k$2, leaving $l_e=F_0/k$3 and $l_e=F_0/k$4 as the two control parameters (Lie et al., 2 Oct 2025).

The discrete implementation used a circular patch with $l_e=F_0/k$5 or $l_e=F_0/k$6 beads, heterogeneous bead diameters drawn uniformly from $l_e=F_0/k$7, Voronoi-defined neighbors, free boundaries, $l_e=F_0/k$8, $l_e=F_0/k$9, $\zeta/k$0, and turning rate $\zeta/k$1–$\zeta/k$2. During polar-ordering kinetics, the velocity field develops multiscale fluctuations with classic signatures of turbulence, even though the sheet remains only weakly deformed elastically. In a fully developed ordering state with $\zeta/k$3, the energy spectrum satisfies

$\zeta/k$4

over roughly one decade in wavenumber. Longitudinal velocity increments show near-Gaussian wings at very small separations but strongly non-Gaussian, fat-tailed PDFs at intermediate scales, with excess kurtosis typically in the range $\zeta/k$5–$\zeta/k$6, corresponding to kurtosis $\zeta/k$7–$\zeta/k$8.

A key conceptual point is that AES turbulence is not organized by an inertial cascade. Spectral decompositions of activity injection $\zeta/k$9 and frictional dissipation $\pi=l_e/l_a$0 show that, once mesoscale order has formed, the two spectra coincide for all $\pi=l_e/l_a$1, so energy is dissipated essentially at the same scale it is injected. The dominant carriers of order are not vortices but domain walls or fronts across which the velocity flips abruptly. These walls propagate at

$\pi=l_e/l_a$2

often exceeding the single-particle propulsion speed $\pi=l_e/l_a$3, and the ordering time scales as

$\pi=l_e/l_a$4

This linear dependence on system size contrasts with diffusive $\pi=l_e/l_a$5 coarsening for Vicsek-type polarity coupling.

5. Variational formulations, Ritz methods, and active-solid statics

A complementary AES literature treats static active solids through energy methods rather than time-dependent bead dynamics. In that formulation, an active elastic solid is described by a displacement field $\pi=l_e/l_a$6 on a domain $\pi=l_e/l_a$7, with passive elastic energy density $\pi=l_e/l_a$8, strain tensor $\pi=l_e/l_a$9, and a prescribed internal active stress ${\bf u}_i$00. The restricted free-energy functional is

${\bf u}_i$01

Stationarity gives the equilibrium equations

${\bf u}_i$02

with ${\bf u}_i$03 (Wang et al., 2022).

This variational statement supports both classical Ritz approximations and deep Ritz methods. In the classical Ritz method, one chooses a finite-parameter trial displacement ${\bf u}_i$04, inserts it into ${\bf u}_i$05, and solves ${\bf u}_i$06. For a thin circular plate of radius ${\bf u}_i$07, thickness ${\bf u}_i$08, and asymmetric active stress ${\bf u}_i$09 acting only on one side ${\bf u}_i$10, a small-deflection ansatz

${\bf u}_i$11

gives the exact solution

${\bf u}_i$12

with ${\bf u}_i$13 and ${\bf u}_i$14. The corresponding closed-form plate profile is

${\bf u}_i$15

so the plate bends toward its contracting side.

In the deep Ritz method, a feed-forward neural network ${\bf u}_i$16 replaces the polynomial ansatz, and the loss is the discretized free energy with a boundary-condition penalty. Random sample points implement Monte-Carlo integration, and stochastic gradient descent such as Adam is used to minimize the loss. The same framework introduces the activogravity length

${\bf u}_i$17

through the dimensionless ratio

${\bf u}_i$18

When the lateral plate dimension is larger than the activogravity length, about ${\bf u}_i$19 micron, gravitational forces become important relative to internal active contraction. This variational AES program is therefore aimed at morphogenetic statics as much as at numerical solution methodology.

6. Shell AES: hydrostatic pressure, local excitation, and shape transitions

AES has also been extended from planar lattices to thin elastic shells with internal pressure. In the network model studied by Maji and Rabin, a shell is discretized as a closed triangular network of ${\bf u}_i$20 point-mass nodes connected by nonlinear springs. In the C${\bf u}_i$21-based realization, the triangulation has ${\bf u}_i$22 vertices and ${\bf u}_i$23 springs. For a spring ${\bf u}_i$24 with instantaneous length ${\bf u}_i$25 and rest length ${\bf u}_i$26, the elongation-dependent spring constant

${\bf u}_i$27

leads to the nonlinear force

${\bf u}_i$28

Uniform hydrostatic pressure ${\bf u}_i$29 acts on each node through the areas and normals of adjacent triangles, and the overdamped equation of motion reads

${\bf u}_i$30

with ${\bf u}_i$31 in the simulations. Activity is implemented by randomly exciting individual springs according to

${\bf u}_i$32

using ${\bf u}_i$33, ${\bf u}_i$34, and excitation times ${\bf u}_i$35; stochastic inter-event intervals are drawn uniformly from ${\bf u}_i$36. The model shows that pressure-induced stretching couples local and global behavior: a single excited spring contracts rapidly, the global shell area dips more slowly, periodic excitation produces undamped oscillations in ${\bf u}_i$37, and random excitation yields an exponential decay with ${\bf u}_i$38. Negative area-pressure feedback narrows the area distribution and reduces ${\bf u}_i$39 to about ${\bf u}_i$40, while positive feedback broadens the distribution and yields ${\bf u}_i$41 (Maji et al., 2022).

A second shell formulation uses a globe-like spherical network with ${\bf u}_i$42, ${\bf u}_i$43, ${\bf u}_i$44 vertices, ${\bf u}_i$45 springs, ${\bf u}_i$46 quadrilaterals, and ${\bf u}_i$47 polar triangles. Here each spring has quartic elastic energy

${\bf u}_i$48

and the active rule is curvature sensitive. For a spring ${\bf u}_i$49, the discrete curvature

${\bf u}_i$50

is compared to a threshold ${\bf u}_i$51. In the two-state rule, the spring stiffness switches between ${\bf u}_i$52 and ${\bf u}_i$53; in the three-state rule, both softening and stiffening are allowed. Pressure acts as a quasi-static control parameter, and a critical pressure

${\bf u}_i$54

separates distinct stationary shapes. Below ${\bf u}_i$55, only longitudinal springs soften and the shell becomes a prolate ellipsoid; above ${\bf u}_i$56, many latitudinal springs also soften and the shell becomes an oblate spheroid. In the three-state model, shrunken oblate spheroids, extended prolate ellipsoids, and inflated oblate spheroids all appear in different pressure regimes. The transitions are described as discontinuous, or first-order-like, because the stiffness jumps discretely when subsets of springs cross ${\bf u}_i$57 (Maji et al., 2023).

7. Scope, misconceptions, and relation to broader active-elastic modeling

AES is not a single equation set but a modeling family unified by one mechanism: elastic stresses generated by active units feed back on the active degrees of freedom, which in turn alter stresses, mode amplitudes, or material parameters. Within that family, the planar lattice model emphasizes selective and collective actuation; the rigid-limit theory emphasizes zero-energy modes, metastability, and mode competition; the continuum sheet model emphasizes polar ordering and turbulent signatures; and the shell models emphasize pressure regulation, curvature sensing, and autonomous shape transitions.

Several recurrent simplifications are explicitly contradicted by the literature. AES does not merely excite the softest mode: the selected oscillatory pair can be modes ${\bf u}_i$58 and ${\bf u}_i$59 rather than the lowest-energy modes. Its oscillatory onset is not a standard local Hopf bifurcation but a global collapse of a manifold of marginal equilibria. Its turbulent regime does not imply an energy cascade: the injection and dissipation spectra can coincide at all ${\bf u}_i$60. Its ordering mechanism is not necessarily vortex-mediated: domain walls can be the primary carriers of order.

A related but distinct line of work studies an active Ornstein-Uhlenbeck particle bound to a generalized elastic system that includes flexible and semiflexible polymers, fluid membranes, and fluctuating interfaces. That framework yields a fractional Langevin equation with a Riemann–Liouville derivative, a force propagator ${\bf u}_i$61, and three dynamical regimes for the active particle: a pseudo-ballistic initial phase, a drastic decrease of the mobility, and an asymptotic subdiffusive regime. This is not the AES model in the narrow bead-spring or force-alignment sense, but it places AES in a broader active-elastic landscape in which memory kernels, hydrodynamic interactions, and viscoelastic propagation can also control active motion (Taloni, 2024).

A plausible implication of the combined AES literature is that geometry, elastic spectra, and feedback architecture are the decisive organizing variables. In planar networks they determine which collective modes self-oscillate; in rigid active solids they determine which zero mode is actuated; in active sheets they determine front speeds and ordering times; and in shells they determine whether pressure stabilizes, amplifies, or switches shape. Under that interpretation, AES functions as a unifying framework for active biological solids, cell collectives, colloidal clusters, and active metamaterials, while retaining sufficient minimality to permit exact reductions, explicit thresholds, and direct numerical implementation.