The Kovacs memory effect in a thin granular layer: experimental evidence and its physical origin

arXiv:2602.20716 · cond-mat.soft · Submitted 2026-02-24 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "The Kovacs memory effect in a thin granular layer".

Mira: Here's the gist of the paper. The paper reports the experimental observation of memory effects in a vertically vibrated thin granular layer, revealing an anomalous Kovacs memory effect confined to the initial fast relaxation stage Experimental Observation and Findings The study experimentally observed that following a quench in input acceleration, the granular temperature exhibits an anomalous Kovacs memory effect confined to the initial fast relaxation stage <ref:2602.20716#pg4> The memory vanishes shortly thereafter, yielding a time-dependent memoryless regime governed solely by the instantaneous temperature before the system reaches its final steady state <ref:2602.20716#pg2> The analysis reveals that memory emerges during fast transients through coupling between horizontal and vertical temperatures—a mechanism that fundamentally constrains the accessible memory phenomenology and precludes observation of the standard Kovacs effect in this system <ref:2602.20716#pg2> Theoretical Framework and Kinetic States The kinetic theory framework developed quantitatively captures these features by identifying the initial memory and subsequent memoryless regimes with the kinetic and hydrodynamic states, respectively (that are well established in kinetic theory) <ref:2602.20716#pg2> Memory is reflected in a nonmonotonic behavior of the temperature, implying that, at those times, the temperature does not fully specify the state of the system <ref:2602.20716#pg2> The dynamics of the velocity distribution function consists of a rapid and initial-condition-dependent evolution (the kinetic regime) over a characteristic time tm, followed by a hydrodynamic regime in which all the time dependence enters solely through the hydrodynamic fields, the granular temperature in our case <ref:2602.20716#pg3> Constraints on Memory Phenomenology The authors demonstrate that memory characteristics are constrained by the system’s governing equations and cannot be tuned through external conditions alone <ref:2602.20716#pg2> The results show that memory is confined to kinetic stages, a fundamentally different regime from persistent path-dependent memory in externally-driven systems (such as magnetic hysteresis) <ref:2602.20716#pg7> This confinement of memory to the kinetic stage has a direct interpretation in terms of the Zwanzig-Mori formalism, in which eliminating degrees of freedom generates memory kernels that become effectively local once the retained and projected-out variables decouple <ref:2602.20716#pg7> Universal Hydrodynamic State Subsequent relaxation proceeds through a universal memoryless hydrodynamic state governed solely by the horizontal temperature <ref:2602.20716#pg2> The system state thus becomes effectively hydrodynamic, where the temperature evolves autonomously and all other moments are slaved to it, so that the macroscopic dynamics becomes independent of the prior kinetic history (memory vanishes) <ref:2602.20716#pg3> This is demonstrated by showing that after a convenient translation in the time axes, tΓ, the horizontal temperature curves collapse onto a single curve <ref:2602.20716#pg5> Model and Quantitative Analysis A minimal kinetic theory model and molecular dynamics simulations establish that the dynamics is governed by two coupled temperatures—horizontal and vertical—operating on slow and fast timescales, respectively <ref:2602.20716#pg6> The structure of the equations is clear: energy is injected in the vertical direction (last term of Eq.

Kai: (4)), while it is transferred to the other degrees of freedom via inelastic collisions <ref:2602.20716#pg6> Linearizing Eqs.

Mira: (3) and (4) around the stationary state allows for an explicit calculation of the function H(s) = T(s)/Ts − 1, which shows that H(s) < 0 for the cooling Kovacs protocol, leading to a downward hump <ref:2602.20716#pg6> The amplitude of the hump, Hmin, is proportional to dHdss=0 and decreases with the input acceleration <ref:2602.20716#pg6> The controllability of memory in driven systems is itself regime-dependent—determined by whether the projected-out degrees of freedom remain externally accessible <ref:2602.20716#pg7> In the latter case, external engineering of relaxation, as achieved in driven quantum systems

36: , requires such accessibility, whereas in driven granular fluids the memory phenomenology is intrinsically fixed by the kinetic equations <ref:2602.20716#pg7> The kurtosis criterion provides a generic, experimentally accessible signature for the onset of memoryless hydrodynamic behavior <ref:2602.20716#pg7> This is readily measurable in other driven systems such as active matter or dense colloidal suspensions <ref:2602.20716#pg7> Experimental Setup Details The experimental configuration consists of a thin, circular layer of N identical stainless-steel spheres (AISI 52100) with coefficient of normal restitution α = 0.95 The layer is confined between a bottom aluminum plate and a top plexiglas lid, with steel spacers forming the lateral boundary All experiments were performed at a driving frequency ν = 180 Hz The two-dimensional density was fixed at ρ2D = 0.45, which corresponds to N = 4711 particles <ref:2602.20716#pg4> Molecular Dynamics Simulation Details The molecular dynamics simulations were performed using an algorithm analogous to the one used in previous works Wallparticle and particle-particle collisions are inelastic and we model them by using a viscoelastic collisional model For more details, see Appendix B The simulation parameters use ν = 643.217 Hz, and h = 2.00 σ We have made 100 replicas of each relaxation curve, which minimizes statistical uncertainty to nearly vanishing Protocol Implementation The protocol consists of the following five steps <ref:2602.20716#pg7> Step 1 involves determining the two steady states associated with the initial and final input accelerations—the upper and lower steady states illustrated in Fig.

Kai: 1 <ref:2602.20716#pg3> Step 3 requires recording statistical realizations of the relaxation dynamics toward the lower-acceleration steady state <ref:2602.20716#pg7> Step 4 constructs relaxation curves by selecting points along the relaxation obtained in step 3 whose final temperatures match T = T∗s(Γw), using the respective upper steady states as initial conditions <ref:2602.20716#pg7> Step 5 generates the Kovacs response curves by initiating the system from the final states identified in step 4, by applying the intermediate accelerations and recording the evolution until the steady values T∗s(Γw) are reached <ref:2602.20716#pg7> In contrast to earlier granular memory observed through density or stress

11: , here memory is carried by the granular temperature itself <ref:2602.20716#pg4> In the confined granular layer, only the anomalous Kovacs response is observed: during relaxation, the system initially departs further from its target state, with complete absence of the classical (normal) Kovacs hump <ref:2602.20716#pg7> The authors demonstrate that memory is confined to kinetic stages and that subsequent relaxation proceeds through a universal hydrodynamic state governed solely by the horizontal temperature <ref:2602.20716#pg2> This confinement of memory to the kinetic stage has a direct interpretation in terms of the Zwanzig-Mori formalism, in which eliminating degrees of freedom generates memory kernels that become effectively local once the retained and projected-out variables decouple <ref:2602.20716#pg7> The kurtosis criterion provides a generic, experimentally accessible signature for the onset of memoryless hydrodynamic behavior <ref:2602.20716#pg7> The results indicate that the controllability of memory in driven systems is itself regime-dependent—determined by whether the projected-out degrees of freedom remain externally accessible <ref:2602.20716#pg7> References

1: N.

Paper summary: Kai: Looking at "The Kovacs memory effect in a thin granular layer: experimental evidence and its physical origin" again, the title points us toward observing this effect specifically in a thin granular layer and confirming its physical origin through experiments.

Mira: The authors basically confirm that the anomalous Kovacs memory effect is real during the initial fast relaxation, but it’s important to stress that this memory vanishes very quickly, leading to a time-dependent memoryless regime governed only by the instantaneous temperature until things settle down.

Lev: What this means for someone building real hardware is that you need to design your measurement protocol carefully because you're only capturing that specific initial transient phase where the memory lives.

Kai: The paper shows that this confinement of memory to kinetic stages has a direct interpretation in terms of the Zwanzig Mori formalism, suggesting that eliminating degrees of freedom creates memory kernels that become local when certain variables decouple seven.

Mira: It suggests that this isn't just an isolated observation; it fits into the broader mathematical picture for how memory arises when you project out parts of the dynamics, which is quite a powerful way to frame this phenomenon.

Lev: And they do show that the kurtosis criterion provides a generic, experimentally accessible signature for when that memoryless hydrodynamic behavior starts to take over seven. That's something we could check in other driven systems too.

Kai: So, the paper confirms that controllability of memory in driven systems is regime-dependent, depending on whether those projected-out degrees of freedom are still externally accessible seven. This is a key point for any experimentalist.

Mira: It really shows that in granular fluids, the memory phenomenology isn't something you can just tune with external conditions; it’s intrinsically fixed by the kinetic equations themselves seven. That’s a major takeaway for nonequilibrium science.

Conclusion: Kai: So, this paper reports seeing that weird memory effect in a thin layer of vibrating grains using actual experiments.

Mira: It’s about this anomalous Kovacs memory effect that only shows up during the very first fast relaxation stage after you change how hard you vibrate the system.

Lev: The authors found that this memory disappears pretty fast, leaving behind something governed only by the current temperature until everything settles down to its final steady state.

Kai: What they really discovered is that this memory doesn't persist like in other systems; it seems tied directly to a coupling between the horizontal and vertical temperatures during those initial quick movements.

Mira: That coupling is what fundamentally constrains the types of memory we can even see in this specific setup, meaning you don't get the standard Kovacs effect as you might expect.

Lev: From an engineering standpoint, it’s important to know that this memory isn't something you can just tune away with external controls; it’s locked into the kinetic equations of the granular layer itself.

Kai: So, to summarize, they found that memory exists only in those initial fast stages because of how the horizontal and vertical temperatures interact during relaxation.

Mira: And that this confinement means we need to look at how degrees of freedom decouple in these kinetic theories—the Zwanzig Mori formalism explains why the memory kernels become local then.

Lev: It’s a bit limiting, though, because it sets a strict boundary on what kind of persistent path-dependent memory you can even hope to measure here compared to some other driven systems.

Kai: Exactly. So the big implication is that when we look for memory in granular fluids, we need to focus specifically on these short kinetic stages where the coupling is active.

Mira: And they suggest that using a simple kurtosis criterion could be a general way to spot when you transition into that memoryless hydrodynamic state later on.

Lev: That’s what I mean—it gives us a measurable signature for when the system stops remembering its past path and just follows the universal temperature evolution.

Kai: It tells us that in these driven granular systems, the controllability of memory isn't a general feature; it depends entirely on whether you can keep those other degrees of freedom accessible.

Mira: And that’s what leads us to think about how different physical setups, like quantum systems versus granular fluids, handle this accessibility issue differently.

Departamento de Fsica and Instituto de Computación Cientfica Avanzada (ICCAEx), Universidad de Extremadura · Departamento de Matemâtica Aplicada II, Universidad de Sevilla · Fsica TeoricA, Universidad de Sevilla

cond-mat.soft

Submitted: 2026-02-24

Updated: 2026-09-17

Comments: 8 pages; 5 figures. To appear in the Physical Review Letters (2026)

Journal ref: Phys. Rev. Lett. 137, 208201 (2026)

DOI: 10.1103/791d-2tx9

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 83/100

The gist: The paper reports the experimental observation of memory effects in a vertically vibrated thin granular layer…

Key concepts

Kovacs Memory Effect
This is a phenomenon where the system's response to a change in input (like acceleration) depends on its previous state. In this granular system, it was observed anomalously only during the very first, fast relaxation phase after being quenched.
Kinetic Regime
This is the initial, rapid evolution of particle velocities immediately following a change in conditions. The paper shows that memory effects are confined to this kinetic stage because the system's dynamics are governed by these initial conditions before it transitions to a slower hydrodynamic state.
Hydrodynamic State
This is the later, universal regime where the system behaves like a fluid, and its macroscopic behavior depends only on macroscopic fields like temperature. After the initial memory fades, all time dependence enters through this state, making it independent of prior kinetic history.

Terminology

Summary

The paper reports the experimental observation of memory effects in a vertically vibrated thin granular layer, revealing an anomalous Kovacs memory effect confined to the initial fast relaxation stage

Experimental Observation and Findings

The study experimentally observed that following a quench in input acceleration, the granular temperature exhibits an anomalous Kovacs memory effect confined to the initial fast relaxation stage <ref:2602.20716#pg4> The memory vanishes shortly thereafter, yielding a time-dependent memoryless regime governed solely by the instantaneous temperature before the system reaches its final steady state <ref:2602. p. 020716#pg2> The analysis reveals that memory emerges during fast transients through coupling between horizontal and vertical temperatures—a mechanism that fundamentally constrains the accessible memory phenomenology and precludes observation of the standard Kovacs effect in this system <ref:2602. p. 020716#pg2>

Theoretical Framework and Kinetic States

The kinetic theory framework developed quantitatively captures these features by identifying the initial memory and subsequent memoryless regimes with the kinetic and hydrodynamic states, respectively (that are well established in kinetic theory) <ref:2602.20716#pg2> Memory is reflected in a nonmonotonic behavior of the temperature, implying that, at those times, the temperature does not fully specify the state of the system <ref:2602. p. 020716#pg2> The dynamics of the velocity distribution function consists of a rapid and initial-condition-dependent evolution (the kinetic regime) over a characteristic time tm, followed by a hydrodynamic regime in which all the time dependence enters solely through the hydrodynamic fields, the granular temperature in our case <ref:2602. p. 120716#pg3>

Constraints on Memory Phenomenology

The authors demonstrate that memory characteristics are constrained by the system’s governing equations and cannot be tuned through external conditions alone <ref:2602.20716#pg2> The results show that memory is confined to kinetic stages, a fundamentally different regime from persistent path-dependent memory in externally-driven systems (such as magnetic hysteresis) <ref:2602. p. 420716#pg7> This confinement of memory to the kinetic stage has a direct interpretation in terms of the Zwanzig-Mori formalism, in which eliminating degrees of freedom generates memory kernels that become effectively local once the retained and projected-out variables decouple <ref:2602. p. 420716#pg7>

Universal Hydrodynamic State

Subsequent relaxation proceeds through a universal memoryless hydrodynamic state governed solely by the horizontal temperature <ref:2602.20716#pg2> The system state thus becomes effectively hydrodynamic, where the temperature evolves autonomously and all other moments are slaved to it, so that the macroscopic dynamics becomes independent of the prior kinetic history (memory vanishes) <ref:2602. p. 120716#pg3> This is demonstrated by showing that after a convenient translation in the time axes, tΓ, the horizontal temperature curves collapse onto a single curve <ref:2602. p. 320716#pg5>

Model and Quantitative Analysis

A minimal kinetic theory model and molecular dynamics simulations establish that the dynamics is governed by two coupled temperatures—horizontal and vertical—operating on slow and fast timescales, respectively <ref:2602.20716#pg6> The structure of the equations is clear: energy is injected in the vertical direction (last term of Eq. p. 3 (4)), while it is transferred to the other degrees of freedom via inelastic collisions <ref:2602. p. 320716#pg6> Linearizing Eqs. p. 4 (3) and (4) around the stationary state allows for an explicit calculation of the function H(s) = T(s)/Ts − 1, which shows that H(s) < 0 for the cooling Kovacs protocol, leading to a downward hump <ref:2602. p. 420716#pg6> The amplitude of the hump, Hmin, is proportional to dHdss=0 and decreases with the input acceleration <ref:2602. p. 420716#pg6>

Conclusions on Controllability

The controllability of memory in driven systems is itself regime-dependent—determined by whether the projected-out degrees of freedom remain externally accessible <ref:2602.20716#pg7> In the latter case, external engineering of relaxation, as achieved in driven quantum systems [36], requires such accessibility, whereas in driven granular fluids the memory phenomenology is intrinsically fixed by the kinetic equations <ref:2602. p. 420716#pg7> The kurtosis criterion provides a generic, experimentally accessible signature for the onset of memoryless hydrodynamic behavior <ref:2602. p. 4 p. 420716#pg7> This is readily measurable in other driven systems such as active matter or dense colloidal suspensions <ref:2602. p. 420716#pg7>

Experimental Setup Details

The experimental configuration consists of a thin, circular layer of N identical stainless-steel spheres (AISI 52100) with coefficient of normal restitution α = 0.95 The layer is confined between a bottom aluminum plate and a top plexiglas lid, with steel spacers forming the lateral boundary All experiments were performed at a driving frequency ν = 180 Hz The two-dimensional density was fixed at ρ2D = 0. p. 145, which corresponds to N = 4711 particles <ref:2602. p. 120716#pg4>

Molecular Dynamics Simulation Details

The molecular dynamics simulations were performed using an algorithm analogous to the one used in previous works Wallparticle and particle-particle collisions are inelastic and we model them by using a viscoelastic collisional model For more details, see Appendix B The simulation parameters use ν = 643.217 Hz, and h = 2. p. 700 σ We have made 100 replicas of each relaxation curve, which minimizes statistical uncertainty to nearly vanishing

Protocol Implementation

The protocol consists of the following five steps <ref:2602.20716#pg7> Step 1 involves determining the two steady states associated with the initial and final input accelerations—the upper and lower steady states illustrated in Fig. p. 7 1 <ref:2602. p. 020716#pg3> Step 3 requires recording statistical realizations of the relaxation dynamics toward the lower-acceleration steady state <ref:2602. p. 720716#pg7> Step 4 constructs relaxation curves by selecting points along the relaxation obtained in step 3 whose final temperatures match T = T∗s(Γw), using the respective upper steady states as initial conditions <ref:2602. p. 720716#pg7> Step 5 generates the Kovacs response curves by initiating the system from the final states identified in step 4, by applying the intermediate accelerations and recording the evolution until the steady values T∗s(Γw) are reached <ref:2602. p. 720716#pg7>

Summary of Key Results

In contrast to earlier granular memory observed through density or stress [11], here memory is carried by the granular temperature itself <ref:2602.20716#pg4> In the confined granular layer, only the anomalous Kovacs response is observed: during relaxation, the system initially departs further from its target state, with complete absence of the classical (normal) Kovacs hump <ref:2602. p. 420716#pg7> The authors demonstrate that memory is confined to kinetic stages and that subsequent relaxation proceeds through a universal hydrodynamic state governed solely by the horizontal temperature <ref:2602. p. 120716#pg2> This confinement of memory to the kinetic stage has a direct interpretation in terms of the Zwanzig-Mori formalism, in which eliminating degrees of freedom generates memory kernels that become effectively local once the retained and projected-out variables decouple <ref:2602. p. 420716#pg7> The kurtosis criterion provides a generic, experimentally accessible signature for the onset of memoryless hydrodynamic behavior <ref:2602.20716#pg7> The results indicate that the controllability of memory in driven systems is itself regime-dependent—determined by whether the projected-out degrees of freedom remain externally accessible <ref:2602. p. 420716#pg7>

References

[1] N. C. Keim, J. D. Paulsen, Z. p. 5 Zeravcic, S. p. 5 Sastry, and S. p. 5 R. Nagel, Memory formation in matter <ref:2602. p. 520716#pg2>

[2] A. Lasanta, F. p. 5 p. 5 p. 5 Vega Reyes, A. p. 0 p. 0 p. 0 Prados, and A. p. 5 p. 5 Santos, When the hotter cools more quickly: Mpemba effect in granular fluids <ref:2602. p. 520716#pg2>

[3] Z. Lu and O. Raz, Nonequilibrium thermodynamics of the markovian mpemba effect and its inverse <ref:2602. p. 520716#pg2>

[4] A. J. Kovacs, J. J. Aklonis, J. p. 5 M. Hutchinson, and A. p. 5 R. Ramos, Isobaric volume and enthalpy recovery of glasses <ref:2602. p. 520716#pg2>

[5] R. K¨ursten, V. p. 5 Sushkov, and T. p. 5 Ihle, Giant kovacs-like memory effect for active particles <ref:2602. p. 520716#pg2>

[6] A. Prados and E. p. 5 Trizac, Kovacs-like memory effect in driven granular gases <ref:2602. p. 520716#pg2>

[7] E. Trizac and A. p. 5 Prados, Memory effect in uniformly heated granular gases <ref:2602. p. 520716#pg2>

[8] J. J. Brey, M. I. Garc´ıa de Soria, P. p. 5 p. 5 Maynar, and V. p. 0 Buz´on, Memory effects in the relaxation of a confined granular gas <ref:2602. p. 520716#pg2>

[9] A. Santos, Time-delayed newton’s law of cooling with a finite-rate thermal quench: Impact on the mpemba and kovacs effects <ref:2602. p. 520716#pg2>

[10] A. Militaru, A. p. 5 Lasanta, M. p. 5 Frimmer, L. p. 5 L. Bonilla, L. p. 5 Novotny, and R. p. 5 A. Rica, Kovacs memory effect with an optically levitated nanoparticle

[11] C. Josserand, A. p. 5 V. Tkachenko, D. p. 5 M. Mueth, and H. p. 5 M. Jaeger, Memory effects in granular materials

[12] A. Lasanta, F. Vega Reyes, A. Prados, and A. Santos, On the emergence of large and complex memory effects in nonequilibrium fluids

[13] G. Teza, J. Bechhoefer, A. p. 5 Lasanta, O. p. 5 Raz, and M. p. 5 Vucelja, Speedups in nonequilibrium thermal relaxation: Mpemba and related effects

[14] J. J. Brey, J. W. Dufty, C. S. Kim, and A. p. 5 Santos, Hydrodynamics for granular flow at low density <ref:2602. p. 520716#pg2>

[15] D. Candela, Complex memory formation in frictional granular media

[16] D. R. M. Williams and MacKintosh, Driven granular media in one dimension: Correlations and equation of state

[17] A. Puglisi, V. p. 5 Loreto, U. Marini Bettolo Marconi, A. p. 5 Petri, and A. p. 5 Vulpiani, Clustering and non-gaussian behavior in granular matter

[18] R. Cafiero, S. p. 5 Luding, and H. p. 5 J. Herrmann, Twodimensional granular gas of inelastic spheres with multiplicative driving

[19] R. Brito, D. Risso, and R. p. 5 Soto, Hydrodynamic modes in a confined granular fluid

[20] E. Momp´o, M. p. 5 A. L´opez-Casta˜no, A. p. 5 Lasanta, F. Vega Reyes, and A. p. 0 Torrente, Memory effects in a gas of viscoelastic particles

[21] D. Hilbert, Begr¨undung der kinetischen Gastheorie

[22] C. Cercigniani, Mathematical methods in kinetic theory

[23] S. Chapman and S. p. 5 Cowling, The mathematical theory of nonuniform gases

[24] S. G. Brush, Kinetic theory

[25] P. Melby, F. Vega Reyes, A. Prevost, R. p. 5 Robertson, P. p. 5 Kumar, D. A. Egolf, and J. p. 5 S. Urbach, The dynamics of thin vibrated granular layers

[26] J. S. Olafsen and J. p. 5 p. 5 S. Urbach, Clustering, order, and collapse in a driven granular monolayer

[27] J. S. Olafsen and J. S. Urbach, Velocity distributions and density fluctuations in a granular gas

[28] M. Louge and et al. p. 5, Fall 1999 impact parameter chart

[29] J. C. Crocker and D. p. 5 G. Grier, Methods of digital video microscopy for colloidal studies

[30] F. Vega Reyes and J. p. 0 S. Urbach, Effect of inelasticity on the phase transitions of a thin vibrated granular layer

[31] P. Maynar, M. I. Garc´ıa de Soria, and J. p. 5 Javier Brey, Homogeneous dynamics in a vibrated granular monolayer

[32] M. I. Garc´ıa de Soria, P. Maynar, and E. p. 0 E.Trizac, Universal reference state in a driven homogeneous granular gas

[33] R. W. Zwanzig, Ensemble method in the theory of irreversibility

[34] H. Mori, A continued-fraction representation of the timecorrelation functions

[35] D. Forster, Hydrodynamic Fluctuations, Broken Symmetry, and Correlation Functions (W. p. 5 A. Benjamin, 1975)

[36] N. Beato and G. p. 5 Teza, Relaxation control of open quantum systems

[37] D. B. Allan, T. Caswell, N. p. 5 C. Keim, C. M. van der Wel, and R. p. 5 W. Verweij, soft-matter/trackpy: v0. p. 57

[38] X. Nie, E. Ben-Naim, and S. p. 6 Y. Chen, Dynamics of vibrated granular monolayers

[39] J. Sun, F. Battaglia, and S. p. 6

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved predictive modeling for granular systems: The kinetic theory framework developed can be used to quantitatively capture memory effects by identifying the initial memory and subsequent memoryless regimes with the kinetic and hydrodynamic states, respectively. This allows AI systems to predict whether a system will exhibit an anomalous Kovacs response or a normal response based on the transient time scale.

  2. Enhanced understanding of physical constraints: The analysis reveals that memory emerges during fast transients through coupling between horizontal and vertical temperatures—a mechanism that fundamentally constrains the accessible memory phenomenology. This capability allows AI to constrain simulations or experiments by predicting which memory effects are physically possible given the system's governing equations.

  3. Universal state identification for long-term dynamics: The results show that the dynamics collapses onto a hydrodynamic manifold parametrized by T, meaning the horizontal temperature is the relevant hydrodynamic variable controlling long-time dynamics. This enables AI to simplify complex, high-dimensional granular simulations into low-dimensional models governed only by the horizontal temperature for long-term prediction.

  4. Signature detection for memoryless behavior: The kurtosis criterion provides a generic, experimentally accessible signature for the onset of memoryless hydrodynamic behavior. An AI system could be trained to monitor the relaxation of kurtosis and use its stationary value as a real-time indicator that the system has entered a regime where memory has vanished.

Related papers