Kovacs Effect¶

A slow-relaxing system can leave a matched reference value and return nonmonotonically, revealing memory of its preparation history.

Version: v1 · 2026-10-03 · History
Domain-specific #: 13366
Domain group: Natural Sciences
Origin domain: Physics
Subdomain: Nonequilibrium Statistical Mechanics → Physics
Aliases: Kovacs Memory Effect, Kovacs Hump

Core Idea¶

The Kovacs effect is a memory response in which a slow-relaxing system has a monitored property equal to its reference value under a final condition, yet that property subsequently moves away from the value before returning. A standard preparation first changes a control, allows partial aging, and then switches to a final control at the moment the monitored property matches the final reference. The later nonmonotonic excursion is the Kovacs hump. It shows that matching one macroscopic measurement is not enough to specify the state that determines future relaxation.^[1]^[2]

Kovacs's original setting concerned volume after temperature changes in glassy polymers. Related behavior has been analyzed in model glasses and, under different physics, in driven granular gases. For the latter, the reference can be a nonequilibrium steady state rather than a thermodynamic equilibrium state. The transferable relation is therefore matched observable but unmatched internal state, not “all systems are below a glass transition.”^[2]^[3]

Structural Signature¶

Sig role-phrases:

Slow dynamics: the system's internal configuration does not instantly follow a change in its external condition.
Preparation history: an earlier condition leaves a state that depends on how the system arrived there.
Reference value: the final condition has an equilibrium or steady reference for the measured observable.
Matched switch: the control changes when the observable equals that final reference value.
Transient departure: despite the apparent match, the observable moves away and then returns nonmonotonically.
History-dependent amplitude or timing: changing the preparation can change the transient even if the observable at the switch is the same.^[2]^[3]

Condensed: different preparation histories + same measured current value → different future trajectory under the final condition.

What It Is Not¶

Not simple monotonic relaxation. Starting visibly away from equilibrium and moving steadily toward it lacks the matched-value paradox.
Not any overshoot. An externally forced overshoot does not by itself show memory of an earlier preparation at an apparently settled observable.
Not proof of one particular hidden variable. The hump demonstrates that the monitored scalar is insufficient, but models differ on the distribution of relaxation modes or microscopic states responsible.^[2]
Not confined to glass-transition polymers. The original material frame is narrow; related matched-reference humps occur in other slow nonequilibrium systems.^[3]
Not always departure from thermodynamic equilibrium. A driven granular gas can display the pattern relative to a nonequilibrium steady temperature.^[3]
Not identical to hysteresis. Hysteresis broadly concerns path-dependent response; this effect adds a distinctive two-stage matching protocol and subsequent nonmonotonic transient.

Scope of Application¶

The historical polymer protocol changes temperature and follows specific volume. Bertin and colleagues reconstruct it from Kovacs's work: a high-temperature sample is quenched to a lower waiting temperature, and then raised to a final temperature when its post-jump volume equals the final-temperature equilibrium reference. The post-jump qualifier removes the immediate thermal-expansion contribution; the later volume hump is the memory observation. Their model paper uses related protocols to probe distributions of relaxation times or structural heterogeneity, not to infer full equilibrium from one volume measurement. The original 1963 chapter remains bibliographically identified but has not been directly page-checked here.^[1]^[2]

Prados and Trizac studied an analogous dilute driven granular gas by theory and simulation. Starting at the steady state for one drive, they reduce the stochastic drive during a waiting interval, then select a final drive whose steady granular temperature equals the temperature at that switch. Temperature nevertheless departs and returns. Their Fig. 2 compares weak and strong inelasticity and finds a reversed hump under strong dissipation. The reference is a driven nonequilibrium steady state, and the paper identifies the gas's excess velocity kurtosis as a relevant additional state variable; neither this mechanism nor the hump's direction transfers automatically to polymers.^[3]

Clarity¶

Specify the controlled variable, monitored observable, preparation sequence, reference state, and switch time. A report of “memory” alone is too broad. The crucial comparison is that the observable already equals the final reference at the switch, yet continues to evolve. Likewise, “equilibrium value” should not be used for a driven system whose reference is a nonequilibrium steady state. The sign of the hump and its peak time are outcomes to measure, not part of a single universal formula.^[2]^[3]

Manages Complexity¶

The effect exposes the inadequacy of a one-variable state description. If volume or granular temperature were the complete state variable under the final control, equal current values would imply the same future. The hump contradicts that prediction for the prepared system. This is a compact diagnostic of hidden state or distributed relaxation, but the diagnostic is not a full microscopic explanation. Extra measurements or controlled preparation comparisons are needed to identify what history information persists.^[2]^[3]

Abstract Reasoning¶

Construct two histories that arrive at the same monitored value under the same final condition. Track the observable after the switch. If one trajectory departs nonmonotonically while a true reference state remains stable, the current scalar does not determine future dynamics. To go further, compare hump sign, amplitude and timing across preparation histories, and test candidate models. Avoid inferring that the same mechanism operates in polymer glasses and inelastic granular gases simply because the traces have a similar shape.^[2]^[3]

Knowledge Transfer¶

What transfers is a state-sufficiency test: match a visible property, vary the route to it, and compare the subsequent relaxation. This reasoning can guide experiments in polymers, molecular liquids and driven granular systems. The material physics does not transfer wholesale. Molecular glass dynamics, granular energy injection and collision dissipation call for different state variables and constitutive models. A hump shows a history-sensitive trajectory; the route from trajectory to mechanism must be demonstrated case by case.^[2]^[3]

Examples¶

Polymer volume after temperature shifts¶

A polyvinyl-acetate sample in the protocol reconstructed by Bertin and colleagues begins at high temperature $T_0$. A separate direct quench to final temperature $T_2$ establishes the equilibrium reference volume $V_{eq}(T_2)$. For the memory run the sample is quenched instead to $T_1<T_2$, waits for time $t_1$, and is raised quickly to $T_2$. The wait is selected so $V(t_1^+)=V_{eq}(T_2)$ after the jump, not before it. Although the observed volume is then apparently correct for $T_2$, it rises above that value and later returns. This is a source-described experimental protocol, not a claim that the full Kovacs chapter was directly read.^[1]^[2]

Mapped back: polymer with slow structural recovery → $T_0\to T_1$, wait, then $T_1\to T_2$ → post-jump $V(t_1^+)=V_{eq}(T_2)$ → positive volume hump and return. Matching the pre-jump volume would confuse the fast thermal step with the memory response.

Driven granular gas¶

Prados and Trizac begin a dilute inelastic hard-particle gas in a nonequilibrium steady state at drive $\xi_0$. They lower the stochastic drive to $\xi_1$ for a waiting time $t_w$. At the switch they choose the final drive $\xi$ so its own steady temperature $T_s(\xi)$ equals the observed $T(t_w)$. The temperature does not stay at this matched value. In their Fig. 2 simulations, coefficient of restitution $\alpha=0.8$ produces a temperature maximum, whereas $\alpha=0.3$ produces a minimum. Their model explains the difference using excess kurtosis of the velocity distribution, which need not match its final steady value when temperature does.^[3]

Mapped back: inelastic gas with velocity-distribution memory → $\xi_0\to\xi_1$, wait, then final $\xi$ → $T(t_w)=T_s(\xi)$ → maximum at $\alpha=0.8$ or minimum at $\alpha=0.3$ before return. This is a driven-steady reference, not polymer equilibrium.

One-variable exponential relaxation, near miss¶

Suppose a system's entire relevant state really is one scalar that relaxes exponentially to a fixed value. If the scalar already equals the fixed value at the switch, it stays there. If it begins elsewhere, it returns monotonically. Such behavior cannot reproduce a Kovacs hump without extra state or different dynamics.

Mapped back: matched complete state → no departure; missing history dependence.

Structural Tensions¶

One-scalar economy versus richer-state fidelity. Measuring and matching only volume or granular temperature gives a compact, reproducible protocol, but treating that scalar as the complete state predicts no later hump and misses the observed trajectory. Tracking distributions or additional moments can account for history, but demands more measurements and a model whose extra variables are actually discriminated by the data. In the granular paper, excess kurtosis adds a relevant variable; that does not license calling it the polymer's microscopic cause. Diagnostic: when two preparations match the scalar, does a measured extra state descriptor predict the different subsequent curves?^[2]^[3]

Shared phenomenological name and mechanism-specific attribution are compatible claims at different levels. A polymer and a driven granular model can exhibit the matched-value/later-excursion signature without sharing a microscopic memory carrier. A mechanism claim needs a separate, material-specific inference: relaxation distributions in one family of glass models and velocity-distribution kurtosis in the Prados–Trizac gas. This is a scope boundary, not an opposed-cost tension. Diagnostic: is the claim about the trace protocol alone, or does an independently tested state variable explain its amplitude and sign?^[2]^[3]

Equilibrium versus driven steady reference is another typing boundary. The original polymer interpretation uses an equilibrium reference; the driven granular comparison uses a stable state under maintained drive. One can name the reference type and test its stability without sacrificing either comparison, so these are not competing objectives. Diagnostic: under the final fixed control, is the comparator demonstrably stable, and is it equilibrium or a drive-maintained nonequilibrium steady state?^[2]^[3]

Structural–Framed Character¶

The effect is mixed, with a strong physical frame. The same-observable/different-future relation is structural, but its defining hump requires slow relaxation after a controlled preparation. Evaluative weight is low: the observed response is not intrinsically good or bad. Human experimental practice makes the phenomenon legible by arranging a matched switch; it does not cause the underlying relaxation law. Naming traces to Kovacs's polymer work, while no present institution is a constituent of the effect. The vocabulary travels literally to model glasses and some driven granular systems when their protocols meet the matching and hump tests, though their mechanisms differ. Calling a workplace's temporary morale rebound a “Kovacs effect” would import a metaphor, not recognize the physical effect; recognizing path dependence alone also falls short. Its character: a transferable nonequilibrium response signature with material-specific mechanisms.

Structural Core vs. Domain Accent¶

The skeletal relation is same observable, different internal state and future, a possible future-prime state-sufficiency pattern. The domain-bound mechanism is a slow physical relaxation process probed by a two-stage temperature or drive protocol and diagnosed by a nonmonotonic excursion. The named effect does not clear the prime bar: in another domain one might use the skeletal question, but without a physical reference state and measurable post-switch hump the identity is only an analogy. Hysteresis is a live neighbor, not an automatically valid parent; removing the hump leaves broader history dependence rather than this effect.

None of the encyclopedia's broader entries is a kind it falls under, so it stands without a parent for now.

Hysteresis is a semantic neighbor, but as the encyclopedia defines it, it requires a response loop that a returning Kovacs transient need not form. Memory is broader conceptual vocabulary rather than an entry it falls under. A distinct slow-relaxation memory-response entry could later serve as an intermediate.

Neighborhood in Abstraction Space¶

Kovacs Effect sits in a sparse region of the domain-specific corpus (66^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Thermodynamics & Dissipative Systems (19 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Ordinary overshoot may be directly driven by the current forcing. Aging is slow state evolution without necessarily producing the matching paradox. Hysteresis is broader path dependence. Mpemba-like behavior compares relaxation rates from different starting values, whereas the Kovacs test deliberately begins at a matched measured value under the final condition.^[2]^[3]

References¶

[1] A. J. Kovacs, Transition vitreuse dans les polymères amorphes: Étude phénoménologique, Fortschritte der Hochpolymeren-Forschung 3 (1963), historical original, bibliographically verified but full original passage not accessed. registry ↩a ↩b ↩c

[2] E. M. Bertin et al., The Kovacs Effect in Model Glasses, original model analysis, Introduction and Fig. 1 reconstructing Kovacs's polyvinyl-acetate protocol, including the post-jump match. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n

[3] A. Prados and E. Trizac, Kovacs-Like Memory Effect in Driven Granular Gases, Physical Review Letters 112 (2014), 198001, Fig. 1 drive protocol and Fig. 2/associated discussion of $\alpha=0.3$ and $0.8$. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n