Finite-Time Quantum Carnot Info Engine

• Finite-Time Quantum Carnot Information Engine is a cyclic quantum machine that replaces the cold bath with an information reservoir via reversible measurement-feedback to convert information into work.
• It operates in finite time by balancing work output, power, and fluctuations using low-dissipation, shortcut-to-equilibrium, and precise measurement strategies.
• The engine leverages quantum resources such as coherence, superposition, and entanglement to optimize performance and push efficiency bounds beyond classical Carnot limits.

A finite-time quantum Carnot information engine is a Carnot-like cyclic quantum machine in which the usual cold thermal bath is replaced by an information reservoir, typically implemented by a reversible measurement-and-feedback operation, while the engine is driven away from the quasistatic limit so that power, work fluctuations, and measurement duration become essential performance variables. In this literature, the term also borders several adjacent constructions: finite-time quantum Carnot-analog heat engines powered by coherence, one-shot entangling engines, and information-theoretic efficiency bounds that are not Maxwell-demon models in the strict sense. The common thread is that entropy exchange, rather than only two-bath heat exchange, becomes an explicit thermodynamic resource (Fadler et al., 2023, Aguilar et al., 5 Sep 2025).

1. Conceptual scope and nomenclature

In the strict sense used in recent Carnot-information-engine papers, the defining modification of a quantum Carnot heat engine is that the cold isotherm is replaced by an information reservoir. The information reservoir exchanges entropy but no net energy with the working substance, so the cycle converts information into work rather than heat from a cold bath into work (Fadler et al., 2023, Aguilar et al., 5 Sep 2025). Within this definition, the cycle is Maxwell-demon-like: measurement acquires information, feedback exploits it, and the thermodynamic bookkeeping differs from that of an ordinary two-bath engine.

This strict usage should be distinguished from several nearby but nonidentical notions. The finite-time harmonic-trap Carnot-analog engine based on shortcut-to-equilibrium and shortcut-to-adiabaticity is a quantum heat engine, not a Maxwell-demon information engine; its “information-like” feature is that monitoring energy induces dephasing, destroys coherence, and can terminate the engine’s power in the fast-driving regime (Dann et al., 2019). Likewise, the rigorous Lieb-Robinson speed–efficiency theorem concerns ordinary cyclic quantum heat engines with hot and cold baths and does not explicitly formulate a measurement-feedback demon, although its logic is highly relevant to local implementations of information engines (Shiraishi et al., 2017). A further distinction is needed for correlation-based generalized Carnot bounds, where “information” refers to state–Hamiltonian correlations rather than memory-bearing feedback (Gabetti et al., 12 Apr 2026).

A useful operational shorthand is therefore an “information-reservoir Carnot cycle” (Editor’s term): a quantum Carnot-like cycle in which the entropy sink is supplied by measurement and feedback rather than by a conventional cold bath. This term matches the formulations developed for arbitrary working media with scaling Hamiltonian $H_t=\omega_t \mathcal P$ (Fadler et al., 2023).

2. Idealized cycle architecture

The generalized finite-time Carnot information engine developed in the low-dissipation literature has four branches. First, a measurement-plus-feedback stroke at fixed Hamiltonian $H=\omega_{\mathrm{fb}}\mathcal P$ maps an initial thermal state $\rho(\omega_{\mathrm{fb}},T_1)$ to a colder thermal state $\rho(\omega_{\mathrm{fb}},T_2)$ with $T_2<T_1$. Second, an isentropic compression maps $\rho(\omega_{\mathrm{fb}},T_2)$ to $\rho(\omega_3,T_h)$ with $\omega_3>\omega_{\mathrm{fb}}$. Third, a hot isothermal expansion couples the system weakly to a bath at $T_h$ and drives $\rho(\omega_3,T_h)\to \rho(\omega_4,T_h)$ with $H=\omega_{\mathrm{fb}}\mathcal P$0. Fourth, an isentropic expansion returns the system to $H=\omega_{\mathrm{fb}}\mathcal P$1 (Aguilar et al., 5 Sep 2025).

The reversibility requirements are specific. The Hamiltonian is assumed to commute with itself at all times,

$H=\omega_{\mathrm{fb}}\mathcal P$2

so the two isentropic branches are frictionless for arbitrary duration (Fadler et al., 2023). The measurement is described by Kraus operators $H=\omega_{\mathrm{fb}}\mathcal P$3 with $H=\omega_{\mathrm{fb}}\mathcal P$4, and thermodynamic reversibility requires

$H=\omega_{\mathrm{fb}}\mathcal P$5

so the measurement is diagonal in the energy basis, while feedback maps each post-measurement state to a common final thermal state and preserves entropy at the trajectory level (Aguilar et al., 5 Sep 2025).

With these assumptions, the cold branch contributes entropy exchange but no net energetic input. For the working medium alone one has $H=\omega_{\mathrm{fb}}\mathcal P$6, so the total cycle work reduces to the hot-isotherm contribution,

$H=\omega_{\mathrm{fb}}\mathcal P$7

in the low-dissipation regime, where $H=\omega_{\mathrm{fb}}\mathcal P$8 is the entropy change on the hot isotherm, $H=\omega_{\mathrm{fb}}\mathcal P$9 is the low-dissipation coefficient, and $\rho(\omega_{\mathrm{fb}},T_1)$0 is the hot-isotherm duration (Fadler et al., 2023). In the later fluctuation theory this is written as

$\rho(\omega_{\mathrm{fb}},T_1)$1

because the isentropic strokes are taken to have zero duration under the commuting-Hamiltonian assumption (Aguilar et al., 5 Sep 2025).

3. Finite-time thermodynamics, power, and efficiency at maximum power

For the low-dissipation Carnot information engine, the information-to-work efficiency is defined as

$\rho(\omega_{\mathrm{fb}},T_1)$2

and the power is

$\rho(\omega_{\mathrm{fb}},T_1)$3

Optimizing over the hot-isotherm duration at fixed feedback time gives

$\rho(\omega_{\mathrm{fb}},T_1)$4

or, in terms of the dissipation time $\rho(\omega_{\mathrm{fb}},T_1)$5,

$\rho(\omega_{\mathrm{fb}},T_1)$6

This is the Curzon–Ahlborn-like efficiency-at-maximum-power formula for the finite-time quantum Carnot information engine and is stated to be valid for arbitrary working media under the model assumptions (Fadler et al., 2023).

For a qubit working medium with $\rho(\omega_{\mathrm{fb}},T_1)$7, weak energy measurements provide a microscopic realization. In that case the hot-isotherm dynamics yields

$\rho(\omega_{\mathrm{fb}},T_1)$8

and therefore

$\rho(\omega_{\mathrm{fb}},T_1)$9

The parameter $\rho(\omega_{\mathrm{fb}},T_2)$0 is the bath-coupling constant appearing in the master-equation description of the hot isotherm (Fadler et al., 2023).

The 2025 fluctuation analysis extends this finite-time picture from mean work to work statistics. Using the end-point measurement scheme rather than a two-point measurement protocol, because TPM would disturb the feedback dynamics, the total work variance is obtained as

$\rho(\omega_{\mathrm{fb}},T_2)$1

with $\rho(\omega_{\mathrm{fb}},T_2)$2 and $\rho(\omega_{\mathrm{fb}},T_2)$3 the equilibrium heat capacity (Aguilar et al., 5 Sep 2025). The associated exact trade-off equality is

$\rho(\omega_{\mathrm{fb}},T_2)$4

so the Fano factor is exactly anticorrelated with efficiency (Aguilar et al., 5 Sep 2025).

A distinctive result of that analysis is a fast-thermalization scaling regime,

$\rho(\omega_{\mathrm{fb}},T_2)$5

in which

$\rho(\omega_{\mathrm{fb}},T_2)$6

while $\rho(\omega_{\mathrm{fb}},T_2)$7 remains finite (Aguilar et al., 5 Sep 2025). Within that model, maximum efficiency is therefore compatible with finite nonzero power and finite work fluctuations.

4. Measurement, mutual information, and finite-time measurement cost

A finite-time information engine is constrained not only by the hot-isotherm dissipation but also by the physical duration and cost of measurement. In a generalized von Neumann measurement model, a working system $\rho(\omega_{\mathrm{fb}},T_2)$8 interacts for a finite time $\rho(\omega_{\mathrm{fb}},T_2)$9 with a quantum meter $T_2<T_1$0 through

$T_2<T_1$1

with $T_2<T_1$2, so the measurement stroke is an explicit unitary correlation-building process rather than an instantaneous abstract POVM (Kirchberg et al., 1 May 2025). The acquired information $T_2<T_1$3 is monotone in $T_2<T_1$4, satisfies $T_2<T_1$5, and saturates only asymptotically (Kirchberg et al., 1 May 2025).

In the concrete two-level-system plus free-particle-meter example, the measurement incurs an energetic cost $T_2<T_1$6 associated with creating system–meter correlations. The paper states that the resulting power is bounded by

$T_2<T_1$7

and that the total extractable work obeys the exact equality

$T_2<T_1$8

when ergotropy and idealized quasistatic thermalization work are both included (Kirchberg et al., 1 May 2025). Because the Carnot-like thermalization part is quasistatic, a plausible implication is that this equality is an upper-bound statement for practical finite-power operation rather than a generic finite-power identity.

A related finite-time quantum Szilard engine with a spinful particle measured by a Maxwell demon makes the information limitation explicit through

$T_2<T_1$9

where $\rho(\omega_{\mathrm{fb}},T_2)$0 is the mutual information acquired during the finite measurement time $\rho(\omega_{\mathrm{fb}},T_2)$1. In that model, power scales as $\rho(\omega_{\mathrm{fb}},T_2)$2 in the short-time regime and $\rho(\omega_{\mathrm{fb}},T_2)$3 in the long-time regime, implying an optimal intermediate measurement time, and positive work requires a threshold measurement ideality $\rho(\omega_{\mathrm{fb}},T_2)$4 (Zhou et al., 2023).

Measurement-feedback refrigeration provides the complementary Clausius-side statement. For a finite-time two-stroke measurement-based quantum cooler, the generalized entropy-production law is

$\rho(\omega_{\mathrm{fb}},T_2)$5

which yields

$\rho(\omega_{\mathrm{fb}},T_2)$6

This does not directly define a heat-engine efficiency, but it shows how mutual information modifies Carnot-style bounds in finite-time feedback-controlled cycles (Fu et al., 2021).

5. Quantum resources beyond measurement: coherence, superposition, and entanglement

Not all finite-time quantum Carnot-like advantages arise from demon-style information reservoirs. In the harmonic-trap Carnot-analog engine, the working medium is a single particle in a driven harmonic trap with

$\rho(\omega_{\mathrm{fb}},T_2)$7

and finite-time operation is achieved by shortcut to equilibrium on the open strokes and shortcut to adiabaticity on the isolated strokes (Dann et al., 2019). The key quantum variable is coherence in the instantaneous energy basis, represented by the observables $\rho(\omega_{\mathrm{fb}},T_2)$8 and $\rho(\omega_{\mathrm{fb}},T_2)$9 and measured by

$\rho(\omega_3,T_h)$0

In the Endo-global cycle, persistent global coherence yields positive power at short cycle times where the branch-coherence shortcut cycles become dissipators; if energy is continuously monitored, the induced dephasing suppresses this coherence and can terminate the power output (Dann et al., 2019).

A much older pure-state quantum-mechanical Carnot engine based on a particle in a one-dimensional infinite square well shows a different quantum resource effect. Replacing endpoint eigenstates by fixed superpositions changes the quasistatic efficiency from

$\rho(\omega_3,T_h)$1

to

$\rho(\omega_3,T_h)$2

The paper concludes that superposition can enhance efficiency, but in the finite-time power optimization the efficiency at maximum power,

$\rho(\omega_3,T_h)$3

is lower than the no-superposition value $\rho(\omega_3,T_h)$4 (Abe et al., 2012). This directly separates coherence-enhanced quasistatic performance from efficiency-at-maximum-power performance.

A third neighboring notion is the one-shot finite-size engine based on semi-local thermal operations. There the engine reaches

$\rho(\omega_3,T_h)$5

with power

$\rho(\omega_3,T_h)$6

by using entangling, semi-local thermal operations and coherent heat transfer between hot and cold baths (Bera et al., 2021). This construction is information-theoretic in the resource-theory sense, not in the Maxwell-demon sense.

6. Fundamental bounds, generalized Carnot theorems, and apparent super-Carnot behavior

For ordinary local quantum heat engines, the rigorous speed–efficiency theorem based on the Lieb-Robinson bound states

$\rho(\omega_3,T_h)$7

with $\rho(\omega_3,T_h)$8, so exact Carnot efficiency is excluded at any finite cycle time under the theorem’s assumptions (Shiraishi et al., 2017). The paper does not prove the same theorem for Maxwell-demon engines, but it explicitly argues that if the controller or memory is absorbed into an enlarged local engine sector, the same speed–efficiency obstruction should apply to the total thermodynamic device (Shiraishi et al., 2017). This is the main no-go backdrop for finite-time two-bath Carnot claims.

A different line of work derives a sharper information-theoretic efficiency bound using state–Hamiltonian correlations: $\rho(\omega_3,T_h)$9 together with a generalized Carnot theorem

$\omega_3>\omega_{\mathrm{fb}}$0

Here “information” refers to covariance structure involving $\omega_3>\omega_{\mathrm{fb}}$1, $\omega_3>\omega_{\mathrm{fb}}$2, and $\omega_3>\omega_{\mathrm{fb}}$3, not to measurement-feedback mutual information (Gabetti et al., 12 Apr 2026). The paper states that the bound can be saturated even in finite-time cycles and that a quantum-dot engine realizes this saturation beyond the quasistatic regime (Gabetti et al., 12 Apr 2026).

Claims of “beyond Carnot” require especially careful bookkeeping. In microscopic Carnot cycles with finite heat reservoirs, the internal-cycle efficiency can exceed the standard benchmark $\omega_3>\omega_{\mathrm{fb}}$4 because the reservoir temperatures drift and an external reset cycle is required; the work does not identify this as a violation of the second law, but as a consequence of finite reservoir size and internal/external cycle separation (Yan et al., 2024). In an explicitly information-assisted Carnot engine, the efficiency is written as

$\omega_3>\omega_{\mathrm{fb}}$5

and the paper proves $\omega_3>\omega_{\mathrm{fb}}$6 within its bookkeeping, with the special case $\omega_3>\omega_{\mathrm{fb}}$7 giving $\omega_3>\omega_{\mathrm{fb}}$8 and positive work output (Xiao et al., 17 Jul 2025). That model, however, is not a finite-time optimization theory; its contribution is the thermodynamic architecture and a trapped-$\omega_3>\omega_{\mathrm{fb}}$9 implementation scheme rather than a finite-time power–efficiency analysis (Xiao et al., 17 Jul 2025).

The recurrent misconception is therefore that every super-Carnot-looking number signals a breakdown of thermodynamics. The literature instead identifies at least three distinct mechanisms: information reservoirs and demon work (Xiao et al., 17 Jul 2025), finite-reservoir bookkeeping (Yan et al., 2024), and generalized correlation-based bounds sharper than standard Carnot (Gabetti et al., 12 Apr 2026).

7. Physical realizations and adjacent platforms

The most explicit implementation proposal for a quantum Carnot information engine uses a single trapped $T_h$0 ion. The two-level system is encoded in Zeeman sublevels, the motional harmonic mode acts as battery or work storage, and the total Hamiltonian is

$T_h$1

with

$T_h$2

The adiabatic branches are implemented over finite durations such as $T_h$3 and $T_h$4, while the isotherms are approximated by $T_h$5-segment sequences of adiabatic shifts and isochoric thermalization (Xiao et al., 17 Jul 2025). This proposal belongs on the implementation side of the subject even though the associated thermodynamic theory is mostly quasistatic.

Two non-information platforms are frequently used as thermodynamic benchmarks. The minimal universal quantum heat machine is a continuously driven two-level system with Floquet sidebands $T_h$6, permanently coupled to two spectrally separated baths. In its simplest regimes the critical modulation rate $T_h$7 recovers the Carnot efficiency at zero power, while finite-power optimization gives $T_h$8 and explicit efficiency-at-maximum-power formulas (Gelbwaser-Klimovsky et al., 2012). The harmonic-trap Carnot-analog engine provides a complementary finite-time benchmark based on Gaussian dynamics, coherence generation, and shortcut control (Dann et al., 2019).

These neighboring platforms matter because they separate the thermodynamic core from the informational layer. A plausible synthesis is that a practical finite-time quantum Carnot information engine will combine three ingredients already developed in separate literatures: a controllable quantum working medium and hot-isotherm model (Gelbwaser-Klimovsky et al., 2012), an explicit finite-time measurement-and-feedback module with nonzero duration and energetic cost (Kirchberg et al., 1 May 2025), and an information-reservoir Carnot bookkeeping that replaces the cold bath by reversible entropy exchange (Fadler et al., 2023, Aguilar et al., 5 Sep 2025).