Information-Assisted Carnot Engine Surpasses Standard Thermodynamic Bounds

Jin Wang; Qian Zeng; Yang Xiao

REVIEW 3 major objections 4 minor 62 references

Information-Assisted Carnot Engine Surpasses Standard Thermodynamic Bounds

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read By grafting an imperfect measurement-and-feedback stroke onto the standard Carnot cycle, this paper derives exact relations tying the engine's output work and efficiency to the change in mutual information $\Delta I$, proving the engine…

desk verdict The algebra is fine and the model is a legitimate extension of information-thermodynamics, but the headline claims of surpassing Carnot and reaching 100% efficiency rest on an efficiency definition that excludes the demon's memory cost, so the abstract overstates what the paper actually shows. read the letter →

arxiv 2507.13412 v1 pith:DO7MBOOL submitted 2025-07-17 cond-mat.stat-mech physics.data-anquant-ph

Yang Xiao , Qian Zeng , Jin Wang This is my paper · ORCID

classification cond-mat.stat-mechphysics.data-anquant-ph

keywords informationengineMaxwell'sdemonCarnotcyclefeedbackcontrolmutualquantumthermodynamicstwo-levelsystemefficiencybound

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 standard Carnot cycle, upgraded with an imperfect classical measurement and a feedback-control stroke, can exceed the efficiency limits of conventional heat engines. The central result is a set of exact relations expressing the engine's output work and efficiency in terms of the change in mutual information between the working system and the demon's memory: the engine can operate as a heat engine in regimes where the standard Carnot cycle is thermodynamically forbidden, its efficiency is always at least the Carnot efficiency, and it can reach 100% efficiency with positive work output for arbitrary two-level systems. These features are demonstrated explicitly on a spin-1/2 working substance, and an experimental implementation with a trapped calcium ion is proposed. If the derivation is correct, information becomes a quantified, controllable resource that can be traded for work beyond the Carnot bound.

What carries the argument

Load-bearing is the five-stroke Carnot information engine (CIE) cycle, which appends a measurement-and-feedback stroke $C \to C'$ (with imperfect measurement error $\epsilon$), an adiabatic expansion $C' \to D'$, and a cold-reservoir compression $D' \to A$ to the standard four-stroke Carnot cycle. The central quantity is the mutual information between the working system and the demon's memory, $I_C = S_Y - S_\epsilon$ before the control stroke and $I_{C'} = S_Y - S_h$ after it, so the information change is $\Delta I = I_{C'} - I_C = S_\epsilon - S_h$. All performance measures are expressed through $\Delta I$ and the demon's work $W_d = \omega_C(\langle n_d\rangle - \langle n_h\rangle)$: the cold-reservoir heat decomposes as $Q_c^d = Q_c - \beta_c^{-1}\Delta I$, the net information-cycle work is $W_{\rm net} = W_d - \beta_c^{-1}\Delta I$, and the efficiency bound $\eta_C \le \eta_d \le \eta_{\rm up}$ follows from the inequality $\beta_h W_d \ge \Delta I \ge \beta_d W_d$ proved in Appendix C.

What would settle it

In the proposed trapped-$^{40}\mathrm{Ca}^+$ implementation, measure over many cycles the average work stored in the motional mode and the heats exchanged with the hot and cold reservoirs, and check whether $\eta_d = W_{\rm tot}^d/(Q_h + W_d)$ is always at least the Carnot efficiency and reaches 1 with $W_{\rm tot}^d = Q_h^d$ at the $S_\epsilon = S_c$ point; a single violation rejects the central claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that adding a demon-controlled stroke to a Carnot cycle converts mutual information between system and demon into a tunable thermodynamic resource. The efficiency is derived as $\eta_d = \eta_C + \frac{(\beta_h/\beta_c)W_d - \Delta I}{Q_h^d}$, where $W_d$ is the work the demon does on the system, $\Delta I = S_\epsilon - S_h$ is the change in mutual information (the measurement-error entropy minus the hot-reservoir entropy), and $Q_h^d = Q_h + W_d$ is the total input energy. Together with the inequality $\beta_h W_d \ge \Delta I \ge \beta_d W_d$, this gives the efficiency bounds $\eta_C \le \eta_d \le \eta_{\rm up}$. When the measurement error is tuned so that $S_\epsilon = S_c$, the engine absorbs zero heat from the cold reservoir, $Q_c^d = 0$, and the total output work equals the total input energy, $W_{\rm tot}^d = Q_h^d \ge 0$, so $\eta_d = 1$ with positive work. The same relations show that in the regime $S_h < S_c < S_\epsilon$ the CIE produces positive total work where the standard Carnot engine's work $W_{\rm tot}$ is negative.

Load-bearing premise

The results assume that the measurement, the memory that stores it, and the erasure of that memory are free, so the 100% efficiency counts only the hot-reservoir heat plus the demon's direct work as input; if the information reservoir carries any thermodynamic cost, the efficiency relative to all inputs cannot reach one.

Editorial extensions

If this is right

In the parameter regime $S_h < S_c < S_\epsilon$, the CIE produces positive total work while the standard Carnot cycle would consume work, so information feedback extends the operating range of heat engines.
The efficiency inequality $\eta_d \ge \eta_C$ holds for all measurement errors, with equality only when $\Delta I = 0$; any nonzero information change improves efficiency.
At the special point $S_\epsilon = S_c$, the cold-reservoir heat vanishes and the engine converts all input energy $Q_h + W_d$ into output work, giving $\eta_d = 1$ with positive work for any two-level system.
For the spin-1/2 example, the efficiency reaches a minimum at $\beta_h = \beta_c \omega_A / \omega_C$ and increases as $|\Delta I|$ grows away from zero, and the paper proposes a trapped-$^{40}\mathrm{Ca}^+$ ion experiment that could realize the cycle.

Reading between the lines

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

The paper's 100% efficiency is reached under its definition that counts only $Q_h + W_d$ as input; if the full thermodynamic cost of measurement, memory, and erasure is included, the effective efficiency would be lower, so the claim should be read as a bound on converting information-assisted input, not on all physical resources.
Since the derivation works with population entropies and level spacings rather than any spin-specific detail, the same formulas should extend to multi-level or continuous working systems; a direct test would be measuring $\eta_d$ versus $\Delta I$ in a trapped ion and comparing with Eq. (8).
A practical design rule suggested by the result is to tune the measurement error $\epsilon$ so that $S_\epsilon$ matches the cold-reservoir entropy $S_c$, which the paper shows is the condition for reaching unit efficiency.

Share X Bluesky LinkedIn Reddit HN

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 short version: the model and algebra are fine, but the central claim that the engine surpasses the Carnot bound and reaches 100% efficiency rests on an efficiency definition that leaves the information reservoir out of the energy balance. Appendix A itself shows the total entropy production of the composite system is σ = I_C' ≠ 0, i.e., there is information dissipation. So the cycle is not dissipationless; the 100% point just trades heat dumped to the cold reservoir for information dumped into the demon's memory. If you include the Landauer erasure cost of that memory, the efficiency drops below 1 and the "beyond Carnot" claim becomes a bookkeeping artifact rather than a new physical bound.

What's genuinely useful: the paper constructs a concrete five-stroke feedback Carnot cycle and derives clean explicit relations between mutual information changes, work output, and efficiency. The bound η_d ≥ η_C follows from a second-law inequality β_h W_d ≥ ΔI, and the derivation is internally consistent. The spin-1/2 example and trapped-ion implementation give the work a concreteness that similar papers often lack. The citation pattern is fine; Sagawa-Ueda and Fadler et al. are acknowledged, and the paper is best read as an extension of those results rather than a departure.

The soft spot is exactly where the reader puts it: the efficiency definition in Eq. (8) counts Q_h + W_d as input but not the cost of measurement, memory storage, or erasure. Appendix B's justification leans on an analogy to ergotropy, but that analogy doesn't cover the demon's memory over repeated cycles. The authors are aware of this; Appendix A even flags σ = I_C' ≠ 0. So the overstatement is in the abstract and conclusions, not in the technical core. I'd suggest the authors rephrase the claims: the CIE can exceed the standard Carnot efficiency relative to a subsystem accounting that excludes information costs, not in a closed-cycle accounting.

Verdict: worth a serious referee. The algebra is consistent, the experimental scheme is detailed, and the issue is interpretational; a competent referee could get the authors to fix the overclaims. But I would not cite the 100% efficiency claim as stated, and I would not bring it to a reading group as a breakthrough.

Referee Report

3 major / 4 minor

Summary. The manuscript introduces a 'Carnot information engine' (CIE), a two-level system operating through two standard Carnot strokes followed by a measurement-feedback control and a cold isothermal return. The central results are Eq. (8), η_d = W_tot^d/(Q_h+W_d) = η_C + (β_h W_d − ΔI)/(β_c Q_d^h), and the inequalities η_C ≤ η_d ≤ η_up derived from β_h W_d ≥ ΔI ≥ β_d W_d. The authors argue that information changes allow the engine to operate in a regime forbidden to the standard Carnot cycle and to reach η_d = 1 when Q_d^c = 0. A spin-1/2 example and a trapped-ion implementation are presented.

Significance. The algebraic framework is transparent and largely self-consistent: the key inequality β_h W_d ≥ ΔI follows from a second-law argument in Appendix C, and the spin-1/2 example gives explicit curves for work, heat, and efficiency. The proposed trapped-ion experiment is a concrete and potentially falsifiable setup. However, the paper's significance as stated is undermined by the efficiency accounting: the information reservoir is not included in the energy balance, and the 100% efficiency claim is therefore a statement about a subsystem with an unpaid information debt rather than a full-cycle thermodynamic bound.

major comments (3)

[Conclusions; Eq. (8); Appendix A, Eq. (A5)] The headline claims that the CIE 'surpasses standard thermodynamic bounds' and 'achieves 100% efficiency' are not supported as full-cycle statements. The efficiency in Eq. (8) counts only Q_h and W_d in the denominator and omits the cost of the information reservoir. Appendix A explicitly computes the total microscopic entropy production of the complete system as σ = I_C' ≠ 0 and attributes it to 'information dissipation.' For a cyclic engine the demon's memory must be reset, and the minimum erasure work is at least β_c^{-1} I_C' (or the corresponding Landauer cost). When this cost is included, the point S_ε = S_c no longer gives η = 1: the heat formerly rejected to the cold reservoir is replaced by information discarded from the memory, so the cycle is not dissipationless. The results should be explicitly presented as a partial efficiency of the working fluid relative to a free information reservoir, with the full accounting made elsewhere.
[Eq. (8); Appendix B] The definition of Q_d^h = Q_h + W_d as 'total invested energy' is problematic when the demon's work is negative, which can occur at the 100% point (in the spin-1/2 example, S_ε = S_c gives a small ε and W_d = ω_C(ε − P_C^e) < 0). The same W_d enters the numerator W_tot^d through W_net, so a negative W_d simultaneously reduces the denominator and the numerator, and η_d = 1 is an artifact of subtracting the demon's extracted work from the input while still counting it (with opposite sign) in the output. If the demon is a work source, W_d should be an input; if it extracts work, that extraction should be counted as output without subtracting it from the heat input. The efficiency definition should be justified for both signs of W_d.
[Eq. (9); Appendix C] The inequality β_h W_d ≥ ΔI, which drives η_d ≥ η_C, is proved by considering a fictitious isochoric thermalization of the post-feedback distribution back to the hot reservoir at frequency ω_C. This is a free-energy inequality for the working system alone and does not involve the information reservoir. The statement that 'information changes' cause the efficiency to exceed the Carnot value is therefore a subsystem-level result; it does not represent a violation of a global second-law bound. The authors should state this distinction explicitly.

minor comments (4)

[Main text, Eqs. (1)-(5)] The notation S_c for the entropy at points A and B and S_h for the entropy at point C is not defined explicitly in the main text, which is confusing because the subscript 'c' also denotes the cold reservoir; please add a sentence defining these symbols.
[Appendix E, Eq. (E2)] Equation (E2) contains a misplaced parenthesis in 'ωC(⟨nd⟩) − ⟨nh⟩)' which should read 'ωC(⟨nd⟩ − ⟨nh⟩)'; in the same appendix, '⟨nd⟩ ≤ ⟨nd⟩' should be '⟨nd⟩ ≤ ⟨nh⟩' and '⟨nd⟩ > ⟨nd⟩' should be '⟨nd⟩ > ⟨nh⟩'.
[Fig. 2 caption] The caption calls Q_d^c the 'absorbed heat from the cold reservoir,' but in the extended-regime discussion Q_d^c < 0; please reword to avoid sign confusion (e.g., 'heat exchange with the cold reservoir').
[Eq. (5), Appendix A] The last equality of Eq. (5) uses the stochastic entropy-flow formula of Ref. [32]; because that reference is a self-citation, a one-line derivation or statement of the formula would make the paper more self-contained.

Circularity Check

0 steps flagged · score 2.0 of 10

No substantive circularity; the main derivation is self-contained, with only a minor non-load-bearing self-citation and a bookkeeping-dependent 100% efficiency claim.

full rationale

The core derivation is not circular. Eqs. (1)-(9) are closed-form consequences of the model definitions; the bound η_d ≥ η_C follows from Eq. (8) together with the separately proved second-law inequality β_h W_d ≥ ΔI (Appendix C), not from the definition of efficiency. No fitted parameter is relabeled as a prediction. The one self-citation, ref. [32] by Zeng and Wang, appears in Eq. (5)/Appendix A for the stochastic-entropy-flow expression of the cold heat; it is jointly attributed to ref. [47] and the expression is re-derived in Eqs. (A1)-(A5), so it is not load-bearing. The 100% efficiency claim is a bookkeeping identity under the chosen convention η_d = W_tot^d/(Q_h+W_d): when S_ε=S_c, Eq. (5) gives Q_c^d=0 and cyclic energy conservation for the working system forces W_tot^d=Q_d^h, so η_d=1 by substitution. The nontrivial model content is that this operating point is reachable with W_d−ΔI/β_h≥0. Appendix A explicitly flags that the full system is not dissipationless by computing σ=I_C'≠0, so the headline should be read as a statement about the working system with the information reservoir excluded; this is a limitation of interpretation, not a circular step.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The central claim relies on the assumption that information processing is free and that feedback can be represented by an effective temperature. These are not derived from first principles and constitute the main unacknowledged inputs.

free parameters (5)

ω_A (initial frequency) = 1.5
Chosen for the spin-1/2 numerical example in Fig. 2.
ω_C (frequency at point C) = 2
Chosen for the numerical example.
β_c (cold inverse temperature) = 1
Chosen for the numerical example.
measurement error ε = 0.24, 0.28, 0.32 (and ε ≈ 0.183 for 100% efficiency)
Varied in the numerical plots; the 100% efficiency case uses S_ε = S_c, which selects ε ≈ 0.183 for the example parameters.
spin populations n_e, n_g = n_e = 1/2, n_g = -1/2
Used to define the two-level energy spacing.

assumptions (5)

standard math Second law of thermodynamics applied to isochoric thermalization
Used in Appendix C to derive β_h W_d ≥ ΔI ≥ β_d W_d.
standard math Boltzmann distribution and detailed balance at thermal equilibrium
Used to relate populations, frequencies, and temperatures throughout the cycle.
domain assumption Quantum adiabatic theorem: slow driving keeps populations constant
Used for strokes A→B and C'→D'; assumes no non-adiabatic transitions.
domain assumption Measurement and feedback operations have no thermodynamic cost other than the work W_d
Unstated assumption behind the efficiency definition; the demon's memory is not reset or costed, so the cycle is not closed in the information reservoir.
ad hoc to paper Feedback process can be modeled as thermalization with an effective inverse temperature β_d
Introduced in the main text after the control stroke; used to prove the upper bound ΔI ≥ β_d W_d.

invented entities (1)

Information reservoir
purpose: Serves as an external resource that supplies information changes ΔI to the engine, enabling efficiency above the standard Carnot bound.
The paper treats information as a fuel-like reservoir but gives no independent handle or cost for this reservoir; it is a conceptual device without falsifiable predictions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Information-Assisted Carnot Engine Surpasses Standard Thermodynamic Bounds." pith.science (2026). https://pith.science/paper/DO7MBOOL

@misc{pith2026250713412,
  author       = {Pith},
  title        = {Pith review of: Information-Assisted Carnot Engine Surpasses Standard Thermodynamic Bounds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DO7MBOOL}},
  note         = {Machine review of arXiv:2507.13412}
}

abstract

Information can improve heat engine performance, but the underlying principles are still not so clear. Here we introduce a Carnot information machine (CIE) and obtain a quantitative relationship between the engine performance and information. We demonstrate that the presence of information changes allows the CIE to operate as a heat engine in the regime where the standard Carnot cycle is prohibited, ensures that the efficiency of the CIE is greater than or equal to the standard Carnot efficiency, and significantly enables it to achieve 100\% efficiency with positive work extraction for arbitrary two-level systems. We explicitly demonstrate these features using a spin-1/2 system and propose an experimental implementation scheme based on a trapped $^{40}\mathrm{Ca}^+$ ion.

Figures

Figures reproduced from arXiv: 2507.13412 by the authors.

**Figure 2.** FIG. 2: Performance characteristics of the spin-1/2 CIE and [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** FIG. 3: The standard Carnot efficiency, the CIE efficiency [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: FIG. 4: (a) Schematic illustration for discretization of is [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 57 canonical work pages

[1]

Szilard, ¨uber die entropieverminderung in einem ther- modynamischen system bei eingriﬀen intelligenter, We- sen

L. Szilard, ¨uber die entropieverminderung in einem ther- modynamischen system bei eingriﬀen intelligenter, We- sen. Z. Physik 53, 840-856 (1929)

work page 1929
[2]

Sagawa and M

T. Sagawa and M. Ueda, Generalized Jarzynski Equal- ity under Nonequilibrium Feedback Control, Phys. Rev. Lett. 104, 090602 (2010)

work page 2010
[3]

J. M. Horowitz and J. M. R. Parrondo, Designing optimal discrete-feedback thermodynamic engines, New J. Phys. 13, 123019 (2019)

work page 2019
[4]

Yamamoto, S

S. Yamamoto, S. Ito, N. Shiraishi, and T. Sagawa, Lin- ear irreversible thermodynamics and Onsager reciprocity for information-driven engines, Phys. Rev. E 94, 052121 (2016)

work page 2016
[5]

In the ﬁrst subsection, the system releases a net heat ∆ I/β c into the cold reservoir, which compensates for the demon-induced entropy change dur- ing the ﬁrst control stroke

allows us to decompose the isothermal compression stroke D ′ → A into two distinct subsections: the demon- induced compression D ′ → D, and the standard Carnot compression D → A. In the ﬁrst subsection, the system releases a net heat ∆ I/β c into the cold reservoir, which compensates for the demon-induced entropy change dur- ing the ﬁrst control stroke. T...

work page
[6]

Moreover, Figs

This extension stems from two mechanisms: (a) the net work generated by information changes, Wnet, oﬀsets the negative work produced by the standard Carnot cy- cle, thereby enabling the CIE to perform positive work overall; (b) The dissipated heat ∆ I/β c caused by the in- formation changes counterbalance the positive heat Qc, eﬀectively resulting in a po...

work page
[7]

is that when the net work Wnet is positive, the CIE produces greater output work than the standard Carnot engine. Moreover, this reveals the ﬁrst advantage of the CIE: in the regime Sh < S c < S ǫ (Sc < S ǫ ensures Qd c < 0 ), the operation of the standard Carnot engine as a heat engine is forbidden due to Qc > 0, Qh < 0, and Wtot < 0. However, if the net...

work page
[8]

In Appendix C, we have proven that βhWd ≥ ∆I ≥ βdWd

indicates that the condition for the CIE eﬃ- ciency to defeat the standard Carnot eﬃciency is that the work performed by the demon on the system sur- passes the heat required to ﬂow from the hot reservoir to the system to produce an equivalent entropy change as that generated by the work. In Appendix C, we have proven that βhWd ≥ ∆I ≥ βdWd. By incorporati...

work page

Show all 62 references

[9]

The ﬁrst inequality in Eq

will be satisﬁed: ηd = ηup = ηC, reﬂecting the enhancement of CIE eﬃciency through the information changes. The ﬁrst inequality in Eq. ( 9) reveals the second advantage of the CIE: Its eﬃciency is greater than or 4 (a) ǫ = 0.24 (b) ǫ = 0.28 (c) ǫ = 0.32 βh βh βh Wtot Wtot Wtot...
[10]

Buﬀoni, A.Solfanelli, P

L. Buﬀoni, A.Solfanelli, P. Verrucchi, A. Cuccoli, and M. Campisi, Quantum Measurement Cooling, Phys. Rev. Lett. 122, 070603 (2019)

2019
[11]

Z. Y. Lin, S. H. Su, J. Y. Chen, J. C. Chen, and J. F. G. Santos, Suppressing coherence eﬀects in quantum- measurement-based engines, Phys. Rev. A 104, 062210 (2021)

2021
[12]

X. H. Ding, J. Y. Yi, Y. W. Kim, and P. Talkner, Measurement-driven single temperature engine, Phys. 11 Rev. E 98, 042122 (2018)

2018
[13]

J. Y. Yi, P. Talkner, and Y. W. Kim, Single-temperature quantum engine without feedback control, Phys. Rev. E 96, 022108 (2017)

2017
[14]

J. J. Park, K.H. Kim, T. Sagawa, and S. W. Kim, Heat Engine Driven by Purely Quantum Information, Phys. Rev. Lett. 111, 230402 (2013)

2013
[15]

Elouard, D

C. Elouard, D. Herrera-Mart ´ ı, B. Huard, and A. Auﬀ` eves, Extracting work from quantum measurement in Maxwell’s demon engines, Phys. Rev. Lett. 118, 260603 (2017)

2017
[16]

Elouard and A

C. Elouard and A. N. Jordan, Eﬃcient quantum mea- surement engines, Phys. Rev. Lett. 120, 260601 (2018)

2018
[17]

Fadler, A

P. Fadler, A. Friedenberger, and E. Lutz, Eﬃciency at Maximum Power of a Carnot Quantum Information En- gine, Phys. Rev. Lett. 130, 240401 (2023)

2023
[18]

S. Seah, S. Nimmrichter, and V. Scarani, Maxwell’s Lesser Demon: A Quantum Engine Driven by Pointer Measurements, Phys. Rev. Lett. 124, 100603 (2020)

2020
[19]

Chida, S

K. Chida, S. Desai, K. Nishiguchi, A. Fujiwara, Power generator driven by Maxwell’s demon, Nat Commun 8, 15301 (2017)

2017
[20]

J. V. Koski, A. Kutvonen, I. M. Khaymovich, T. Ala- Nissila, and J. P. Pekola, On-Chip Maxwell’s Demon as an Information-Powered Refrigerator, Phys. Rev. Lett. 115, 260602 (2015)

2015
[21]

Bresque, P

L. Bresque, P. A. Camati, S. Rogers, K. Murch, A. N. Jordan, and A. Auﬀ´ eves, Two-Qubit Engine Fueled by Entanglement and Local Measurements, Phys. Rev. Lett. 126, 120605 (2021)

2021
[22]

Mandal, H

D. Mandal, H. T. Quan, and C. Jarzynski, Maxwell’s Re- frigeator: An Exactly Solvable Model, Phys. Rev. Lett. 111, 030602 (2013)

2013
[23]

Mandal and C

D. Mandal and C. Jarzynski, Work and information pro- cessing in a solvable model of Maxwell’s demon, Proc. Natl. Acad. Sci. U. S. A. 109, 11641 (2012)

2012
[24]

T. K. Saha, J. N. E. Lucero, J. Ehrich, D. A. Sivak, and J. Bechhoefer, Maximizing power and velocity of an information engine, Proc. Natl. Acad. Sci. U. S. A. 118, e2023356118 (2021)

2021
[25]

T. K. Saha, J.Ehrich, M. Gavrilov, S. Still, D, A. Sivak, and J. Bechhoefer, Information Engine in a Nonequilib- rium Bath, Phys. Rev. Lett. 131, 057101 (2023)

2023
[26]

J. C. Maxwell, Theory of heat (Longmans London, 1871)
[27]

Maruyama, F

K. Maruyama, F. Nori, and V. Vedral, Colloquium: The physics of Maxwell’s demon and information, Rev. Mod. Phys. 81, 1 (2009)

2009
[28]

Mizraji, The biological Maxwell’s demons: explorin g ideas about the information processing in biological sys- tems

E. Mizraji, The biological Maxwell’s demons: explorin g ideas about the information processing in biological sys- tems. Theory Biosci. 140, 307 (2021)

2021
[29]

Wen, Maxwell’s demon at work: mitochondria, the or- ganelles that convert in formation into energy? Chronic Dis

Y. Wen, Maxwell’s demon at work: mitochondria, the or- ganelles that convert in formation into energy? Chronic Dis. Transl. Med. 4, 135 (2018)

2018
[30]

Schirber, Information Flow in Molecular Machines, Physics 1 7, 162 (2024)

M. Schirber, Information Flow in Molecular Machines, Physics 1 7, 162 (2024)

2024
[31]

Flatt, D

S. Flatt, D. M. Busiello, S. Zamuner, and P. De Los Rios, ABC transporters are billion-year-old Maxwell Demons, Commun Phys 6, 205 (2023)

2023
[32]

Serreli, C

V. Serreli, C. F. Lee, E. R. Kay, and D. A. Leigh, A molecular information ratchet, Nature 445, 523 (2007)

2007
[33]

Sagawa and M

T. Sagawa and M. Ueda, Fluctuation theorem with infor- mation exchange: role of correlations in stochastic ther- modynamics, Phys. Rev. Lett. 109, 180602 (2012)

2012
[34]

Sagawa and M

T. Sagawa and M. Ueda, Second Law of Thermodynamics with Discrete Quantum Feedback Control, Phys. Rev. Lett. 100, 080403 (2008)

2008
[35]

Paneru, S

G. Paneru, S. Dutta, T. Tlusty, and H. K. Pak, Reach- ing and violating thermodynamic uncertainty bounds in information engines, Phys. Rev. E 102, 032126 (2020)

2020
[36]

T. V. Vu and Y. Hasegawa, Uncertainty relation under information measurement and feedback control, J. Phys. A: Math. Theor. 53, 075001 (2020)

2020
[37]

Q. Zeng, J. Wang, New ﬂuctuation theorems on Maxwell’s demon, Sci. Adv. 7, 23 (2021)

2021
[38]

B. J. Lopez, N. J. Kuwada, E. M. Craig, B. R. Long, and H. Linke, Realization of a feedback controlled ﬂashing ratchet, Phys. Rev. Lett. 101, 220601 (2008)

2008
[39]

Toyabe, T

S. Toyabe, T. Sagawa, M. Ueda, E. Muneyuki, and M. Sano, Experimental demonstration of information- to-energy conversion and validation of the generalized Jarzynski equality, Nat. Phys. 6, 988 (2010)

2010
[40]

Paneru, D

G. Paneru, D. Y. Lee, T. Tlusty, and H. K. Pak, Loss- less Brownian information engine, Phys. Rev. Lett. 120, 020601 (2018)

2018
[41]

Y. Jun, M. Gavrilov, and J. Bechhoefer, High-Precision Test of Landauer’s Principle in a Feedback Trap, Phys. Rev. Lett. 113, 190601 (2014)

2014
[42]

J. V. Koski, V. F. Maisi, J. P. Pekola, and D. V. Averin, Experimental realization of a Szilard engine with a sin- gle electron, Proc. Natl. Acad. Sci. U. S. A. 111, 13786 (2014)

2014
[43]

J. V. Koski, V. F. Maisi, T. Sagawa, and J. P. Pekola, Experimental observation of the role of mutual informa- tion in the nonequilibrium dynamics of a Maxwell demon, Phys. Rev. Lett. 113, 030601 (2014)

2014
[44]

M. D. Vidrighin, O. Dahlsten, M. Barbieri, M. S. Kim, V. Vedral, and I. A. Walmsley, Photonic Maxwell’s Demon, Phys. Rev. Lett. 116, 050401 (2016)

2016
[45]

Cottet, S

N. Cottet, S. Jezouin, L. Bretheau, P. Campagne-Ibarcq , Q. Ficheux, J. Anders, A. Auﬀ` eves, R. Azouit, P. Rou- chon, and B. Huard, Observing a quantum Maxwell de- mon at work, Proc. Natl. Acad. Sci. U. S. A. 114, 7561 (2017)

2017
[46]

Carnot, in Reﬂections on the Motive Power of Heat and on Machines Fitted to Develop that Power, edited by R

S. Carnot, in Reﬂections on the Motive Power of Heat and on Machines Fitted to Develop that Power, edited by R. H. Thurston (Wiley, New York, 1890)
[47]

H. B. Callen, Thermodynamics and an Introduction to Thermostatistics (John Wiley & Sons, New York, 1985)

1985
[48]

Deﬀner and C

S. Deﬀner and C. Jarzynski, Information Processing and the Second Law of Thermodynamics: An Inclusive, Hamiltonian Approach, Phys. Rev. X, 3, 041003 (2013)

2013
[49]

Rana and A

S. Rana and A. M. Jayannavar, A multipurpose informa- tion engine that can go beyond the Carnot limit, J. Stat. Mech. (2016) 103207

2016
[50]

H. T. Quan, Y. X. Liu, C. P. Sun, and F. Nori, Quantum thermodynamic cycles and quantum heat engines, Phys. Rev. E 75, 031105 (2007)

2007
[51]

At point C ′ , the system state xC′ is determined by the system state xC and the measurement outcome y. The post-control state xC′ can be expressed as a function of the state xC and measurement outcome y following the relations: xC′ (xC = e, y = e) = g, xC′ (xC = e, y = g) = e...
[52]

Schreiber, Measuring information transfer

T. Schreiber, Measuring information transfer. Phys. R ev. Lett. 85, 461 (2000)

2000
[53]

W. E. Hou, X. Y. Zhao, K. Rehan, Y. Li, Y. Li, E. Lutz, Y. H. Lin, and J. F. Du, An energy eﬃcient quantum- enhanced machine, arXiv:2404.15075

arXiv
[54]

O. V. Ivakhnenko, S. N. Shevchenko, and F. Nori, Nonadiabatic Landau-Zener-St¨ uckelberg-Majorana tran- sitions, dynamics, and interference, Phys. Rep. 995, 1 (2023)

2023
[55]

R. J. de Assis, T. M. de Mendonca, C. J. Villas-Boas, A. M. de Souza, R. S. Sarthour, I. S. Oliveira, and N. G. de Almeida, Eﬃciency of a quantum Otto heat engine operating under a reservoir at eﬀective negative temper- atures, Phys. Rev. Lett. 122, 240602 (2019)

2019
[56]

H. T. Quan, S. Yang, and C. P. Sun, Microscopic work distribution of small systems in quantum isothermal pro- cesses and the minimal work principle, Phys. Rev. E 78, 021116 (2008)

2008
[57]

Geva and R

E. Geva and R. Kosloﬀ, A quantum-mechanical heat en- gine operating in ﬁnite time. A model consisting of spin- 1/2 systems as the working ﬂuid, J. Chem. Phys. 96, 3054 (1992)

1992
[58]

T. M. Cover, J. A. Thomas, Elements of information theory (Wiley, ed. 2, 2006)

2006
[59]

Niedenzu, D

W. Niedenzu, D. Gelbwaser-Klimovsky, A. G. Kofman, and G. kurizki, On the operation of machines powered by quantum non-thermal baths, New J. Phys. 18, 083012 (2016)

2016
[60]

Niedenzu, V

W. Niedenzu, V. Mukherjee, A. Ghosh, A. G. Kofman, and G. Kurizki, Quantum engine eﬃciency bound beyond the second law of thermodynamics, Nat. Commun. 9, 165 (2018)

2018
[61]

J. P. S. Peterson, T. B. Batalh˜ ao, M. Herrera, A. M. Souza, R. S. Sarthour, I. S. Oliveira, and R. M. Serra, Experimental characterization of a spin quantum heat engine, Phys. Rev. Lett. 123, 240601 (2019)

2019
[62]

Yan, J.-W

L.-L. Yan, J.-W. Zhang, M.-R. Yun, J.-C. Li, G.-Y. Ding, J.-F. Wei, J.-T. Bu, B. Wang, L. Chen, S.-L. Su et al., Experimental veriﬁcation of dissipation-time uncertaint y relation, Phys. Rev. Lett. 128, 050603 (2022)

2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Szilard, ¨uber die entropieverminderung in einem ther- modynamischen system bei eingriﬀen intelligenter, We- sen

L. Szilard, ¨uber die entropieverminderung in einem ther- modynamischen system bei eingriﬀen intelligenter, We- sen. Z. Physik 53, 840-856 (1929)

work page 1929

[2] [2]

Sagawa and M

T. Sagawa and M. Ueda, Generalized Jarzynski Equal- ity under Nonequilibrium Feedback Control, Phys. Rev. Lett. 104, 090602 (2010)

work page 2010

[3] [3]

J. M. Horowitz and J. M. R. Parrondo, Designing optimal discrete-feedback thermodynamic engines, New J. Phys. 13, 123019 (2019)

work page 2019

[4] [4]

Yamamoto, S

S. Yamamoto, S. Ito, N. Shiraishi, and T. Sagawa, Lin- ear irreversible thermodynamics and Onsager reciprocity for information-driven engines, Phys. Rev. E 94, 052121 (2016)

work page 2016

[5] [5]

In the ﬁrst subsection, the system releases a net heat ∆ I/β c into the cold reservoir, which compensates for the demon-induced entropy change dur- ing the ﬁrst control stroke

allows us to decompose the isothermal compression stroke D ′ → A into two distinct subsections: the demon- induced compression D ′ → D, and the standard Carnot compression D → A. In the ﬁrst subsection, the system releases a net heat ∆ I/β c into the cold reservoir, which compensates for the demon-induced entropy change dur- ing the ﬁrst control stroke. T...

work page

[6] [6]

Moreover, Figs

This extension stems from two mechanisms: (a) the net work generated by information changes, Wnet, oﬀsets the negative work produced by the standard Carnot cy- cle, thereby enabling the CIE to perform positive work overall; (b) The dissipated heat ∆ I/β c caused by the in- formation changes counterbalance the positive heat Qc, eﬀectively resulting in a po...

work page

[7] [7]

is that when the net work Wnet is positive, the CIE produces greater output work than the standard Carnot engine. Moreover, this reveals the ﬁrst advantage of the CIE: in the regime Sh < S c < S ǫ (Sc < S ǫ ensures Qd c < 0 ), the operation of the standard Carnot engine as a heat engine is forbidden due to Qc > 0, Qh < 0, and Wtot < 0. However, if the net...

work page

[8] [8]

In Appendix C, we have proven that βhWd ≥ ∆I ≥ βdWd

indicates that the condition for the CIE eﬃ- ciency to defeat the standard Carnot eﬃciency is that the work performed by the demon on the system sur- passes the heat required to ﬂow from the hot reservoir to the system to produce an equivalent entropy change as that generated by the work. In Appendix C, we have proven that βhWd ≥ ∆I ≥ βdWd. By incorporati...

work page