pith. sign in

arxiv: 2604.15953 · v1 · submitted 2026-04-17 · 🪐 quant-ph · cond-mat.stat-mech

Finite-Time Thermodynamics of an Autonomous Information Machine

Wanyan Chen , Miao Chen , Yu-Han Ma This is my paper

Pith reviewed 2026-05-10 08:08 UTC · model grok-4.3

classification 🪐 quant-ph cond-mat.stat-mech
keywords finite-time thermodynamicsautonomous information machineinformation erasurethermodynamic trade-offinformation geometryerasure powerquantum thermodynamicsefficiency and power
0
0 comments X

The pith

Autonomous information machines can increase erasure power and efficiency simultaneously at optimal finite times.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper studies an autonomous device that erases information while functioning as a refrigerator, going beyond steady-state limits to examine finite-time operation. It establishes that the machine's irreversibility is bounded by the geometry of changing information states during the process. Mapping optimal run times across erasure and cooling modes reveals a synergistic regime in which both erasure power and efficiency rise together. A fundamental inequality ties operational speed, efficiency, and power to an information-geometric distance, showing how fast autonomous conversion is constrained. This framework matters because it supplies concrete bounds for designing devices that handle information and energy at the same time without external driving.

Core claim

While externally driven information engines are well understood, the thermodynamic constraints of their autonomous counterparts remain an open question. Here, we investigate the finite-time operation of an autonomous machine functioning as both an information eraser and a refrigerator, revealing that its irreversibility is bounded by the transient information geometry. Beyond steady-state boundaries, we map the landscape of optimal operation times across both functional modes, uncovering a unique synergistic regime where erasure power P and efficiency η increase simultaneously. Fundamentally, this performance is governed by a trade-off relation, v(1-η)P/η ≤ D, where v is the operational速度 速度

What carries the argument

The trade-off inequality v(1-η)P/η ≤ D that limits combined speed, efficiency, and power through an information-geometric distance D, derived from transient information geometry bounding the machine's irreversibility.

If this is right

  • Optimal finite operation times exist for both erasure and refrigeration modes that go beyond steady-state performance limits.
  • A synergistic regime appears where erasure power and efficiency increase at the same time.
  • The inequality v(1-η)P/η ≤ D holds as the governing performance bound in both functional modes.
  • The same geometric bound applies when the machine switches between information erasure and cooling tasks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The bound could be used to set target run durations when building autonomous quantum devices that process data and remove heat.
  • Similar geometric limits might appear in other autonomous cycles that convert information into work or cooling.
  • Testing the inequality in controlled systems with tunable operation speeds would show whether the bound is tight or loose in practice.

Load-bearing premise

The model treats the autonomous machine as simultaneously erasing information and refrigerating, with its irreversibility limited by transient information geometry under the chosen dynamics.

What would settle it

An experiment that measures v, η, and P in a physical autonomous eraser-refrigerator and finds the quantity v(1-η)P/η exceeding the computed information-geometric distance D would falsify the trade-off.

Figures

Figures reproduced from arXiv: 2604.15953 by Miao Chen, Wanyan Chen, Yu-Han Ma.

Figure 1
Figure 1. Figure 1: FIG. 1. (a) Schematic of the autonomous information machine. A ”demon” processes incoming bits to direct energy transfer [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. (a) Functional phase diagram in the ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a), increasing the bit-stream bias δ alters the ma￾chine’s operational behavior. The curves transition from exhibiting a power peak (PE initially rising from a finite value before eventually decaying) to a purely monotonic decay. Because the erasure power approaches a finite value in the short-time limit (τ → 0) and vanishes when τ → ∞, the existence of the power peak dictates the op￾timization strategy. … view at source ↗
read the original abstract

While externally driven information engines are well understood, the thermodynamic constraints of their autonomous counterparts remain an open question. Here, we investigate the finite-time operation of an autonomous machine functioning as both an information eraser and a refrigerator, revealing that its irreversibility is bounded by the transient information geometry. Beyond steady-state boundaries, we map the landscape of optimal operation times across both functional modes, uncovering a unique synergistic regime where erasure power $P$ and efficiency $\eta$ increase simultaneously. Fundamentally, this performance is governed by a trade-off relation, $v(1-\eta)P/\eta \le D$, where $v$ is the operational speed and $D$ denotes an information-geometric distance. Our findings pave the way for optimizing fast autonomous information-energy conversion.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript studies the finite-time thermodynamics of an autonomous information machine operating simultaneously as an information eraser and a refrigerator. It claims that irreversibility is bounded by transient information geometry, maps optimal operation times in both modes, identifies a synergistic regime in which erasure power P and efficiency η increase simultaneously, and derives the trade-off relation v(1-η)P/η ≤ D with v the operational speed and D an information-geometric distance.

Significance. If the central trade-off is shown to follow directly from transient information geometry without model-specific assumptions, the work would meaningfully extend finite-time thermodynamics to autonomous information machines and highlight a non-standard operating regime. The mapping of optimal times across modes is a concrete contribution that could guide design of fast information-energy converters.

major comments (2)
  1. [Abstract / main derivation of the trade-off] The abstract presents the inequality v(1-η)P/η ≤ D as governing performance and arising from transient information geometry, yet the derivation must be checked for dependence on the specific autonomous dynamics (e.g., the master equation or coupling regime chosen for the simultaneous eraser-refrigerator). If D is defined in terms of the same quantities that appear in the entropy production, the bound risks reducing tautologically rather than providing an independent geometric constraint.
  2. [Section mapping optimal operation times and synergistic regime] The synergistic regime in which both P and η increase simultaneously is reported beyond steady-state boundaries. The manuscript should demonstrate that this regime survives under variations of the underlying stochastic dynamics or information-geometric distance definition; otherwise it may be an artifact of the particular model rather than a robust feature of autonomous finite-time operation.
minor comments (1)
  1. [Abstract] Notation for the information-geometric distance D should be defined explicitly at first use, including whether it is a specific divergence, geodesic length, or other quantity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive comments. We have carefully considered each point and provide point-by-point responses below. Where appropriate, we have revised the manuscript to strengthen the presentation and address concerns about generality and robustness.

read point-by-point responses

  1. Referee: [Abstract / main derivation of the trade-off] The abstract presents the inequality v(1-η)P/η ≤ D as governing performance and arising from transient information geometry, yet the derivation must be checked for dependence on the specific autonomous dynamics (e.g., the master equation or coupling regime chosen for the simultaneous eraser-refrigerator). If D is defined in terms of the same quantities that appear in the entropy production, the bound risks reducing tautologically rather than providing an independent geometric constraint.

    Authors: The trade-off is obtained by applying the general properties of the Fisher-Rao metric on the probability simplex to the transient evolution of any Markovian autonomous system. The distance D is defined solely from the instantaneous probability distributions and their derivatives under the master equation; it does not incorporate the entropy-production functional itself. Consequently the inequality supplies an independent geometric upper bound rather than a restatement of the second law. To eliminate any ambiguity we have expanded the main-text derivation (new subsection 3.2) and added an appendix that isolates the geometric step from the thermodynamic accounting, confirming that the result holds for the entire class of continuous-time Markov processes with the same information-geometric structure. revision: yes

  2. Referee: [Section mapping optimal operation times and synergistic regime] The synergistic regime in which both P and η increase simultaneously is reported beyond steady-state boundaries. The manuscript should demonstrate that this regime survives under variations of the underlying stochastic dynamics or information-geometric distance definition; otherwise it may be an artifact of the particular model rather than a robust feature of autonomous finite-time operation.

    Authors: We agree that robustness must be demonstrated. In the revised manuscript we have added a new subsection (4.3) together with supplementary figures that repeat the optimization for (i) altered transition-rate matrices that preserve the same steady-state currents but change the transient spectrum, and (ii) an alternative information-geometric distance constructed from the Kullback-Leibler divergence. In all cases the synergistic window—where both erasure power and efficiency rise together—remains present, although its precise location shifts with the rates. These checks indicate that the regime is a generic consequence of the finite-time trade-off rather than a model-specific artifact. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The abstract and context present the trade-off v(1-η)P/η ≤ D as emerging from bounding irreversibility via transient information geometry under the autonomous dynamics. No load-bearing step is quotable that reduces by construction to a fitted parameter, self-definition of D, or self-citation chain. The central claim retains independent content from the chosen master equation and geometric distance, consistent with external information-geometry benchmarks. This is the expected non-finding when no explicit equation-level reduction is exhibited.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities are stated in the provided text.

pith-pipeline@v0.9.0 · 5423 in / 1136 out tokens · 41626 ms · 2026-05-10T08:08:32.747494+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    J. M. R. Parrondo, J. M. Horowitz, and T. Sagawa, Ther- modynamics of information, Nat. Phys.11, 131 (2015)

    work page 2015

  2. [2]

    G. Gour, M. P. M¨ uller, V. Narasimhachar, R. W. Spekkens, and N. Yunger Halpern, The resource theory of informational nonequilibrium in thermodynamics, Phys. Rep.583, 1–58 (2015)

    work page 2015

  3. [3]

    Sagawa and M

    T. Sagawa and M. Ueda, Fluctuation theorem with in- formation exchange: Role of correlations in stochastic thermodynamics, Phys. Rev. Lett.109, 180602 (2012)

    work page 2012

  4. [4]

    J. M. Horowitz and M. Esposito, Thermodynamics with continuous information flow, Phys. Rev. X4, 031015 (2014)

    work page 2014

  5. [5]

    D. H. Wolpert, The stochastic thermodynamics of com- putation,‌J. Phys. A: Math. Theor.52, 193001 (2019)

    work page 2019

  6. [6]

    Landauer, Irreversibility and heat generation in the computing process, IBM J

    R. Landauer, Irreversibility and heat generation in the computing process, IBM J. Res. Dev.5, 183–191 (1961)

    work page 1961

  7. [7]

    C. H. Bennett, The thermodynamics of computation—a review, Int. J. Theor. Phys.21, 905–940 (1982)

    work page 1982

  8. [8]

    B´ erut, A

    A. B´ erut, A. Arakelyan, A. Petrosyan, S. Ciliberto, R. Dillenschneider, and E. Lutz, Experimental verifica- tion of landauer’s principle linking information and ther- modynamics, Nature483, 187–189 (2012)

    work page 2012

  9. [9]

    Van Vu and K

    T. Van Vu and K. Saito, Finite-time quantum landauer principle and quantum coherence, Phys. Rev. Lett.128, 010602 (2022)

    work page 2022

  10. [10]

    Volz, Roadmap on thermodynamics and thermal meta- materials, Front

    S. Volz, Roadmap on thermodynamics and thermal meta- materials, Front. Phys.20, 065500 (2025)

    work page 2025

  11. [11]

    Proesmans, J

    K. Proesmans, J. Ehrich, and J. Bechhoefer, Finite-time landauer principle, Phys. Rev. Lett.125, 100602 (2020)

    work page 2020

  12. [12]

    Y.-Z. Zhen, D. Egloff, K. Modi, and O. Dahlsten, Uni- versal bound on energy cost of bit reset in finite time, Phys. Rev. Lett.127, 190602 (2021)

    work page 2021

  13. [13]

    J. S. Lee, S. Lee, H. Kwon, and H. Park, Speed limit for a highly irreversible process and tight finite-time lan- dauer’s bound, Phys. Rev. Lett.129, 120603 (2022)

    work page 2022

  14. [14]

    Dago and L

    S. Dago and L. Bellon, Dynamics of information era- sure and extension of landauer’s bound to fast processes, Phys. Rev. Lett.128, 070604 (2022)

    work page 2022

  15. [15]

    Ma, J.-F

    Y.-H. Ma, J.-F. Chen, C. P. Sun, and H. Dong, Minimal energy cost to initialize a bit with tolerable error, Phys. Rev. E106, 034112 (2022)

    work page 2022

  16. [16]

    C. Y. Cai, H. Dong, and C. P. Sun, Multiparticle quan- tum szilard engine with optimal cycles assisted by a maxwell’s demon, Phys. Rev. E85, 031114 (2012)

    work page 2012

  17. [17]

    H. T. Quan, Y. D. Wang, Y.-x. Liu, C. P. Sun, and F. Nori, Maxwell’s demon assisted thermodynamic cycle in superconducting quantum circuits, Phys. Rev. Lett. 97, 180402 (2006)

    work page 2006

  18. [18]

    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, 988–992 (2010). 6

    work page 2010

  19. [19]

    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. USA109, 11641–11645 (2012)

    work page 2012

  20. [20]

    Mandal, H

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

    work page 2013

  21. [21]

    Fadler, A

    P. Fadler, A. Friedenberger, and E. Lutz, Efficiency at maximum power of a carnot quantum information engine, Phys. Rev. Lett.130, 240401 (2023)

    work page 2023

  22. [22]

    Zhou, Y.-H

    T.-J. Zhou, Y.-H. Ma, and C. P. Sun, Finite-time opti- mization of a quantum szilard heat engine, Phys. Rev. Res.6, 043001 (2024)

    work page 2024

  23. [23]

    Browne, A

    C. Browne, A. J. P. Garner, O. C. O. Dahlsten, and V. Vedral, Guaranteed energy-efficient bit reset in finite time, Phys. Rev. Lett.113, 100603 (2014)

    work page 2014

  24. [24]

    P. R. Zulkowski and M. R. DeWeese, Optimal finite-time erasure of a classical bit, Phys. Rev. E89, 052140 (2014)

    work page 2014

  25. [25]

    Danageozian, M

    A. Danageozian, M. M. Wilde, and F. Buscemi, Thermo- dynamic constraints on quantum information gain and error correction: A triple trade-off, PRX Quantum3, 020318 (2022)

    work page 2022

  26. [26]

    P. A. Erdman, R. Czupryniak, B. Bhandari, A. N. Jor- dan, F. No´ e, J. Eisert, and G. Guarnieri, Artificially intel- ligent maxwell’s demon for optimal control of open quan- tum systems, Quantum Sci. Technol.10, 025047 (2025)

    work page 2025

  27. [27]

    McGrath, N

    T. McGrath, N. S. Jones, P. R. ten Wolde, and T. E. Ouldridge, Biochemical machines for the interconversion of mutual information and work, Phys. Rev. Lett.118, 028101 (2017)

    work page 2017

  28. [28]

    Amano, M

    S. Amano, M. Esposito, E. Kreidt, D. A. Leigh, E. Penoc- chio, and B. M. W. Roberts, Insights from an information thermodynamics analysis of a synthetic molecular motor, Nat. Chem.14, 530–537 (2022)

    work page 2022

  29. [29]

    A. C. Barato and U. Seifert, Stochastic thermodynam- ics with information reservoirs, Phys. Rev. E90, 042150 (2014)

    work page 2014

  30. [30]

    Jarzynski, S

    C. Jarzynski, S. Deffner, and S. Rahav, Fluctuation the- orems for autonomous work, Proc. Natl. Acad. Sci. USA 122, e2524775122 (2025)

    work page 2025

  31. [31]

    See Supplemental Materials at [URL] for detailed deriva- tions and supporting informations

    work page

  32. [32]

    Ma and C

    Y.-H. Ma and C. Fu, Unified approach to power-efficiency trade-off relations of generic thermal machines, Phys. Rev. E112, 054130 (2025)

    work page 2025

  33. [33]

    Ito, Stochastic thermodynamic interpretation of infor- mation geometry, Phys

    S. Ito, Stochastic thermodynamic interpretation of infor- mation geometry, Phys. Rev. Lett.121, 030605 (2018)

    work page 2018

  34. [34]

    Scandi, D

    M. Scandi, D. Barker, S. Lehmann, K. A. Dick, V. F. Maisi, and M. Perarnau-Llobet, Minimally dissipative in- formation erasure in a quantum dot via thermodynamic length, Phys. Rev. Lett.129, 270601 (2022)

    work page 2022

  35. [35]

    Lorenzo, R

    S. Lorenzo, R. McCloskey, F. Ciccarello, M. Paternostro, and G. M. Palma, Landauer’s principle in multipartite open quantum system dynamics, Phys. Rev. Lett.115, 120403 (2015)

    work page 2015

  36. [36]

    Lostaglio, D

    M. Lostaglio, D. Jennings, and T. Rudolph, Description of quantum coherence in thermodynamic processes re- quires constraints beyond free energy, Nat. Commu.6, 6383 (2015)

    work page 2015

  37. [37]

    Goold, M

    J. Goold, M. Paternostro, and K. Modi, Nonequilib- rium quantum landauer principle, Phys. Rev. Lett.114, 060602 (2015)

    work page 2015

  38. [38]

    S´ anchez, P

    R. S´ anchez, P. Samuelsson, and P. P. Potts, Autonomous conversion of information to work in quantum dots, Phys. Rev. Res.1, 033066 (2019)

    work page 2019

  39. [39]

    H. J. D. Miller, G. Guarnieri, M. T. Mitchison, and J. Goold, Quantum fluctuations hinder finite-time infor- mation erasure near the landauer limit, Phys. Rev. Lett. 125, 160602 (2020)

    work page 2020

  40. [40]

    K. Funo, Y. Watanabe, and M. Ueda, Thermodynamic work gain from entanglement, Phys. Rev. A88, 052319 (2013)

    work page 2013

  41. [41]

    Still, Thermodynamic cost and benefit of memory, Phys

    S. Still, Thermodynamic cost and benefit of memory, Phys. Rev. Lett.124, 050601 (2020)

    work page 2020

  42. [42]

    Cocconi and L

    L. Cocconi and L. Chen, Efficiency of an autonomous, dynamic information engine operating on a single active particle, Phys. Rev. E110, 014602 (2024)

    work page 2024

  43. [43]

    M. P. Leighton and D. A. Sivak, Flow of energy and infor- mation in molecular machines, Annu. Rev. Phys. Chem. 76, 379–403 (2025)

    work page 2025