Pith. sign in

REVIEW 2 major objections 5 minor 45 references

Splitting of nonequilibrium phase transitions in driven Ising models

T0 review · 2 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read In a three-state Ising model driven by two opposite thermal baths, the order-disorder transition splits into two distinct transition points.

desk verdict The splitting of the transition into two distinct points is a real and interesting effect, but the paper's claim that one of them is a continuous critical point doesn't survive a look at the paper's own linear stability analysis. read the letter →

arxiv 2412.09343 v1 pith:DKIVTC4V submitted 2024-12-12 cond-mat.stat-mech cond-mat.mes-hallcond-mat.othercond-mat.soft

Gustavo A. L. Forão , Fernando S. Filho , André P. Vieira , Bart Cleuren , Daniel M. Busiello , Carlos E. Fiore This is my paper · ORCID
classification cond-mat.stat-mechcond-mat.mes-hallcond-mat.othercond-mat.soft PACS 05.70.Ln05.50.+q05.70.Fh
keywords nonequilibriumphasetransitiondrivenIsingmodeltwothermalbathsentropyproductionpowerfluctuationscriticalexponentcollectiveheatengine
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

The paper introduces a three-state Ising model in which each spin is coupled to a hot and a cold bath that drive transitions in opposite directions. It claims that, unlike equilibrium order-disorder transitions, the critical point splits: the phase dominated by −1 spins orders continuously at ε1c, while the phase dominated by +1 spins orders discontinuously at a different point, ε2c. The continuous transition has an unusual mean-field exponent β=1, and the entropy production, power, and power fluctuations all scale with the same exponent δ=2. The two ordered phases also differ thermodynamically, with the less dissipative phase supporting heat-engine operation near maximum power and efficiency. If correct, the work shows that coupling to multiple driving baths can create new universality classes of nonequilibrium phase transitions.

What carries the argument

The load-bearing object is the Landau-type expansion of the order-parameter dynamics, dm/dt ≈ a(ε−ε1c)m + b m² + c m³, obtained from the exact master equations for the densities n−, n0, n+ in the all-to-all limit. The coefficient b ∝ sinh[F(β1−β2)/4] vanishes only when the two baths are identical; its presence makes the transition at ε1c continuous with β=1 and shifts the critical point for the other ordered phase, producing ε2c≠ε1c. The same expansion is used to derive the scaling of entropy production, power fluctuations, and efficiency near ε1c.

What would settle it

Solve the exact all-to-all steady-state equations at F=2, β1=2, β2=1, and plot the stable order parameter m against ε just below ε1c; if the data follow (ε1c−ε)^1/2 rather than (ε1c−ε), or if a stable m=0 solution coexists with the m>0 branch over a finite interval, the transition at ε1c is not the continuous β=1 transition claimed. A Binder cumulant crossing would also distinguish the two cases.

Watch

Extended reading notes

Core claim

The central discovery is that the interplay of two thermal baths with opposite nonconservative driving forces splits the order-disorder transition of a minimal Ising model into two separate transition points, one for each ordered phase. In the all-to-all mean-field limit, which is solved exactly, the order parameter m obeys dm/dt ≈ a(ε−ε1c)m + b m² + c m³ near the first transition. Because b is nonzero whenever the baths are asymmetric (F≠0 and β1≠β2), the m² term breaks the up-down symmetry and forces m ∼ (ε1c−ε), i.e., β=1 instead of the standard β=1/2. The transition at the second point, ε2c, is discontinuous and has a spinodal region where final states depend on initial conditions. Near ε1c, entropy production, power variance, and efficiency all exhibit the same scaling exponent δ=2β=2. The same qualitative behavior is found numerically for nearest-neighbor couplings on a square lattice.

Load-bearing premise

The paper relies on the assumption that the expansion dm/dt ≈ a(ε−ε1c)m + b m² + c m³ faithfully describes the full dynamics near the 'up' transition, meaning that ε1c is a genuine continuous critical point rather than a spinodal limit of metastability.

Editorial extensions

If this is right

  • The order-disorder transition in this nonequilibrium Ising model is characterized by two distinct transition points, ε1c and ε2c, rather than a single critical point.
  • The continuous transition at ε1c has mean-field exponent β=1 and thermodynamic scaling exponent δ=2, both distinct from the equilibrium values β=1/2 and δ=1.
  • The two ordered phases have different thermodynamic signatures: phase A has lower entropy production and power fluctuations than phase B.
  • Heat-engine operation occurs only in phase A, whose maximum-power and maximum-efficiency operating points converge to the discontinuous transition point as the temperature asymmetry increases.
  • These results hold beyond the all-to-all mean-field limit, as shown by square-lattice simulations.

Reading between the lines

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

  • If ε1c turns out to be a spinodal rather than a true critical point, the claimed exponent β=1 would characterize the limit of metastability rather than a critical point, which would change the universality statement; this distinction is testable with a Binder-cumulant analysis.
  • The mechanism of a nonzero m² term from opposing baths is generic, so similar split transitions and phase-dependent thermodynamic advantages might appear in driven chemical reaction networks or active matter models.
  • The convergence of optimal power and efficiency conditions to the discontinuous transition point suggests that collective heat engines could generally exploit proximity to a nonequilibrium transition for performance, without requiring engineered control protocols.
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.

Referee Report

2 major / 5 minor

Summary. The paper studies a three-state Ising model in which spins are coupled to two thermal baths with opposite driving forces, and it claims that the order-disorder transition splits into two distinct transition points, one for each ordered phase. For the all-to-all interaction case, the authors derive an effective Landau-type equation for the magnetization near the symmetric state and identify the m>0 transition as a continuous critical point with exponent β=1, while the m<0 transition is said to be discontinuous and to feature a spinodal region. From the same expansion they obtain scaling laws δ=2 for entropy production, power fluctuations, and efficiency, and they argue that the m>0 phase is less dissipative and supports efficient heat-engine operation. The same scenario is asserted to hold for a square lattice on the basis of Gillespie simulations with N=102.

Significance. The claimed splitting of a single order-disorder transition into two phase-transition points with different classifications and distinct thermodynamic scaling would be a novel nonequilibrium phenomenon and would be of interest to the statistical-physics and stochastic-thermodynamics communities. The all-to-all analysis is a genuine strength: the master equation is stated, the expansion coefficients are provided in the Supplemental Material, and no parameter is fitted to data, making the mean-field results internally checkable. The prediction that phase A is less dissipative and has lower power fluctuations than phase B is falsifiable by simulation and experiment. However, the central conceptual claim—that ε1c is a continuous critical point with a true exponent β=1—is undermined by the linear stability of Eq. (2), as detailed below.

major comments (2)
  1. [Emergence of phase transitions..., Eq. (2)] Equation (2) is not consistent with the claim that ε1c is a continuous critical point. The text states a>0 and the ordered phase corresponds to ε<ε1c; hence for all ε<ε1c the linear coefficient a(ε−ε1c) is negative, so m=0 is a linearly stable fixed point on the ordered side of the transition. In a Landau equation of the form dm/dt = L m + b m^2 + c m^3 with L<0, the nonzero branch that vanishes linearly at L=0, m* ≈ −L/b, has linearized growth rate f'(m*) ≈ −L > 0, so it is an unstable (separatrix) branch, not the stable order parameter of phase A. The stable ordered branch, if present, must terminate instead at a saddle-node spinodal point. Thus ε1c is at best a transcritical bifurcation or limit of metastability of the m>0 branch, and the exponent β=1 is not a critical exponent for the stable order parameter. The subsequent scaling laws for entropy production, power fluctuations, and efficiency (δ=2β=2) are derived from this same expansion and inherit the same mischaracterization. This also contradicts the paper's own distinction between a critical transition and a spinodal region: the coexistence of a stable disordered state and an ordered branch on the same side of ε1c is exactly the kind of spinodal behavior the text reserves for ε2c. The central claim of a continuous critical transition at ε1c therefore needs to be reanalyzed; if the stability calculation is correct, the two transitions are both spinodal-like.
  2. [Emergence of phase transitions..., Figs. 1e, 2a-b, 3] The square-lattice results, which are used to claim robustness beyond the all-to-all case, are based on Gillespie simulations for N=102 with no error bars and no finite-size scaling analysis. A single small lattice size cannot support the quantitative claims that the two-transition scenario, the scaling exponents, and the thermodynamic asymmetries persist for short-range interactions. In particular, the caption of Fig. 1(e) itself notes a finite-size jump from m≈1 to −1, which signals strong finite-size effects. Please provide error bars, at least two or three system sizes, and a finite-size-scaling or extrapolation analysis, or explicitly restrict the claims to the all-to-all model.
minor comments (5)
  1. [Thermodynamics section] In the sentence defining ⟨σc⟩, the phrase "for ϵ1≥ ϵ1c" appears to be a typo for "ε ≥ ε1c".
  2. [Scaling of power fluctuations] The expression "γ(A)_P −γ2(c)_P" should read "γ(A)_P − γ_P^{(c)}" for consistency with the notation introduced for the critical value of the variance.
  3. [Eq. for ε1c in the main text] The displayed formula for ε1c is typeset with a leading negative sign and a denominator that are difficult to parse; please check the formatting so that readers can verify the sign and value.
  4. [Supplemental Material, coefficient c] The expression for the coefficient c in Eq. (B2) contains several unbalanced parentheses and repeated opening brackets, which make the formula effectively unreadable and prevent verification; please rewrite it cleanly, possibly with intermediate definitions.
  5. [Fig. 1(e) caption] The finite-size jump from m≈1 to −1 is not explained; a brief statement about the simulation protocol (initial conditions, number of runs) would help the reader assess the reliability of the square-lattice data.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: all expansion coefficients are derived analytically from the model, and the square-lattice simulations are independent checks rather than fitted inputs.

full rationale

The paper's central derivation is self-contained. The normal-form expansion dm/dt ≈ a(ε−ε1c)m + b m^2 + c m^3 in Eq. (2) is obtained by expanding the exact all-to-all master equation around the disordered state m=0, with explicit analytic expressions for a, b, and c given in the main text and Supplemental Material. No parameter is fitted to simulation data; the square-lattice Gillespie results are used only as independent numerical verification. The scaling laws for entropy production, power fluctuations, and efficiency follow from the same analytically derived expansion and are not adjusted to match numerics. Self-citations to Refs. [7,8,15] are methodological and do not carry the central claim. The possible concern that ε1c may be a spinodal rather than a true critical point is a physical-interpretation or correctness issue, not a circularity: it does not reduce the claimed prediction to an input by construction. Therefore no circular step is identified.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard master-equation modeling, a mean-field limit, a smoothness assumption for the expansion, and a strong-coupling approximation for the closed-form thermodynamic expressions. No constants are fitted to data and no new physical entities are introduced.

assumptions (5)
  • domain assumption The transition rates obey local detailed balance with a nonconservative driving force F, as given in the main text.
    Invoked in 'Minimal collective model and thermodynamics'; this is a standard stochastic-thermodynamics modeling choice, not proven from first principles.
  • standard math In the all-to-all (k=N→∞) limit, the dynamics is exactly described by the density master equation for n±, n0.
    Standard mean-field/Curie-Weiss limit assumption; used to derive Eq. (2).
  • standard math The steady-state relation q=q(m) is a smooth function near m=0, allowing elimination of q in the expansion of dm/dt.
    Needed for Eq. (2); assumed by the local-expansion analysis.
  • ad hoc to paper For the phenomenological expressions, the system is in the strong-coupling limit |βν(ε+Δ)|≫1 with |m|≈1, so density dependence in rates can be neglected.
    Used in Appendix A to obtain closed forms for entropy production and power; validated only by comparison to the full model in a limited parameter range.
  • ad hoc to paper The N=102 square-lattice Gillespie simulation results are representative of the thermodynamic limit.
    Implicit in claiming robustness for nearest-neighbor couplings; no finite-size scaling analysis is provided, and finite-size effects are acknowledged in Fig. 1e.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Splitting of nonequilibrium phase transitions in driven Ising models." pith.science (2026). https://pith.science/paper/DKIVTC4V

@misc{pith2026241209343,
  author       = {Pith},
  title        = {Pith review of: Splitting of nonequilibrium phase transitions in driven Ising models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DKIVTC4V}},
  note         = {Machine review of arXiv:2412.09343}
}
read the original abstract

Spontaneous symmetry breaking occurs in various equilibrium and nonequilibrium systems, where phase transitions are typically marked by a single critical point that separates ordered and disordered regimes. We reveal a novel phenomenon in which the interplay between different temperatures and driving forces splits the order-disorder transition into two distinct transition points depending on which ordered state initially dominates. Crucially, these two emerging phases have distinct scaling behaviors and thermodynamic properties. To study this, we propose a minimal variant of the Ising model where spins are coupled to two thermal baths and subjected to two opposite driving forces associated to them. Our findings, robust both for all-to-all interactions (where exact solutions are possible) and nearest-neighbor couplings on a square lattice, uncover unique nonequilibrium behaviors and scaling laws for crucial thermodynamic quantities, such as efficiency, dissipation, power and its fluctuations, that are different between the two ordered phases. We also highlight that one of these emerging phases enables heat-engine operations that are less dissipative and show reduced fluctuations. In this setup, the system can also operate near maximum power and efficiency over a wide parameter range. Our results offer new insights into the relevance of phase transitions under nonequilibrium conditions.

Figures

Figures reproduced from arXiv: 2412.09343 by the authors.

Figure 1
Figure 1. FIG. 1. (a) For all-to-all interactions, we show the order parameter for di [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. (a) Average entropy production as functions of [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. (a) Power [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Upper panels: Plots of e [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 40 canonical work pages

  1. [1]

    Tom ´e and M

    T. Tom ´e and M. J. de Oliveira, Phys. Rev. Lett. 108, 020601 (2012)

    work page 2012

  2. [2]

    C. E. F. Noa, P. E. Harunari, M. J. de Oliveira, and C. E. Fiore, Phys. Rev. E 100, 012104 (2019)

    work page 2019

  3. [3]

    Aguilera, M

    M. Aguilera, M. Igarashi, and H. Shimazaki, Nature Commu- nications 14, 3685 (2023)

    work page 2023

  4. [4]

    S. A. Loos, S. H. Klapp, and T. Martynec, Physical review letters 130, 198301 (2023)

    work page 2023

  5. [5]

    Herpich, J

    T. Herpich, J. Thingna, and M. Esposito, Phys. Rev. X 8, 031056 (2018)

    2018

  6. [6]

    Herpich and M

    T. Herpich and M. Esposito, Phys. Rev. E 99, 022135 (2019)

    work page 2019

  7. [7]

    F. S. Filho, G. A. L. For ˜ao, D. M. Busiello, B. Cleuren, and C. E. Fiore, Phys. Rev. Res. 5, 043067 (2023)

    work page 2023

  8. [8]

    I. N. Mamede, K. Proesmans, and C. E. Fiore, Phys. Rev. Res. 5, 043278 (2023)

    2023

Show all 45 references
  1. [9]

    C. G. Feyisa and H. H. Jen, New Journal of Physics 26, 093039 (2024)

    2024

  2. [10]

    L. P. Bettmann, M. J. Kewming, and J. Goold, Phys. Rev. E 107, 044102 (2023)

    2023

  3. [11]

    Liang, Y .-H

    S. Liang, Y .-H. Ma, D. M. Busiello, and P. D. L. Rios, arXiv preprint arXiv:2312.02323 (2023)

    2023 arXiv

  4. [12]

    Rolandi, P

    A. Rolandi, P. Abiuso, and M. Perarnau-Llobet, Phys. Rev. Lett. 131, 210401 (2023)

    2023

  5. [13]

    Campisi and R

    M. Campisi and R. Fazio, Nature communications 7, 1 (2016)

    2016

  6. [14]

    L. d. S. Souza, G. Manzano, R. Fazio, and F. Iemini, Physical Review E 106, 014143 (2022)

    2022

  7. [15]

    Vroylandt, M

    H. Vroylandt, M. Esposito, and G. Verley, EPL (Europhysics Letters) 120, 30009 (2017)

    2017

  8. [16]

    Vroylandt, M

    H. Vroylandt, M. Esposito, and G. Verley, Phys. Rev. Lett.124, 250603 (2020)

    2020

  9. [17]

    R. N. Valani, Phys. Rev. E 105, L012101 (2022)

    2022

  10. [18]

    R. N. Valani and B. S. Dandogbessi, Phys. Rev. E110, L052203 (2024)

    2024

  11. [19]

    Liepelt and R

    S. Liepelt and R. Lipowsky, Phys. Rev. Lett.98, 258102 (2007)

    2007

  12. [20]

    Liepelt and R

    S. Liepelt and R. Lipowsky, Phys. Rev. E 79, 011917 (2009)

    2009

  13. [21]

    Berton, D

    C. Berton, D. M. Busiello, S. Zamuner, E. Solari, R. Scopelliti, F. Fadaei-Tirani, K. Severin, and C. Pezzato, Chemical Science 11, 8457 (2020)

    2020

  14. [22]

    Liang, P

    S. Liang, P. De Los Rios, and D. M. Busiello, Physical Review Letters 132, 228402 (2024)

    2024

  15. [23]

    D. M. Busiello, S. Liang, F. Piazza, and P. De Los Rios, Com- munications Chemistry 4, 16 (2021)

    2021

  16. [24]

    Rao and M

    R. Rao and M. Esposito, Physical Review X 6, 041064 (2016)

    2016

  17. [25]

    De Los Rios and A

    P. De Los Rios and A. Barducci, Elife 3, e02218 (2014)

    2014

  18. [26]

    Flatt, D

    S. Flatt, D. M. Busiello, S. Zamuner, and P. De Los Rios, Com- munications Physics 6, 205 (2023)

    2023

  19. [27]

    D.-J. Liu, X. Guo, and J. W. Evans, Phys. Rev. Lett.98, 050601 (2007)

    2007

  20. [28]

    Guo, D.-J

    X. Guo, D.-J. Liu, and J. W. Evans, The Journal of Chemical Physics 130, 074106 (2009), https: //pubs.aip.org/aip/jcp/article- pdf/doi/10.1063/1.3074308/15423883/074106 1 online.pdf

    2009 doi

  21. [29]

    Tom´e, Brazilian journal of physics 36, 1285 (2006)

    T. Tom´e, Brazilian journal of physics 36, 1285 (2006)

    2006

  22. [30]

    Aguilera, S

    M. Aguilera, S. A. Moosavi, and H. Shimazaki, Nature com- munications 12, 1197 (2021)

    2021

  23. [31]

    Pancotti, M

    N. Pancotti, M. Scandi, M. T. Mitchison, and M. Perarnau- Llobet, Physical Review X 10, 031015 (2020)

    2020

  24. [32]

    P. A. Erdman and F. No ´e, arXiv preprint arXiv:2204.04785 (2022)

    2022 arXiv

  25. [33]

    P. A. Erdman, A. Rolandi, P. Abiuso, M. Perarnau-Llobet, and F. No´e, Physical Review Research 5, L022017 (2023)

    2023

  26. [34]

    J. M. Yeomans, Statistical mechanics of phase transitions 6 (Clarendon Press, 1992)

    1992

  27. [35]

    Schnakenberg, Reviews of Modern physics 48, 571 (1976)

    J. Schnakenberg, Reviews of Modern physics 48, 571 (1976)

    1976

  28. [36]

    Touchette, Physics Reports 478, 1 (2009)

    H. Touchette, Physics Reports 478, 1 (2009)

    2009

  29. [37]

    Kumar, C

    N. Kumar, C. Van den Broeck, M. Esposito, and K. Lindenberg, Physical Review E 84, 051134 (2011)

    2011

  30. [38]

    Meibohm and M

    J. Meibohm and M. Esposito, Phys. Rev. E 110, L042102 (2024)

    2024

  31. [39]

    Wachtel, J

    A. Wachtel, J. V ollmer, and B. Altaner, Physical Review E 92, 042132 (2015)

    2015

  32. [40]

    D. T. Gillespie, The journal of physical chemistry 81, 2340 (1977)

    1977

  33. [41]

    Curzon and B

    F. Curzon and B. Ahlborn, American Journal of Physics 43, 22 (1975)

    1975

  34. [42]

    e F 4 (3β1+β2) 2 cosh F 2 (β1−β2) + 1 2 cosh F 4 (β1 +β2) × β2 1− 4β1β2 +β2 2 cosh F 4 (β1−β2) −β2 1 cosh F 4 (3β1 +β2) −β2 2 cosh F 4 (β1 + 3β2) !# ×

    C. Van den Broeck, Physical Review Letters95, 190602 (2005). 7 Supplemental Material: Splitting of nonequilibrium phase transitions in driven Ising models Gustavo A. L. For˜ao, Fernando S. Filho, Andr´e P. Vieira, Bart Cleuren, Daniel M. Busiello and Carlos E. Fiore This suppl...

    2005

  35. [43]

    Although ηCA is not a universal bound, it depends solely on the temperature ratio and has been verified for various heat engines operating out of equilibrium [41, 42]

    The efficiencies at maximum power,ηMP, are constrained between the Curzon-Ahlborn efficiency,ηCA = 1− p β2/β1, and the maximum efficiency,ηME . Although ηCA is not a universal bound, it depends solely on the temperature ratio and has been verified for various heat engines oper...

  36. [44]

    However, this optimization of power and e fficiency comes at the cost of increased dissipation, as indicated by the rise inPMP (see the inset)

    As β1 increases, both ηMP andηME converge, eventually matching the ideal Carnot efficiencyηc = 1−β2/β1 asϵ and F increase. However, this optimization of power and e fficiency comes at the cost of increased dissipation, as indicated by the rise inPMP (see the inset)

  37. [45]

    This reveals a striking relationship between optimization and the occurrence of a phase transition

    The points of maximum power, ∆MP, and maximum e fficiency, ∆ME , approach the discontinuous phase transition point ∆1c asβ1 increases. This reveals a striking relationship between optimization and the occurrence of a phase transition. 12 0.0200.0 0.1 0.3 0.5 0.7 4 2 0 2 4 6 84...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.