pith. sign in

arxiv: 2606.29751 · v1 · pith:NQCQIXA3new · submitted 2026-06-29 · ❄️ cond-mat.stat-mech · math-ph· math.MP

Finite-resolution exhaustive traversal of thermodynamic state spaces has divergent thermodynamic length

Satori Tsuzuki This is my paper

Pith reviewed 2026-06-30 04:32 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech math-phmath.MP
keywords thermodynamic lengthexhaustive traversalfinite resolutionexcess workfriction tensorMarkov jump processharmonic trapstate space coverage
0
0 comments X

The pith

Any ε-dense exhaustive path through a d-dimensional thermodynamic state space has length at least order ε to the power 1 minus d.

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

The paper establishes that rectifiable finite-resolution paths cannot surject onto a d-dimensional thermodynamic window without their length in the metric g growing at least as C ε^{1-d}. When the physical friction tensor ζ coincides with or dominates g, this length lower bound converts via Cauchy-Schwarz into a quadratic excess-work lower bound that diverges as ε^{2(1-d)} divided by protocol duration. Maintaining a fixed excess-work budget then requires the duration itself to scale as ε^{2(1-d)}. Explicit friction tensors derived from a three-state Markov jump process and from an overdamped harmonic trap confirm that the geometric divergence is realized as a physical cost. A laboratory or simulation resolution floor replaces the continuum divergence with a cutoff scaling as the maximum of ε and that floor to the power 1-d.

Core claim

Continuous space-filling maps can be surjective onto higher-dimensional regions, but thermodynamic protocols are rectifiable finite-resolution paths. We study exhaustive traversal of a compact d-dimensional thermodynamic state-space window (M,g) by curves H_ε whose images are ε-dense in intrinsic distance. A standard covering/tube estimate gives L_g[H_ε] ≥ C_g ε^{1-d} - O(ε) for every regular d>1 window. When the physical friction tensor ζ coincides with, or uniformly dominates, the coverage metric g, Cauchy-Schwarz for the quadratic slow-driving action gives W_ex^{(2)} ≥ L_ζ²/τ = Ω(ε^{2(1-d)}/τ). Equivalently, at fixed quadratic excess-work budget, maintaining slow driving requires τ = Ω(ε^

What carries the argument

The covering-tube lower bound on the g-length of any ε-dense curve in a d-dimensional manifold, converted to an excess-work bound by Cauchy-Schwarz once the friction tensor dominates the geometric metric.

If this is right

  • At fixed protocol time the quadratic excess work for exhaustive coverage diverges as ε to the power 2(1-d).
  • In the overdamped harmonic trap a raster scan produces length scaling as the inverse of the metric diameter and fixed-time excess work scaling as the inverse square of that diameter.
  • Morton or Z-order curves preserve the same exponent while altering locality-dependent prefactors.
  • A physical resolution floor Δ_g replaces the continuum divergence by a cutoff scaling as max(ε, Δ_g) to the power 1-d.

Where Pith is reading between the lines

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

  • The scaling suggests that exhaustive dense sampling of thermodynamic landscapes becomes prohibitively expensive in dimensions greater than one unless an explicit resolution cutoff is imposed.
  • The same geometric obstruction may limit the practicality of high-resolution parameter scans in optimization or landscape exploration routines that rely on slow driving.
  • Response-proxy metrics that are not derived from explicit microscopic dynamics remain diagnostic tools rather than physical friction tensors.

Load-bearing premise

The physical friction tensor must coincide with or uniformly dominate the coverage metric g, allowing the length lower bound to be turned into an excess-work lower bound.

What would settle it

A numerical simulation or laboratory measurement in a two-dimensional thermodynamic parameter space that keeps quadratic excess work bounded while resolution ε is driven to zero would falsify the claimed divergence.

Figures

Figures reproduced from arXiv: 2606.29751 by Satori Tsuzuki.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic of finite-resolution space-filling traversals. Hilbert and Peano approximants refine the traversal locally, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Operational finite-resolution resource laws in the overdamped harmonic-trap friction metric. (a) A finite observation or [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Finite-resolution thermodynamic-length scaling for two-dimensional state-space windows. (a) Representative normal [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Critical integrability classification for the van der Waals response-proxy metric [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Morton/Z-order traversal as a locality-control benchmark. (a) Fitted resolution exponents for Hilbert and Morton [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Uniform comparability of the mean-field Ising Model [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Dimensional dependence of the finite-resolution exponent, tested on the flat three-dimensional ideal-gas Ruppeiner [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Baseline critical-cutoff diagnostics. (a) Highest-order prefactors [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
read the original abstract

Continuous space-filling maps can be surjective onto higher-dimensional regions, but thermodynamic protocols are rectifiable finite-resolution paths. We study exhaustive traversal of a compact $d$-dimensional thermodynamic state-space window $(\mathcal{M},g)$ by curves $H_\varepsilon$ whose images are $\varepsilon$-dense in intrinsic distance. A standard covering/tube estimate gives $L_g[H_\varepsilon]\ge C_g\varepsilon^{1-d}-O(\varepsilon)$ for every regular $d>1$ window. The geometry is classical; the contribution is to turn it into an operational resource law for thermodynamic coverage. When the physical friction tensor $\zeta$ coincides with, or uniformly dominates, the coverage metric $g$, Cauchy--Schwarz for the quadratic slow-driving action gives $W_{\rm ex}^{(2)}\ge L_\zeta^2/\tau=\Omega(\varepsilon^{2(1-d)}/\tau)$. Equivalently, at fixed quadratic excess-work budget, maintaining slow driving requires $\tau=\Omega(\varepsilon^{2(1-d)})$. We derive microscopic friction metrics for a detailed-balance three-state Markov jump process, $\zeta_{ij}=(\beta/\gamma)(\pi_i\delta_{ij}-\pi_i\pi_j)$, and for an overdamped harmonic trap, $\mathrm d\ell_\zeta^2=\mu^{-1}\mathrm da^2+(4\beta\mu k^3)^{-1}\mathrm dk^2$. In the trap, a raster scan gives $L_\zeta\sim\Delta_g^{-1}$ and fixed-time $W_{\rm ex}^{(2)}\sim\Delta_g^{-2}$, while fixed dwell time shifts the cost to acquisition time. A laboratory or simulation floor cuts off the continuum divergence as $L_{\rm op}=\Theta(\max\{\varepsilon,\Delta_g\}^{1-d})$. Controlled singular response-proxy metrics diagnose critical prefactors and directional integrability, but are not physical friction tensors unless derived from microscopic dynamics. Morton/Z-order preserves the exponent while increasing locality-dependent amplitudes.

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

0 major / 3 minor

Summary. The paper claims that exhaustive traversal of a compact d-dimensional thermodynamic state space (M,g) by rectifiable ε-dense curves H_ε yields the lower bound L_g[H_ε] ≥ C_g ε^{1-d} - O(ε) for d>1 via standard covering/tube estimates. When the physical friction tensor ζ coincides with or uniformly dominates g, Cauchy-Schwarz on the quadratic slow-driving excess-work functional implies W_ex^{(2)} ≥ L_ζ²/τ = Ω(ε^{2(1-d)}/τ), or equivalently τ = Ω(ε^{2(1-d)}) at fixed excess-work budget. Explicit microscopic derivations are given for ζ in a detailed-balance three-state Markov jump process (ζ_ij = (β/γ)(π_i δ_ij - π_i π_j)) and an overdamped harmonic trap (dℓ_ζ² = μ^{-1} da² + (4βμ k³)^{-1} dk²), with numerical verification via raster scan in the trap showing L_ζ ~ Δ_g^{-1} and W_ex^{(2)} ~ Δ_g^{-2}.

Significance. If the result holds, the work converts a classical geometric covering bound into an operational thermodynamic resource law, demonstrating that finite-resolution exhaustive coverage in d>1 requires divergent length and excess work under slow driving. Credit is given for the explicit microscopic derivations of the friction tensors from dynamics (MJP and trap) rather than fitted quantities, and for the direct numerical demonstration of the scaling in the trap example. The conditional nature of the ζ-domination assumption is stated clearly, and the result highlights resolution-time trade-offs with a laboratory cutoff L_op = Θ(max{ε, Δ_g}^{1-d}).

minor comments (3)
  1. [Abstract] Abstract: the final sentence on Morton/Z-order curves preserving the exponent but affecting locality-dependent amplitudes would benefit from a brief parenthetical explanation or citation to space-filling curve literature for readers unfamiliar with the locality effect.
  2. [Trap example] Trap example: the statement that fixed dwell time shifts the cost to acquisition time is clear in intent but would be improved by an explicit formula relating total time to the raster parameters.
  3. Notation: the symbol Δ_g in the numerical results is used without prior definition in the provided abstract; ensure it is introduced as the grid spacing in the raster scan section.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive assessment of the manuscript, accurate summary of the geometric covering argument and its thermodynamic implications, and recommendation to accept. No revisions are required.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The derivation imports a standard geometric covering/tube lower bound L_g[H_ε] ≥ C_g ε^{1-d} - O(ε) from external Riemannian geometry and applies Cauchy-Schwarz once the independently derived friction tensor ζ is shown to dominate the coverage metric g. Microscopic expressions for ζ are obtained directly from the detailed-balance Markov jump process and overdamped Langevin dynamics; the raster-scan numerics confirm the scaling without fitting parameters to the target bound. No self-definitional steps, fitted inputs renamed as predictions, or load-bearing self-citations appear in the chain.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The paper relies on standard geometric axioms and domain assumptions about thermodynamic metrics and protocols; no new entities are postulated and no free parameters are fitted.

axioms (3)
  • domain assumption Compact d-dimensional thermodynamic state-space window (M,g)
    The setting for the traversal study.
  • domain assumption Thermodynamic protocols are rectifiable finite-resolution paths H_ε that are ε-dense in intrinsic distance
    Definition of the curves whose length is bounded.
  • standard math Standard covering/tube estimate applies to give the length lower bound
    Basis for L_g[H_ε] ≥ C_g ε^{1-d} - O(ε).

pith-pipeline@v0.9.1-grok · 5901 in / 1555 out tokens · 74098 ms · 2026-06-30T04:32:30.343606+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

51 extracted references · 5 canonical work pages

  1. [1]

    it separates topological state-space filling from finite-resolution thermodynamic traversability

  2. [2]

    it converts the codimension-one length cost, un- der explicit friction-metric hypotheses, into a dissipation–duration tradeoff

  3. [3]

    it realizes the tradeoff in microscopic Markov-jump and Langevin control models rather than only in diagnostic geometries

  4. [4]

    The finite-time interpretation uses the slow-driving quadratic action

    it identifies an integrability criterion for critical prefactors and distinguishes controlled response- proxy metrics from transport-derived friction ten- sors. The finite-time interpretation uses the slow-driving quadratic action. If the physical friction tensorζco- incides with, or uniformly dominates, the coverage met- ric, Cauchy–Schwarz givesW (2) ex...

  5. [5]

    Ideal gas For a single-component ideal gas at fixed particle num- ber, use molar or per-particle variablesx= (u, v), where u >0 is internal energy per particle andv >0 is vol- ume per particle. In dimensionless units, the entropy per particle may be written σ(u, v) = s(u, v) kB =σ 0 +clnu+ lnv,(76) wherec=C V /(N kB)>0 is the dimensionless heat ca- pacity...

  6. [6]

    In dimensionless per-particle vari- ables, let q(u, v) =u+ a v , r(v) =v−b,(81) witha >0,b >0,q >0, andr >0

    van der Waals fluid A van der Waals fluid supplies a curved state-space ex- ample and introduces singular sets associated with me- chanical instability. In dimensionless per-particle vari- ables, let q(u, v) =u+ a v , r(v) =v−b,(81) witha >0,b >0,q >0, andr >0. A convenient entropy representation is σ(u, v) =σ 0 +clnq(u, v) + lnr(v).(82) The Ruppeiner met...

  7. [7]

    ForNspinsσ i =±1, the Curie–Weiss Hamiltonian is HN(σ;J, h) =− J 2N NX i=1 σi !2 −h NX i=1 σi,(88) withJ >0

    Mean-field Ising model The mean-field Ising model [21] provides a minimal setting in which a thermodynamic metric becomes large near a continuous phase transition. ForNspinsσ i =±1, the Curie–Weiss Hamiltonian is HN(σ;J, h) =− J 2N NX i=1 σi !2 −h NX i=1 σi,(88) withJ >0. The partition function is ZN(β, h) = X {σi} exp   βJ 2N X i σi !2 +βh X i σi   ....

  8. [8]

    friction-like

    A mean-field Model-A friction ansatz The model classes above supply coverage metrics, but the dissipation bounds of Sec. III C require a friction ten- sorζthat is uniformly comparable to the coverage metric on the working window. For the mean-field Ising exam- ple we therefore use a scaling-motivated Model-A friction ansatz. In slow-driving linear respons...

  9. [9]

    Choose a compact regular thermodynamic window Mand a thermodynamic metricgfor coverage

  10. [10]

    The Markov-jump and harmonic-trap examples in Secs

    Choose a friction metricζ, or setζ=gif the same metric is used for both coverage and dissipation. The Markov-jump and harmonic-trap examples in Secs. III E and III D are microscopic cases in which this choice is direct

  11. [11]

    Construct anε-dense Hilbert/Peano-type grid traversalH ε in coordinates adapted either to the experimental controls or to the Riemannian volume element dµg

  12. [12]

    (17) or a geodesic quadrature rule

    Compute or estimateL g[Hε] andL ζ[Hε] using Eq. (17) or a geodesic quadrature rule

  13. [13]

    For regular compact state spaces, the expected leading exponent is fixed by dimension

    Compare the observed scaling with the universal lower boundL g[Hε]≥C gε1−d−O(ε), the quadratic finite-time dissipation boundW (2) ex ≥ L 2 ζ/τ, and, when a finite observation floor is present, the oper- ational cutoff lawL op = Θ(max{ε,∆ g}1−d). For regular compact state spaces, the expected leading exponent is fixed by dimension. Deviations from this be-...

  14. [14]

    Define the metricg(x) on that window

    Select a model, a cutoff ∆ if required, and a rect- angular coordinate window from Table III. Define the metricg(x) on that window

  15. [15]

    For Hilbert traversal, order the 2 n ×2 n cell centers by the finite Hilbert ordering

    Select a traversal typeT. For Hilbert traversal, order the 2 n ×2 n cell centers by the finite Hilbert ordering. For Morton traversal, order the same cen- ters by the bit-interleaved Morton key in Eq. (15)

  16. [16]

    , xN and the coordinate covering radiusε n from Eq

    For each ordernlisted in Table IV, compute the cell centersx 0, . . . , xN and the coordinate covering radiusε n from Eq. (A2)

  17. [17]

    For each consecutive pair (xk, xk+1), set ∆x k =x k+1 −x k andx k+1/2 = (xk +x k+1)/2, then add h ∆xi kgij(xk+1/2)∆xj k i1/2 toL n

    InitializeL n = 0. For each consecutive pair (xk, xk+1), set ∆x k =x k+1 −x k andx k+1/2 = (xk +x k+1)/2, then add h ∆xi kgij(xk+1/2)∆xj k i1/2 toL n. For the ideal-gas subdivision check, split each Morton jump into four equal coordinate sub- segments and apply the same midpoint rule to each subsegment

  18. [18]

    Storep, its slope standard error, the log- spaceR 2, and the residualsr n

    Fit logL n =α−plogε n +r n over the specified fit orders. Storep, its slope standard error, the log- spaceR 2, and the residualsr n

  19. [19]

    Estimate local critical slopes from adjacent values of logC hi ∆ versus log ∆

    For each critical cutoff, compute the highest-order prefactorC hi ∆ =L εmin εmin. Estimate local critical slopes from adjacent values of logC hi ∆ versus log ∆

  20. [20]

    (A4) to iden- tify finite-resolution cutoffs whose critical core is not resolved by the grid

    Apply the core-resolution score in Eq. (A4) to iden- tify finite-resolution cutoffs whose critical core is not resolved by the grid. 20 TABLE III. Model parameters, coordinate windows, and metric components used in the numerical analysis. Heret=T−T c, β= 1/T,mis the real solution ofbm 3 +atm−h= 0,χ −1 =at+ 3bm 2, ands(T, v) = (v−b) −2 −2a/(T v 3). Case Co...

  21. [21]

    Seta∈[0,1],k∈[1,4],β=µ= 1, and use the friction metricζ aa = 1,ζ kk = (4k3)−1

  22. [22]

    ,9, form a 2 n ×2 n serpen- tine raster over cell centers

    For each ordern= 3, . . . ,9, form a 2 n ×2 n serpen- tine raster over cell centers. The covering scale is the coordinate half-diagonal

  23. [23]

    For each segment, compute its midpoint friction lengthℓ i and accumulateL ζ =P i ℓi andP i ℓ2 i

  24. [24]

    Report the quadratic fixed-time estimate W (2) fixed time =L 2 ζ/τwithτ= 1, and Wdwell = P i ℓ2 i /tdwell,τ dwell =N segtdwell with tdwell = 1

  25. [25]

    Full tail

    FitL ζ,W (2) fixed time,W dwell, andτ dwell as powers of the covering scale. 21 TABLE V. Baseline Hilbert fit diagnostics by model family. The table reports the largest standard error ofp, the smallest log-spaceR 2, and the largest absolute relative residual over the fitted tail orders and, where applicable, over the listed critical cutoffs. Model family ...

  26. [26]

    Weinhold, Metric geometry of equilibrium thermody- namics, J

    F. Weinhold, Metric geometry of equilibrium thermody- namics, J. Chem. Phys.63, 2479 (1975)

    1975

  27. [27]

    Ruppeiner, Riemannian geometry in thermody- namic fluctuation theory, Rev

    G. Ruppeiner, Riemannian geometry in thermody- namic fluctuation theory, Rev. Mod. Phys.67, 605 (1995), erratum: Rev. Mod. Phys.68, 313 (1996), 22 doi:10.1103/RevModPhys.68.313

    work page doi:10.1103/revmodphys.68.313 1995

  28. [28]

    C. R. Rao, Information and the accuracy attainable in the estimation of statistical parameters, Bull. Calcutta Math. Soc.37, 81 (1945)

    1945

  29. [29]

    Amari,Information Geometry and Its Applications, Applied Mathematical Sciences, Vol

    S. Amari,Information Geometry and Its Applications, Applied Mathematical Sciences, Vol. 194 (Springer, Tokyo, 2016)

    2016

  30. [30]

    Salamon and R

    P. Salamon and R. S. Berry, Thermodynamic length and dissipated availability, Phys. Rev. Lett.51, 1127 (1983)

    1983

  31. [31]

    G. E. Crooks, Measuring thermodynamic length, Phys. Rev. Lett.99, 100602 (2007)

    2007

  32. [32]

    D. A. Sivak and G. E. Crooks, Thermodynamic metrics and optimal paths, Phys. Rev. Lett.108, 190602 (2012)

    2012

  33. [33]

    Blaber and D

    S. Blaber and D. A. Sivak, Optimal control in stochastic thermodynamics, J. Phys. Commun.7, 033001 (2023), arXiv:2212.00706 [cond-mat.stat-mech]

    work page arXiv 2023

  34. [34]

    Loutchko, Y

    D. Loutchko, Y. Sughiyama, and T. J. Kobayashi, Riemannian geometry of optimal driving and thermo- dynamic length and its application to chemical reac- tion networks, Phys. Rev. Research4, 043049 (2022), arXiv:2205.03829 [cond-mat.stat-mech]

    work page arXiv 2022

  35. [35]

    Scandi and M

    M. Scandi and M. Perarnau-Llobet, Thermodynamic length in open quantum systems, Quantum3, 197 (2019)

    2019

  36. [36]

    Peano, Sur une courbe, qui remplit toute une aire plane, Math

    G. Peano, Sur une courbe, qui remplit toute une aire plane, Math. Ann.36, 157 (1890)

  37. [37]

    Hilbert, ¨Uber die stetige Abbildung einer Linie auf ein Fl¨ achenst¨ uck, Math

    D. Hilbert, ¨Uber die stetige Abbildung einer Linie auf ein Fl¨ achenst¨ uck, Math. Ann.38, 459 (1891)

  38. [38]

    Sagan,Space-Filling Curves, Universitext (Springer, New York, 1994)

    H. Sagan,Space-Filling Curves, Universitext (Springer, New York, 1994)

    1994

  39. [39]

    Bader,Space-Filling Curves: An Introduction with Applications in Scientific Computing, Texts in Computa- tional Science and Engineering, Vol

    M. Bader,Space-Filling Curves: An Introduction with Applications in Scientific Computing, Texts in Computa- tional Science and Engineering, Vol. 9 (Springer, Berlin, Heidelberg, 2013)

    2013

  40. [40]

    Weyl, On the volume of tubes, Am

    H. Weyl, On the volume of tubes, Am. J. Math.61, 461 (1939)

    1939

  41. [41]

    Gray,Tubes, 2nd ed., Progress in Mathematics, Vol

    A. Gray,Tubes, 2nd ed., Progress in Mathematics, Vol. 221 (Birkh¨ auser, Basel, 2004)

    2004

  42. [42]

    Mattila,Geometry of Sets and Measures in Euclidean Spaces: Fractals and Rectifiability, Cambridge Studies in Advanced Mathematics, Vol

    P. Mattila,Geometry of Sets and Measures in Euclidean Spaces: Fractals and Rectifiability, Cambridge Studies in Advanced Mathematics, Vol. 44 (Cambridge University Press, Cambridge, 1995)

    1995

  43. [43]

    G. M. Morton,A Computer Oriented Geodetic Data Base and a New Technique in File Sequencing, Technical Re- port (IBM Ltd., Ottawa, Ontario, Canada, 1966)

    1966

  44. [44]

    B. Moon, H. V. Jagadish, C. Faloutsos, and J. H. Saltz, Analysis of the clustering properties of the Hilbert space- filling curve, IEEE Trans. Knowl. Data Eng.13, 124 (2001)

    2001

  45. [45]

    P. R. Zulkowski, D. A. Sivak, G. E. Crooks, and M. R. DeWeese, The geometry of thermodynamic control, Phys. Rev. E86, 041148 (2012)

    2012

  46. [46]

    G. M. Rotskoff and G. E. Crooks, Optimal control in nonequilibrium systems: Dynamic riemannian geometry of the Ising model, Phys. Rev. E92, 060102 (2015)

    2015

  47. [47]

    Skilling, Programming the Hilbert curve, inBayesian Inference and Maximum Entropy Methods in Science and Engineering, AIP Conf

    J. Skilling, Programming the Hilbert curve, inBayesian Inference and Maximum Entropy Methods in Science and Engineering, AIP Conf. Proc., Vol. 707 (AIP, 2004) pp. 381–387

    2004

  48. [48]

    O. M. Basri and O. Raz, Thermodynamic geometry through second order phase transitions (2025), arXiv preprint, arXiv:2512.01936 [cond-mat.stat-mech]

    work page arXiv 2025

  49. [49]

    P. C. Hohenberg and B. I. Halperin, Theory of dynamic critical phenomena, Rev. Mod. Phys.49, 435 (1977)

    1977

  50. [50]

    Chennakesavalu and G

    S. Chennakesavalu and G. M. Rotskoff, Unified, geomet- ric framework for nonequilibrium protocol optimization, Phys. Rev. Lett.130, 107101 (2023)

    2023

  51. [51]

    M. C. Engel, J. A. Smith, and M. P. Brenner, Op- timal control of nonequilibrium systems through auto- matic differentiation, Phys. Rev. X13, 041032 (2023), arXiv:2201.00098 [physics.comp-ph]

    work page arXiv 2023