pith. sign in

arxiv: 2605.17838 · v1 · pith:QULKCDW4new · submitted 2026-05-18 · ❄️ cond-mat.stat-mech · cond-mat.mes-hall

Network analysis for steady-state current fluctuations under finite affinity: Application to Brownian computation

Yasuhiro Utsumi This is my paper

Pith reviewed 2026-05-20 01:20 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech cond-mat.mes-hall
keywords Brownian computationcurrent fluctuationsFano factorthermodynamic affinitymaster equationsnetwork analysisirreversible computationsteady-state noise
0
0 comments X

The pith

In a tree-like Brownian computation model, the Fano factor of reset current transitions from noiseless to Poissonian at affinity ln α.

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

The paper develops a graph-theoretic method with twisted circuit matrices to handle current noise in master equations driven by finite thermodynamic affinity. When applied to a Brownian computation setup on a tree-like diagram that has exponential backward branching and one reset cycle, the method identifies a sharp change in the noise properties of the reset current. In the limit of very long computation paths, this change occurs when the affinity per step equals the logarithm of the branching factor α. The transition aligns with a known easy-hard switch in computational complexity and shows how much driving force is needed to overcome the effects of logical irreversibility.

Core claim

In the limit of an infinitely long intended computation path ℓ, the Fano factor of the reset current undergoes a transition from noiseless to Poissonian behavior at an affinity equal to the logarithm of the number of immediate predecessors α. This corresponds to an easy-hard transition in the computational time complexity, which is not captured by the thermodynamic uncertainty relation. This transition point precisely characterizes the thermodynamic costs of logically irreversible computation: in the absence of affinity, the reset cost scales as ln ℓ, whereas reaching the transition point requires a thermodynamic force of order ln α per step to counteract backward branching.

What carries the argument

Twisted-cycle matrices obtained from the affinity-twisted incidence matrix, used to express the signal-to-noise ratio as a quadratic optimization problem over twisted-cycle currents on the single-reset-cycle tree diagram.

If this is right

  • The transition marks the minimum thermodynamic force required per step to suppress backward branching in logically irreversible steps.
  • In the absence of affinity the reset cost grows as ln ℓ with computation length.
  • The identified transition matches the easy-hard change in computational time complexity from related analyses.
  • The standard thermodynamic uncertainty relation does not predict or locate this particular transition.

Where Pith is reading between the lines

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

  • The same twisted-matrix approach might reveal analogous noise transitions in other branching nonequilibrium systems such as molecular search or motor models.
  • Physical devices performing irreversible computation could tune driving affinity near ln α to balance noise suppression against energy cost.
  • Relaxing the single-cycle or pure-tree assumption could uncover broader links between fluctuation statistics and computational hardness.

Load-bearing premise

The state-transition diagram is exactly tree-like with exponential backward branching and only a single reset cycle, with the Fano-factor transition corresponding exactly to the complexity transition in the infinite-ℓ limit.

What would settle it

Numerical integration of the master equation on successively longer finite tree graphs with fixed branching α, checking whether the Fano factor of the reset current changes from near zero to near one precisely when the affinity reaches ln α.

Figures

Figures reproduced from arXiv: 2605.17838 by Yasuhiro Utsumi.

Figure 1
Figure 1. Figure 1: FIG. 1. Typical three configurations of the twig [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Dependence of the Fano factor on [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. State-transition diagram illustrating a Brownian com [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. The average reset current (a) and its Fano factor (b) [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
read the original abstract

A graph-theoretic analysis of the steady-state current noise in master equations under a finite thermodynamic force (affinity) is presented. The incidence matrix twisted by a finite affinity is not orthogonal to the standard cycle space, motivating the introduction of twisted circuit matrices to restore the orthogonality. The resulting twisted-cycle matrix yields an interference-like effect, enabling us to express the signal-to-noise ratio as a quadratic optimization problem in terms of twisted-cycle currents. We apply this framework to a Brownian computation model on a tree-like state-transition diagram with exponential backward branching, finite affinity at each step, and a single reset cycle. In the limit of an infinitely long intended computation path $\ell$, the Fano factor of the reset current undergoes a transition from noiseless to Poissonian behavior at an affinity equal to the logarithm of the number of immediate predecessors $\alpha$. This corresponds to an easy-hard transition in the computational time complexity [K. Okajima, K. Hukushima, arXiv:2512.24728 ], which is not captured by the thermodynamic uncertainty relation. This transition point precisely characterizes the thermodynamic costs of logically irreversible computation: in the absence of affinity, the reset cost scales as $\ln \ell$, whereas reaching the transition point requires a thermodynamic force of order $\ln \alpha$ per step to counteract backward branching.

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 / 2 minor

Summary. The manuscript develops a graph-theoretic framework for analyzing steady-state current fluctuations in master equations driven by finite thermodynamic affinity. It introduces twisted circuit matrices to restore orthogonality between the affinity-twisted incidence matrix and the cycle space, allowing the signal-to-noise ratio of reset currents to be expressed as a quadratic optimization over twisted-cycle currents. The framework is applied to a Brownian computation model whose state-transition graph is a tree with exponential backward branching factor α, finite affinity per step, and exactly one reset cycle. In the infinite-path-length limit ℓ → ∞ the Fano factor of the reset current is shown to undergo a sharp transition from noiseless to Poissonian statistics precisely at affinity = ln α; this point is asserted to coincide with the easy-hard complexity threshold of Okajima & Hukushima and to furnish a thermodynamic characterization of logically irreversible computation (reset cost ∼ ln ℓ without affinity versus ∼ ln α per step at the transition).

Significance. If the central derivation and the exact correspondence hold, the work supplies a concrete, falsifiable link between current-noise properties and computational-complexity transitions that is not captured by the thermodynamic uncertainty relation. The twisted-matrix construction and the quadratic-optimization formulation are potentially reusable for fluctuation analysis in other driven Markov networks. The explicit mapping to a Brownian-computation model gives a clear physical interpretation of the thermodynamic cost of logical irreversibility.

major comments (2)
  1. [§4 and Eq. (27)] §4 (Application to Brownian computation) and the paragraph following Eq. (27): the exact location of the Fano-factor transition at affinity = ln α and its claimed one-to-one correspondence with the Okajima-Hukushima complexity threshold are derived under the assumption that the state-transition diagram is exactly tree-like, possesses exponential backward branching with factor α, and contains precisely one reset cycle. The manuscript does not demonstrate that the transition survives the addition of even a single extra cycle or a deviation from pure exponential branching; because this assumption is load-bearing for the central claim that the transition “precisely characterizes” thermodynamic costs, a robustness check or an explicit statement of the necessary graph-theoretic conditions is required.
  2. [§3.2] §3.2 (Twisted circuit matrices): the construction of the twisted-cycle matrix is presented as restoring orthogonality after affinity twisting, yet the subsequent quadratic optimization for the reset-current SNR is performed only in the infinite-ℓ limit. It is unclear whether the interference effect that produces the sharp noiseless-to-Poissonian transition remains well-defined for finite ℓ or for graphs that are only approximately tree-like; a finite-ℓ expansion or a counter-example on a small non-tree graph would clarify the scope of the result.
minor comments (2)
  1. [§3] The notation for the twisted incidence matrix and the twisted-cycle matrix is introduced without an explicit comparison table to the untwisted objects; adding such a table would improve readability.
  2. [Figure 2] Figure 2 (state-transition diagram) does not indicate the numerical value of α used in the plotted trajectories; labeling the branching factor on the figure would help readers connect the schematic to the analytic result.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. We respond to each major comment below and indicate the revisions we will make to clarify the scope of the results.

read point-by-point responses

  1. Referee: [§4 and Eq. (27)] §4 (Application to Brownian computation) and the paragraph following Eq. (27): the exact location of the Fano-factor transition at affinity = ln α and its claimed one-to-one correspondence with the Okajima-Hukushima complexity threshold are derived under the assumption that the state-transition diagram is exactly tree-like, possesses exponential backward branching with factor α, and contains precisely one reset cycle. The manuscript does not demonstrate that the transition survives the addition of even a single extra cycle or a deviation from pure exponential branching; because this assumption is load-bearing for the central claim that the transition “precisely characterizes” thermodynamic costs, a robustness check or an explicit statement of the necessary graph-theoretic conditions is required.

    Authors: We agree that the sharp Fano-factor transition at affinity = ln α and its correspondence to the Okajima-Hukushima threshold are derived specifically for a tree-like state-transition graph with exact exponential backward branching factor α and a single reset cycle. These features define the Brownian computation model under consideration and enable the exact mapping to the complexity transition. In the revised manuscript we will add an explicit paragraph in §4 stating the necessary graph-theoretic conditions: the diagram must be a tree with pure exponential branching and precisely one reset cycle. The general twisted-matrix framework of §3 applies to arbitrary graphs, but the precise thermodynamic characterization of logical irreversibility is tied to this model class. A full numerical robustness analysis against added cycles or non-exponential branching lies beyond the present scope. revision: partial

  2. Referee: [§3.2] §3.2 (Twisted circuit matrices): the construction of the twisted-cycle matrix is presented as restoring orthogonality after affinity twisting, yet the subsequent quadratic optimization for the reset-current SNR is performed only in the infinite-ℓ limit. It is unclear whether the interference effect that produces the sharp noiseless-to-Poissonian transition remains well-defined for finite ℓ or for graphs that are only approximately tree-like; a finite-ℓ expansion or a counter-example on a small non-tree graph would clarify the scope of the result.

    Authors: The twisted-cycle matrix construction and the associated quadratic optimization are formulated generally in §3.2 for any finite affinity and any graph. The sharp noiseless-to-Poissonian transition, however, emerges only in the ℓ → ∞ limit because the interference among twisted-cycle currents becomes dominant under exponential branching. For finite ℓ the Fano factor approaches the limiting value asymptotically; we will add a short asymptotic expansion in the revised manuscript showing this convergence. For graphs that deviate mildly from the pure tree structure the interference effect persists as long as backward branching remains the dominant feature. We will include a brief remark to this effect in the discussion section. revision: partial

Circularity Check

0 steps flagged

Fano-factor transition derived independently via twisted-cycle matrices; correspondence to complexity result is external citation only

full rationale

The paper constructs a graph-theoretic framework starting from the incidence matrix twisted by finite affinity, introduces twisted circuit matrices to restore orthogonality, and reduces the signal-to-noise ratio of the reset current to a quadratic optimization over twisted-cycle currents. This yields an explicit transition in the Fano factor from noiseless to Poissonian at affinity = ln α for the stated tree-like model with single reset cycle in the ℓ → ∞ limit. The model assumptions (exponential backward branching factor α, single reset cycle) are stated up front and enter the derivation directly through the cycle-space basis; the location of the transition follows from the quadratic form without reference to the Okajima-Hukushima result. The claimed correspondence to the easy-hard complexity transition is asserted by citation to an external arXiv preprint whose authors do not overlap with the present work, but this citation is not load-bearing for the noise calculation itself. No step reduces to a self-definition, fitted input renamed as prediction, or self-citation chain. The derivation is therefore self-contained against the paper's own equations.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 1 invented entities

The framework rests on standard master-equation assumptions plus model-specific choices of tree topology and single reset cycle; the twisted matrices are introduced to handle finite affinity.

free parameters (2)
  • α
    Branching factor of immediate predecessors; sets the location of the noise transition.
  • ℓ
    Intended computation path length; taken to infinity to obtain the sharp transition.
axioms (2)
  • domain assumption The dynamics obey a continuous-time Markov master equation on a finite state space.
    Standard starting point for steady-state current and noise calculations in non-equilibrium statistical mechanics.
  • ad hoc to paper The state-transition graph is a tree with exponential backward branching and exactly one reset cycle.
    Specific modeling choice for the Brownian computation application.
invented entities (1)
  • twisted circuit matrices no independent evidence
    purpose: Restore orthogonality between incidence matrix and cycle space when affinity is finite.
    New objects defined to enable the quadratic optimization of the signal-to-noise ratio.

pith-pipeline@v0.9.0 · 5767 in / 1618 out tokens · 61456 ms · 2026-05-20T01:20:10.017712+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

39 extracted references · 39 canonical work pages

  1. [1]

    In each subtreeT d, one can choose a path from the root nodev d to a bottom leaf node ˜v0 ∈V(T d) [dashed arcs in Fig

    Node-state probability The node-state probability for a nodev d =∂ −td (d= 0,· · ·, ℓ, wheret 0 =c) on the cycleC c, is computed as [5, 26], nst vd =Wd/Z ,(D1) where the total weight of directed spanning trees directed towardv d is calculated as, Wd = ˜ℓX d′=d+1 Γ−td+1 · · ·Γ −td′ −1 = β˜ℓ−d −1 β−1 .(D2) Here we exploited the fact that for an isolated sub...

    work page

  2. [2]

    Edge current and traffic The steady-state edge current on a twig that lies on the cyclet d ∈C c (d= 1,2,· · ·, ℓ) is jtd(nst v ) =n st vd−1 −βn st vd = 1/Z ,(D7) 10 where we used unit forward rate and backward rateβ, ∂+td =v d−1,∂ −td =v d and (D1). It is equal to the average reset current at the chordc, (48), ⟨ ⟨w⟩ ⟩=jc(nst v ) =n st vℓ = 1/Z .(D8) The e...

    work page

  3. [3]

    Then the (t d, c) element of the matrix ˜Fis ˜Ftd,c =− Γ−c Γ−td eAc+APvℓ ←vd =− η β 1 η 1 β ℓ−d =−β d−˜ℓ , (D12) which is independent ofη

    Matrices and vectors We set the backward transition rate of the chordcto a positive infinitesimal, Γ −c =η. Then the (t d, c) element of the matrix ˜Fis ˜Ftd,c =− Γ−c Γ−td eAc+APvℓ ←vd =− η β 1 η 1 β ℓ−d =−β d−˜ℓ , (D12) which is independent ofη. From this expression, the Gram matrix (B3) and the weighted cycle Laplacian (B6) become ˜BB T = ˜ℓ−1X d=0 β−d ...

    work page

  4. [4]

    (D25) 11 In the limitℓ→ ∞, (D25) diverges forβ <1

    Steady-state TURs For the pseudo-entropy bound, σpseudo 2⟨ ⟨w⟩ ⟩= ℓX d=1 jtd gtd + jc gc = ℓX d=1 β−1 2β˜ℓ−d+1 −β−1 + 1. (D25) 11 In the limitℓ→ ∞, (D25) diverges forβ <1. Forβ >1, by replacing summation with integral, we obtain, σpseudo 2⟨ ⟨w⟩ ⟩≈1 + β−1 (β+ 1) lnβ ln 2β2 (β−1)(1 + 2β) .(D26) Asβ→1 +, the pseudo-entropy exhibits a logarithmic divergence,σ...

    work page

  5. [5]

    Bryant, The algebra and topology of electrical net- works, Proceedings of the IEE - Part C: Monographs108, 215 (1961)

    P. Bryant, The algebra and topology of electrical net- works, Proceedings of the IEE - Part C: Monographs108, 215 (1961)

    work page 1961

  6. [6]

    Branin, Computer methods of network analysis, Pro- ceedings of the IEEE55, 1787 (1967)

    F. Branin, Computer methods of network analysis, Pro- ceedings of the IEEE55, 1787 (1967)

    work page 1967

  7. [7]

    Takahashi,Theory of Linear Lumped Parameter Sys- tem I, Kisokougaku, Iwanami kouza No

    H. Takahashi,Theory of Linear Lumped Parameter Sys- tem I, Kisokougaku, Iwanami kouza No. 6 (Iwanami shoten, Tokyo, 1969)

    work page 1969

  8. [8]

    Rasmussen, K

    S. Rasmussen, K. Christensen, S. Pedersen, L. Kris- tensen, T. Bækkegaard, N. Loft, and N. Zinner, Su- perconducting circuit companion—an introduction with worked examples, PRX Quantum2, 040204 (2021)

    work page 2021

  9. [9]

    Schnakenberg, Network theory of microscopic and macroscopic behavior of master equation systems, Re1978v

    J. Schnakenberg, Network theory of microscopic and macroscopic behavior of master equation systems, Re1978v. Mod. Phys.48, 571 (1976)

    work page 1976

  10. [10]

    Dal Cengio, V

    S. Dal Cengio, V. Lecomte, and M. Polettini, Geometry of nonequilibrium reaction networks, Phys. Rev. X13, 021040 (2023)

    work page 2023

  11. [11]

    Yoshimura, A

    K. Yoshimura, A. Kolchinsky, A. Dechant, and S. Ito, Housekeeping and excess entropy production for general nonlinear dynamics, Phys. Rev. Res.5, 013017 (2023)

    work page 2023

  12. [12]

    Andrieux and P

    D. Andrieux and P. Gaspard, Fluctuation theorem for currents and schnakenberg network theory, Journal of Statistical Physics127, 107 (2007)

    work page 2007

  13. [13]

    Polettini and M

    M. Polettini and M. Esposito, Transient fluctuation the- orems for the currents and initial equilibrium ensembles, Journal of Statistical Mechanics: Theory and Experi- ment2014, P10033 (2014)

    work page 2014

  14. [14]

    Pietzonka, A

    P. Pietzonka, A. C. Barato, and U. Seifert, Affinity- and topology-dependent bound on current fluctuations, Journal of Physics A: Mathematical and Theoretical49, 34LT01 (2016)

    work page 2016

  15. [15]

    Polettini, A

    M. Polettini, A. Lazarescu, and M. Esposito, Tightening the uncertainty principle for stochastic currents, Phys. Rev. E94, 052104 (2016)

    work page 2016

  16. [16]

    Utsumi, Network analysis for the steady-state thermo- dynamic uncertainty relation, Phys

    Y. Utsumi, Network analysis for the steady-state thermo- dynamic uncertainty relation, Phys. Rev. E110, 044103 (2024)

    work page 2024

  17. [17]

    C. H. Bennett, The thermodynamics of computation—a review, International Journal of Theoretical Physics21, 905 (1982)

    work page 1982

  18. [18]

    D. H. Wolpert, J. Korbel, C. W. Lynn, F. Tasnim, J. A. Grochow, G. Karde¸ s, J. B. Aimone, V. Balasub- ramanian, E. D. Giuli, D. Doty, N. Freitas, M. Marsili, T. E. Ouldridge, A. W. Richa, P. Riechers, ´Edgar Rold´ an, B. Rubenstein, Z. Toroczkai, and J. Paradiso, Is stochastic thermodynamics the key to understanding the energy costs of computation?, Proc...

    work page doi:10.1073/pnas.2321112121 2024

  19. [19]

    Manzano, G

    G. Manzano, G. Karde¸ s, E. Rold´ an, and D. H. Wolpert, Thermodynamics of computations with absolute irre- versibility, unidirectional transitions, and stochastic com- putation times, Phys. Rev. X14, 021026 (2024)

    work page 2024

  20. [20]

    Utsumi, Y

    Y. Utsumi, Y. Ito, D. Golubev, and F. Peper, Com- putation time and thermodynamic uncertainty relation of brownian circuits, (2022), arXiv:2205.10735 [cond- mat.stat-mech]

    work page arXiv 2022

  21. [21]

    Utsumi, D

    Y. Utsumi, D. Golubev, and F. Peper, Thermodynamic cost of brownian computers in the stochastic thermody- namics of resetting, The European Physical Journal Spe- cial Topics232, 3259 (2023)

    work page 2023

  22. [22]

    Nagaosa, Problems related to superconductivity, in Quantum Field Theory in Condensed Matter Physics (Springer Berlin Heidelberg, Berlin, Heidelberg, 1999) pp

    N. Nagaosa, Problems related to superconductivity, in Quantum Field Theory in Condensed Matter Physics (Springer Berlin Heidelberg, Berlin, Heidelberg, 1999) pp. 113–160

    work page 1999

  23. [23]

    Fazio and G

    R. Fazio and G. Sch¨ on, Charge and vortex dynamics in arrays of tunnel junctions, Phys. Rev. B43, 5307 (1991)

    work page 1991

  24. [24]

    Ulrich and F

    J. Ulrich and F. Hassler, Dual approach to circuit quan- tization using loop charges, Phys. Rev. B94, 094505 (2016)

    work page 2016

  25. [25]

    Polettini, Nonequilibrium thermodynamics as a gauge theory, EPL97, 30003 (2012)

    M. Polettini, Nonequilibrium thermodynamics as a gauge theory, EPL97, 30003 (2012)

    work page 2012

  26. [26]

    Okajima and K

    K. Okajima and K. Hukushima, Phase transitions in time complexity of brownian circuits, (2026), arXiv:2512.24728 [cond-mat.stat-mech]

    work page arXiv 2026

  27. [27]

    Fuchs, S

    J. Fuchs, S. Goldt, and U. Seifert, Stochastic thermo- dynamics of resetting, Europhysics Letters113, 60009 (2016)

    work page 2016

  28. [28]

    J. D. Norton, Brownian computation is thermodynami- cally irreversible, Foundations of Physics43, 1384 (2013)

    work page 2013

  29. [29]

    Strasberg, J

    P. Strasberg, J. Cerrillo, G. Schaller, and T. Brandes, Thermodynamics of stochastic turing machines, Phys. Rev. E92, 042104 (2015)

    work page 2015

  30. [30]

    Weidlich, On the structure of exact solutions of discrete masterequations, Zeitschrift f¨ ur Physik B Con- 12 densed Matter30, 345 (1978)

    W. Weidlich, On the structure of exact solutions of discrete masterequations, Zeitschrift f¨ ur Physik B Con- 12 densed Matter30, 345 (1978)

    work page 1978

  31. [31]

    P. W. Gross and P. R. Kotiuga,Electromagnetic theory and computation: a topological approach, Mathematical Sciences Research Institute Publications, Vol. 48 (Cam- bridge University Press, Cambridge, 2004)

    work page 2004

  32. [32]

    Flindt, T

    C. Flindt, T. c. v. Novotn´ y, A. Braggio, and A.-P. Jauho, Counting statistics of transport through coulomb blockade nanostructures: High-order cumulants and non- markovian effects, Phys. Rev. B82, 155407 (2010)

    work page 2010

  33. [33]

    J. M. Horowitz and T. R. Gingrich, Thermodynamic uncertainty relations constrain non-equilibrium fluctua- tions, Nature Physics16, 15 (2020)

    work page 2020

  34. [34]

    Shiraishi, Optimal thermodynamic uncertainty rela- tion in markov jump processes, Journal of Statistical Physics185, 19 (2021)

    N. Shiraishi, Optimal thermodynamic uncertainty rela- tion in markov jump processes, Journal of Statistical Physics185, 19 (2021)

    work page 2021

  35. [35]

    A. Pal, S. Reuveni, and S. Rahav, Thermodynamic un- certainty relation for systems with unidirectional transi- tions, Phys. Rev. Res.3, 013273 (2021)

    work page 2021

  36. [36]

    T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein, Introduction to Algorithms, fourth edition, fourth edition ed. (The MIT Press, 2022) p. 1312

    work page 2022

  37. [37]

    (2026), see Supplemental Material at [URL], a Mathe- matica notebook (.nb) file containing detailed derivations and numerical calculations for the second-order pertur- bation theory

    work page 2026

  38. [38]

    J. Lee, F. Peper, S. Cotofana, M. Naruse, M. Ohtsu, T. Kawazoe, Y. Takahashi, T. Shimokawa, L. B. Kish, and T. Kubota, Brownian circuits: Designs, Interna- tional Journal of Unconventional Computing12, 341 (2016)

    work page 2016

  39. [39]

    Utsumi, Join gate with memory in token-conserving brownian circuits and the thermodynamic cost, Phys

    Y. Utsumi, Join gate with memory in token-conserving brownian circuits and the thermodynamic cost, Phys. Rev. E113, 024129 (2026)

    work page 2026