Pith. sign in

REVIEW 2 major objections 3 minor 48 references

Thermodynamic geometry of friction on graphs: Resistance, commute times, and optimal transport

T0 review · 2 major / 3 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read For slowly driven Markov chains, the thermodynamic friction metric is the same quadratic form as the graph Laplacian pseudoinverse, effective resistance, and mean commute time; and the thermodynamic distance is a discrete optimal-transport

desk verdict The core reversible-chain result is sound and worth publishing after fixes; the NESS generalization and the continuous-space extension both overreach. read the letter →

arxiv 2601.01273 v3 pith:5EIMALVM submitted 2026-01-03 cond-mat.stat-mech

Jordan R Sawchuk , David A Sivak This is my paper · ORCID
classification cond-mat.stat-mech
keywords thermodynamicgeometryfrictionmetricMarkovchainslinearresponsegraphLaplacianresistancedistancecommutetimeoptimaltransport
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 claims that the thermodynamic friction metric governing energy dissipation in slowly driven continuous-time Markov chains is, as a quadratic form, exactly the pseudoinverse of the weighted graph Laplacian. Because that Laplacian also determines effective electrical resistance and mean commute (round-trip) times, linear-response dissipation in any such network is equivalent to Joule heating in a resistor circuit and to a Euclidean embedding in which distance between states is commute time. The same framework identifies the linear-response thermodynamic distance as a discrete L2-Wasserstein optimal-transport cost evaluated along paths of equilibrium distributions. If these equivalences hold, a wide class of thermodynamic metric calculations reduce to circuit algebra, and dissipation acquires a concrete physical picture as the cost of routing probability through dynamical bottlenecks.

What carries the argument

The object that carries the whole equivalence is the weighted graph Laplacian L = −W Dπ, whose off-diagonal entries are the negative equilibrium fluxes w(x|y)π(y) and whose diagonal entries are the total outgoing fluxes. Its pseudoinverse L+ acts as the inverse on the subspace of probability-conserving directions; the projectors I − π1ᵀ strip the non-physical null direction and make βg, L+, −½ Reff, and −½ C numerically interchangeable as metrics on the simplex. The same Laplacian supplies the continuity equation π̇ = Lφ used in the optimal-transport representation, with φ the velocity (node) potential and edge currents given by Ohm's law.

What would settle it

Numerically solve the master equation for a three-state Markov chain driven through a cycle by time-varying energies, compute the exact mean excess work, and check whether the leading term as protocol duration τ goes to infinity matches the circuit prediction βg ≃ L+; the equivalence should hold to first order in 1/τ, with deviations appearing only at higher order. For the non-equilibrium steady-state extension, drive a three-state cycle with a fixed non-conservative force and compare the ratio of linear-response excess work to the detailed-balanced case against the predicted factor α = (a0+a1

Watch

Extended reading notes

Core claim

The paper's central claim is that, for a reversible continuous-time Markov chain driven slowly by conservative changes to state energies, the thermodynamic friction metric on the space of equilibrium distributions is the same quadratic form as the pseudoinverse of the weighted graph Laplacian, the electrical effective resistance, and the symmetrized mean first-passage (commute) time: βg ≃ L+ ≃ −½ Reff ≃ −½ C on the probability simplex. The argument uses the fact that the edge weights of the Laplacian are precisely the equilibrium fluxes, and that the generalized inverse of the rate matrix can be expressed through mean first-passage times, so all four geometries share one object. The paper al

Load-bearing premise

The load-bearing premise is the linear-response approximation: the protocol must be slow enough that the lag between the actual and instantaneous equilibrium distributions is small and proportional to the driving speed, so that the excess work is exactly the quadratic form assumed; if this fails for fast or strong driving, the metric equivalence and the circuit and optimal-transport identities no longer describe the dissipation.

Editorial extensions

If this is right

  • Friction metrics for arbitrary Markov graphs can be computed with standard circuit rules (series and parallel reduction, Kron reduction) instead of matrix inversions, and adding an edge always lowers the linear-response dissipation by a classical monotonicity property of effective resistance.
  • The linear-response work cost of moving a small probability mass between two states is kBT times the squared commute-time distance between them, so bottlenecks appear as large embedded distances that are genuinely costly to traverse.
  • The squared thermodynamic distance between equilibrium distributions is a discrete optimal-transport cost evaluated along equilibrium paths, linking stochastic thermodynamics to discrete Wasserstein geometry and providing variational principles for minimal-dissipation protocols.
  • For non-equilibrium steady states with fixed non-conservative forces, the equivalence survives with the metric given by the symmetric part of the generalized Laplacian pseudoinverse, and stationary currents reduce the linear-response excess work compared with a detailed-balanced system with the same edge traffic.

Reading between the lines

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

  • The correspondence suggests a practical route the paper does not develop: because commute times and effective resistances can be estimated from trajectory data and spectral methods, thermodynamic friction metrics could be inferred directly from simulations or experiments on molecular and biological networks without reconstructing the full energy landscape.
  • If the discrete optimal-transport structure extends beyond the equilibrium-restricted setting, minimal-dissipation protocols on finite graphs may be obtainable as geodesics in a discrete Wasserstein geometry, potentially leading to finite-graph analogues of counterdiabatic driving that go beyond linear response.
  • The resistor-network picture implies that entropic bottlenecks—sparse connectivity—impose a dissipation cost that no conservative control can eliminate; this could be tested by comparing the excess work of two graphs that have identical equilibrium energies and populations but differ only in how edges are wired.
  • The NESS result that stationary currents reduce dissipation regardless of orientation is suggestive for autonomous molecular machines: it hints that background nonequilibrium flows can assist externally driven probability transport at no additional linear-response cost, a prediction that could be checked in a three-state cycle experiment.
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 / 3 minor

Summary. The paper establishes an equivalence, for slowly driven continuous-time Markov chains in the linear-response regime, between the thermodynamic friction metric on the probability simplex and three graph-theoretic objects: the Moore–Penrose pseudoinverse of the weighted graph Laplacian, the effective resistance (via a resistor-network mapping), and the symmetrized mean first-passage (commute) time. The central chain is Eq. (12): βg ~ L^+ ~ −(1/2)R_eff ~ −(1/2)C. The paper also derives closed-form friction metrics for linear and cyclic graphs, interprets linear-response dissipation as Joule heating, connects the thermodynamic distance to a restricted discrete L^2-Wasserstein cost, and claims extensions to nonequilibrium steady states (Appendix A) and continuous state spaces (Section VII and Appendix C).

Significance. The discrete reversible result is valuable and appears sound: it unifies linear-response thermodynamics with classical random-walk and circuit theory, and it provides practical computational tools (circuit reduction) for friction metrics. The derivation builds on classical identities (deviation matrix vs. MFPTs, resistance distance vs. Laplacian pseudoinverse) and the reader's hand-checked verification of the 2- and 3-state examples supports the core equivalence. The resistor-network picture, including the exact formulas for linear and cyclic topologies, is a useful contribution. However, the paper's claims beyond the detailed-balance discrete setting are not supported: the NESS generalization rests on a false response identity, and the continuous-space commute-time kernel is not finite in d≥2. These overclaims need to be corrected before the paper can be recommended for publication.

major comments (2)
  1. [Appendix A, Eq. (A2)] The NESS generalization rests on Eq. (A2), ∂π/∂(βV) = ππ^T − Dπ, cited to Ref. [41] for fixed nonconservative forces. This identity is not valid for general irreducible rate matrices satisfying local detailed balance with fixed affinities. For example, take the 3-state cycle with k01=e^{1−ε}, k10=1, k12=1, k21=e^{−1−ε}, k20=1, k02=e^{−1} and energies V=(0,ε,0). Differentiating Wπ=0 at ε=0 gives dπ/dε≈(0.163,−0.175,0.012), whereas Eq. (A2) predicts (0.098,−0.240,0.142). Consequently Eqs. (A3)–(A4) and the claimed survival of Eq. (12) for NESS do not follow. The NESS statements in Section III and the Introduction must be removed or rederived under assumptions that actually guarantee the response identity.
  2. [Sec. VII / App. C, Eqs. (32)/(C1)] The continuous-space commute-time kernel C(x,y) as defined is not finite for reversible diffusions in d≥2. The heat-kernel short-time asymptotics give p_t(x|x)−p_t(x|y) ∼ (4πDt)^{-d/2} for t→0 when x≠y, so the t-integral diverges at t=0. Exponential relaxation of the generator (the 'standard assumptions' cited) controls only the large-t tail, not the short-time singularity. Thus the statement that 'the integral in (32) remains finite' is incorrect, and the proof of βg ∼ −(1/2)C for continuous spaces is not well-defined. The authors should restrict the continuous claim to d=1 (where the expression matches the classical MFPT) or introduce a regularized kernel and clearly state the resulting approximation.
minor comments (3)
  1. [Section IV, Eq. (15)] The infimum in Eq. (15) is written only with the constraint ˙π_s = L_s φ_s. Please specify explicitly that the infimum is over equilibrium paths π_s connecting π_0 to π_1, i.e., π_0 and π_1 fixed and π_s ∈ Δ_n for all s, so the 'equilibrium-path-restricted' nature of the OT cost is unambiguous.
  2. [Section V A, Eq. (18)] Please check the physical dimensions in Eq. (18). The commute-time distance C(x,y) has units of time (the MFPTs are times), so ε^2 k_B T C(x,y) does not have units of energy unless an implicit rate/frequency scale is specified. If rescaled variables or dimensionless rates are used, state this directly.
  3. [General notation] The equivalence symbol used in Eq. (12) and elsewhere is compact but can confuse: R_eff is defined as a two-point function, while the equivalence is on the tangent space of the simplex. For readability, explicitly define the quadratic-form equivalence for R_eff and C, e.g., ˙π^T βg ˙π = −(1/2) Σ_{x,y} ˙π(x) R_eff(x,y) ˙π(y), before using the shorthand.

Circularity Check

1 steps flagged · score 3.0 of 10

Main resistance/commute-time chain is independently grounded; the discrete-OT equivalence is chosen by construction.

  1. self definitional [Appendix B, Eq. (B9); see also Sec. IV, Eq. (15)]
    "Under the restriction p_s = π_s, and ω, θπ chosen such that ω(x, y)θπ(x, y) = wπ(x|y)π(y), the L2-Wasserstein distance (B4) coincides exactly with the expression (15) for the thermodynamic distance."

    The discrete OT metric (B4) is defined only after choosing a flux function θ_p and edge weights ω. The paper imposes condition (B9) so that the OT cost equals the thermodynamic cost. Since θ is a free ingredient in the definition of the discrete OT metric, this equality is manufactured by selecting θπ = wπ(x|y)π(y) to reproduce the target; it is not an independent theorem. The advertised conclusion that thermodynamic distance is a discrete L2-Wasserstein cost is thus true by construction, although Eq. (12) itself is unaffected because it relies on external graph identities.

full rationale

Equation (12) is not circular: it follows from the published friction metric (4), the algebraic relation L = −W Dπ, the pseudoinverse relation (8), the classical effective-resistance formula (9), and the external deviation-matrix identity (11) from [21]. The circuit results in Sec. VI are derived from those identities plus Thompson's principle, not assumed. Self-citations [6,17,32] are prior published parameter-free derivations whose assumptions do not include the graph-resistance/commute-time equivalence, so they count as independent support. The one definitional element is the discrete-OT correspondence: Appendix B chooses the OT flux function θπ and weights ω so that the OT cost coincides with the thermodynamic cost, making that claimed unification true by construction rather than by a derived theorem. The Appendix A NESS extension relies on an external response identity, Eq. (A2) from [41]; whether that identity is correct is a correctness concern, not a circularity concern. Overall, the central resistance/commute-time result has substantial independent content; the circularity score is moderate only because of the construction-based OT identification.

Assumptions & free parameters 1 free parameters · 7 assumptions · 0 invented entities

No parameter fitting anywhere in the paper: the friction metric is computed from the given rate matrix and stationary distribution, and the cycle/NESS closed forms are parameter-free derivations. The only hand-picked numbers are conventions of the OT representation (omega=1, gauge <phi>_pi=0), which drop out of all physical statements. The load-bearing axioms are the LR premise (Eqs 2, 23), the equilibrium response formula (A2), the deviation-matrix identity (11), and the classical resistance/commute-time theorems; Thompson's principle supplies the cycle calculation. The continuous extension rests on an additional assertion (finiteness of (C1) for d>=2) that is incorrect, so that part should be read as d=1-only, where it is the classical Bicout-Szabo identity. No invented entities: the potentials phi and the resistor network are exact mathematical images of the linear-response lag (Eqs 24-25), not new physical entities.

free parameters (1)
  • omega(x,y): edge weights of the Wasserstein graph in the OT representation = 1 for (x,y) in E, 0 otherwise (Eq B10)
    Representational convention in App B: with omega fixed and theta_pi := w_pi(x|y)pi(y)/omega, the OT cost (B4) reproduces the thermodynamic distance. The choice does not affect the physical metric; not fitted to any data.
assumptions (7)
  • domain assumption Linear-response quadratic excess work (Eq 2) with friction tensor zeta = -beta W^D D_pi and lag approximation (23): delta p ~ tau_prot^-1 W^D D_pi pi_dot
    Central framework taken from [6], [17], [32]; restricts every result in the paper to slow driving (small lag). Without it, the 'metric' reading of dissipation and the Joule-heating/OT rephrasings do not hold.
  • domain assumption Equilibrium response relation d pi / d(beta V) = pi pi^T - D_pi under conservative driving (Eq A2)
    Exact at equilibrium; for NESS driving cited from [41]. It converts the energy-space metric (2) into the simplex metric (4), underwriting the whole change of variables.
  • standard math Deviation-matrix identity W^D = D_pi T_mfp (I - pi 1^T) (Eq 11)
    Known theorem for ergodic continuous-time Markov chains [21]; the bridge from the friction metric to mean first-passage times and commute times.
  • standard math Classical resistance-distance and commute-time identities: Reff(x,y) = L+(x,x) + L+(y,y) - 2 L+(x,y), the quadratic-form identity pi_dot^T R pi_dot = -2 pi_dot^T L+ pi_dot on the tangent space, and the squared-Euclidean commute-time embedding (Eq 17)
    Classical — Klein-Randic [16], Chandra et al. [15], Doyle-Steiner [14]; the paper relies on these to close the equivalence chain (12). Verified by the reviewer on 2- and 3-state examples.
  • standard math Thompson's principle (dissipated power minimized by the true currents under KCL) and Rayleigh's monotonicity (adding edges lowers effective resistance)
    Used in Sec VI B to fix the cycle current i_cyc and to interpret the sign of the correction (29)-(31); cited to [8].
  • domain assumption The rate law must make theta_pi(x,y) = w_pi(x|y) pi(y) a symmetric generalized mean of pi(x), pi(y) for the canonical discrete-Wasserstein identification (B9)-(B13)
    The OT equivalence (15)=(B4) is exact only when the flux function is a proper generalized mean — shown for the sqrt-pi-product rates (geometric mean) and Glauber rates (harmonic mean), not for arbitrary rate laws. This restricts the OT section's scope; disclosed in App B.
  • ad hoc to paper Finiteness of the continuous commute-time kernel C(x,y) (C1)/(32) for d>=2 under exponential relaxation
    Asserted in Sec VII ('the integral in (32) remains finite'); false for d>=2 because the short-time heat-kernel difference diverges like t^(-d/2) (exponential relaxation controls only the t->infinity tail). The App C equivalence is therefore vacuous in d>=2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Thermodynamic geometry of friction on graphs: Resistance, commute times, and optimal transport." pith.science (2026). https://pith.science/paper/5EIMALVM

@misc{pith2026260101273,
  author       = {Pith},
  title        = {Pith review of: Thermodynamic geometry of friction on graphs: Resistance, commute times, and optimal transport},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5EIMALVM}},
  note         = {Machine review of arXiv:2601.01273}
}
abstract

We demonstrate that the thermodynamic friction metric governing dissipation in slowly driven continuous-time Markov chains is equivalent to the commute-time embedding and the resistance distance. This equivalence yields complementary insights: The commute-time embedding demonstrates the intrinsic cost of transporting probability across dynamical bottlenecks, while the resistance distance maps thermodynamic dissipation to Joule heating in an electrical network. We further demonstrate that the linear-response thermodynamic distance is a discrete $L^2$-Wasserstein optimal transport cost evaluated along paths of equilibrium distributions, extending a continuous-state correspondence to discrete networks. This conceptual synthesis of linear-response thermodynamics, random walks on graphs, electrical circuits, and optimal-transport theory connects independently developed geometric frameworks, reduces complex metric calculations to simple circuit algebra, and provides a clear physical picture of dissipation as the energetic cost of routing probability through the state space network.

Figures

Figures reproduced from arXiv: 2601.01273 by the authors.

Figure 1
Figure 1. FIG. 1. (a) The Markov graph for a linear chain of states and [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Three-state cycle with stationary current [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 2 linked inside Pith

  1. [41]

    G. A. Pavliotis,Stochastic Processes and Applications: Diffusion Processes, the Fokker-Planck and Langevin Equations, Vol. 60 (Springer Nature, New York, 2014)

    2014

  2. [1]

    Three-state cycle For a three-state cycle driven by fixed edge affinities A(x, y) maintaining a stationary currentj ss (Fig. 2), the dissipation is scaled down compared to the detailed- balance case: ⟨ ˙Wex⟩LR =α⟨ ˙Wex⟩LR db (A12) with α= Rcyc Rcyc + 1 4 j2ss r(0)r(1)r(2) .(A13) HereR cyc andr(x) are defined as in Sec. VI B. More transparently, define the...

  3. [2]

    IV, following [25] and later taking the conventions of [26]

    Discrete calculus Here we provide the formal definitions for discrete cal- culus used in Sec. IV, following [25] and later taking the conventions of [26]. The need for a careful treatment can be seen in the expressionρ s∇ϕs in the continuity equa- tion (14): because continuous vector fields map to edge functions and scalars to node functions, the product ...

  4. [3]

    IV is a special case of the met- ric for probability transport on finite graphs [26–28, 44]

    Connections to previous work on discrete OT We show here that the restrictedL 2-Wasserstein met- ric (15) defined in Sec. IV is a special case of the met- ric for probability transport on finite graphs [26–28, 44]. Consider a weighted graphG= (Ω,E, ω) with vertex set 9 Ω, edge setE, and symmetric edge weightsω(x, y) = ω(y, x)>0 for (x, y)∈ E. The discrete...

  5. [4]

    Weinhold, Metric geometry of equilibrium thermody- namics, J

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

    1975

  6. [5]

    Salamon and R

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

    1983

  7. [6]

    Ruppeiner, Riemannian geometry in thermodynamic fluctuation theory, Rev

    G. Ruppeiner, Riemannian geometry in thermodynamic fluctuation theory, Rev. Mod. Phys.67, 605 (1995)

    1995

  8. [7]

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

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

    2018

Show all 48 references
  1. [8]

    Dechant, S

    A. Dechant, S. Sasa, and S. Ito, Geometric decomposi- tion of entropy production in out-of-equilibrium systems, Phys. Rev. Res.4, L012034 (2022)

    2022

  2. [9]

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

    2012

  3. [10]

    Zhong and M

    A. Zhong and M. DeWeese, Beyond linear response: Equivalence between thermodynamic geometry and op- timal transport, Phys. Rev. Lett.133, 057102 (2024)

    2024

  4. [11]

    P. G. Doyle and J. L. Snell, Random walks and electric networks (2000), arXiv:math/0001057 [math.PR]

    2000 arXiv

  5. [12]

    Ghosh, S

    A. Ghosh, S. Boyd, and A. Saberi, Minimizing effec- tive resistance of a graph, SIAM Rev. 10.1137/050645452 (2008)

    2008 doi

  6. [13]

    D. A. Spielman and N. Srivastava, Graph sparsifica- tion by effective resistances, SIAM J. Comput.40, 1913 (2011)

    1913

  7. [14]

    Y. Deng, Q. Dai, R. Wang, and Z. Zhang, Commute time guided transformation for feature extraction, Comput. Vis. Image Underst.116, 473 (2012)

    2012

  8. [15]

    Fitch, Effective resistance preserving directed graph symmetrization, SIAM J

    K. Fitch, Effective resistance preserving directed graph symmetrization, SIAM J. Matrix Anal. Appl.40, 49 (2019)

    2019

  9. [16]

    J. R. Sato, C. M. Sato, M. K. d. C. Silva, and C. E. Bi- azoli, Commute time as a method to explore brain func- tional connectomes, Brain Connect.9, 155 (2019)

    2019

  10. [17]

    P. G. Doyle and J. Steiner, Commuting time geometry of ergodic Markov chains (2017), arXiv:1107.2612 [math]

    2017 arXiv

  11. [18]

    A. K. Chandra, P. Raghavan, W. L. Ruzzo, and R. Smolensky, The electrical resistance of a graph cap- tures its commute and cover times, inProceedings of the twenty-first annual ACM symposium on Theory of com- puting - STOC ’89(ACM Press, Seattle, Washington, United States, 198...

    1989

  12. [19]

    D. J. Klein and M. Randi´ c, Resistance distance, J. Math. Chem.12, 81 (1993)

    1993

  13. [20]

    J. R. Sawchuk and D. A. Sivak, Global thermodynamic manifold for conservative control of stochastic systems, Phys. Rev. Res.8, 013004 (2026)

    2026

  14. [21]

    Mirzaev and J

    I. Mirzaev and J. Gunawardena, Laplacian dynamics on general graphs, Bull. Math. Biol.75, 2118 (2013)

    2013

  15. [22]

    G. Wang, Y. Wei, and S. Qiao,Generalized Inverses: Theory and Computations, Developments in Mathemat- ics, Vol. 53 (Springer, Singapore, 2018)

    2018

  16. [23]

    Yadav and K

    M. Yadav and K. Thulasiraman, Network science meets circuit theory: Kirchhoff index of a graph and the power of node-to-datum resistance matrix, inISCAS, Vol. 2015- (IEEE, New York, 2015) pp. 854–857

    2015

  17. [24]

    Coolen-Schrijner and E

    P. Coolen-Schrijner and E. A. van Doorn, The deviation matrix of a continuous-time Markov chain, Probab. Eng. Inf. Sci.16, 351 (2002)

    2002

  18. [25]

    Van Vu and K

    T. Van Vu and K. Saito, Thermodynamic unification of optimal transport: Thermodynamic uncertainty relation, minimum dissipation, and thermodynamic speed limits, Phys. Rev. X13, 011013 (2023)

    2023

  19. [26]

    Benamou and Y

    J.-D. Benamou and Y. Brenier, A computational fluid mechanics solution to the Monge-Kantorovich mass transfer problem, Numer. Math.84, 375 (2000)

    2000

  20. [27]

    Otto, The geometry of dissipative evolution equations: 11 The porous medium equation, Commun

    F. Otto, The geometry of dissipative evolution equations: 11 The porous medium equation, Commun. Partial Differ. Equ.26, 101 (2001)

    2001

  21. [28]

    L. J. Grady and J. R. Polimeni,Discrete Calculus: Applied Analysis on Graphs for Computational Science (Springer-Verlag London, 2010)

    2010

  22. [29]

    Maas, Gradient flows of the entropy for finite Markov chains, J

    J. Maas, Gradient flows of the entropy for finite Markov chains, J. Funct. Anal.261, 2250 (2011)

    2011

  23. [30]

    S.-N. Chow, W. Huang, Y. Li, and H. Zhou, Fokker– Planck equations for a free energy functional or Markov process on a graph, Arch. Ration. Mech. Anal.203, 969 (2012)

    2012

  24. [31]

    Li, Transport information geometry: Riemannian cal- culus on probability simplex, Inf

    W. Li, Transport information geometry: Riemannian cal- culus on probability simplex, Inf. Geom.5, 161 (2022)

    2022

  25. [32]

    Wang, Classical Multidimensional Scaling, inGeomet- ric Structure of High-Dimensional Data and Dimension- ality Reduction(Springer Berlin Heidelberg, Berlin, Hei- delberg, 2012) pp

    J. Wang, Classical Multidimensional Scaling, inGeomet- ric Structure of High-Dimensional Data and Dimension- ality Reduction(Springer Berlin Heidelberg, Berlin, Hei- delberg, 2012) pp. 115–129

    2012

  26. [33]

    Zwanzig, Dynamical disorder: Passage through a fluc- tuating bottleneck, J

    R. Zwanzig, Dynamical disorder: Passage through a fluc- tuating bottleneck, J. Chem. Phys.97, 3587 (1992)

    1992

  27. [34]

    Chakrabarti and B

    D. Chakrabarti and B. Bagchi, Waiting time distribution and nonexponential relaxation in single molecule spec- troscopic studies: Realization of entropic bottleneck in a simple model, J. Chem. Phys.118, 7965 (2003)

    2003

  28. [35]

    Mandal and C

    D. Mandal and C. Jarzynski, Analysis of slow transitions between nonequilibrium steady states, J. Stat. Mech.: Theory Exp.2016(6), 063204

    2016

  29. [36]

    J. E. Avron, M. Fraas, G. M. Graf, and P. Grech, Adi- abatic theorems for generators of contracting evolutions, Commun. Math. Phys.314, 163 (2012)

    2012

  30. [37]

    M. M. Lin, Circuit reduction of heterogeneous nonequi- librium systems, Phys. Rev. Lett.125, 218101 (2020)

    2020

  31. [38]

    Dorfler and F

    F. Dorfler and F. Bullo, Kron reduction of graphs with applications to electrical networks, IEEE Trans. Circuits Syst. I: Regul. Pap.60, 150 (2013)

    2013

  32. [39]

    P. R. Zulkowski and M. R. DeWeese, Optimal control of overdamped systems, Phys. Rev. E92, 032117 (2015)

    2015

  33. [40]

    D. J. Bicout and A. Szabo, First passage times, correla- tion functions, and reaction rates, J. Chem. Phys.106, 10292 (1997)

    1997

  34. [42]

    No´ e, R

    F. No´ e, R. Banisch, and C. Clementi, Commute maps: Separating slowly mixing molecular configurations for kinetic modeling, J. Chem. Theory Comput.12, 5620 (2016)

    2016

  35. [43]

    Maes, Local detailed balance, SciPost Phys

    C. Maes, Local detailed balance, SciPost Phys. Lect. Notes , 32 (2021)

    2021

  36. [44]

    J. A. Owen, T. R. Gingrich, and J. M. Horowitz, Univer- sal thermodynamic bounds on nonequilibrium response with biochemical applications, Phys. Rev. X10, 011066 (2020)

    2020

  37. [45]

    Seabrook and L

    E. Seabrook and L. Wiskott, A tutorial on the spec- tral theory of Markov chains, Neural Comput.35, 1713 (2023)

    2023

  38. [46]

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

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

    1976

  39. [47]

    S.-N. Chow, W. Li, and H. Zhou, Entropy dissipation of Fokker-Planck equations on graphs, Disc. Cont. Dyn. Sys.38, 4929 (2018)

    2018

  40. [48]

    P. D. Dixit, A. Jain, G. Stock, and K. A. Dill, In- ferring transition rates of networks from populations in continuous-time Markov processes, J. Chem. Theory Comput.11, 5464 (2015)

    2015

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.