pith. sign in

arxiv: 2605.17082 · v1 · pith:A76BA47Vnew · submitted 2026-05-16 · 🧮 math.PR

Quantitative Spectral Rigidity and Finite-Time Spectral Thermodynamics in Reversible Markov Chains

Qiao Wang This is my paper

Pith reviewed 2026-05-20 15:10 UTC · model grok-4.3

classification 🧮 math.PR
keywords reversible Markov chainsspectral rigidityfinite-time dynamicsspectral entropypower iterationeigenvalue separationrelaxation operator
0
0 comments X

The pith

Reversible Markov chains reach spectral rigidity in finite time bounded by the ratio of their second and third eigenvalues rather than the usual spectral gap.

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

The paper establishes that reversible Markov chains develop a precise finite-time structure in which the slowest decaying mode comes to dominate the spectral energy at a well-defined moment. The timing of this dominance, called the rigidity time, is controlled by the ratio between the second and third eigenvalues rather than the overall mixing rate alone. Two-sided bounds that differ by at most one step are given for this time when the slowest mode holds all but a small fraction δ of the energy. The same organization yields an exact entropy balance law that tracks how spectral entropy changes and reaches a maximum exactly when the covariance term switches sign. These structures give exact error formulas and stopping criteria when the framework is applied to power iteration on the chain.

Core claim

For reversible Markov chains satisfying λ₂ > λ₃ the first time T_rigid(δ) at which the slowest spectral mode accounts for a (1-δ) fraction of the total spectral energy obeys explicit upper and lower bounds differing by at most one step; these bounds are governed by the separation ratio λ₃/λ₂. The paper further derives the exact balance law ΔS = Cov/ρ − D_KL for the evolution of spectral entropy and shows that the covariance term changes sign precisely at the half-rigidity threshold where entropy equals log 2 in the two-mode setting. When applied to power iteration the same spectral organization supplies the observable variance identity Var_{p_k}[λ²] = ρ_k(ρ_{k+1} − ρ_k) together with a data-

What carries the argument

The relaxation operator G = I − P, which organizes the exact spectral relaxation dynamics and reveals the finite-time rigidity and entropy balance structures.

If this is right

  • Power iteration admits an exact error identity based on the observable spectral variance formula.
  • In the two-mode case spectral entropy reaches its global maximum of log 2 exactly at the half-rigidity threshold.
  • A sharp sufficient rigidity criterion guarantees monotone decay of spectral entropy for general chains.
  • The rigidity time is insensitive to the classical gap 1-λ₂ and is instead set by the separation ratio λ₃/λ₂.

Where Pith is reading between the lines

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

  • The finite-time rigidity picture could be used to design adaptive stopping rules in Monte Carlo sampling that halt once spectral purification is achieved rather than waiting for full asymptotic mixing.
  • The covariance representation of entropy transfer may supply a practical diagnostic for convergence that does not require explicit knowledge of the stationary measure.

Load-bearing premise

The Markov chain is reversible so that the transition operator is self-adjoint on L² with respect to the stationary distribution.

What would settle it

A concrete reversible Markov chain with λ₂ > λ₃ in which the computed rigidity time T_rigid(δ) for small δ lies strictly outside the explicit two-sided bounds that differ by at most one step.

Figures

Figures reproduced from arXiv: 2605.17082 by Qiao Wang.

Figure 1
Figure 1. Figure 1: Emergence of spectral rigidity. The slow-mode fraction α2(k) rises from 0.1 toward 1 as fast modes dissipate. For the threshold δ = 0.1, the theoretical bounds L(δ) and L(δ) + 1 from Theorem 4.4 are shown as vertical dashed lines. The actual crossing of α2(k) = 0.9 occurs at k = 18, well within the predicted interval [L(δ), L(δ) + 1]. For the threshold δ = 0.1, the theoretical bounds give L(0.1) = log(0.9/… view at source ↗
Figure 2
Figure 2. Figure 2: Monotonic decay of the spectral entropy-energy G(k). Left: dispersed initial condition (α2(0) = 0.1). Right: slow-mode concentrated initial condition (α2(0) = 0.7). In both cases, G(k) (solid red curve, normalized by G(0)) decays strictly monotonically, while the total energy Ek (dashed blue) and the spectral entropy Sspec(k) (dotted green) each decay but may cross. The strict monotonicity of G(k) is the c… view at source ↗
Figure 3
Figure 3. Figure 3: Covariance sign flip and spectral entropy. The covariance crosses zero at k ∗ = 4, where α2(k) = 1/2 and Sspec(k) attains 0.6932 ≈ log 2. 8.5. Power iteration: the adaptive stopping criterion [PITH_FULL_IMAGE:figures/full_fig_p025_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Two-phase structure of the entropy change. Cov/ρk (blue) and DKL (red) cross at k ∗ = 4. 10 20 30 40 50 60 Step k 0.4 0.2 0.0 0.2 0.4 0.6 Cumulative k 1 j = 0 Cov/ j k 1 j = 0 DKL S(0) + (Cov/ DKL) Sspec(k) (direct) [PITH_FULL_IMAGE:figures/full_fig_p026_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Verification of the spectral Clausius equality to machine precision. Γk = ρk+1/ρk − 1, computed entirely from the energy ratios ρk = Ek+1/Ek. The red curve shows the true (but in practice unobservable) eigenvector error ∥vk − ϕ2∥π. The close correspondence between Γk and the true error is a consequence of the exact identities (42) and (45): both quantities are controlled by 1 − α2(k), and the observable va… view at source ↗
Figure 6
Figure 6. Figure 6: Adaptive stopping criterion for power iteration. The observable indicator Γk (blue, left axis) and the true eigenvector error ∥vk − ϕ2∥π (red, right axis) decay in close correspondence. Horizontal dashed lines mark error thresholds ε = 0.1 and ε = 0.05. The stopping rule of Theorem 7.6 monitors Γk alone and guarantees the error bound without knowledge of the true eigenvectors. derived in Theorem 7.5. The s… view at source ↗
Figure 7
Figure 7. Figure 7: Chebyshev acceleration of rigidity emergence. The slow-mode fraction α2(k) is shown for the basic power iteration (blue) and the Chebyshev-accelerated iteration with degree m = 4 (red). The horizontal axis counts equivalent power iteration steps (m = 4 per Chebyshev step). Circles mark the first step at which α2(k) ≥ 0.9. The Chebyshev scheme reaches the threshold in approximately 60% of the steps required… view at source ↗
read the original abstract

We study finite-time spectral rigidity in reversible Markov chains via exact spectral relaxation dynamics. While the underlying identities follow classically from self-adjointness on $L^2(\pi)$, organizing the dynamics around the relaxation operator $\mathcal{G}=I-P$ reveals finite-time structures invisible to traditional asymptotic estimates. For chains with $\lambda_2>\lambda_3$, we establish explicit two-sided bounds on the rigidity time $T_{\mathrm{rigid}}(\delta)$, the first moment the slowest mode captures a fraction $1-\delta$ of the total spectral energy. The bounds differ by at most one step and show that rigidity emergence is controlled by the spectral separation ratio $\lambda_3/\lambda_2$, not the classical gap $1-\lambda_2$ alone. We develop a spectral entropy theory governed by the exact balance law $\Delta S=\mathrm{Cov}/\rho-D_{\mathrm{KL}}$ and a canonical covariance representation of entropy transfer. In the two-mode case, the covariance changes sign precisely at the half-rigidity threshold $T_{\mathrm{rigid}}(1/2)$, where spectral entropy attains its maximum $\log 2$. For general chains, we obtain a sharp sufficient rigidity criterion for monotone entropy decay. Applied to power iteration, the framework yields an exact error identity, the observable spectral variance formula $\mathrm{Var}_{p_k}[\lambda^2]=\rho_k(\rho_{k+1}-\rho_k)$, and a fully data-driven adaptive stopping criterion with provable guarantees. These results demonstrate that reversible Markov chains possess a precise finite-time rigidity structure governing spectral purification, entropy dynamics, and observable convergence beyond classical asymptotic theory.

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 / 4 minor

Summary. The manuscript develops a quantitative theory of finite-time spectral rigidity for reversible Markov chains. For chains with λ₂ > λ₃ it derives explicit two-sided bounds on the rigidity time T_rigid(δ) (the first time the slowest mode captures fraction 1-δ of total spectral energy) showing that these bounds differ by at most one step and are controlled by the separation ratio λ₃/λ₂ rather than the classical gap 1-λ₂. It further introduces a spectral entropy theory governed by the exact balance law ΔS = Cov/ρ - D_KL together with a covariance representation of entropy transfer, a sharp sufficient criterion for monotone entropy decay, and an application to power iteration that yields an exact error identity, the observable spectral variance formula Var_{p_k}[λ²] = ρ_k(ρ_{k+1}-ρ_k), and a fully data-driven adaptive stopping criterion.

Significance. If the derivations hold, the work supplies a precise finite-time refinement of spectral analysis for reversible chains that goes beyond classical asymptotic gap estimates. The explicit two-sided bounds, the exact entropy balance law, and the data-driven stopping rule with provable guarantees constitute concrete strengths. These results could usefully inform both theoretical probability and practical MCMC or power-method implementations.

minor comments (4)
  1. The relaxation operator G = I - P is central to the finite-time viewpoint yet is introduced only briefly; a short dedicated paragraph or subsection early in the paper would clarify how it organizes the dynamics differently from the usual transition operator P.
  2. The definition of spectral energy and the precise meaning of the fraction 1-δ captured by the slowest mode should be stated as an explicit equation (with a numbered display) rather than left implicit in the prose description of T_rigid(δ).
  3. A brief comparison with classical references on the spectral theory of reversible chains (e.g., standard texts by Diaconis or Levin-Peres-Wilmer) would help situate the finite-time results against the existing asymptotic literature.
  4. In the power-iteration section, the transition from the exact error identity to the adaptive stopping criterion would be easier to follow if the dependence on the observable spectral variance were summarized in a single displayed equation.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary of the manuscript, recognition of its contributions to finite-time spectral analysis, and recommendation for minor revision. No specific major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper derives explicit two-sided bounds on the rigidity time T_rigid(δ) and the entropy balance law ΔS = Cov/ρ - D_KL directly from the spectral decomposition of the self-adjoint transition operator on L²(π), which is a standard consequence of reversibility and requires no reference to the target quantities themselves. The dominance estimate ||higher modes|| ≤ λ₃^t C and the resulting threshold t* = log(·)/log(λ₂/λ₃) follow from orthonormality and the given spectral ordering λ₂ > λ₃; the integer bounds differing by at most one step are a direct arithmetic consequence of this estimate applied to fixed initial coefficients. The entropy identity is obtained by exact differentiation of the KL functional along the spectral flow and holds independently of the rigidity analysis. No load-bearing self-citations, fitted inputs renamed as predictions, or ansatzes smuggled via prior work appear in the derivation chain; the results remain self-contained against classical spectral theory.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on the standard domain assumption of reversibility and the introduction of the relaxation operator as an organizing device; no free parameters or new postulated entities are evident from the abstract.

axioms (1)
  • domain assumption The Markov chain is reversible, ensuring the transition operator P is self-adjoint on L²(π).
    Invoked at the outset to obtain the spectral decomposition and exact relaxation dynamics around G = I - P.

pith-pipeline@v0.9.0 · 5827 in / 1298 out tokens · 54894 ms · 2026-05-20T15:10:05.374748+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages · 1 internal anchor

  1. [1]

    D. A. Levin, Y. Peres, and E. L. Wilmer,Markov Chains and Mixing Times, 2nd edition, American Mathematical Society, 2017

    work page 2017

  2. [2]

    Aldous and J

    D. Aldous and J. A. Fill,Reversible Markov Chains and Random Walks on Graphs, unfinished monograph, 2002. Available athttps://www.stat.berkeley.edu/ ~{}aldous/RWG/book.html

    work page 2002

  3. [3]

    Saloff-Coste, Lectures on finite Markov chains, inLectures on Probability Theory and Statistics(Saint-Flour, 1996), Lecture Notes in Mathematics, vol

    L. Saloff-Coste, Lectures on finite Markov chains, inLectures on Probability Theory and Statistics(Saint-Flour, 1996), Lecture Notes in Mathematics, vol. 1665, Springer, 1997, pp. 301–413

    work page 1996

  4. [4]

    Saad,Iterative Methods for Sparse Linear Systems, 2nd edition, SIAM, 2003

    Y. Saad,Iterative Methods for Sparse Linear Systems, 2nd edition, SIAM, 2003

    work page 2003

  5. [5]

    L. N. Trefethen,Approximation Theory and Approximation Practice, SIAM, 2013

    work page 2013

  6. [6]

    B. T. Polyak, Some methods of speeding up the convergence of iteration methods,USSR Computational Mathematics and Mathematical Physics,4(1964), 1–17

    work page 1964

  7. [7]

    T. M. Cover and J. A. Thomas,Elements of Information Theory, 2nd edition, Wiley-Interscience, 2006

    work page 2006

  8. [8]

    Esposito and C

    M. Esposito and C. Van den Broeck, Three faces of the second law. I. Master equation formulation,Physical Review E,82 (2010), 011143

    work page 2010

  9. [9]

    Seifert, Stochastic thermodynamics, fluctuation theorems and molecular machines,Reports on Progress in Physics,75 (2012), 126001

    U. Seifert, Stochastic thermodynamics, fluctuation theorems and molecular machines,Reports on Progress in Physics,75 (2012), 126001

    work page 2012

  10. [10]

    Chafa¨ ı, From Boltzmann to random matrices and beyond,Annales de la Facult´ e des Sciences de Toulouse,24(2015), 641–689

    D. Chafa¨ ı, From Boltzmann to random matrices and beyond,Annales de la Facult´ e des Sciences de Toulouse,24(2015), 641–689. 34 QIAO WANG

    work page 2015

  11. [11]

    Hartich and A

    D. Hartich and A. Godec, Interlacing relaxation and first-passage phenomena in reversible discrete and continuous space Markovian dynamics,Journal of Statistical Mechanics: Theory and Experiment,2019(2019), 024002

    work page 2019

  12. [12]

    J. B. Voits and U. S. Schwarz, Emergence of generic first-passage time distributions for large Markovian networks, arXiv:2602.18265v2, 2026

    work page arXiv 2026

  13. [13]

    A. J. F. Siegert, On the first passage time probability problem,Physical Review,81(1951), 617–623

    work page 1951

  14. [14]

    Diaconis and M

    P. Diaconis and M. Shahshahani, Time to reach stationarity in the Bernoulli-Laplace diffusion model,SIAM Journal on Mathematical Analysis,18(1987), 208–218

    work page 1987

  15. [15]

    Bovier and F

    A. Bovier and F. den Hollander,Metastability: A Potential-Theoretic Approach, Grundlehren der mathematischen Wis- senschaften, vol. 351, Springer, 2015

    work page 2015

  16. [16]

    Jiang and F

    D. Jiang and F. Zhang, The Green–Kubo formula and power spectrum of reversible Markov processes,Journal of Mathe- matical Physics,44(2003), 4681–4689

    work page 2003

  17. [17]

    Relaxation Kernel and Global Convergence of the Blahut-Arimoto Dynamics

    Q. Wang, Relaxation kernel and global convergence of the Blahut-Arimoto dynamics, arXiv:2604.25106v2, 2026. Available athttps://doi.org/10.48550/arXiv.2604.25106 School of Information Science and Engineering, and School of Economics and Management, Southeaast Uni- versity, Nanjing, 211189, China Email address:qiaowang@seu.edu.cn

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.25106 2026