Observation cost of nonequilibrium in nonnormal networks: a matched exponent for an explicit four-state generator

A mix of human-written and AI-generated textHuman understanding: all partsmath.PR — Probabilitymath.ST — Statistics Theory

Contributed by Samuil Petkov ↗

Version 1 / Nov 19, 2026 / CC BY 4.0

Abstract

We consider detection of nonequilibrium from passive, causally scheduled snapshots of a stationary continuous-time Markov chain, with gaps between snapshots of at least τ\tau. False alarms must be controlled uniformly over every finite stationary detailed-balance chain observed through a deterministic labelling, with no restriction on hidden dimension, rates, stationary label law or mixing time. Aznagulov showed that for normal generators the optimal snapshot cost grows like e2γirrτe^{2\gamma_{\rm irr}\tau}, and that for nonnormal generators its exponent lies in a spectral window. For an explicit nonnormal four-state generator Q4Q_4, with a real slow mode and residual damping β\beta, we prove that τ−1log⁡N∗→2γ\tau^{-1}\log N^*\to2\gamma, where γ=(701−158761)/126∈(2,β)\gamma=(701-\sqrt{158761})/126\in(2,\beta) is a relaxation rate of an explicit reversible birth–death comparator; for Q4Q_4 this exponent lies strictly inside the window [2,2β][2,2\beta]. The comparator comes from an ordered-eigenfunction completion lemma: among π\pi-reversible generators with a prescribed strictly monotone eigenfunction and eigenvalue, a birth–death generator maximizes the entire Dirichlet form. The upper bound uses symmetric two-lag scores whose stationary mean is nonpositive under every hidden reversible model, together with a positive-operator tilt lemma that gives the moment-generating bound e32nθ2e^{32n\theta^2} for a deterministic number nn of dependent transitions, uniformly over the null class. We also show that the two-lag pair-law approximation exponent of Q4Q_4 by hidden reversible models is γ\gamma, that the cost exponent 2γ2\gamma holds uniformly over a four-parameter box of nonnormal deformations, and that it extends to interacting path families on {1,2,3,4}d\{1,2,3,4\}^d built from tensor lifts. More generally, if uu has distinct entries and is an eigenvector of QQ and of its L2(π)L^2(\pi)-adjoint with eigenvalue −λ-\lambda, the exponent is 2γ∗2\gamma^* whenever γ∗<β\gamma^*<\beta, where γ∗\gamma^* is the restricted gap of the completion.

Provenance statement

OpenAI GPT-5.6 Sol, accessed through ChatGPT and Codex, was used to develop and audit the proofs for the explicit generator and its extensions, and for drafting. Anthropic Claude Opus 5.5, used through Claude Code, was used to rewrite the note, verify the constants exactly, and derive the general class theorem (Section 10), including the case of small restricted gap. The author reviewed and takes responsibility for every statement and proof.

Tools used

OpenAI
ChatGPTVersion GPT-5.6 Sol
OpenAI
CodexVersion GPT-5.6 Sol
Anthropic
Claude CodeVersion Claude Opus 5.5

References

  1. Murad Aznagulov, The spectral cost of detecting nonequilibrium at finite temporal resolution , 2026, arXiv:2609.37205, version 1 (29 September 2026).arXiv
  2. George Bakewell-Smith, Federico Girotti, M ă d ă lin Gu ţ ă , and Juan P. Garrahan, General upper bounds on fluctuations of trajectory observables , Physical Review Letters 131 (2023), 197101.
  3. Ryuna Nagayama, Artemy Kolchinsky, and Sosuke Ito, Oscillatory-nonnormal decomposition of dissipation in O rnstein- U hlenbeck processes , 2026, arXiv:2606.07263, version 2 (1 July 2026).arXiv

Version history

  1. v1 · 17 pagesSubmitted by Samuil PetkovInitial depositCurrentNov 19, 2026