Entropy Production under Hidden-State Reduction: Exact Safety and Non-Testability
Version 1 / Nov 19, 2026 / CC BY 4.0
Abstract
Restricting hidden-state complexity can raise the least entropy production compatible with partial observations. We determine exactly when a two-hidden-state fit preserves this lower bound for finite irreducible stationary reciprocal continuous-time Markov chains with two individually observed states and one hidden label. We begin with an explicit chain with three hidden states whose observed data admit exact two-hidden-state fits, every one of which has entropy production more than sixty times that of the true chain. For strictly positive consistent stationary masses, flows, excursion counts and first duration moments, the unrestricted infimum is an explicit count functional, and four hidden states approach it while preserving every datum exactly. We classify finite attainment and safety of feasible two-hidden-state fits in every determinant regime; in the positive-determinant regime, safety is equivalent to equality of the conditional cross-excursion means, and explicit bounds quantify the reduction bias. Exact classification does not imply finite-data detectability: rare long excursions can conceal a fixed positive bias on any prescribed finite interval of stationary label observation, even under a common upper rate bound.
Provenance statement
Tools used
- OpenAI
- ChatGPTVersion GPT-5.6 Sol
- OpenAI
- CodexVersion GPT-5.6 Sol
- Anthropic
- Claude CodeVersion Claude Opus 5.5
References
- Murad Aznagulov. Thermodynamic realizability of hidden Markov processes: Attainment, observable certificates, and the price of architecture. arXiv preprint , page 2609.14623, 2026. arXiv:2609.14623v1, submitted 13 September 2026.arXiv
- R. R. Bahadur and Leonard J. Savage. The nonexistence of certain statistical procedures in nonparametric problems. The Annals of Mathematical Statistics , 27 0 (4): 0 1115--1122, 1956. 10.1214/aoms/1177728077.DOI
- Gili Bisker, Matteo Polettini, Todd R. Gingrich, and Jordan M. Horowitz. Hierarchical bounds on entropy production inferred from partial information. Journal of Statistical Mechanics: Theory and Experiment , page 093210, 2017. 10.1088/1742-5468/aa8c0d. arXiv:1708.06769; Sections 3.2, 5 and 6.arXivDOI
- Peter Buchholz and Mikl ó s Telek. On minimal representations of Rational Arrival Processes . Annals of Operations Research , 202 0 (1): 0 35--58, 2013. 10.1007/s10479-011-1001-5. URL https://link.springer.com/article/10.1007/s10479-011-1001-5.DOI
- Julius Deg ü nther, Jann van der Meer, and Udo Seifert. Fluctuating entropy production on the coarse-grained level: Inference and localization of irreversibility. Physical Review Research , 6: 0 023175, 2024. 10.1103/PhysRevResearch.6.023175. arXiv:2309.07665; Sections II.C--II.E: Markovian events and trajectory snippets.arXivDOI
- Jannik Ehrich. Tightest bound on hidden entropy production from partially observed dynamics. Journal of Statistical Mechanics: Theory and Experiment , page 083214, 2021. 10.1088/1742-5468/ac150e. arXiv:2105.08803.arXivDOI
- Massimiliano Esposito. Stochastic thermodynamics under coarse graining. Physical Review E , 85: 0 041125, 2012. 10.1103/PhysRevE.85.041125. URL https://arxiv.org/abs/1112.5410.DOI
- Massimiliano Esposito and Juan M. R. Parrondo. Stochastic thermodynamics of hidden pumps. Physical Review E , 91: 0 052114, 2015. 10.1103/PhysRevE.91.052114. arXiv:1310.2987.arXivDOI
- Jie Gu. Excursion-resolved thermodynamic inference without observing reverse transitions. arXiv preprint , page 2609.39205, 2026. URL https://arxiv.org/abs/2609.39205v1. arXiv:2609.39205v1; unrefereed version 1, submitted 30 September 2026.arXiv
- Ill é s Horv á th and Mikl ó s Telek. A constructive proof of the phase-type characterization theorem. Stochastic Models , 31 0 (2): 0 316--350, 2015. 10.1080/15326349.2015.1012912. URL https://doi.org/10.1080/15326349.2015.1012912.DOI
Show all 24 referencesShow fewer references
- Oleg A. Igoshin, Anatoly B. Kolomeisky, and Dmitrii E. Makarov. Uncovering dissipation from coarse observables: A case study of a random walk with unobserved internal states. The Journal of Chemical Physics , 162: 0 034111, 2025. 10.1063/5.0247331.DOI
- Uri Kapustin, Aishani Ghosal, and Gili Bisker. Utilizing time-series measurements for entropy-production estimation in partially observed systems. Physical Review Research , 6: 0 023039, 2024. 10.1103/PhysRevResearch.6.023039. arXiv:2212.13487v1 is the earlier 2022 preprint.arXivDOI
- Bo Henry Lindqvist. Phase-type models for competing risks, with emphasis on identifiability issues. Lifetime Data Analysis , 29 0 (2): 0 318--341, 2023. 10.1007/s10985-022-09547-7. URL https://link.springer.com/article/10.1007/s10985-022-09547-7.DOI
- Ignacio A. Mart í nez, Gili Bisker, Jordan M. Horowitz, and Juan M. R. Parrondo. Inferring broken detailed balance in the absence of observable currents. Nature Communications , 10: 0 3542, 2019. 10.1038/s41467-019-11051-w. URL https://arxiv.org/abs/1809.02084.DOI
- Eden Nitzan, Aishani Ghosal, and Gili Bisker. Universal bounds on entropy production inferred from observed statistics. Physical Review Research , 5: 0 043251, 2023. 10.1103/PhysRevResearch.5.043251. arXiv:2212.01783.arXivDOI
- Matteo Polettini and Massimiliano Esposito. Effective thermodynamics for a marginal observer. Physical Review Letters , 119: 0 240601, 2017. 10.1103/PhysRevLett.119.240601. arXiv:1703.05715.arXivDOI
- É dgar Rold á n and Juan M. R. Parrondo. Estimating dissipation from single stationary trajectories. Physical Review Letters , 105: 0 150607, 2010. 10.1103/PhysRevLett.105.150607. arXiv:1004.2831.arXivDOI
- É dgar Rold á n and Juan M. R. Parrondo. Entropy production and Kullback--Leibler divergence between stationary trajectories of discrete systems. Physical Review E , 85: 0 031129, 2012. 10.1103/PhysRevE.85.031129. arXiv:1201.5613.arXivDOI
- Udo Seifert. Stochastic thermodynamics, fluctuation theorems and molecular machines. Reports on Progress in Physics , 75: 0 126001, 2012. 10.1088/0034-4885/75/12/126001. arXiv:1205.4176.arXivDOI
- Dominic J. Skinner and J ö rn Dunkel. Improved bounds on entropy production in living systems. Proceedings of the National Academy of Sciences , 118: 0 e2024300118, 2021 a . 10.1073/pnas.2024300118. arXiv:2011.08765.arXivDOI
- Dominic J. Skinner and J ö rn Dunkel. Estimating entropy production from waiting time distributions. Physical Review Letters , 127: 0 198101, 2021 b . 10.1103/PhysRevLett.127.198101. arXiv:2105.08681.arXivDOI
- Mikl ó s Telek and G á bor Horv á th. A minimal representation of Markov arrival processes and a moments matching method. Performance Evaluation , 64 0 (9--12): 0 1153--1168, 2007. 10.1016/j.peva.2007.06.001. URL https://www.sciencedirect.com/science/article/abs/pii/S0166531607000508.DOI
- Gianluca Teza and Attilio L. Stella. Exact coarse graining preserves entropy production out of equilibrium. Physical Review Letters , 125: 0 110601, 2020. 10.1103/PhysRevLett.125.110601. arXiv:2003.08674.arXivDOI
- Jann van der Meer, Benjamin Ertel, and Udo Seifert. Thermodynamic inference in partially accessible Markov networks: A unifying perspective from transition-based waiting time distributions. Physical Review X , 12: 0 031025, 2022. 10.1103/PhysRevX.12.031025. arXiv:2203.12020.arXivDOI
Version history
- v1 · 19 pagesSubmitted by Samuil PetkovThe computer-assisted step in the proof of the family overestimate theorem is performed in exact rational arithmetic by the script anc/certify_I2.py included with the source.CurrentNov 19, 2026