Abstract
The universal typical-signal estimators of entropy and cross-entropy based on the asymptotics of recurrence and waiting times play an important role in information theory. Building on their construction, we introduce and study universal typical-signal estimators of entropy production in the context of nonequilibrium statistical mechanics of one-sided shifts over finite alphabets.
Similar content being viewed by others
Data availability statement
The datasets analyzed during the current study are publicly available in the Genome Reference Consortium Human Build 38 repository, Patch Release 14 [24].
Notes
Some remarks are in order to ease comparisons with the setup of Bradley [6, 7]. First, note that the coefficients do not change if we replace [a] with [A], \(A \subseteq {\mathcal {A}}^n\) and [b] with [B], \(B \subseteq {\mathcal {A}}^m\): For example, if \({\mathbb {P}}([a] \cap \sigma ^{-n-\ell }[b]) \le \psi _{{\mathbb {P}}}^*(\ell ){\mathbb {P}}([a]){\mathbb {P}}([b])\) for all a and b, then summing over \(a \in A\) and \(b\in B\) gives \({\mathbb {P}}([A] \cap \sigma ^{-n-\ell }([B])) \le \psi _{{\mathbb {P}}}^*(\ell ){\mathbb {P}}([A]){\mathbb {P}}([B])\). Second, because m is arbitrary, we have a generating semi-algebra at hand and a standard approximation argument shows that we can replace the requirement that \(B \subseteq {\mathcal {A}}^m\) for some \(m \in \mathbb {N}\) with the requirement that B be Borel measurable. Finally, since we are only interested in \(\sigma \)-invariant measures on \({\mathcal {A}}^\mathbb {N}\)—which are naturally in one-to-one correspondence with \(\sigma \)-invariant measures on \({\mathcal {A}}^\mathbb {Z}\)—the definitions then translate to Bradley’s definitions on \({\mathcal {A}}^\mathbb {Z}\) exploiting \(\sigma \)-invariance and yet another approximation procedure by sets now in the semi-algebra built by shifting by n cylinders naturally associated with sets of the form \(A \subseteq {\mathcal {A}}^n\) for some n.
The original result, Theorem 1 in [6], requires the measure \({\mathbb {P}}\) to be mixing in the sense of ergodic theory. However, later in the same paper it is remarked that this extra assumption is superfluous to derive that \(\psi '(\ell ') > 0\) for some \(\ell '\) implies \(\psi '(\ell ) \rightarrow 1\) as \(\ell \rightarrow \infty \). As noted by Bradley in his later review [7, §4.1], the fact that \(\psi '(\ell ) \rightarrow 0\) in turn implies \(\phi \)-mixing—and thus mixing in the sense of ergodic theory—can be combined with the original result to obtain the variant of the result stated here. According to Bradley in this same review, this version of the result was included in later works at the suggestion of Denker. It is also worth noting that the proof of (any version of) the result relies heavily on the earlier work [5] on \(\phi \)-mixing.
Unlike in the Markov case, this is not a necessary requirement.
The adjoint \(\Phi ^*\) is defined with respect to the inner product \(\langle A, B\rangle ={{\,\textrm{tr}\,}}(A^*B)\) on \({\mathcal {B}}({\mathcal {H}})\).
In the Markov case, irreducibility is equivalent to the ergodicity of \({\mathbb {P}}\).
On the set where \(\tfrac{1}{n} \log {\widehat{{\mathbb {P}}}}_n(x_1^n) \rightarrow -\infty \) as \(n\rightarrow \infty \), the inequality \(\limsup \tfrac{1}{n}\log {\widehat{R}}_n(x) \le h_{{\widehat{{\mathbb {P}}}}}(x)\) is vacuously true (recall Hypothesis i”).
References
Abadi, M.: Sharp error terms and necessary conditions for exponential hitting times in mixing processes. Ann. Probab. 32, 243–264 (2004)
Abadi, M., Chazottes, J.-R., Redig, F., Verbitskiy, E.: Exponential distribution for the occurrence of rare patterns in Gibbsian random fields. Commun. Math. Phys. 246, 269–294 (2004)
Abadi, M., Vergne, N.: Sharp error terms for return time statistics under mixing conditions. J. Theor. Probab. 22, 18–37 (2009)
Cristadoro, G., Degli Esposti, M., Altmann, E.G.: The common origin of symmetry and structure in genetic sequences. Sci. Rep. 8, 15817 (2018)
Bradley, R.C.: On the \(\phi \)-mixing condition for stationary random sequences. Duke Math. J. 47, 421–433 (1980)
Bradley, R.C.: On the \(\psi \)-mixing condition for stationary random sequences. Trans. Am. Math. Soc. 276, 55–66 (1983)
Bradley, R.C.: Basic properties of strong mixing conditions: a survey and some open questions. Probab. Surv. 2, 107–144 (2005)
Benoist, T., Cuneo, N., Jakšić, V., Pillet, C.-A.: On entropy production of repeated quantum measurements II. Examples. J. Stat. Phys. 182(3), 1–71 (2021)
Benoist, T., Jakšić, V., Pautrat, Y., Pillet, C.-A.: On entropy production of repeated quantum measurements I. General theory. Commun. Math. Phys. 357, 77–123 (2018)
Bryc, W., Dembo, A.: Large deviations and strong mixing. Ann. Inst. Henri Poincaré 32, 549–569 (1996)
Climenhaga, V.: The thermodynamic approach to multifractal analysis. Ergod. Theory Dyn. Syst. 34, 1409–1450 (2014)
Cristadoro, G., Degli Esposti, M., Jakšić, V., Raquépas, R.: On a waiting-time result of Kontoyiannis: mixing or decoupling? (2022) Preprint, arXiv:2209.09717
Cuneo, N., Jakšić, V., Pillet, C.-A., Shirikyan, A.: Fluctuation theorem and thermodynamic formalism. (2017) Unpublished report, arXiv:1712.05167
Cuneo, N., Jakšić, V., Pillet, C.-A., Shirikyan, A.: Large deviations and fluctuation theorem for selectively decoupled measures on shift spaces. Rev. Math. Phys. 31, 1950036-1–54 (2019)
Chazottes, J.-R., Olivier, E.: Relative entropy, dimensions and large deviations for g-measures. J. Phys. A Math. Gen. 33, 675–689 (2000)
Chazottes, J.-R., Redig, F.: Testing the irreversibility of a Gibbsian process via hitting and return times. Nonlinearity 18, 2477–2489 (2005)
Dembo, A., Zeitouni, O.: Large Deviations Techniques and Applications, 2nd edn. Springer, Berlin (1998)
Evans, D.J., Cohen, E.G.D., Morriss, G.P.: Probability of second law violation in shearing steady flows. Phys. Rev. Lett. 71, 2401–2404 (1993)
van Enter, A.C.D., Fernández, R., Sokal, A.D.: Regularity properties and pathologies of position-space renormalization-group transformations. J. Stat. Phys. 72, 879–1167 (1993)
Gallavotti, G., Cohen, E.G.D.: Dynamical ensembles in nonequilibrium statistical mechanics. Phys. Rev. Lett. 74, 2694–2697 (1995)
Gallavotti, G., Cohen, E.G.D.: Dynamical ensembles in stationary states. J. Stat. Phys. 80, 931–970 (1995)
Gao, Y., Kontoyiannis, I., Bienenstock, E.: Estimating the entropy of binary time series: methodology, some theory and a simulation study. Entropy 10, 71–99 (2008)
Georgii, H.O.: Gibbs Measures and Phase Transitions, 2nd edn. De Gruyter, Berlin (2011)
Genome Reference Consortium: Human Build 38, Patch Release 14, Available online at https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.40 as of December 2022, (2022)
Jakšić, V.: Lectures on entropy I: information–theoretic notions. In: Bahns et al (Eds) Dynamical Methods in Open Quantum Systems, Tutorials, Schools and Workshops in the Mathematical Sciences, pp. 141–268, Springer (2019)
Johansson, A., Öberg, A., Pollicott, M.: Ergodic theory of Kusuoka measures. J. Fractal Geom. 4(2), 185–214 (2017)
Jakšić, V., Ogata, Y., Pillet, C.-A., Seiringer, R.: Hypothesis testing and non-equilibrium statistical mechanics. Rev. Math. Phys. 24(6), 1–67 (2012)
Jakšić, V., Pillet, C.-A., Shirikyan, A.: Beyond Gibbsianity. In preparation
Jakšić, V., Pillet, C.-A., Rey-Bellet, L.: Entropic fluctuations in statistical mechanics I. Classical dynamical systems. Nonlinearity 24, 699–763 (2011)
Kontoyiannis, I.: Asymptotic recurrence and waiting times for stationary processes. J. Theor. Probab. 11, 795–811 (1998)
Kontoyiannis, I.: Asymptotic recurrence and waiting times in stationary processes, and their application in data compression. Stanford PhD Thesis, (1998)
Kusuoka, S.: Dirichlet forms on fractals and products of random matrices. Publ. Res. Inst. Math. Sci. 25, 659–680 (1989)
Kontoyiannis, I., Algoet, P.H., Suhov, Y.M., Wyner, A.J.: Nonparametric entropy estimation for stationary processes and random fields, with applications to English text. IEEE Trans. Inf. Theory 44, 1319–1327 (1998)
Lebowitz, J.L., Spohn, H.: A Gallavotti-Cohen-type symmetry in the large deviation functional for stochastic dynamics. J. Stat. Phys. 95, 333–365 (1999)
Ziv, L., Lempel, A.: A universal algorithm for sequential data compression. IEEE Trans. Inf. Theory 23, 327–343 (1977)
Ziv, L., Lempel, A.: Compression of individual sequences via variable-rate coding. IEEE Trans. Inf. Theory 24, 530–536 (1978)
Maes, C.: The fluctuation theorem as a Gibbs property. J. Stat. Phys. 95, 367–392 (1999)
Orey, S., Pelikan, S.: Large deviation principles for stationary processes. Ann. Probab. 16, 1481–1495 (1988)
Ornstein, D.S., Weiss, B.: Entropy and data compression schemes. IEEE Trans. Inf. Theory 39, 78–83 (1993)
Pfister, Ch.-É.: Thermodynamical aspects of classical dynamical systems. Progr. Probab. 51, 393–472 (2002)
Raquépas, R.: A gapped generalization of Kingman’s subadditive ergodic theorem. (2022) Preprint, arXiv:2211.13134
Ruelle, D.: Smooth dynamics and new theoretical ideas in nonequilibrium statistical mechanics. J. Stat. Phys. 95, 393–468 (1999)
Rudner, R., Karkas, J.D., Chargaff, E.: Separation of B. subtilis DNA into complementary strands II-III. Proc. Natl. Acad. Sci. USA 60, 915–922 (1968)
Salgado-García, R.: Time-irreversibility test for random-length time series: the matching-time approach applied to DNA. Chaos 31, 123126 (2021)
Salgado-García, R., Moldano, C.: Estimating entropy rate from censored symbolic time series: a test for time-irreversibility. Chaos 31, 013131 (2021)
Schneider, V.A., et al.: Evaluation of GRCh38 and de novo haploid genome assemblies demonstrates the enduring quality of the reference assembly. Genome Res. 21, 849–864 (2017)
Shields, P.C.: The ergodic theory of discrete sample paths. In: Graduate Studies in Mathematics. American Mathematical Society (1996)
Verdú, S.: Empirical estimation of information measures: a literature guide. Entropy 21, 720–736 (2019)
Walters, P.: Regularity conditions and Bernoulli properties of equilibrium states and g-measures. J. Lond. Math. Soc. 71, 379–396 (2005)
Wyner, A.D., Ziv, J.: Some asymptotic properties of the entropy of a stationary ergodic data source with applications to data compression. IEEE Trans. Inf. Theory 35, 1250–1258 (1989)
Acknowledgements
This work was supported by the Agence Nationale de la Recherche through the grant NONSTOPS (ANR-17-CE40-0006-01, ANR-17-CE40-0006-02, ANR-17-CE40-0006-03) and was partly developed during VJ’s and MDE’s stays at the CY Advanced Studies, whose support is gratefully acknowledged. Another part of this work was done during MDE’s stay at McGill University funded by Simons CRM Scholar-in-Residence Program. Additional funding was provided by the CY Initiative of Excellence (Investissements d’Avenir, grant ANR-16-IDEX-0008). GC acknowledges partial support by the PRIN Grant 2017S35EHN “Regular and stochastic behaviour in dynamical systems” of the Italian Ministry of University and Research (MUR) and by the UMI Group “DinAmicI.” VJ acknowledges the support of NSERC. Most of this work was done, while RR was a postdoctoral researcher at CY Cergy Paris Université and supported by the LabEx MME-DII (Investissements d’Avenir). Part of this work was also completed during RR’s stay at the Centre de recherches mathématiques of Université de Montréal, whose support is gratefully acknowledged. The authors wish to thank T. Benoist and N. Cuneo for useful discussions.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A
Appendix A
Proof of Theorem 6.1
One follows the steps of the proof of Theorem 3.4 in Sect. 5, with the following changes. In Step 1, the \({\mathbb {P}}\)-almost sure validity of (4) now relies on Hypothesis i and adaptations of Fekete’s lemma and Kingman’s theorem to the corresponding generalized subadditivity condition which are discussed in [41]. The set \(E_{n, \epsilon }\) is defined in the same way, but one takes
In Step 2, one starts with
for n large enough, and estimates
The final estimate and the assumption that \(c_n\) and \(\tau _n\) are o(n) give the desired summability condition. In Step 3, one first chooses, for n large enough, a natural number m(a) so that
and replaces (17) with the estimates
At this point, one proceeds in exactly the same way as in Sect. 5 to derive the estimate
which, combined with Hypothesis ii, yields the desired summability. \(\square \)
Proof of Theorem 6.2
One again follows the same strategy. In Step 1, the \({\mathbb {P}}\)-almost sure validity of (4) now relies on Hypothesis i”. In Step 2, the sets \(B_{n, \epsilon }\) are the same as in (13) and one starts with the identity
which gives
as soon as n is large enough that \(n - \tau _{n,K} \ge 1\). (This is possible because \(\tau _{n,K}\) is o(n).) Since each of the \(\lfloor \text {e}^{- \log {\widehat{{\mathbb {P}}}}_n(a) - n \epsilon }\rfloor \) sets of the form \(\sigma ^{-\tau _{n,K}}\{ x: {\widehat{R}}_n(a\sigma ^{n}(x)) = j \}\) is also necessarily of the form \(\sigma ^{- n - \tau _{n,K}}(B)\) for some \(B \in {\mathcal {F}}_{\text {fin}}\) and has probability bounded above by \({\widehat{{\mathbb {P}}}}_n(a)\), Hypothesis i’ gives that
where we used \({{\,\textrm{supp}\,}}{\mathbb {P}}_n \subseteq {{\,\textrm{supp}\,}}{\mathbb {P}}_{n-\tau _{n,K}} \times {\mathcal {A}}^{\tau _{n,K}}\) to get the second inequality. If \(K > \log |{\mathcal {A}}|\), the last estimate gives the desired summability since both \(c_{n,K}\) and \(\tau _{n,K}\) are o(n). We now turn to Step 3. With \(E_{n, \epsilon }\) as in (13) and
it suffices to show that for all \(\epsilon > 0\) and all \(K \in \mathbb {N}\) large enough, the summability condition
holds.Footnote 6 The identity
gives
Setting
and using an obvious inclusion, \(\sigma \)-invariance and then Hypothesis i’, one derives
The identities
and
further give
Note that for n large enough,
for all \(x_1^n\in {{{\mathcal {A}}}}^n\). Taking \(K>\log |{\mathcal {A}}|\), the estimates (29)–(30) and Hypothesis ii’ give the summability condition (26). \(\square \)
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Cristadoro, G., Degli Esposti, M., Jakšić, V. et al. Recurrence times, waiting times and universal entropy production estimators. Lett Math Phys 113, 19 (2023). https://doi.org/10.1007/s11005-023-01640-8
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11005-023-01640-8