pith. sign in

arxiv: 2607.00330 · v1 · pith:DJUMOW7Jnew · submitted 2026-07-01 · 🧮 math.ST · math.PR· stat.TH

Ergodicity and High-Frequency Inference for Hybrid Switching L\'{e}vy-Driven Stochastic Differential Equations

Yuzhong Cheng This is my paper

Pith reviewed 2026-07-02 00:34 UTC · model grok-4.3

classification 🧮 math.ST math.PRstat.TH
keywords ergodicityhigh-frequency samplingLévy-driven SDEswitching processesquasi-likelihoodasymptotic normalitylarge deviation inequality
0
0 comments X

The pith

A three-stage estimator for parameters in hybrid switching Lévy-driven SDEs is consistent and jointly asymptotically normal under high-frequency sampling and ergodicity.

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

The paper develops inference for hybrid switching Lévy-driven stochastic differential equations observed at high frequency, where the noise is pure-jump and switching rates depend on state. It proposes a three-stage procedure that pairs staged Gaussian quasi-likelihood estimation for drift and scale with an intensity-type contrast for switching rates. The work first derives checkable conditions for weighted exponential ergodicity of the hybrid process via a fixed skeleton-chain argument that uses small-jump accessibility and regime connectivity without Brownian smoothing. Under these conditions the full estimator is shown to be consistent, jointly asymptotically normal, and to obey a polynomial large deviation inequality, with the limiting covariance coupling drift and scale through the third moment of the Lévy noise while leaving the switching-rate block asymptotically uncorrelated with the continuous blocks.

Core claim

Under ergodicity and the high-frequency sampling scheme, consistency, joint asymptotic normality, and a polynomial-type large deviation inequality hold for the full three-stage estimator; the joint limit exhibits a transparent covariance structure where the drift and scale blocks are coupled through the third moment of the driving Lévy noise while the switching-rate block is asymptotically uncorrelated with the continuous-coefficient blocks.

What carries the argument

Three-stage inference procedure that combines staged Gaussian quasi-likelihood with an intensity-type contrast, supported by weighted exponential ergodicity proved via a fixed skeleton-chain argument.

If this is right

  • The estimator attains joint asymptotic normality whose covariance matrix has an explicit block structure determined by the Lévy noise.
  • Switching-rate estimates remain asymptotically independent of the continuous-parameter estimates.
  • A polynomial large deviation inequality supplies quantitative rates beyond the central limit theorem.
  • The ergodicity conditions can be verified directly from the model coefficients without invoking Brownian smoothing.

Where Pith is reading between the lines

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

  • The skeleton-chain technique for ergodicity may apply to other pure-jump switching models that lack a diffusion component.
  • The asymptotic separation of switching-rate estimates could simplify the construction of separate confidence regions in practice.
  • Numerical checks with normal inverse Gaussian noise already indicate that finite-sample performance aligns with the derived limits.

Load-bearing premise

The hybrid process satisfies weighted exponential ergodicity, proved via a fixed skeleton-chain argument that combines small-jump accessibility and regime connectivity.

What would settle it

High-frequency data from a model satisfying the stated conditions where the sample covariance between the drift-scale estimators fails to converge to the value predicted by the third moment of the Lévy noise.

Figures

Figures reproduced from arXiv: 2607.00330 by Yuzhong Cheng.

Figure 1
Figure 1. Figure 1: Histograms of selected standardized errors Z (r) ξ for representative compo￾nents of ˆζn under (T, h) = (200, 0.005), together with the standard normal density. with jump measures ν s and ν r , respectively. Since ρ := ν s (R) = 2κ0r0, µs(dz) := ρ −1 ν s (dz) = 1 2r0 1(−r0,r0)(z) dz, the process L s is compound Poisson L s t = PNs t k=1 Yk, where Ns has rate ρ, and Yk are i.i.d. with law µs, independent of… view at source ↗
Figure 2
Figure 2. Figure 2: QQ plots of selected standardized errors Z (r) ξ for representative components of ˆζn under (T, h) = (200, 0.005). Model 1: vartheta10 Standardized error Density −2 −1 0 1 2 3 0.0 0.1 0.2 0.3 0.4 mean = 0.00 var = 0.93 N(0,1) Model 1: vartheta11 Standardized error Density −2 −1 0 1 2 3 4 0.0 0.1 0.2 0.3 0.4 mean = −0.05 var = 0.97 N(0,1) Model 1: vartheta20 Standardized error Density −2 −1 0 1 2 3 0.0 0.1 … view at source ↗
Figure 3
Figure 3. Figure 3: Histograms of standardized errors Z (r) ξ for all four components of the ϑ-block of ˆζn, shown for Models 1 and 2 under (T, h) = (200, 0.005), together with the standard normal density. Recall that Φ(h) is a T-chain if there exists a substochastic kernel T satisfying T(·, B) ≤ Ph(·, B) for every Borel B, ζ 7→ T(ζ, B) is lower semicontinuous for every Borel B, and T(ζ, R × S) > 0 for every ζ; [PITH_FULL_IM… view at source ↗
Figure 4
Figure 4. Figure 4: QQ plots of standardized errors Z (r) ξ for all four components of the ϑ-block of ˆζn, shown for Models 1 and 2 under (T, h) = (200, 0.005). and is φ-irreducible if φ(B) > 0 =⇒ X N≥1 P N h (ζ, B) > 0 for every ζ ∈ R × S (see [27]). Fix a deterministic number h0 > 0. Step 1: T-chain property. Fix z = (x, i) ∈ R × S and set Kz := [x − 1, x + 1]. Apply Lemma 7.7 to Kz and regime i. Then there exists h i Kz > … view at source ↗
read the original abstract

Hybrid switching L\'evy-driven stochastic differential equations with pure-jump noise and state-dependent switching rates are studied under high-frequency observation. A three-stage inference procedure is proposed for the drift, scale, and switching-rate parameters, combining a staged Gaussian quasi-likelihood with an intensity-type contrast. Checkable sufficient conditions for weighted exponential ergodicity are established for the hybrid process; the proof does not rely on Brownian smoothing, but uses a fixed skeleton-chain argument combining small-jump accessibility and regime connectivity. Under ergodicity and the high-frequency sampling scheme, consistency, joint asymptotic normality, and a polynomial-type large deviation inequality are proved for the full estimator. The joint limit exhibits a transparent covariance structure: the drift and scale blocks are coupled through the third moment of the driving L\'evy noise, whereas the switching-rate block is asymptotically uncorrelated with the continuous-coefficient blocks. Numerical experiments for models driven by normal inverse Gaussian noise illustrate the finite-sample behavior of the proposed estimators.

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 / 3 minor

Summary. The manuscript develops inference for hybrid switching Lévy-driven SDEs observed at high frequency. It introduces a three-stage estimator combining staged Gaussian quasi-likelihood for drift and scale parameters with an intensity contrast for switching rates. Checkable sufficient conditions for weighted exponential ergodicity of the hybrid process are derived via a fixed skeleton-chain argument that relies on small-jump accessibility and regime connectivity (without Brownian smoothing). Under these conditions, consistency, joint asymptotic normality, and a polynomial large-deviation inequality are established for the full estimator; the limiting covariance couples the drift and scale blocks through the third moment of the driving Lévy noise while rendering the switching-rate block asymptotically orthogonal to the continuous-coefficient blocks. Finite-sample behavior is illustrated numerically for normal-inverse-Gaussian-driven models.

Significance. If the ergodicity conditions and asymptotic expansions hold, the work supplies a rigorous, checkable framework for high-frequency estimation in switching jump-driven processes together with an explicit covariance structure that isolates the effect of the Lévy third moment. The skeleton-chain approach to ergodicity and the polynomial large-deviation bound constitute concrete technical contributions that extend standard quasi-likelihood theory to the hybrid Lévy setting.

minor comments (3)
  1. [Abstract] The abstract states that the joint limit covariance is 'transparent,' yet the precise expression for the cross-term involving the third moment of the Lévy measure is not displayed; a displayed formula would clarify the claimed orthogonality of the switching-rate block.
  2. The numerical section reports finite-sample behavior for NIG-driven models but does not tabulate the estimated covariance matrices or compare them with the theoretical block structure; adding such a comparison would strengthen the illustration of the asymptotic claim.
  3. [Introduction] Notation for the regime-dependent coefficients and the intensity function is introduced gradually; a consolidated table of symbols at the end of the introduction would improve readability.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and the positive assessment. The referee's summary accurately captures the main contributions regarding the three-stage estimator, the skeleton-chain ergodicity conditions, and the joint asymptotic normality with the explicit covariance structure. We appreciate the recommendation for minor revision.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The paper derives weighted exponential ergodicity via a fixed skeleton-chain argument relying on small-jump accessibility and regime connectivity (checkable conditions independent of the estimator). Consistency, joint asymptotic normality, and large-deviation bounds for the three-stage quasi-likelihood estimator then follow from standard ergodic theory and quasi-likelihood expansions under high-frequency sampling. The reported covariance structure (drift/scale blocks coupled via Lévy third moment, switching-rate block orthogonal) is obtained directly from the expansion once ergodicity holds, without any reduction to fitted parameters, self-definitional steps, or load-bearing self-citations. All steps use external probabilistic tools and are falsifiable outside the fitted values.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on domain assumptions about the Lévy driver having sufficient moments and the switching mechanism allowing regime connectivity; no free parameters or invented entities are introduced beyond the model class itself.

axioms (2)
  • domain assumption The driving Lévy process possesses moments up to at least order three
    Invoked to obtain the explicit covariance structure coupling drift/scale through the third moment.
  • domain assumption State-dependent switching rates permit regime connectivity
    Required for the skeleton-chain argument establishing weighted exponential ergodicity.

pith-pipeline@v0.9.1-grok · 5699 in / 1647 out tokens · 38593 ms · 2026-07-02T00:34:06.972881+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

38 extracted references · 1 canonical work pages

  1. [1]

    Aalen,Nonparametric inference for a family of counting processes, The Annals of Statistics6(1978), no

    Odd O. Aalen,Nonparametric inference for a family of counting processes, The Annals of Statistics6(1978), no. 4, 701–726

    1978

  2. [2]

    Andersen, Ørnulf Borgan, Richard D

    Per K. Andersen, Ørnulf Borgan, Richard D. Gill, and Niels Keiding,Statistical models based on counting processes, Springer, New York, 1993

    1993

  3. [3]

    David Applebaum,L´ evy processes and stochastic calculus, Cambridge university press, 2009

    2009

  4. [4]

    2, 3648–3692

    Romain Aza¨ ıs and Aur´ elie Muller-Gueudin,Optimal choice among a class of nonparametric estimators of the jump rate for piecewise-deterministic Markov processes, Electronic Journal of Statistics10(2016), no. 2, 3648–3692

    2016

  5. [5]

    Barndorff-Nielsen,Normal inverse gaussian distributions and stochastic volatility modelling, Scandinavian Journal of Statistics24(1997), no

    Ole E. Barndorff-Nielsen,Normal inverse gaussian distributions and stochastic volatility modelling, Scandinavian Journal of Statistics24(1997), no. 1, 1–13

    1997

  6. [6]

    2, 185–201

    Rabi N Bhattacharya,On the functional central limit theorem and the law of the iterated logarithm for markov processes, Zeitschrift f¨ ur Wahrscheinlichkeitstheorie und verwandte Gebiete60(1982), no. 2, 185–201

    1982

  7. [7]

    3, 395–410

    Mogens Bladt and Michael Sørensen,Statistical inference for discretely observed Markov jump processes, Journal of the Royal Statistical Society Series B: Statistical Methodology67(2005), no. 3, 395–410

    2005

  8. [8]

    Ørnulf Borgan,Maximum likelihood estimation in parametric counting process models, with applications to censored failure time data, Scandinavian Journal of Statistics11(1984), no. 1, 1–16

    1984

  9. [9]

    HYBRID SWITCHING L ´EVY-DRIVEN SDES 39

    Rama Cont and Peter Tankov,Financial modelling with jump processes, Chapman & Hall/CRC Financial Mathematics Series, Chapman & Hall/CRC, Boca Raton, 2004. HYBRID SWITCHING L ´EVY-DRIVEN SDES 39

    2004

  10. [10]

    4, 1671–1691

    Douglas Down, Sean P Meyn, and Richard L Tweedie,Exponential and uniform ergodicity of markov processes, The Annals of Probability23(1995), no. 4, 1671–1691

    1995

  11. [11]

    Avner Friedman,Stochastic differential equations and applications, Courier Corporation, 2006

    2006

  12. [12]

    Probabilit´ es et statistiques38(2002), no

    Emmanuel Gobet,Lan property for ergodic diffusions with discrete observations, Annales de l’I.H.P. Probabilit´ es et statistiques38(2002), no. 5, 711–737 (eng)

    2002

  13. [13]

    8, 1354–1374

    El Houcine Hibbah, Hamid El Maroufy, Christiane Fuchs, and Taib Ziad,An MCMC computational approach for a continuous time state-dependent regime switching diffusion process, Journal of Applied Statistics47(2020), no. 8, 1354–1374

    2020

  14. [14]

    2, 211–229

    Mathieu Kessler,Estimation of an ergodic diffusion from discrete observations, Scandinavian Journal of Statistics24 (1997), no. 2, 211–229

    1997

  15. [15]

    ,Estimation of an ergodic diffusion from discrete observations, Scand. J. Statist.24(1997), no. 2, 211–229

    1997

  16. [16]

    Mathieu Kessler, Alexander Lindner, and Michael Sørensen,Statistical methods for stochastic differential equations, Monographs on Statistics and Applied Probability124(2012), 7–12

    2012

  17. [17]

    4, 2362–2388

    Nathalie Krell and Emeline Schmisser,Nonparametric estimation of jump rates for a specific class of piecewise deter- ministic Markov processes, Bernoulli27(2021), no. 4, 2362–2388

    2021

  18. [18]

    Kulik,Exponential ergodicity of the solutions to SDE’s with a jump noise, Stochastic Processes and their Applications119(2009), no

    Alexey M. Kulik,Exponential ergodicity of the solutions to SDE’s with a jump noise, Stochastic Processes and their Applications119(2009), no. 2, 602–632

    2009

  19. [19]

    Xuerong Mao and Chenggui Yuan,Stochastic differential equations with Markovian switching, Imperial college press, 2006

    2006

  20. [20]

    1, 35–56

    Hiroki Masuda,Ergodicity and exponentialβ-mixing bounds for multidimensional diffusions with jumps, Stochastic Processes and their Applications117(2007), no. 1, 35–56

    2007

  21. [21]

    Statist.41(2013), no

    ,Convergence of Gaussian quasi-likelihood random fields for ergodic L´ evy driven SDE observed at high fre- quency, Ann. Statist.41(2013), no. 3, 1593–1641

    2013

  22. [22]

    Carlo Mari and Emiliano Mari,Deep learning based regime-switching models of energy commodity prices, Energy Systems14(2023), 913–934

    2023

  23. [23]

    1, 105–137

    Hiroki Masuda and Yuma Uehara,Two-step estimation of ergodic L´ evy driven SDE, Statistical Inference for Stochastic Processes20(2017), no. 1, 105–137

    2017

  24. [24]

    2, 914–960

    Florence Merlev` ede and Magda Peligrad,Rosenthal-type inequalities for the maximum of partial sums of stationary processes and examples, The Annals of Probability41(2013), no. 2, 914–960

    2013

  25. [25]

    Meyn and R

    Sean P. Meyn and R. L. Tweedie,Stability of markovian processes i: Criteria for discrete-time chains, Advances in Applied Probability24(1992), no. 3, 542–574

    1992

  26. [26]

    3, 518–548

    Sean P Meyn and Richard L Tweedie,Stability of markovian processes iii: Foster–lyapunov criteria for continuous-time processes, Advances in Applied Probability25(1993), no. 3, 518–548

    1993

  27. [27]

    ,Markov chains and stochastic stability, Springer Science & Business Media, 2012

    2012

  28. [28]

    68, Cambridge University Press, Cambridge, 1999, Translated from the 1990 Japanese original, Revised by the author

    Ken-iti Sato,L´ evy processes and infinitely divisible distributions, Cambridge Studies in Advanced Mathematics, vol. 68, Cambridge University Press, Cambridge, 1999, Translated from the 1990 Japanese original, Revised by the author

    1999

  29. [29]

    8, 2885–2924

    Masayuki Uchida and Nakahiro Yoshida,Adaptive estimation of an ergodic diffusion process based on sampled data, Stochastic Processes and their Applications122(2012), no. 8, 2885–2924

    2012

  30. [30]

    4, 467–492

    Gabrielle Viennet,Inequalities for absolutely regular sequences: application to density estimation, Probability Theory and Related Fields107(1997), no. 4, 467–492

    1997

  31. [31]

    F. B. Xi,Asymptotic properties of jump-diffusion processes with state-dependent switching, Stochastic Processes and their Applications119(2009), no. 7, 2198–2221

    2009

  32. [32]

    F. B. Xi and G. Yin,Jump-diffusions with state-dependent switching: existence and uniqueness, feller property, linearization, and uniform ergodicity, Science China Mathematics54(2011), no. 12, 2651–2667

    2011

  33. [33]

    Fubao Xi and G. George Yin,On feller and strong feller properties and exponential ergodicity of regime-switching jump diffusion processes with countable regimes, SIAM Journal on Control and Optimization55(2017), no. 3, 1789–1818

    2017

  34. [34]

    63, Springer Science & Business Media, 2009

    G George Yin and Chao Zhu,Hybrid switching diffusions: properties and applications, vol. 63, Springer Science & Business Media, 2009

    2009

  35. [35]

    Yoshida,Simplified quasi-likelihood analysis for a locally asymptotically quadratic random field, arXiv preprint arXiv:2102.12460, 2021

    N. Yoshida,Simplified quasi-likelihood analysis for a locally asymptotically quadratic random field, arXiv preprint arXiv:2102.12460, 2021

    work page arXiv 2021

  36. [36]

    10, 3910–3940

    Yuzhong Cheng and Hiroki Masuda,Statistical inference for ergodic diffusion with Markovian switching, Discrete and Continuous Dynamical Systems - B30(2025), no. 10, 3910–3940

    2025

  37. [37]

    Nakahiro Yoshida,Polynomial type large deviation inequalities and quasi-likelihood analysis for stochastic differential equations, Ann. Inst. Statist. Math.63(2011), no. 3, 431–479

    2011

  38. [38]

    Yin, and Nicholas A

    Chao Zhu, G. Yin, and Nicholas A. Baran,Feynman–kac formulas for regime-switching jump diffusions and their applications, SIAM Journal on Control and Optimization55(2017), no. 2, 1045–1085. Kyushu University, Institute of Mathematics for Industry, 744 Motooka Fukuoka, Japan Email address:cheng.yuzhong.451@m.kyushu-u.ac.jp

    2017