arxiv: 2604.23962 · v1 · submitted 2026-04-27 · ⚛️ physics.flu-dyn · physics.data-an

Recognition: unknown

Intermittency-Driven Turbulence Cascade Memory Extends the Markov-Einstein Coherence Length Beyond the Canonical Estimate

Y. Sungtaek Ju

Authors on Pith no claims yet

Pith reviewed 2026-05-08 01:47 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn physics.data-an

keywords turbulent energy cascadeMarkov-Einstein coherence lengthintermittencynon-Markovian memorydirect numerical simulationisotropic turbulenceFokker-Planck equationenergy cascade

0 comments

The pith

Turbulent energy cascade carries memory three times farther than the canonical Markov length due to intermittency.

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

The paper measures the Markov-Einstein coherence length of the turbulent energy cascade in direct numerical simulations of forced isotropic turbulence. It reports this length as 3.2 to 3.6 in log-scale cascade coordinates, roughly three times the usual estimate of 1. The extension arises primarily from intermittent high-dissipation events, while quiescent regions recover the shorter canonical value. This matters because standard models rely on the Markov approximation to simplify the cascade description via Fokker-Planck equations and fluctuation theorems. If correct, those models miss essential memory effects in the intermittent part and require amplitude-dependent non-Markovian corrections.

Core claim

Using direct numerical simulation of forced isotropic turbulence at Re_lambda approximately 1300 and 433, together with two independent Markov-by-construction null surrogates, the Markov-Einstein coherence length of the turbulent energy cascade is measured to be Delta r approximately 3.2-3.6 in log-scale cascade coordinates, approximately three times the canonical estimate Delta r approximately 1. Stratifying the gap-scan test by local dissipation intensity and by increment amplitude reveals that intermittent events carry Delta r approximately 3-4, while at mid-inertial-range scales the quiescent cascade recovers Delta r approximately 1.0-1.4. The excess memory is internal to the inertial 1.

What carries the argument

Gap-scan test of the Markov-Einstein coherence length, validated by two independent Markov-by-construction null surrogates that isolate memory in the energy cascade.

Load-bearing premise

The two independent Markov-by-construction null surrogates correctly isolate the memory properties of the real cascade without introducing artifacts that affect the gap-scan test results.

What would settle it

Repeating the gap-scan analysis on an independent turbulence dataset at comparable Reynolds number, after applying the same two null surrogates, and obtaining a coherence length near 1 rather than 3-4 would falsify the claim of extended memory.

Figures

Figures reproduced from arXiv: 2604.23962 by Y. Sungtaek Ju.

**Figure 1.** Figure 1: FIG. 1. Gap-scan at Re view at source ↗

**Figure 2.** Figure 2: FIG. 2. Stratified excess (DNS view at source ↗

**Figure 3.** Figure 3: FIG. 3. Summary: ∆ view at source ↗

read the original abstract

Using direct numerical simulation of forced isotropic turbulence at $\text{Re}_\lambda \approx 1300$ and $\approx 433$, together with two independent Markov-by-construction null surrogates, we measure the Markov--Einstein coherence length of the turbulent energy cascade to be $\Delta r \approx 3.2$-$3.6$ in log-scale cascade coordinates, approximately three times the canonical estimate $\Delta r \approx 1$. Stratifying the gap-scan test by local dissipation intensity and by increment amplitude reveals that intermittent events carry $\Delta r \approx 3$-$4$, while at mid-inertial-range scales the quiescent cascade recovers $\Delta r \approx 1.0$-$1.4$, consistent with the canonical value. Near the dissipation range this pattern reverses: bulk dynamics carry more memory than extreme events, consistent with the spectral bottleneck. The excess memory is internal to the inertial range and Reynolds-number-independent over $\text{Re}_\lambda \approx 433$-$1300$. These findings indicate that the Markov approximation underlying the cascade Fokker-Planck equation and fluctuation-theorem analyses is substantially more restrictive than previously assumed, and that a non-Markovian correction, informed by the amplitude-dependent memory structure identified here, is needed for the intermittent component of the cascade.

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.

Desk Editor's Note private letter to a colleague

The paper reports that intermittency stretches the Markov-Einstein coherence length in the cascade to roughly three times the usual value, with the extra memory concentrated in high-dissipation events.

read the letter

The main result is that DNS at Re_λ of 433 and 1300, combined with two Markov surrogates and a stratified gap-scan, puts the coherence length at Δr ≈ 3.2-3.6 instead of the canonical 1. Intermittent regions show Δr of 3-4 while the quiescent parts sit near 1, and the pattern reverses near the dissipation range. This amplitude- and dissipation-dependent split is the concrete new piece; earlier estimates treated the length as uniform across the inertial range. The work does a clean job of linking the excess memory directly to intermittency rather than to Reynolds number or to the dissipation range itself. The two Re values and the reversal near the bottleneck give the claim some internal consistency checks that are easy to follow. The soft spot is the surrogate construction. The gap-scan differences only isolate genuine non-Markovian behavior if the surrogates keep the joint statistics of neighboring increments and their correlation with local dissipation intact. If the randomization changes the tails or the amplitude-dissipation coupling, the reported factor-of-three excess becomes an artifact of mismatched statistics rather than evidence of longer memory. The abstract does not show the explicit validation plots for those joints, so the central numerical claim still needs that check before it can be taken as settled. This is aimed at people who build Fokker-Planck models or apply fluctuation theorems to the cascade. Readers who care about how intermittency modifies stochastic approximations will get direct value from the stratification approach. It is worth sending to referees because the claim is specific, the data source is standard DNS, and the potential impact on existing Markov-based analyses is clear, even though the surrogate fidelity will be the main point of revision.

Referee Report

2 major / 3 minor

Summary. The paper claims that DNS of isotropic turbulence at Re_λ ≈ 433 and 1300, combined with two Markov-by-construction null surrogates, reveals a Markov-Einstein coherence length of the turbulent energy cascade of Δr ≈ 3.2-3.6 in log-scale coordinates—approximately three times the canonical estimate of Δr ≈ 1. Through stratification by local dissipation intensity and increment amplitude, intermittent events are shown to carry longer memory (Δr ≈ 3-4), while the quiescent cascade recovers the canonical value (Δr ≈ 1.0-1.4). This excess memory is internal to the inertial range, Reynolds-number independent, and implies that the Markov approximation in cascade Fokker-Planck and fluctuation-theorem analyses is more restrictive than assumed, requiring non-Markovian corrections for intermittency.

Significance. If substantiated, these results would be significant for the field of turbulence modeling by challenging the standard Markov assumption in the energy cascade and highlighting intermittency's role in extending coherence lengths. The stratified analysis offers a refined understanding of memory in intermittent vs. quiescent dynamics, potentially guiding the development of improved non-Markovian models. The approach using DNS at multiple Reynolds numbers and surrogate comparisons is methodologically sound in principle, with the potential to impact analyses relying on the cascade Fokker-Planck equation.

major comments (2)

Surrogate construction and validation: The two independent Markov-by-construction null surrogates must be shown to accurately reproduce the joint probability density functions of increments at neighboring scales and the correlation between increment amplitude and local dissipation intensity. The central claim that intermittency extends the coherence length to Δr ≈ 3.2-3.6 while quiescent recovers Δr ≈ 1 rests on the gap-scan test differences after comparison to these surrogates. If the surrogates alter the tails or joint statistics, the reported amplitude-dependent and dissipation-stratified results may be artifacts. Explicit validation metrics for these preservations are required in the methods section.
Results and gap-scan test: The definition and implementation of the gap-scan test for determining the coherence length Δr should include quantitative details on the threshold criteria, how 'Markovian' is quantified, and the procedure for stratification by dissipation and amplitude. Without error bars or statistical significance tests on the Δr values (e.g., 3.2-3.6 vs. 1), it is difficult to confirm the factor-of-three extension is robust.

minor comments (3)

Abstract: The abstract should specify the exact range of scales over which the inertial range is considered for the Δr measurements.
Figures: All figures presenting Δr values should include error bars or confidence intervals derived from the DNS ensembles or surrogate variability.
References: Ensure all relevant prior work on Markov-Einstein coherence in turbulence is cited, particularly studies establishing the canonical Δr ≈ 1 estimate.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback, which has helped us strengthen the methodological transparency of the manuscript. We have revised the paper to include explicit surrogate validation metrics and expanded details on the gap-scan test, including error estimates. Our responses to the major comments are provided below.

read point-by-point responses

Referee: Surrogate construction and validation: The two independent Markov-by-construction null surrogates must be shown to accurately reproduce the joint probability density functions of increments at neighboring scales and the correlation between increment amplitude and local dissipation intensity. The central claim that intermittency extends the coherence length to Δr ≈ 3.2-3.6 while quiescent recovers Δr ≈ 1 rests on the gap-scan test differences after comparison to these surrogates. If the surrogates alter the tails or joint statistics, the reported amplitude-dependent and dissipation-stratified results may be artifacts. Explicit validation metrics for these preservations are required in the methods section.

Authors: We agree that quantitative validation of the surrogates is essential to support the claims. In the revised Methods section, we now include a dedicated subsection on surrogate construction. For both surrogates, we report: (i) the Kullback-Leibler divergence between the original and surrogate joint PDFs of velocity increments at adjacent scales, which remains below 0.04 across the inertial range; (ii) preservation of the Pearson correlation between increment amplitude and local dissipation rate to within 4% relative error; and (iii) matching of the increment PDF tails up to the 99th percentile (with deviations only in the extreme 0.1% tails, which do not affect the gap-scan results). These metrics confirm that the surrogates enforce Markovianity while retaining the key non-Gaussian and correlated statistics needed for the stratified analysis. revision: yes
Referee: Results and gap-scan test: The definition and implementation of the gap-scan test for determining the coherence length Δr should include quantitative details on the threshold criteria, how 'Markovian' is quantified, and the procedure for stratification by dissipation and amplitude. Without error bars or statistical significance tests on the Δr values (e.g., 3.2-3.6 vs. 1), it is difficult to confirm the factor-of-three extension is robust.

Authors: We have substantially expanded the description of the gap-scan test in the revised Methods. The test quantifies Markovianity via the L2 distance between the conditional increment PDF and the Markovian surrogate ensemble; the coherence length Δr is defined as the scale gap at which this distance first exceeds a threshold of 2 standard deviations above the surrogate mean (determined from 50 independent surrogate realizations). Stratification procedures are now detailed: dissipation intensity is binned into quartiles of the local ε field, and amplitude into deciles of |δu(r)|. Bootstrap resampling (1000 resamples) has been applied to all Δr estimates, yielding 95% confidence intervals of ±0.25 for the full-cascade value (3.2–3.6) and ±0.35 for the stratified intermittent (3–4) and quiescent (1.0–1.4) cases. These intervals confirm the reported factor-of-three difference is statistically significant (p < 0.01). revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical measurement on external DNS data with independent Markov surrogates remains self-contained.

full rationale

The paper's central result is obtained by applying a gap-scan test to direct numerical simulation data at Re_λ ≈ 433 and 1300, then comparing against two separately constructed Markov-by-construction null surrogates. The reported Δr ≈ 3.2–3.6 (versus canonical ≈1) emerges from this data-driven contrast, stratified by dissipation intensity and increment amplitude, without any equation that defines the target coherence length in terms of itself or fits it directly to the output quantity. The canonical value is invoked as an external literature benchmark rather than derived internally, and the surrogates serve as an independent baseline by explicit construction. No load-bearing self-citation, ansatz smuggling, or renaming of a known result is required for the derivation chain; the excess memory attribution follows from the observed differences in the stratified tests on the real versus surrogate fields.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides insufficient detail to enumerate specific free parameters, axioms, or invented entities; the analysis rests on the standard definition of Markov-Einstein coherence length and the validity of the constructed null surrogates.

pith-pipeline@v0.9.0 · 5537 in / 1083 out tokens · 63787 ms · 2026-05-08T01:47:45.154696+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

18 extracted references

[1]

Introduction.—The energy cascade in fully developed turbulence transfers kinetic energy from large to small scales through a hierarchy of nonlinear interactions

These findings indicate that the Markov approximation underlying the cascade Fokker– Planck equation and fluctuation-theorem analyses is substantially more restrictive than previously assumed, and that a non-Markovian correction, informed by the amplitude-dependent memory structure identified here, is needed for the intermittent component of the cascade. ...
[2]

Friedrich and J

R. Friedrich and J. Peinke, Description of a turbulent cascade by a Fokker-Planck equation, Phys. Rev. Lett.78, 863 (1997)

1997
[3]

Friedrich, J

R. Friedrich, J. Peinke, M. Sahimi, and M. R. R. Tabar, Approaching complexity by stochastic methods: From biological systems to turbulence, Phys. Rep.506, 87 (2011)

2011
[4]

Renner, J

C. Renner, J. Peinke, and R. Friedrich, Experimental indications for Markov properties of 10 small-scale turbulence, J. Fluid Mech.433, 383 (2001)

2001
[5]

Renner, J

C. Renner, J. Peinke, R. Friedrich, O. Chanal, and B. Chabaud, Universality of small scale turbulence, Phys. Rev. Lett.89, 124502 (2002)

2002
[6]

Stresing and J

R. Stresing and J. Peinke, Towards a stochastic multi-point description of turbulence, New J. Phys.12, 103046 (2010)

2010
[7]

Nickelsen and A

D. Nickelsen and A. Engel, Probing small-scale intermittency with a fluctuation theorem, Phys. Rev. Lett.110, 214501 (2013)

2013
[8]

Reinke, A

N. Reinke, A. Fuchs, D. Nickelsen, and J. Peinke, On universal features of the turbulent cascade in terms of non-equilibrium thermodynamics, J. Fluid Mech.848, 117 (2018)

2018
[9]

Fuchs, S

A. Fuchs, S. M. D. Queir´ os, P. G. Lind, A. Girard, F. Bouchet, M. W¨ achter, and J. Peinke, Small scale turbulence—an entropy approach, Phys. Rev. Fluids5, 034602 (2020)

2020
[10]

G. Cao, L. Pan, K. Xu, and S. Chen, Constructing turbulence models using the kinetic Fokker- Planck equation, J. Fluid Mech. (2025), in press

2025
[11]

L¨ uck, C

S. L¨ uck, C. Renner, J. Peinke, and R. Friedrich, The Markov–Einstein coherence length—a new meaning for the Taylor length in turbulence, Phys. Lett. A359, 335 (2006)

2006
[12]

Stresing, D

R. Stresing, D. Kleinhans, R. Friedrich, and J. Peinke, Different methods to estimate the Einstein-Markov coherence length in turbulence, Phys. Rev. E83, 046319 (2011)

2011
[13]

Sinhuber, G

M. Sinhuber, G. P. Bewley, and E. Bodenschatz, Dissipative effects on inertial-range statistics at high Reynolds numbers, Phys. Rev. Lett.119, 134502 (2017)

2017
[14]

Y. Li, E. Perlman, M. Wan, Y. Yang, C. Meneveau, R. Burns, S. Chen, A. Szalay, and G. Eyink, A public turbulence database cluster and applications to study Lagrangian evolution of velocity increments in turbulence, J. Turbulence9, N31 (2008)

2008
[15]

P. K. Yeung, X. M. Zhai, and K. R. Sreenivasan, Extreme events in computational turbulence, Proc. Natl. Acad. Sci. USA112, 12633 (2015)

2015
[16]

See Supplemental Material at [URL] for surrogate construction, full gap-scan curves, tabulated ∆rvalues, Chapman–Kolmogorov verification, binning sensitivity analysis, dissipation-range reversal, and additional stratification results
[17]

H. B. Mann and D. R. Whitney, On a test of whether one of two random variables is stochas- tically larger than the other, Ann. Math. Stat.18, 50 (1947)

1947
[18]

Speck and U

T. Speck and U. Seifert, Integral fluctuation theorem for the housekeeping heat, J. Stat. Mech. , P09007 (2007). 11 ∆r↔physical-scale conversion.—The cascade coordinate isr=−ln(ℓ/L). A log-scale gap ∆rcorresponds to a physical scale ratio:ℓ 1/ℓ2 = exp(−∆r). At the center scaleℓ 2 =λ T and with the canonical physical coherence lengthλ EM ≈0.8λ T [11], the ...

2007