Pith. sign in

REVIEW 4 major objections 7 minor 45 references

The prediction of extreme uncertainty-production events in three-dimensional Navier-Stokes turbulence

T0 review · 4 major / 7 minor · reviewed 2026-08-08 · deepseek-v4-flash

Pith's one-line read In lower-Reynolds-number turbulence, strain rate is the best single-field predictor of extreme uncertainty production, while vorticity is weak and near-average states are unpredictable.

desk verdict New evolution equation and honest statistics, but the claim of an unpredictable near-average region likely rests on a scale mismatch between box-trained committors and pointwise inputs. read the letter →

arxiv 2608.05208 v1 pith:2NCC224Q submitted 2026-08-05 physics.flu-dyn

classification physics.flu-dyn
keywords turbulencepredictabilityuncertaintyproductioncommittorfunctionanalogueMarkovchainstrainratevorticityBrierscorevelocitygradientinvariants
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper asks which local flow quantities reveal, before it happens, that a small uncertainty in a turbulent flow will be violently amplified. It derives an evolution equation for the local uncertainty-production term from the Navier-Stokes equations, then estimates the probability of an extreme production event as a function of candidate predictor fields using a committor function built from an analogue Markov chain. In the lower-Reynolds-number simulations, the magnitude of the strain rate is the most informative single-flow predictor and vorticity is the weakest; at higher Reynolds numbers the ranking flattens. The paper also identifies a region near the spatial averages of strain rate and vorticity where even probabilistic forecasts are effectively impossible with these predictors. If this holds, local strain-rate measurements could serve as an early-warning diagnostic for where turbulence will decorrelate fastest, and forecasters should distrust predictions made near the average state.

What carries the argument

The central object is the committor function $q(z_0) = P(T_B < T_A \mid z(0)=z_0)$, the probability that a coarse-grained flow state reaches the extreme positive production set $B=\{P_\Delta>1\}$ before the normal negative-production set $A=\{P_\Delta<0\}$. The machinery that makes it computable is the analogue Markov chain: sampled box-averaged predictor states $\{Z_n\}$ define a discrete phase space, each state can jump with probability $1/K$ to the successor of one of its $K$ nearest neighbours, the sets $A$ and $B$ are made absorbing, and the committor is obtained from the eigenvectors of the resulting transition matrix by solving the linear system (3.9)-(3.11). The Brier score separates the intrinsic probabilistic floor from the error of the approximate committor, so comparing scores across predictor choices measures how much information each predictor retains about future extremes.

What would settle it

Run the same committor training on a fresh DNS realization at the same Reynolds number and compare out-of-sample Brier scores for $\|S\|_F$ and $\|\Omega\|_F$: the central claim fails if vorticity matches or beats strain rate at low Reynolds number, or if adding a nonlocal predictor such as pressure or two-point velocity correlations removes the near-average unpredictability band.

Watch

Extended reading notes

Core claim

This paper studies the local production of uncertainty in forced three-dimensional Navier-Stokes turbulence, defined as $P_\Delta = -w_i S_{ij} w_j$, where $w$ is the difference between two nearly identical velocity fields and $S$ is the strain-rate tensor of the reference flow. From the Navier-Stokes equations it derives an evolution equation for $P_\Delta$ (Eq. 2.4), identifying the inertial contributions of strain rate, vorticity, and vortex deformation. To forecast extreme positive $P_\Delta$ events before they occur, the paper estimates the committor function --- the probability that a state reaches the extreme-production set $B$ before the normal set $A$ --- from DNS data using an analogue Markov chain, and scores the resulting probabilistic forecasts with the Brier score. It finds that in lower-Reynolds-number cases the strain-rate magnitude $\|S\|_F$ has the lowest Brier score among single-flow predictors, while vorticity $\|\Omega\|_F$ has weak predictive power; at higher Reynolds numbers the three velocity-gradient predictors become nearly indistinguishable. It also finds a band of states near the spatial-average values of strain rate and vorticity where the committor varies rapidly and no stable probabilistic forecast can be made from these one-point fields.

Load-bearing premise

The whole forecast construction assumes that the box-averaged, normalized predictor state evolves as a Markov chain with time step $\Delta t$, so that the analogue neighbours and transition probabilities trained from DNS are a faithful representation of the flow's future; if the reduced dynamics has longer memory or the analogues do not represent the flow, the committor is biased and the predictor ranking could change.

Editorial extensions

If this is right

  • A single measured field --- the local strain-rate magnitude --- can flag where uncertainty will be produced, without needing to know the perturbed flow simultaneously.
  • Forecast systems for turbulence should either avoid, or supplement with extra information, regions where strain rate and vorticity sit near their spatial averages.
  • At higher Reynolds numbers, no one-point velocity-gradient scalar dominates; predictive skill depends on the alignment of uncertainty with strain and on fuller flow information.
  • The $Q$-$R$ topology maps extreme uncertainty production onto known strain self-amplification and vortex-stretching regions, giving testable spatial signatures of future decorrelation.
  • The analogue-Markov-committor method transfers to other localized extreme events anywhere in a turbulent flow, not only events at fixed positions or windows.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: if strain rate is the dominant precursor, velocity-gradient measurements in experimental or operational flows could be processed in real time to issue probabilistic warnings of uncertainty bursts, a step the paper does not take.
  • Beyond the paper: the near-average unpredictability band is probably a signature of missing state variables (pressure, alignment, two-point structure); a testable extension is to add the pressure Hessian or a second-point strain correlation as a predictor and see whether the band shrinks.
  • Beyond the paper: the same pipeline could rank precursors for other intermittent extremes in turbulence --- local dissipation, enstrophy bursts, or Reynolds-stress events --- and would reveal whether strain-rate dominance is specific to uncertainty production or generic.
  • Beyond the paper: the box-size dependence of the Brier ranking is untested; reducing $L_D$ toward the Taylor scale should sharpen or wash out the strain-rate advantage, and that would be a clean numerical experiment.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The manuscript derives an evolution equation for the local uncertainty-production term P_Delta from the three-dimensional incompressible Navier-Stokes equations, identifies strain rate, vorticity, and vortex deformation as candidate local predictors, and estimates committor functions from DNS data using an analogue Markov chain. The predictor ranking is evaluated with Brier scores in four flow configurations (two forcing types, two Reynolds numbers each). The two headline claims are that at lower Reynolds numbers the strain rate is the best single-flow-field predictor of future extreme P_Delta events while vorticity has weak predictive power, and that near the spatial means of strain rate and vorticity the committor varies so rapidly that stable probabilistic forecasts are effectively impossible.

Significance. If the claims hold, the Brier-score ranking identifies a cheap single-flow precursor for localized uncertainty bursts, and the near-average unpredictability region would be an important practical caveat for reduced-order forecasting of turbulence. The paper has clear strengths: a complete derivation of the P_Delta evolution equation from Navier-Stokes, four systematically varied DNS configurations, multiple perturbation realizations for each case, a transparent analogue-Markov-chain methodology, and a direct comparison between committor estimates and empirical probabilities for single predictors in Figure 10. The central statistical framework is well suited to the question, and the paper is a plausible step toward probabilistic local predictability of turbulent uncertainty production.

major comments (4)
  1. [Section 6.2, Figure 13; Sections 4.3.1 and 4.3.4; Eq. (3.12)] The committor is trained and Brier-tested on box-averaged fields, as stated in Section 4.3.1 and used consistently in Section 4.3.4, but Figure 13 evaluates the committor at every grid point using pointwise values of the predictors and the nearest-neighbor interpolation formula (3.12). Box averaging reduces variance and changes the temporal autocorrelation of the fields, so pointwise states near the spatial mean may have few close analogues among the box-averaged training states, and the K-nearest-neighbor average in (3.12) may then mix distant, unrepresentative analogues. The rapid committor fluctuations interpreted as probabilistic unpredictability in the near-average region could therefore be a scale-mismatch artifact. Please recompute the Figure 13 field from box-averaged testing boxes, or retrain the committor on pointwise fields, before claiming that one-point kinematic fields cannot resolve predictability near the average.
  2. [Section 3.1 and Section 3.3] The analogue Markov chain assumes that the reduced predictor process is Markovian with time step Delta t, an assumption stated in Section 3.1 and used to build the chain in Sections 3.2 and 3.3. This assumption is not tested. If the true reduced dynamics has memory longer than Delta t, the committor computed from Eq. (3.8) is biased and the Brier-score ranking of predictors could change. The agreement with directly sampled empirical probabilities in Figure 10 validates the single-step conditional distribution but does not validate the multi-step Markov property on which the committor equation rests. Please add a quantitative memory check, for example comparing transition probabilities conditioned on one-step versus two-step histories, or reporting the autocorrelation decay of the predictor fields relative to Delta t.
  3. [Section 2.1, Eq. (2.4)] The analysis drops the third-order term -R_ij w_k d_k S_ij on the basis of |w| << |u|, and then focuses on the inertial contribution (i), dropping the pressure contribution (ii), the viscous contribution (iii), and the forcing contribution (iv) without a quantitative estimate of their magnitudes. Since the choice of strain rate, vorticity, and vortex deformation as predictors is motivated by the retained inertial terms, the theoretical basis for the predictor ranking would be substantially strengthened by a DNS-based estimate of each term in Eq. (2.4) within the exponential-growth regime. Such an estimate would show whether the neglected pressure and viscous contributions could change the relative importance of the candidate predictors.
  4. [Section 6.1, Figures 8 and 9] The headline ranking at lower Reynolds numbers -- strain rate best, vorticity worst -- rests on differences between Brier-score curves, but the text reports no statistical significance test for these differences. The standard-deviation bands shown in Figures 8 and 9 can be comparable to the separation between curves, particularly at intermediate values of alpha. Please provide a significance test, for example a paired bootstrap over testing boxes or over perturbation realizations, and state whether the ranking is statistically robust at each value of alpha and for each case.
minor comments (7)
  1. [Section 3, introductory paragraph] In the sentence 'measure the precision of that estimation using the Brier score in ref 3.4', the cross-reference should read 'Section 3.4' rather than 'ref 3.4'.
  2. [Eq. (3.14)] The typesetting of Eq. (3.14) contains a stray vertical bar and mismatched parentheses in the expression involving the square root; the formula should be displayed so that the decomposition into the irreducible variance term and the squared error term is unambiguous.
  3. [Figure 11 and Figure 12 captions] The word 'normalsed' appears in both captions and should be 'normalised'.
  4. [Section 6.2, final paragraph] The word 'probablities' should be 'probabilities'.
  5. [Conclusion, last paragraph before Acknowledgments] The word 'alignement' should be 'alignment'.
  6. [Figure 7 caption] The caption contains a double comma after 'F2 512' in the listing of panels.
  7. [Abstract and front matter] The 'Key words' section is empty; either provide keywords or remove the heading.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: data-driven committor results contradict the motivating theory, and self-citations are minor and not load-bearing.

full rationale

The paper's derivation chain is self-contained. Eq. (2.4) is derived from the Navier-Stokes equations inside the paper, and the predictor set is motivated by it, but the committor estimates are data-driven and not forced by the theory: Fig. 10 explicitly shows that the probability of extreme events increases with vorticity, which the authors note 'goes counter to our theoretical analysis in equation (2.4)'. The Brier-score ranking is an empirical evaluation on testing boxes, not a fitted parameter renamed as a prediction. The analogue Markov chain method is cited from Lucente et al. (2022b), which includes one of the present authors, but it is a published methodological tool and the paper's physical conclusions do not rest on an unverified self-cited uniqueness claim. The 'probabilistic unpredictability' region in Fig. 13 is an empirical interpolation of the committor to pointwise fields; whether the box-averaged training versus pointwise evaluation introduces a scale-mismatch bias is a correctness risk, not a circularity. Minor self-citations to Ge et al. (2023, 2025a,b) and Goto & Vassilicos (2009) are present, but the observations they cite are either reproduced in this paper (PDF collapse, exponential growth) or are standard derivations, so they are not load-bearing. Score 2 reflects these minor self-citations while affirming that the central derivation and statistical findings are independent.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

No new physical entities are posited. The main free choices are the event threshold, box size, and number of analogues. The analysis rests on Markovianity, ergodicity, and the restriction to inertial effects, all stated in the text.

free parameters (3)
  • Event threshold for set B = P_Delta/sigma_PDelta > 1
    The extreme-event set B is defined by a normalized threshold of one standard deviation; this choice is not fitted but arbitrary and could affect the committor and the ranking of predictors.
  • Number of nearest neighbors K = 100
    K=100 is chosen for the analogue Markov chain; the authors state K=50 and K=150 give no qualitative change, but the value is a hand-selected parameter of the algorithm.
  • Sampling box half-size L_D = 2^n * dx, n chosen so 2L_D just exceeds P(u=0)*lambda
    The box size is selected from a flow-based geometric criterion rather than fitted to the prediction target, but it determines the spatial coarse-graining of all predictors and affects the sampled state space.
assumptions (6)
  • domain assumption Incompressible Navier-Stokes with periodic boundary conditions and two prescribed forcing types
    The DNS solves forced N-S on a periodic box; the results apply to this idealized setup, not to wall-bounded or shear turbulence.
  • domain assumption The analogue Markov chain with time step Delta t models the true coarse-grained dynamics
    Section 3.1 states 'assuming that it is Markovian with a time step Delta t'; this is the load-bearing premise for the committor computation.
  • domain assumption Ergodicity of the sampled process
    Section 3.1 assumes sample means converge to ensemble averages and there are no isolated areas in phase space.
  • ad hoc to paper Neglect of the third-order term -R_ij w_k d_k S_ij
    Section 2.1 justifies this by |w| much smaller than |u| during the exponential growth regime, but no quantitative bound is given.
  • ad hoc to paper Focus on inertial contribution only, dropping pressure and viscous terms
    Section 2.1 states pressure is nonlocal and viscous terms are small-scale, but the effect of these dropped terms on P_Delta is not quantified; the authors flag this as speculative in the conclusion.
  • domain assumption Approximate statistical stationarity of normalized fields
    Section 4.2 shows the normalized P_Delta PDF collapses during the exponential growth, which justifies disregarding chronological order when constructing the Markov chain.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The prediction of extreme uncertainty-production events in three-dimensional Navier-Stokes turbulence." pith.science (2026). https://pith.science/paper/2NCC224Q

@misc{pith2026260805208,
  author       = {Pith},
  title        = {Pith review of: The prediction of extreme uncertainty-production events in three-dimensional Navier-Stokes turbulence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2NCC224Q}},
  note         = {Machine review of arXiv:2608.05208}
}
abstract

We investigate the exponential growth of uncertainty energy in 3D Navier-Stokes turbulence, emphasising the intermittent and highly localized amplification/production of uncertainty, a critical factor in understanding the predictability of turbulent systems. From the Navier-Stokes equations one can identify some key fields contributing to the growth/decay of uncertainty-production term $P_{\Delta}$: strain rate, vorticity, and vortex deformation. The dynamics of these fields are examined in the $Q-R$ plane, where $Q$ and $R$ are the second and third invariants of the velocity gradient tensor, to understand their role in the evolution of uncertainty-production term $P_{\Delta}$. We proceed by estimating committor functions across the entire spatiotemporal domain of direct numerical simulations (DNS) of turbulence in a periodic domain at different Reynolds numbers. Our estimates of the probability of rare extreme events of local uncertainty-production term as a function of uncertainty energy, strain rate, vorticity, and vortex deformation confirm the role of strain rate in driving uncertainty. Where strain rate and vorticity are too close to their space-average values, stable probabilistic forecasts appear impossible solely on the basis of the fields considered here.

Figures

Figures reproduced from arXiv: 2608.05208 by the authors.

Figure 1
Figure 1. An example of probability density function (PDF) of the local uncertainty-production term 𝑃𝛥 obtained as explained in section 4. The sets B and A are schematically represented in this plot: B contains rare extreme events in the positive 𝑃𝛥 tail of the PDF, and A contains all the negative production events. The red line is the PDF of 𝑃𝛥 normalized by the probability at 𝑃𝛥 = 0 and the standard deviation 𝜎𝑃𝛥 of 𝑃𝛥 esti… view at source ↗
Figure 2
Figure 2. Top: Schematic of the analogue method to predict a deterministic evolution (figure reproduced from Miloshevich et al. (2024)). Bottom: Schematic of the analogue Markov chain method taking 𝐾 = 5 as an example. On the left-hand side, the event 𝒁𝑛 (shown in red and not corresponding to the event at the final instant of any time series) is surrounded by its 𝐾 nearest events  𝒁T𝑛𝑘 1≤ T𝑛𝑘 ≤𝑁𝑡 −1 , 1 ≤ 𝑘 ≤ 𝐾 including its… view at source ↗
Figure 3
Figure 3. Time evolution of (a) the average uncertainty energy in a semilogarithmic plot and (b) the corresponding growth exponent in a linear plot. 𝛾 ≡ 1 2 d ln⟨𝐸𝛥⟩/d𝑡, 𝜏𝜂 denotes the Kolmogorov time scale, and ⟨𝐸𝑡𝑜𝑡⟩ = D 𝒖 (1) [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Time evolution of PDFs of 𝑃𝛥 for cases (a) F1128, (b) F1512 and (c) F2128 and (d) F2512. Similarly to Ge et al. (2023), PDFs are normalized by their maximum Pmax and are plotted versus 𝑃𝛥/𝜎𝑃𝛥 , where 𝜎𝑃𝛥 is the standard deviation of 𝑃𝛥 with 𝜎 2 𝑃𝛥 = ∫ 𝑃𝛥max 𝑃𝛥min (𝑃𝛥 −…
Figure 5
Figure 5. Figure 5: Time evolution of the average uncertainty production, normalized by its standard deviation. The top blue 𝑥-axis corresponds to cases F1128 and F2128, whereas the bottom red 𝑥-axis corresponds to cases F1512 and F2512. The size of the sampling box, 2𝐿D, is chosen to be …
Figure 6
Figure 6. Figure 6: (a) One example of spatial distribution of observation boxes (𝑁𝑥 = 28) within the simulation domain. Inside the boxes the streamlines of the velocity of the reference flow are plotted. (b) Spatial distribution of all potential testing data boxes Case [𝑡min/𝜏𝜂, 𝑡max/𝜏𝜂]…
Figure 7
Figure 7. Figure 7: Time averaged distribution of 𝑃𝛥 in the 𝑄 − 𝑅 plane for cases (a) F1128, (b) F1512, (c) F2128 and (d) F2512,, The continuous black line corresponds to (27/4)𝑅 2 + 𝑄 3 = 0 and is dominated by strain self￾amplification in the lower-right quadrant. (e) shows panels (a) an…
Figure 8
Figure 8. Figure 8: For (a)(c)(e) case F1128 in left panel and case F1512 in right panel, the Brier score 𝐵𝑇𝑁 of the approximated committor function versus the number of observation boxes 𝛼𝑁𝑥 , obtained from (a)(b) a single scalar field, (c)(d) two scalar fields and (e)(f) more than two s…
Figure 9
Figure 9. Figure 9: For (a)(c)(e) case F2128 in left panel and case F2512 in right panel, the Brier score 𝐵𝑇𝑁 of the approximated committor function versus the number of observation boxes 𝛼𝑁𝑥 , obtained from (a)(b) a single scalar field, (c)(d) two scalar fields and (e)(f) more than two s…
Figure 10
Figure 10. Figure 10: For cases (a) F1128, (b) F1512, (c) F2128, and (d) F2512, committor functions estimated from different single-field predictors using the analogue Markov chain, namely {𝐸𝛥, ∥ [S, 𝜴] ∥𝐹, ∥S∥𝐹, ∥𝜴∥𝐹 }. The lines show the averages over the 𝑁case realizations, and the shad…
Figure 11
Figure 11. Figure 11: For (a)(c)(e) case F1128 in left panel and (b)(d)(f) case F1512 in right panel, the committor function estimated with different predictors: (a)(b) normalised strain rate ∥S∥𝐹 - normalsed vorticity ∥𝜴∥𝐹. The dashed line indicates 𝑄 = 0 and separates the strain-dominate…
Figure 12
Figure 12. Figure 12: For (a)(c)(e) case F2128 and (b)(d)(f) case F2512, the committor function estimated with different predictors: (a)(b) normalised strain rate ∥S∥𝐹 - normalsed vorticity ∥𝜴∥𝐹. The dashed line indicates 𝑄 = 0 and separates the strain-dominated region (𝑄 < 0) from the rot…
Figure 13
Figure 13. Figure 13: For (a) F1128, (b) F1512, (c) F2128 and (d) F2512, the probability (in colours) based on the estimated committor function, using the input {∥𝜴∥𝐹 , ∥S∥𝐹 } at every grid point of the DNS at its corresponding starting time of sampling. The vertical and horizontal dash-do…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 35 canonical work pages

  1. [1]

    Batchelor, G. K. 1953 The Theory of Homogeneous Turbulence\/ . Cambridge: Cambridge University Press

  2. [2]

    & Ho, R.D.J.G

    Berera, A. & Ho, R.D.J.G. 2018 Chaotic properties of a turbulent isotropic fluid . Phys. Rev. Lett. 120 (2), 024101

  3. [3]

    & Musacchio, S

    Boffetta, G. & Musacchio, S. 2017 Chaos and predictability of homogeneous-isotropic turbulence . Phys. Rev. Lett. 119 (5), 054102

  4. [4]

    Monthly weather review 78 (1), 1--3

    Brier, Glenn W 1950 Verification of forecasts expressed in terms of probability . Monthly weather review 78 (1), 1--3

  5. [5]

    2013 Introduction to Symmetry Analysis\/

    Cantwell, B. 2013 Introduction to Symmetry Analysis\/ . Cambridge: Cambridge University Press

  6. [6]

    & Wong, A.Y

    Cheung, P.Y. & Wong, A.Y. 1987 Chaotic behavior and period doubling in plasmas . Phys. Rev. Lett. 59 (5), 551

  7. [7]

    & Girimaji, S

    Das, R. & Girimaji, S. S. 2019 On the reynolds number dependence of velocity-gradient structure and dynamics . J. Fluid Mech. 861 , 163--179

  8. [8]

    1986 Is navier--stokes turbulence chaotic? Phys

    Deissler, R.G. 1986 Is navier--stokes turbulence chaotic? Phys. Fluids 29 (5), 1453--1457

Show all 45 references
  1. [9]

    , Ren, W

    E, W. , Ren, W. & Vanden-Eijnden, E. 2005 Transition pathways in complex systems: reaction coordinates, isocommittor surfaces, and transition tubes . Chem. Phys. Lett. 413 (1--3), 242--247

  2. [10]

    & Vanden-Eijnden, E

    E, W. & Vanden-Eijnden, E. 2006 Towards a theory of transition paths . J. Stat. Phys. 123 (3), 503--523

  3. [11]

    Elsinga, G. E. , Ishihara, T. , Goto, S. , Boersma, B. J. & Hunt, J. C. R. 2023 Intermittency across reynolds numbers -- the influence of large-scale shear layers on the scaling of the enstrophy and dissipation in homogeneous isotropic turbulence . J. Fluid Mech. 974 , A6

  4. [12]

    & Peinke, J

    Friedrich, R. & Peinke, J. 1997 Description of a turbulent cascade by a Fokker--Planck equation . Physical Review Letters 78 (5), 863--866

  5. [13]

    , Peinke, J

    Friedrich, R. , Peinke, J. , Sahimi, M. & Reza Rahimi Tabar, M. 2011 Approaching complexity by stochastic methods: From biological systems to turbulence . Physics Reports 506 (5), 87--162

  6. [14]

    2009 Stochastic Methods: A Handbook for the Natural and Social Sciences\/

    Gardiner, C. 2009 Stochastic Methods: A Handbook for the Natural and Social Sciences\/ . Springer Berlin Heidelberg

  7. [15]

    , Rolland, J

    Ge, J. , Rolland, J. & Vassilicos, J. C. 2023 The production of uncertainty in three-dimensional navier–stokes turbulence . J. Fluid Mech. 977 , A17

  8. [16]

    , Rolland, J

    Ge, J. , Rolland, J. & Vassilicos, J. C. 2025 a\/ Eulerian--lagrangian scaling of the lyapunov exponent in homogeneous turbulence . Phys. Rev. Fluids 10 (7), L072601

  9. [17]

    , Rolland, J

    Ge, J. , Rolland, J. & Vassilicos, J. C. 2025 b\/ The interscale behaviour of uncertainty in three-dimensional Navier--Stokes turbulence . J. Fluid Mech. 1017 , A29

  10. [18]

    & Vassilicos, J.C

    Goto, S. & Vassilicos, J.C. 2009 The dissipation rate coefficient of turbulence is not universal and depends on the internal stagnation point structure . Phys. Fluids 21 (3)

  11. [19]

    , Armua, A

    Ho, R.D.J.G. , Armua, A. & Berera, A. 2020 Fluctuations of lyapunov exponents in homogeneous and isotropic turbulence . Phys. Rev. Fluids 5 (2), 024602

  12. [20]

    , Wray, A

    Hunt, J. , Wray, A. & Moin, P. 1988 Eddies, streams, and convergence zones in turbulent flows . Center for Turbulence Research Proceeedings of the Summer Program pp. 193--208

  13. [21]

    Johnson, P. L. & Wilczek, M. 2024 Multiscale velocity gradients in turbulence . Annu. Rev. Fluid Mech. 56 (1), 463--490

  14. [22]

    1970 Instability in fully developed turbulence

    Kraichnan, R.H. 1970 Instability in fully developed turbulence . Phys. Fluids 13 (3), 569--575

  15. [23]

    1971 Atmospheric predictability and two-dimensional turbulence

    Leith, C.E. 1971 Atmospheric predictability and two-dimensional turbulence . J. Atmos. Sci. 28 (2), 145--161

  16. [24]

    Physical Review Letters 115 (13), 134502

    Linkmann, Moritz F & Morozov, Alexander 2015 Sudden relaminarization and lifetimes in forced isotropic turbulence . Physical Review Letters 115 (13), 134502

  17. [25]

    1963 Deterministic nonperiodic flow

    Lorenz, E.N. 1963 Deterministic nonperiodic flow . J. Atmos. Sci. 20 (2), 130--141

  18. [26]

    1969 a\/ The predictability of a flow which possesses many scales of motion

    Lorenz, E.N. 1969 a\/ The predictability of a flow which possesses many scales of motion . Tellus 21 (3), 289--307

  19. [27]

    Lorenz, E. N. 1969 b\/ Atmospheric predictability as revealed by naturally occurring analogues . J. Atmos. Sci. 26 (4), 636--646

  20. [28]

    Lorenz, E. N. 1969 c\/ Three approaches to atmospheric predictability . Bull. Am. Meteorol. Soc. 50 (5), 345--349

  21. [29]

    , Herbert, C

    Lucente, D. , Herbert, C. & Bouchet, F. 2022 a\/ Committor functions for climate phenomena at the predictability margin: the example of El Ni \ n o--Southern Oscillation in the jin and timmermann model . J. Atmos. Sci. 79 (9), 2387--2400

  22. [30]

    , Rolland, J

    Lucente, D. , Rolland, J. , Herbert, C. & Bouchet, F. 2022 b\/ Coupling rare event algorithms with data-based learned committor functions using the analogue markov chain . J. Stat. Mech.: Theory Exp. 2022 (8), 083201

  23. [31]

    Journal of Physics A: Mathematical and Theoretical 48 (25), 25FT01

    McComb, WD , Linkmann, MF , Berera, A , Yoffe, SR & Jankauskas, B 2015 Self-organization and transition to turbulence in isotropic fluid motion driven by negative damping at low wavenumbers . Journal of Physics A: Mathematical and Theoretical 48 (25), 25FT01

  24. [32]

    , Cozian, B

    Miloshevich, G. , Cozian, B. , Abry, P. , Borgnat, P. & Bouchet, F. 2023 Probabilistic forecasts of extreme heatwaves using convolutional neural networks in a regime of lack of data . Phys. Rev. Fluids 8 (4), 040501

  25. [33]

    , Lucente, D

    Miloshevich, G. , Lucente, D. , Yiou, P. & Bouchet, F. 2024 Extreme heat wave sampling and prediction with analog markov chain and comparisons with deep learning . Environ. Data Sci. 3 , e9

  26. [34]

    , Sasai, Y

    Nonaka, M. , Sasai, Y. , Sasaki, H. , Taguchi, B. & Nakamura, H. 2016 How potentially predictable are midlatitude ocean currents? Sci. Rep. 6 (1), 20153

  27. [35]

    1938 Initial recombination of ions

    Onsager, L. 1938 Initial recombination of ions . Phys. Rev. 54 (8), 554

  28. [36]

    Pope, S. B. 2000 Turbulent flows\/ . Cambridge university press

  29. [37]

    , Held, M

    Prinz, J. , Held, M. , Smith, J. C. & No \'e , F. 2011 Efficient computation, sensitivity, and error analysis of committor probabilities for complex dynamical processes . Multiscale Model. Simul. 9 (2), 545--567

  30. [38]

    1981 Small random perturbations of dynamical systems and the definition of attractors

    Ruelle, D. 1981 Small random perturbations of dynamical systems and the definition of attractors . Commun. Math. Phys. 82 , 137--151

  31. [39]

    Springer

    Tsinober, Arkady 2019 The Essence of Turbulence as a Physical Phenomenon: With Emphasis on Issues of Paradigmatic Nature\/ , 2nd edn. Springer

  32. [40]

    1982 Local interaction between vorticity and shear in a perfect incompressible fluid

    Vieillefosse, P. 1982 Local interaction between vorticity and shear in a perfect incompressible fluid . J.Phys. 43 (6), 837--842

  33. [41]

    1984 Internal motion of a small element of fluid in an inviscid flow

    Vieillefosse, P. 1984 Internal motion of a small element of fluid in an inviscid flow . Physica A 125 (1), 150--162

  34. [42]

    & Meneguzzi, M

    Vincent, A. & Meneguzzi, M. 1991 The spatial structure and statistical properties of homogeneous turbulence . J. Fluid Mech. 225 , 1--20

  35. [43]

    2014 Anawege: a weather generator based on analogues of atmospheric circulation

    Yiou, P. 2014 Anawege: a weather generator based on analogues of atmospheric circulation . Geosci. Model Dev. 7 (2), 531--543

  36. [44]

    & D \'e andr \'e is, C

    Yiou, P. & D \'e andr \'e is, C. 2019 Stochastic ensemble climate forecast with an analogue model . Geosci. Model Dev. 12 (2), 723--734

  37. [45]

    2012 Investigation of the transfer and dissipation of energy in isotropic turbulence

    Yoffe, S.R. 2012 Investigation of the transfer and dissipation of energy in isotropic turbulence . PhD thesis, University of Edinburgh

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.