Pith.

Fluid Dynamics · arXiv:2611.16512 v1 · submitted 12 Nov 2026 · 63 references

Variational Newton-Krylov descent for periodic orbits of two-dimensional Kolmogorov flow

Davide Lasagna, Thomas Burton

The paper says

“A matrix-free Newton-Krylov method for computing long unstable periodic orbits of turbulent flows is presented. The method arises from the space-time variational formulation of Lan & Cvitanović (Phys. Rev.”

more of the abstract

E, vol. 69, 2004, 016217), and runs on graphics processing units with a preconditioner from time-spectral methods that retains advection. In two-dimensional Kolmogorov flow at Re = 40, near recurrences remain about as frequent at long periods as at moderate ones, and searches from them converge to 183 distinct solutions with periods up to 191, about forty escape times, all but one verified by multiple shooting. The primary question is then whether such a long orbit can be taken, on its own, as a proxy for the attractor, and thus capture the statistics of the turbulent flow. The mean dissipation and energy of the long orbits lie close to the medians of the flow, and the distribution of the dissipation along a long orbit is comparable with that of a turbulent segment of the same duration, except in its tail, which most orbits miss. Their leading Floquet exponents appear to lie slightly below the Lyapunov spectrum of the flow, as may be expected of quieter orbits. The typical distance of the turbulent flow from a long orbit decreases as the period grows, but with an unfavourable scaling set by the correlation dimension of the attractor, which suggests that ever longer orbits may be needed as the Reynolds number grows.

from the abstract

Pith says

Machine review · deepseek-v4-flash · 13 Nov 2026 · what was read was not recorded · not checked by a person

It can be wrong. Challenge this review or reply to it.

Holds up. This paper delivers the longest periodic orbits yet in 2D Kolmogorov flow with a practical GPU-based space-time Newton method, and gives a thorough, honest assessment of how well single long orbits capture turbulent statistics. This line is wrong

5 minor

How often objections from reviews like this one are right: not yet measured

Are you an author? What you can do.

Not checked: figures, code, what the 63 references say.

5 minor objections

The paper says, at Abstract and §8.3

Pith says: minor objection 1 of 5

The decorrelation time τ_c is fitted to the median distance of turbulent segments (Figure 16d) and then used, via equation (8.7), to predict the median distance of the same segments and of the orbits. The paper already acknowledges that this is a fitted parameter (p. 32: 'with this single fitted constant'), but a brief remark in the abstract or conclusions clarifying that the scaling law is not parameter-free would help readers who skim.

The paper says, at §3.3

Pith says: minor objection 2 of 5

The choice of the shift σ (values between 0.5 and 4, with σ=2.5 for short loops and smaller for longer ones) is mentioned but not analysed systematically. A short discussion of how σ was selected or a sensitivity test would strengthen the reproducibility of the method.

show 3 more

The paper says, at Figure 16 and §8.3

Pith says: minor objection 3 of 5

The cumulative distribution in Figure 16(c) uses a sample of 1 152 trajectories for RPO178 and 768 for RPO86. The text states these numbers but does not explain why they differ; a brief justification would clarify the methodology.

The paper says, at §7.1, Table 4

Pith says: minor objection 4 of 5

The sample of seeds for the comparison between space-time and multiple shooting is described as 'stratified in period and chosen by the outcome of their space-time searches'. The paper makes clear that this is only a first indication (p. 25: 'we only sought a first indication'), but the reader would benefit from a statement of how many seeds were attempted in total and what the selection criteria were for inclusion in Table 4.

The paper says, at §9

Pith says: minor objection 5 of 5

The sentence 'Both methods were started from near recurrences…' (p. 37) refers to the space-time method and multiple shooting, but the preceding paragraph discusses adjoint-based methods; a small rephrasing would avoid ambiguity.

What would settle it

As a verification step, one could regenerate the statistics using a different norm (e.g., energy norm instead of vorticity norm) for the distance and correlation integral, to check that the scaling of median distance with period is not an artifact of the specific norm. This would test the robustness of the argument in §8.3.

from the skeptic's reading

How the review went

4 readings by deepseek-v4-flash · 13 Nov 2026

  1. Reader holds up confidence high
  2. Skeptic agrees 44 s · 41,696 tokens read, 2,832 written
  3. Editor wrote the verdict above 29 s · 42,401 tokens read, 1,258 written
  4. Referee accept: 5 minor 37 s · 41,815 tokens read, 2,420 written

Not run on this paper: the simulated authors' defence.

Pith did not check this review's objections against the paper.

The review in full

The review in brief, machine-written

The main news here is that the authors have made the space-time Newton method (Lan & Cvitanović) practical for a turbulent flow at Re=40 by using an approximate-factorisation preconditioner that retains advection, accompanied by a GPU implementation. They converge 183 distinct periodic orbits, the longest at period 191—more than double the previous record—and verify all but one by multiple shooting. That alone is a methodological step forward.

The paper is written with unusual candour. It openly states its limitations: the orbits are biased toward quiet states, the method's wall time scales roughly as the square of the period, the statistical comparison misses the tail of bursts, and sensitivities (the eventual goal) are not computed here. The analysis of shadowing via the correlation integral and the fitted decorrelation time is clever, but that fit is a free parameter. The authors acknowledge this and show the fit works well across periods.

Soft spots: the method has only been tested on one flow at one Reynolds number; scaling to 3D or higher Re is non-trivial and the authors say so. The selection bias toward quiet orbits is real and acknowledged—most orbits miss the rare bursts that raise the mean dissipation above the median. The comparison with multiple shooting is preliminary and limited to near-recurrence seeds, but the authors are appropriately cautious. None of these is a fatal flaw because they are all acknowledged and contextualised.

This paper is for the invariant-solutions and recurrent-flow-analysis community. It is a solid, reproducible computational study that advances the practical computation of long periodic orbits and provides a clear-eyed evaluation of their statistical relevance. The authors deserve credit for shipping code, doing careful verification, and not overselling.

Recommendation: Accept with minor revisions. The main suggestion would be to add a brief discussion of how the scaling issue might be tackled (e.g., projecting onto unstable/neutral modes), but that is not required for publication. The paper is ready for serious referees.

The strongest objection

After careful review, I cannot identify a load-bearing concern that would undermine the central claim. The paper is honest about its limitations (selection bias towards quiet orbits, scaling difficulties, need for longer orbits to confirm trends). The method is validated by multiple shooting, and the statistical comparisons with turbulent segments are appropriately qualified. The weakest assumption about dynamic connectedness of near recurrences is acknowledged and mitigated by empirical convergence of 183 orbits. No single unacknowledged error or gap appears to threaten the main conclusions.

The paper is a careful computational study that supports its claims with multiple lines of evidence. The central claim—that long periodic orbits can be found and that their statistics resemble turbulent segments except for the tail—is supported by convergence tests, multiple-shooting verification, and statistical comparisons. The authors openly discuss the selection bias towards quiet orbits, the need for longer orbits for stronger tests, and the limitations of the KL divergence in penalizing missing tails. I see no critical error or unsupported assertion that would require a change in verdict. The reader's ACCEPT verdict with high confidence is appropriate.

The referee's report

If the results hold, this work advances the computational reach of periodic orbit theory in turbulence by demonstrating that a variational Newton–Krylov approach can converge orbits far beyond the reach of standard shooting methods (periods up to 191 at Re=40). The statistical and instability comparisons with turbulent segments provide the first multi-orbit evidence at this scale that long periodic orbits, found from near recurrences, behave as turbulent segments of the same duration—a necessary condition for using individual orbits as proxies for the attractor in sensitivity analysis. The carefully verified method (all but one orbit confirmed by multiple shooting in double precision) and the honest acknowledgment of selection biases (quiet orbits dominate, tail is missed) strengthen the credibility of the conclusions. The work is a concrete step toward parameter sensitivities of turbulent statistics via periodic orbits, and the preconditioning strategy (time-spectral approximate factorisation on GPUs) is a technical innovation that may transfer to other multiscale periodic-flow problems.

Pith's reading of the paper (a machine's summary, not the authors')

The paper presents a preconditioned, matrix-free Newton-Krylov method that makes Newton's method on the whole space-time loop practical for computing long unstable periodic orbits in two-dimensional Kolmogorov flow at Re = 40. The method converges 183 distinct orbits with periods up to 191 (about 40 Lyapunov escape times). The central finding is that a single long orbit has mean dissipation and energy near the medians of the turbulent flow, and its dissipation distribution is comparable to that of a turbulent segment of the same length, except that it typically misses the rare high-dissipation bursts. The leading Floquet exponents lie slightly below the Lyapunov spectrum, consistent with quieter orbits. The typical distance of the turbulent flow from a long orbit falls as the period grows, but with a scaling set by the attractor's correlation dimension, implying that much longer orbits will be needed as the Reynolds number increases.

What carries the argument. The approximate-factorisation preconditioner for the space-time Newton system, which retains advection and makes Newton's method on the whole loop feasible. It replaces the sum of a temporal operator (dense in the time domain, diagonal in frequency) and a spatial operator (block diagonal in time, but coupling all spatial scales) by a product of two factors, each invertible in its own representation. This preconditioner clusters eigenvalues near a circle centred at 1/2, allowing GMRES to converge within a few hundred Arnoldi steps independent of spatial resolution at this Reynolds number.

Core claim. For two-dimensional Kolmogorov flow at Re = 40, the unstable periodic orbits that can be found from near recurrences have statistics that approach those of turbulent segments of the same length, with mean dissipation and energy near the flow's medians and Floquet exponents slightly below the Lyapunov spectrum. This supports the hypothesis that a single long periodic orbit can serve as a proxy for the bulk statistics of a turbulent flow, but also reveals the limitation that most orbits miss the rare bursting events that raise the mean dissipation above the median.

The premise it rests on. The assumption that near recurrences have dynamically connected ends, such that a small distance in state space implies the start and end of the trajectory lie on the same sheet of the folded attractor and can be linked by a smooth loop that Newton's method can deform into an orbit.

If the paper is right (Pith's inferences, not the authors')

If a single long orbit captures bulk statistics, sensitivities to parameters could be computed via the adjoint on the orbit without exponential growth, opening a route to parameter sensitivity of turbulent flows.

The unfavourable scaling of the typical distance with period (∝ T^{-1/D₂}, where D₂ is the correlation dimension) implies that for stronger turbulence, prohibitively long orbits may be needed to achieve a given closeness to the flow.

The near-recurrence rate at long periods suggests that the exponential proliferation of orbits balances the exponential decay of probability of staying near a specific orbit, so good seeds exist even at long periods.

The method's cost grows quadratically with period due to slow-scale modes that lengthen the GMRES plateau, motivating future work on preconditioners that capture those modes more efficiently.

The missing tail in the dissipation distribution indicates that a single long orbit cannot represent extreme events; a hybrid approach combining a long quiet orbit with a library of short bursting orbits may be needed to cover both bulk and tail.

The slow-scale modes that cause the GMRES plateau are likely associated with nearly neutral Floquet modes of the orbit; understanding their physical origin (e.g., large-scale drift) could lead to improved preconditioners that scale linearly with period.

The correlation-dimension scaling argument predicts that the orbit-proxy approach becomes increasingly demanding at higher Reynolds numbers unless the dimension of the attractor saturates, which is not yet established.

Check this review

Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.

How often reviews like this one are right

Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.

How often objections from reviews like this one are right: not yet measured

Every value above, in the review's public record (JSON)

Reviews by people

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

A matrix-free Newton-Krylov method for computing long unstable periodic orbits of turbulent flows is presented. The method arises from the space-time variational formulation of Lan & Cvitanović (Phys. Rev. E, vol. 69, 2004, 016217), and runs on graphics processing units with a preconditioner from time-spectral methods that retains advection. In two-dimensional Kolmogorov flow at Re = 40, near recurrences remain about as frequent at long periods as at moderate ones, and searches from them converge to 183 distinct solutions with periods up to 191, about forty escape times, all but one verified by multiple shooting. The primary question is then whether such a long orbit can be taken, on its own, as a proxy for the attractor, and thus capture the statistics of the turbulent flow. The mean dissipation and energy of the long orbits lie close to the medians of the flow, and the distribution of the dissipation along a long orbit is comparable with that of a turbulent segment of the same duration, except in its tail, which most orbits miss. Their leading Floquet exponents appear to lie slightly below the Lyapunov spectrum of the flow, as may be expected of quieter orbits. The typical distance of the turbulent flow from a long orbit decreases as the period grows, but with an unfavourable scaling set by the correlation dimension of the attractor, which suggests that ever longer orbits may be needed as the Reynolds number grows.

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-11-13

The authors' main claim, as the classifier picked it out 0.90

“The typical distance of the turbulent flow from a long orbit decreases as the period grows, but with an unfavourable scaling set by the correlation dimension of the attractor, which suggests that ever longer orbits may be needed as the Reynolds number grows.”

How the main claim is stated: hedged possibility 0.60

What the abstract itself states, as probabilities: claims first 0.05 · states limitation 0.94 · human participants 0.01 · controlled comparison 0.20 · randomized assignment 0.01 · states quantitative result 0.98 · compares with prior results 0.48 · mentions code or data availability 0.03

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
A matrix-free Newton-Krylov method for computing long unstable periodic orbits of turbulent flows is presented. contribution 0.66 0.90 0.05 asserted 1.00
The method arises from the space-time variational formulation of Lan & Cvitanovi\'c (Phys. Rev. E, vol. 69, 2004, 016217), and runs on graphics processing units with a preconditioner from time-spectral methods that retains advection. method 0.98 0.64 0.03 asserted 1.00
In two-dimensional Kolmogorov flow at $Re = 40$, near recurrences remain about as frequent at long periods as at moderate ones, and searches from them converge to 183 distinct solutions with periods up to 191, about forty escape times, all but one verified by multiple shooting. result 0.99 0.94 0.10 asserted 1.00
The primary question is then whether such a long orbit can be taken, on its own, as a proxy for the attractor, and thus capture the statistics of the turbulent flow. gap 0.57 0.17 0.06 unclear 0.72
The mean dissipation and energy of the long orbits lie close to the medians of the flow, and the distribution of the dissipation along a long orbit is comparable with that of a turbulent segment of the same duration, except in its tail, which most orbits miss. result 0.99 0.92 0.28 asserted 0.99
Their leading Floquet exponents appear to lie slightly below the Lyapunov spectrum of the flow, as may be expected of quieter orbits. result 0.78 0.86 0.07 hedged possibility 0.97
The typical distance of the turbulent flow from a long orbit decreases as the period grows, but with an unfavourable scaling set by the correlation dimension of the attractor, which suggests that ever longer orbits may be needed as the Reynolds number grows. interpretation 0.61 0.89 0.45 hedged possibility 0.60

What kind of paper the classifier reports this is: method or tool 0.98 · evidence from simulation or computation 1.00

What was examined, and what was not

Sources read

  • arXiv listing abstract v1

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-13: 1 request, 5,954 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Figures (18)
Figure 1 of the paper
Figure 1. The space-time residual for the normal form of a Hopf bifurcation, 𝑥¤ = 𝑥 − 𝑦 − 𝑥(𝑥 2 + 𝑦 2 ), 𝑦¤ = 𝑥 + 𝑦 − 𝑦(𝑥 2 + 𝑦 2 ), whose limit cycle is the unit circle. (a) A loop that is not an orbit, an ellipse with frequency 𝜔 = 0.8, along which its tangent (blue) and the vector field (orange) differ by the residual. (b) The orbit, to which the vector field is everywhere tangent. Grey lines are trajectories of the system. The orbits are then the zeros of the space-time residual 𝑟(𝑝) = 𝜔 𝜕𝑠𝜂 − 𝑐 𝜕𝑥𝜂 − N (𝜂), 𝜔 = e 𝛾 . (2.8) The residual (2.8) maps P into U, since the derivatives and t
Figure 2 of the paper
Figure 2. The effect of the resolution in time, on the periodic orbit P1 of Chandler & Kerswell (2013), of period 5.38 (a, b), and on RPO114, of period 60.80, which visits a burst (c–e). P1 is converged on one, two and four instants per unit time (blue, red and green) and RPO114 on two, four and six (red, green and purple), both also on about ten (grey). (a, c) The dissipation along the period, and (e) the burst of RPO114 in the window marked in (c). (b, d) The temporal spectrum (3.19), relative to ∥𝜂∥ 2 , over that of the turbulent flow (black), with the level 10−6 and the highest frequency…
Figure 3 of the paper
Figure 3. Near recurrences of the turbulent flow. (a) The distance (4.1) for 𝑡 and 𝑇 up to 2 000 along one simulation, with its local minima below the threshold 0.25 (red), and (c) the box marked in (a). (b) The median, the quartiles and the 1% and 0.1% quantiles of 𝑑(𝑡, 𝑇) over 𝑡, in bins of 𝑇, over sixty-four simulations. (d) The correlation integral 𝐶(𝑑), the fraction of pairs of states closer than 𝑑, in three bands of 𝑇, with a power law fitted to the pairs with 100 ≤ 𝑇 ≤ 500 for 0.1 ≤ 𝑑 ≤ 0.4 (dashed). (e) The rate of near recurrences per unit time, unit period and unit distance. Red…
Figure 4 of the paper
Figure 4. Smoothing the jump of an initial loop of period 108.8 that passes through a burst, on eight instants per unit time, by low-pass filters at three cut-offs and by the weighted average (4.2), against the time 𝜏 from the junction, with the window |𝜏| ≤ 𝜃 shaded. (a) The real part of the vorticity coefficient that jumps most at the junction, ˆ𝜂2,3, of wavenumber (𝑘 𝑥 , 𝑘 𝑦) = (2, 3). (b) The relative change of the loop at each instant. (c) The norm of the residual at each instant and (d) its temporal spectrum (3.19) against the frequency 𝑓 = 𝑚/𝑇, both with the weighted average also on…
Figure 5 of the paper
Figure 5. The effect of the preconditioner on the Newton system of the initial loop of period 5.4 that converges to P1. (a) The Ritz values of J after 2 000 Arnoldi steps, over the eigenvalues (3.10) of J0 (grey), which fill the rectangle of width 2𝐾 2 /Re and half-height 𝜔𝑀. (b) The same for J B−1 0 and (c) for J B−1 with 𝜎 = 2.5, with the circle of diameter one centred at one half (dashed). (d) The GMRES relative residual against the Arnoldi step, also with twice the Fourier modes in space (circles) and in time (triangles). derivative, the drift and the viscous term but not convection,…
Figure 6 of the paper
Figure 6. The progress of two searches, one per row, from a loop of period 100.0 that converges to RPO168 (a–d) and from one of period 86.3 that stalls (e–h). (a, e) The residual at each Newton iteration, open for hooksteps on the boundary of the trust region and filled for Newton steps. (b, f) The radius 𝛥 of the trust region and the length of the step, with a cross at each rejected step. (c, g) The ratio 𝜌 of §3.4, with the thresholds 1/4, below which (shaded) the radius shrinks, and 3/4. (d, h) The GMRES relative residual against the Arnoldi step, one line per Newton iteration, light to…
Figure 7 of the paper
Figure 7. The first Newton correction and the gradient in a search from a loop of period 42.9 that converges to RPO78. (a) The residual of the initial loop, (b) the correction of the first Newton iteration and (c) the gradient (5.2) of 𝑅 on the initial loop, at 𝑥 = 𝜋, against 𝑡/𝑇 and 𝑦. (d) The norm of the residual at each instant, at the start (orange) and every third Newton iteration (light to dark), with the window of the weighted average shaded, its end dashed in (a–c). done with shooting (Page et al. 2024b; Page 2025), may thus still save time, and is left to future work. 5.3. Wall time…
Figure 8 of the paper
Figure 8. The Newton iterations, Arnoldi steps and wall time of the searches against the period 𝑇0 of the initial loop. (a) The Newton iterations of the searches that converge. (b) The median number of Arnoldi steps per Newton iteration, for the searches that converge (blue) and those that do not (grey crosses). (c) The wall time of the searches that converge. The dashed lines are proportional to 𝑇0 in (b) and to 𝑇 2 0 in (c). campaign 𝑁𝑠/𝑇 precision 𝜎 Krylov searches converged solutions first 10 double 2.5 600–1 000 494 107 16 second 4 mixed, single 1, 2.5 1 500 537 188 52 third 8, 6–7…
Figure 9 of the paper
Figure 9. The outcome of the 3 962 searches. (a) The fraction of searches that converge, in bins of the period 𝑇0 of the seed, and the distinct orbits reached per search (triangles). (b) The same fraction in bins of the largest dissipation along the seed, for all the seeds (blue) and for 30 ≤ 𝑇0 < 100 (grey), without bins of fewer than five searches. (c) The number of searches that converge to each periodic (diamonds) and relative periodic orbit (discs) against its period, without the travelling wave and the equilibrium, found 308 and 73 times. quietest seeds mostly converge, to orbits in the…
Figure 10 of the paper
Figure 10. The orbits at Re = 40 in the plane of the energy input and the dissipation, over the joint probability density of the turbulent flow (grey, on a logarithmic scale relative to its peak), with the diagonal 𝐼 = 𝐷 (dotted). (a) The orbits found only by our searches (blue), only in the libraries of Page et al. (orange) and by both (black). (b) RPO178, the longest orbit (red), and RPO150, our long orbit of highest mean dissipation (purple). RPO178 appears as two vortex tubes that persist over the whole period and wander slowly across the domain, with small scales only near the peak of…
Figure 11 of the paper
Figure 11. RPO178, the longest orbit of our searches. (a) Its dissipation (solid) and energy input (dashed) along the period, and its vorticity at four instants A–D a quarter of the period apart (dots), the last at the largest dissipation, on a scale from −10 (blue) through 0 (black) to 10 (red), with the contours 𝜂 = ±5 (white). (b) The isosurfaces 𝜂 = 5 (red) and 𝜂 = −5 (blue) over one period, in the frame that drifts with the orbit, lighter where the faces of the box cut a vortex, with the vorticity on the scale of (a) at the instants A–D on semi-transparent slices (dashed) and at 𝑡 = 𝑇 on…
Figure 12 of the paper
Figure 12. The work of the two methods on the 263 seeds of the third campaign that both converge, for the space-time method (discs) and multiple shooting (triangles), against the period 𝑇0 of the seed. (a) The Newton iterations. (b) The evaluations of the vector field or of its linearisation at one instant, over the instants of the products with J and of the inner steps of the preconditioner for the space-time method, and over the stages of the time steps for multiple shooting. method a few more above, and above a period of 100 both converge almost none, but these differences rest on a few…
Figure 13 of the paper
Figure 13. The instability of the orbits against their period. (a) The leading Floquet exponent 𝜆1, (b) the sum 𝜎¯ = Í Re 𝜆𝑘>0 Re 𝜆𝑘 of the unstable exponents and (c) their number 𝑁𝑢, without the two zero exponents, also, where available, for the orbits of the libraries of Page et al., of table S1 in the supplementary material of Page et al. (2024b) and of Redfern et al. (2024), with the same quantities for the turbulent flow (dashed), averaged over the trajectories of table 5. (d) The leading Floquet exponents of our orbits against their index 𝑘, coloured by period, with the mean Lyapunov…
Figure 14 of the paper
Figure 14. Four long orbits against the initial loops of their searches, (a) RPO178, (b) RPO175, (c) RPO174 and (d) RPO169. In each block, from the top, the dissipation of the initial loop at 𝜏(𝑡) and of the orbit, their difference, the distance (8.3) with the recurrence distance of the seed, the threshold 0.075 and the fit (8.4), and 𝜏(𝑡) − 𝑡𝑇0/𝑇. On the right, the leading Floquet exponents of the orbit, up to twenty, from the subspace iteration of Lust (1997) (dots, red if unstable, grey if neutral, blue if stable) and from the Arnoldi iteration on the monodromy matrix (circles), with the…
Figure 15 of the paper
Figure 15. How close the orbits of seed period 𝑇0 ≥ 20 stay to their initial loops, against their period. (a) The time average of the distance (8.3), with the recurrence distance of the seed (crosses) and a line proportional to 1/𝑇. (b) The fraction of the period with 𝑑0 below the threshold 0.075, with the whole period, 80% and half of it (grey) and the line 1 − 𝑇𝑏/𝑇 fitted by least squares. Quiet orbits are open circles and those that approach the region of the bursts filled triangles, and RPO178, RPO175, RPO174 and RPO169 of figure 14 are ringed by a circle, a square, a diamond and a…
Figure 16 of the paper
Figure 16. The distance (8.5) of turbulent trajectories from long orbits. (a) To RPO178, along the trajectory that comes closest to it, over 1 000 time units about its closest approach at 𝑡𝑐, in escape times 1/𝜎¯ of the orbit, with the range of the distance to 32 turbulent segments of the same duration (grey). (b) The closest approach. (c) The cumulative distribution of the distance, the fraction of samples below 𝑑, to RPO178, over 1 152 trajectories of 2 500 time units (red, with a star at the median 𝑑𝑇 ), to RPO86 over 768 trajectories (green, with a star at its median), to the 32 turbulent…
Figure 17 of the paper
Figure 17. The statistics of single orbits against those of the turbulent flow. (a, b) The Kullback–Leibler divergence (8.10) of each orbit from the flow against its period, KL(𝐷) for the dissipation (a) and KL(𝐸) for the kinetic energy (b), coloured as in (c), with the median (grey line) and the quartiles (grey band) of turbulent segments of duration 𝑇, and a line ∝ 1/𝑇 through the median at 𝑇 = 2 000 (dashed). Orbits beyond the largest value of the flow have an infinite divergence (triangles at the top). (c) The mean dissipation of each orbit against its period, with vertical lines spanning…
Figure 18 of the paper
Figure 18. The performance of the operators on GPUs, for Kolmogorov flow at 𝐾 = 33, against the number of instants 𝑁𝑠 , for the residual, the product with J and an application of B −1 with the settings of the searches, in double (solid) and single precision (dashed). (a) The wall time on an NVIDIA H200 GPU. (b) Its speed-up over one core of an AMD EPYC 7502 processor. (c) The speed-up of the product with J on NVIDIA L4, L40S, A100 and H200 GPUs. The dotted line marks the largest loop of the searches. which this paper rests were developed in this way, through a careful analysis of the…

Figures are the authors', reproduced from the paper.

References (63)
  1. [1] Ashtari, O. & Schneider, T. M.2023 Identifying invariant solutions of wall-bounded three-dimensional shear flows using robust adjoint-based variational techniques.J. Fluid Mech.977, A7 on Pith
  2. [2] Ashtari, O., Tuckerman, L. S. & Schneider, T. M.2027 Computing simple invariant solutions in chaotic fluid flows.Annu. Rev. Fluid Mech.(in press) on Pith
  3. [3] Azimi, S., Ashtari, O. & Schneider, T. M.2022 Constructing periodic orbits of high-dimensional chaotic systems by an adjoint-based variational method.Phys. Rev. E105, 014217 on Pith
  4. [4] Beck, P. & Schneider, T. M.2026 Identifying recurrent flows in high-dimensional dissipative chaos from low-dimensional embeddings. arXiv:2601.01590 on PitharXiv
  5. [5] Benettin, G., Galgani, L., Giorgilli, A. & Strelcyn, J.-M.1980 Lyapunov characteristic exponents for smooth dynamical systems and for Hamiltonian systems; a method for computing all of them. Part 1: theory. Meccanica15, 9–20 on Pith
  6. [6] Budanur, N. B., Short, K. Y., Farazmand, M., Willis, A. P. & Cvitanovi´c, P.2017 Relative periodic orbits form the backbone of turbulent pipe flow.J. Fluid Mech.833, 274–301 on Pith
  7. [7] Burton, T., Symon, S. & Lasagna, D.2026 Searching for invariant solutions to wall-bounded flows using resolvent-based optimisation.J. Fluid Mech.1038, A12 on Pith
  8. [8] Cavaglieri, D. & Bewley, T.2015 Low-storage implicit/explicit Runge–Kutta schemes for the simulation of stiff high-dimensional ODE systems.J. Comput. Phys.286, 172–193 on Pith
  9. [9] Chandler, G. J. & Kerswell, R. R.2013 Invariant recurrent solutions embedded in a turbulent two- dimensional Kolmogorov flow.J. Fluid Mech.722, 554–595 on Pith
  10. [10] Chandramoorthy, N. & Wang, Q.2021 On the probability of finding nonphysical solutions through shadowing.J. Comput. Phys.440, 110389 on Pith
  11. [11] Crowley, C. J., Pughe-Sanford, J. L., Toler, W., Krygier, M. C., Grigoriev, R. O. & Schatz, M. F.2022 Turbulence tracks recurrent solutions.Proc. Natl. Acad. Sci. USA119, e2120665119. Cvitanovi´c, P., Artuso, R., Mainieri, R., Tanner, G. & Vattay, G.2025Chaos: Classical and Qu... on Pith
  12. [12] Ding, X., Chat ´e, H., Cvitanovi´c, P., Siminos, E. & Takeuchi, K. A.2016 Estimating the dimension of an inertial manifold from unstable periodic orbits.Phys. Rev. Lett.117, 024101 on Pith
  13. [13] Douglas, C. M., Emerson, B. L. & Lieuwen, T. C.2021 Nonlinear dynamics of fully developed swirling jets. J. Fluid Mech.924, A14 on Pith
  14. [14] Edwards, W. S., Tuckerman, L. S., Friesner, R. A. & Sorensen, D. C.1994 Krylov methods for the incompressible Navier–Stokes equations.J. Comput. Phys.110, 82–102 on Pith
  15. [15] Farazmand, M.2016 An adjoint-based approach for finding invariant solutions of Navier–Stokes equations.J. Fluid Mech.795, 278–312 on Pith
  16. [16] Farazmand, M. & Sapsis, T. P.2016 Dynamical indicators for the prediction of bursting phenomena in high-dimensional systems.Phys. Rev. E94, 032212 on Pith
  17. [17] Farazmand, M. & Sapsis, T. P.2017 A variational approach to probing extreme events in turbulent dynamical systems.Sci. Adv.3, e1701533 on Pith
  18. [18] Fazendeiro, L., Boghosian, B. M., Coveney, P. V. & L¨att, J.2010 Unstable periodic orbits in weak turbulence. J. Comput. Sci.1, 13–23 on Pith
  19. [19] Frame, P. & Towne, A.2027 Nonlinear space-time model reduction in the frequency domain.J. Comput. Phys. 568, 115403 on Pith
  20. [20] Freitas, A. C. M., Freitas, J. M. & Todd, M.2010 Hitting time statistics and extreme value theory.Probab. Theory Relat. Fields147, 675–710 on Pith
show 43 more
  1. [21] Golden, M.2026 A catalog of exact steady states in Kolmogorov flow. arXiv:2609.27062 on PithPith's reviewarXiv
  2. [22] Graham, M. D. & Floryan, D.2021 Exact coherent states and the nonlinear dynamics of wall-bounded turbulent flows.Annu. Rev. Fluid Mech.53, 227–253 on Pith
  3. [23] Grassberger, P. & Procaccia, I.1983 Characterization of strange attractors.Phys. Rev. Lett.50, 346–349. 0X0-40 Journal of Fluid Mechanics on Pith
  4. [24] Hall, K. C., Thomas, J. P. & Clark, W. S.2002 Computation of unsteady nonlinear flows in cascades using a harmonic balance technique.AIAA J.40, 879–886. Jim´enez, J.2020 Computers and turbulence.Eur. J. Mech. B/Fluids79, 1–11 on Pith
  5. [25] Kawahara, G. & Kida, S.2001 Periodic motion embedded in plane Couette turbulence: regeneration cycle and burst.J. Fluid Mech.449, 291–300 on Pith
  6. [26] Kawahara, G., Uhlmann, M. & van Veen, L.2012 The significance of simple invariant solutions in turbulent flows.Annu. Rev. Fluid Mech.44, 203–225 on Pith
  7. [27] Kazantsev, E.2001 Sensitivity of the attractor of the barotropic ocean model to external influences: approach by unstable periodic orbits.Nonlinear Process. Geophys.8, 281–300 on Pith
  8. [28] Knoll, D. A. & Keyes, D. E.2004 Jacobian-free Newton–Krylov methods: a survey of approaches and applications.J. Comput. Phys.193, 357–397 on Pith
  9. [29] Krygier, M. C., Pughe-Sanford, J. L. & Grigoriev, R. O.2021 Exact coherent structures and shadowing in turbulent Taylor–Couette flow.J. Fluid Mech.923, A7 on Pith
  10. [30] Lan, Y. & Cvitanovi´c, P.2004 Variational method for finding periodic orbits in a general flow.Phys. Rev. E 69, 016217 on Pith
  11. [31] Lasagna, D.2018 Sensitivity analysis of chaotic systems using unstable periodic orbits.SIAM J. Appl. Dyn. Syst.17, 547–580 on Pith
  12. [32] Lasagna, D.2020 Sensitivity of long periodic orbits of chaotic systems.Phys. Rev. E102, 052220 on Pith
  13. [33] Lea, D. J., Allen, M. R. & Haine, T. W. N.2000 Sensitivity analysis of the climate of a chaotic system.Tellus A52, 523–532 on Pith
  14. [34] Leffell, J. I., Murman, S. M. & Pulliam, T. H.2017 Time-spectral method for overlapping meshes.AIAA J. 55, 3381–3398 on Pith
  15. [35] Li, X. & Lasagna, D.2026 A space–time projection framework for approximate periodic orbit computation. J. Fluid Mech.1037, A32 on Pith
  16. [36] Lucas, D. & Kerswell, R. R.2015 Recurrent flow analysis in spatiotemporally chaotic 2-dimensional Kolmogorov flow.Phys. Fluids27, 045106 on Pith
  17. [37] Lust, K.1997 Numerical bifurcation analysis of periodic solutions of partial differential equations. PhD thesis, Katholieke Universiteit Leuven on Pith
  18. [38] McMullen, M. S., Jameson, A. & Alonso, J. J.2006 Demonstration of nonlinear frequency domain methods. AIAA J.44, 1428–1435 on Pith
  19. [39] Mundis, N. L. & Mavriplis, D. J.2017 Toward an optimal solver for time-spectral fluid-dynamic and aeroelastic solutions on unstructured meshes.J. Comput. Phys.345, 132–161 on Pith
  20. [40] Padovan, A., Otto, S. E. & Rowley, C. W.2020 Analysis of amplification mechanisms and cross-frequency interactions in nonlinear flows via the harmonic resolvent.J. Fluid Mech.900, A14 on Pith
  21. [41] Page, J.2025 Computation of simple invariant solutions in fluid turbulence with the aid of deep learning. Nonlinear Dyn.113, 31895–31915 on Pith
  22. [42] Page, J. & Kerswell, R. R.2020 Searching turbulence for periodic orbits with dynamic mode decomposition. J. Fluid Mech.886, A28 on Pith
  23. [43] Paninski, L.2003 Estimation of entropy and mutual information.Neural Comput.15, 1191–1253 on Pith
  24. [44] Parker, J. P. & Schneider, T. M.2022 Variational methods for finding periodic orbits in the incompressible Navier–Stokes equations.J. Fluid Mech.941, A17 on Pith
  25. [45] Platt, N., Sirovich, L. & Fitzmaurice, N.1991 An investigation of chaotic Kolmogorov flows.Phys. Fluids A3, 681–696 on Pith
  26. [46] Poulain, A., Content, C., Schioppa, A., Nibourel, P., Rigas, G. & Sipp, D.2024 Adjoint-based optimisation of time- and span-periodic flow fields with space–time spectral method: application to non-linear instabilities in compressible boundary layer flows.Comput. Fluids282, 106386 on Pith
  27. [47] Ramezanian, D., Mavriplis, D. J. & Ahrabi, B. R.2020 An order𝑁log𝑁parallel solver for time-spectral problems.J. Comput. Phys.411, 109319 on Pith
  28. [48] Redfern, E. M., Lazer, A. L. & Lucas, D.2024 Dynamically relevant recurrent flows obtained via a nonlinear recurrence function from two-dimensional turbulence.Phys. Rev. Fluids9, 124401 on Pith
  29. [49] Rigas, G., Sipp, D. & Colonius, T.2021 Nonlinear input/output analysis: application to boundary layer transition.J. Fluid Mech.911, A15 on Pith
  30. [50] Ruelle, D.1999 Smooth dynamics and new theoretical ideas in nonequilibrium statistical mechanics.J. Stat. Phys.95, 393–468 on Pith
  31. [51] Saad, Y.1993 A flexible inner-outer preconditioned GMRES algorithm.SIAM J. Sci. Comput.14, 461–469. 0X0-41 D. Lasagna and T. Burton on Pith
  32. [52] Saad, Y. & Schultz, M. H.1986 GMRES: a generalized minimal residual algorithm for solving nonsymmetric linear systems.SIAM J. Sci. Stat. Comput.7, 856–869. S´anchez, J. & Net, M.2010 On the multiple shooting continuation of periodic orbits by Newton–Krylov methods.Int. J. Bifu... on Pith
  33. [53] Sierra, J., Jolivet, P., Giannetti, F. & Citro, V.2021 Adjoint-based sensitivity analysis of periodic orbits by the Fourier–Galerkin method.J. Comput. Phys.440, 110403. Sierra-Aus´ın, J., Citro, V., Giannetti, F. & Fabre, D.2022 Efficient computation of time-periodic compressi... on Pith
  34. [54] Suri, B., Kageorge, L., Grigoriev, R. O. & Schatz, M. F.2020 Capturing turbulent dynamics and statistics in experiments with unstable periodic orbits.Phys. Rev. Lett.125, 064501 on Pith
  35. [55] Thomas, J. P., Custer, C. H., Dowell, E. H., Hall, K. C. & Corre, C.2013 Compact implementation strategy for a harmonic balance method within implicit flow solvers.AIAA J.51, 1374–1381 on Pith
  36. [56] Tuckerman, L. S.1989 Steady-state solving via Stokes preconditioning; recursion relations for elliptic operators. In11th International Conference on Numerical Methods in Fluid Dynamics(ed. D. L. Dwoyer, M. Y. Hussaini & R. G. Voigt),Lecture Notes in Physics, vol. 323, pp. 573–... on Pith
  37. [57] Tuckerman, L. S., Langham, J. & Willis, A. P.2019 Order-of-magnitude speedup for steady states and traveling waves via Stokes preconditioning in Channelflow and Openpipeflow. InComputational Modelling of Bifurcations and Instabilities in Fluid Dynamics(ed. A. Gelfgat),Computat... on Pith
  38. [58] Viswanath, D.2007 Recurrent motions within plane Couette turbulence.J. Fluid Mech.580, 339–358 on Pith
  39. [59] Wang, B., Ayats, R., Deguchi, K., Meseguer, A. & Mellibovsky, F.2025 Mathematically established chaos and forecast of statistics with recurrent patterns in Taylor–Couette flow.J. Fluid Mech.1011, R2 on Pith
  40. [60] Wang, D. & Lan, Y.2023 A reduced variational approach for searching cycles in high-dimensional systems. Nonlinear Dyn.111, 5579–5592 on Pith
  41. [61] Yang, H.-l., Takeuchi, K. A., Ginelli, F., Chat ´e, H. & Radons, G.2009 Hyperbolicity and the effective dimension of spatially extended dissipative systems.Phys. Rev. Lett.102, 074102 on Pith
  42. [62] Yang, J.2015 A numerical method for computing time-periodic solutions in dissipative wave systems.Stud. Appl. Math.134, 420–455 on Pith
  43. [63] Zheng, Z., Beck, P., Yang, T., Ashtari, O., Parker, J. P. & Schneider, T. M.2025 Ghost states underlying spatial and temporal patterns: how nonexistent invariant solutions control nonlinear dynamics.Phys. Rev. E 112, 024212. 0X0-42 on Pith

Challenges and discussion

Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.

No comments yet.

Sign in to comment