Separation of periodic orbits in the delay embedded space of chaotic attractors

Eurika Kaiser; J Nathan Kutz; Prerna Patil; Steven Brunton

REVIEW 4 major objections 4 minor 17 references

Separation of periodic orbits in the delay embedded space of chaotic attractors

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper shows that in long time-delay embeddings of chaotic attractors, unstable periodic orbits separate into clusters whose positions are set by a single ratio $f(\rho)=(\rho-1)/(\rho+1)$, with $\rho$ the time spent in the two…

desk verdict A genuinely new visual phenomenon—UPO clusters ordered by lobe time ratio in long delay embeddings—with an unproven quantitative formula and fixable proof gaps. read the letter →

arxiv 2411.13103 v1 pith:EEIYXHUG submitted 2024-11-20 nlin.CD

Prerna Patil , Eurika Kaiser , J Nathan Kutz , Steven Brunton This is my paper · ORCID

classification nlin.CD MSC 37D4537C2737M10

keywords unstableperiodicorbitstime-delayembeddingHankelmatrixsymbolicdynamicsLorenzattractorRösslersingularvaluedecompositionRedfield-Pólyaenumeration

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 tries to establish that unstable periodic orbits (UPOs) of a chaotic attractor do not sit arbitrarily in a long time-delay embedding: as the height of the Hankel matrix grows, the orbits separate into clusters, and the position of each orbit is set by a single scalar, the ratio $\rho$ of time the orbit spends in the two symbolic lobes of the attractor. The proposed position law is $f(\rho)=(\rho-1)/(\rho+1)$, so an orbit spending equal time in both lobes sits in the center, while A-heavy and B-heavy orbits move to opposite sides of the unfolded attractor. If correct, the geometry of the delay-embedded attractor becomes a simple ordering device: the embedding does not just reconstruct the attractor, it classifies its building-block orbits by a dynamical invariant. This matters because UPOs are the organizing skeleton of chaotic dynamics, and a quantitative handle on their placement in embedding space could aid identifying, comparing, and controlling orbits from time series alone.

What carries the argument

The machine that carries the argument is the Hankel matrix $H_{p,q}$ built from time-shifted scalar measurements of one periodic orbit, projected onto the SVD basis of a Hankel matrix built from a long chaotic trajectory via $V_{\mathrm{proj}}^{\mathsf{T}} = \Sigma_c^{-1}U_c^{\mathsf{T}}H^{(k)}$. Theorem 1 is the load-bearing result: for a periodic orbit sampled with irrational $\Delta t/\tilde{t}$, Birkhoff's ergodic theorem makes the row-space average $(1/p)(1_p\otimes\sigma)^{\mathsf{T}}H$ converge to a constant row vector, so the projection onto the direction that repeats the separation-plane normal $\sigma$ across all $p$ delays is a pure function of the dwell-time ratio. The explicit function is $f(\rho)=(\rho-1)/(\rho+1)$, with $\rho = |I_A|/|I_B|$; Lemma 1 and Theorem 2 supply the range-containment and lifted-projection facts needed for the SVD basis to realize this separation. A secondary piece is the modified Redfield-Pólya enumeration $P(n) = \frac{1}{n}\sum_{i=1}^{n} k^{\gcd(i,n)} - \sum_{i|n}P(i) - k$, which counts the unique non-cyclic, multi-symbol sequences.

What would settle it

Compute the centroids of UPO clusters in the embedded space at large $t_{\mathrm{height}}$ for several Lorenz orbits with known symbol counts and check whether the centroids lie on the predicted curve $\alpha(\rho-1)/(\rho+1)$ along the $v_3$ axis with the same constant $\alpha$; a nonlinear or non-monotone relationship, or a mismatch between the Theorem 1 direction and $v_3$, would refute the claim. Equivalently, take two Lorenz UPOs with the same symbol ratio $\rho$ but different symbol order, such as the ratio-1 orbits AABABB and BBABAA, and check whether their cluster centroids coincide at large $t_{\mathrm{height}}$ within the within-cluster spread.

Watch

Extended reading notes

Core claim

The central claim is that delay embedding with a sufficiently tall Hankel matrix unravels the UPOs of the Lorenz and Rössler attractors so that each orbit maps to a point whose location reflects the fraction of its period spent on each side of a separation plane. For the Lorenz system the plane is $x=0$, the symbolic lobes A and B match the two sides, and Theorem 1 shows that the long-time average of the projection $\sigma^{\mathsf{T}} x$ along the repeated direction $(1_p\otimes\sigma)$ converges to $\alpha |I_A|/\tilde{t} + \beta |I_B|/\tilde{t}$. Because $\alpha \approx -\beta$ by symmetry and the lobe dwell times are proportional to symbol counts, this limit collapses to $\alpha(\rho-1)/(\rho+1)$, which is the formula the paper uses to order the clusters. The same mechanism works for Rössler, but with $\rho$ redefined as the ratio of time spent in $y>0$ versus $y<0$ rather than the symbolic A/B ratio. The paper also derives the number of distinct UPOs of sequence length $n$ by a constrained Redfield-Pólya enumeration, $P(n)$, that removes cyclic repeats and mono-symbolic sequences.

Load-bearing premise

The quantitative position formula is derived for a specific averaged projection direction, but the paper does not prove that this direction coincides with the SVD axis along which the clusters are plotted and measured.

Editorial extensions

If this is right

UPOs with the same symbol ratio $\rho$ cluster together in the embedded space regardless of the order of symbols in their symbolic name, so the geometric placement ignores the order of A and B symbols.
Symmetric orbits with equal A and B counts ($\rho=1$) sit in the central cluster, while A-heavy and B-heavy orbits move monotonically to opposite ends as $f(\rho)$ approaches $+1$ and $-1$.
For the Rössler attractor, the ordering is governed by the time spent above and below $y=0$, not by the symbolic partition at $y_c \approx -3.04$, so the same law applies with a recalculated $\rho$.
The constrained Redfield-Pólya count gives the number of unique UPOs for sequence length $n$ with $k$ symbols, matching the completeness of the symbolic dynamics for the parameter values studied.

Reading between the lines

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

Inference: if the position law holds along a principal axis, the delay-embedded coordinate of a UPO is a direct observable estimate of its dwell-time ratio $\rho$, turning cluster geometry into a data-driven symbol-ratio estimator for unseen orbits.
Inference: the paper leaves open whether Theorem 1's projection direction $(1_p\otimes\sigma)$ coincides with the SVD axis used in the figures; a direct test would be to regress measured cluster centroids along $v_3$ against $\alpha f(\rho)$ and check that the slope is one.
Inference: for attractors with more than two symbols, the scalar ratio would have to be replaced by a vector of per-lobe dwell fractions, so the clustering would become a higher-dimensional diagram rather than a one-dimensional ordering.

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.

Desk Editor's Note

read the letter

Patil et al. report a visually striking and, as far as I can tell, new observation: in long delay embeddings, unstable periodic orbits of the Lorenz and Rössler attractors separate into clusters ordered by the ratio of time spent in the two lobes. The figures are convincing—A-heavy and B-heavy orbits land on opposite ends of the unfolded attractor, and the symmetric ones sit in the middle. The paper also ships data and code, and uses the high-precision Divakar Lorenz dataset, which is good practice.

What is genuinely new is the geometry: UPOs in a Hankel-SVD embedding, sorted by a single scalar ratio. The combinatorial section (modified Redfield-Pólya) is a useful minor extension, though it is a standard necklace-count with extra pruning constraints.

The soft spots are in the theory, not the observation. Theorem 1 is a straightforward ergodic-average limit: projecting a periodic orbit's Hankel matrix along (1_p⊗σ) converges to a constant depending on the lobe time fractions. But that direction is not the one you actually plot. The embedded coordinates are V_proj = Σ_c^{-1} U_c^T H^{(k)}, and nothing shows that the direction (1_p⊗σ) aligns with v1 or v3. The paper even says 'plane of separation v1=0' in one place and 'v3=0' in another. So the quantitative formula f(ρ) is not tested against the plotted axes; it is an unverified conjecture. There is also a dropped factor: the theorem gives α(ρ-1)/(ρ+1) when β≈−α, not f(ρ) alone.

The proof issues are concrete: Lemma 1 uses approximate columns to conclude exact range inclusion, and Theorem 2 uses the false equality ŨŨ^T=I (it's Ũ^T Ũ=I for economy SVD). The identification ρ=M/N is only within ~2% (Table 2) and that error is not propagated. These are fixable but not cosmetic.

In sum: the paper is worth engaging with, and a serious referee should see it. The central observation likely survives; the theory needs repair before the formula can be believed. I'd recommend major revision and a check that the theorem's direction actually matches the plotted SVD axis, or a clear statement that the plotted axis is empirically that direction.

Referee Report

4 major / 4 minor

Summary. The paper studies how unstable periodic orbits (UPOs) of the Lorenz and Rössler attractors separate into clusters when their time series are projected into the row space of a Hankel matrix built from a long chaotic trajectory. It proposes Lemma 1 (range inclusion of UPO Hankel matrices in the chaotic Hankel matrix), Theorem 1 (an ergodic limit showing a left projection equals α|IA|/t̃ + β|IB|/t̃), Theorem 2 (a representability result), and a modified Redfield-Pólya enumeration formula. The central empirical claim is that, as the Hankel height theight increases, UPOs form clusters whose distance from the separation plane is given by f(ρ) = (ρ−1)/(ρ+1), where ρ is the ratio of A-to-B symbol counts for Lorenz or the ratio of time spent in y>0 versus y<0 for Rössler.

Significance. If correct, the paper would establish a simple scalar descriptor that controls the geometry of UPOs in delay-embedded coordinates, with implications for periodic-orbit organizing centers and for Hankel/Koopman-based reconstructions. The paper has notable strengths: the data and code are made available, the cluster-separation phenomenon is visually clear and reproducible in Figures 6, 8, 10-12, and Theorem 1 is a clean and essentially correct ergodic-theory limit. However, the quantitative prediction f(ρ) is not actually tested against the plotted coordinates, and two of the three supporting theoretical results have incomplete proofs; these issues are load-bearing for the paper's headline claim.

major comments (4)

[§3.1, Theorem 2 proof] The proof uses the identity ŨŨᵀ = I for the economy SVD, but in an economy SVD with rank r < m, Ũ has orthonormal columns and only ŨᵀŨ = I_r holds; ŨŨᵀ is the orthogonal projection onto R(Ũ), not the identity. Consequently, the step ρᵀH̃ = ρᵀŨŨᵀH̃ is not justified, and the displayed derivation of ρᵀH̃ = ρ̂ᵀΣ⁻¹UᵀH̃ fails as written. The theorem may be provable directly from R(H̃) ⊂ R(H), but the supplied proof does not establish it.
[§3.1, Lemma 1 proof] The proof shows that, for every ε, a column of H^(k) is within ε of a column of H^c once a sufficiently long chaotic trajectory is available. This establishes approximate inclusion in the closure of the union of ranges, not exact inclusion R(H^(k)_{p,q}) ⊂ R(H^c_{p,q}); the limit of ε-close columns need not lie exactly in the range of the finite matrix. The subsequent monotone-dimension argument is also stated only at the level of dimensions rather than subspaces, although a finite-dimensional nested-subspace argument could be repaired. As written, the lemma is unproved.
[§4.1.3 vs §4.1.1 and Eqs. (8)-(9)] The quantitative claim that cluster distance is f(ρ) = (ρ−1)/(ρ+1) is not connected to a specific plotted coordinate. Theorem 1 computes a left projection along (1_p ⊗ σ), whereas the embedded coordinates are the row-space coordinates V_projᵀ = Σ_c⁻¹U_cᵀH^(k) defined in Eq. (9). Lemma 2 only shows that some linear functional on those coordinates can represent ρᵀH^(k); it does not show that this functional is v1 or v3 or a principal-component axis. The paper itself refers to the plane of separation v1 = 0 in §4.1.1 and v3 = 0 in §4.1.3. No figure plots measured cluster position against f(ρ), so the functional form is not actually tested against the data.
[§3.1, derivation after Table 2] Even if the projection direction were aligned with a plotted SVD axis, Theorem 1 combined with β ≈ −α gives α(ρ−1)/(ρ+1), not f(ρ) alone; for Lorenz the prefactor is α ≈ 6 (Table 1), and this factor is dropped in the stated distance formula. In addition, the identification ρ = M/N uses symbol-count ratios that differ from measured time ratios by up to 2.3% (Table 2), and this uncertainty is not propagated into any predicted distance. For Rössler, ρ is recomputed directly from the UPO time series that is then used for sorting and color-coding in Figure 12, so the monotone arrangement may be, at least in part, a restatement of the ordering variable rather than an independent quantitative test.

minor comments (4)

[§3.2, Eq. (17)] The sum over i|n in Eq. (17) must be over proper divisors of n; as written it includes i = n, which makes the recurrence circular. The table values match the aperiodic-necklace count with the mon-symbolic sequences removed, so this is likely a typographical issue, but it should be corrected explicitly.
[Abstract and §3.2] The name of the enumeration theorem is written inconsistently: the abstract uses 'Polyá-Redfield' while Section 3.2 uses 'Redfield-Polyá'; the latter is the standard name and should be used throughout.
[§4.3, Figure 12 caption] The caption lists only five τ values for the six panels of Figure 12; the sixth value should be supplied.
[Throughout] There are several typos that should be corrected: 'Singular singular vectors' in Section 3, 'Birkoff' for Birkhoff in the proof of Theorem 1, and 'tend=1000' should be defined with units. Also, Theorem 1 uses H_{pd,q} for a state-vector Hankel matrix while Eq. (4) uses H_{pd,q} for a block Hankel matrix; the two notations should be reconciled.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the cluster-separation formula is an empirical correlation between SVD-embedded coordinates and symbol/time ratios, not a quantity forced by construction; the theorem-to-axis gap and dropped alpha factor are correctness concerns, not circularity.

full rationale

The paper's derivation chain is not circular. The embedded coordinates are constructed independently of the ratio rho: V_proj^T = Sigma_c^{-1} U_c^T H^(k) (Eq. 9), and the cluster positions are obtained by plotting these projected coordinates. Theorem 1 computes a limit for a specific left projection (1_p ⊗ sigma)^T H^(k), and Theorem 2 shows only existence of some linear functional on the embedded coordinates that can represent such a row-space functional. The claimed distance formula f(rho) = (rho-1)/(rho+1) in Section 4.1.3 is not used to build V_proj; it is compared with the observed cluster locations after the embedding is computed. Thus the 'prediction' is not equivalent to its input by construction: rho is a scalar ratio derived from symbol counts or time spent in half-spaces, while the SVD coordinates of the embedded UPO are a different function of the same time series. For the Rossler case, rho is recomputed from the time series and used only to color-code and sort the already-computed embedded points; the reported monotonic arrangement is an empirical observation that could have failed. The modified Redfield-Polya enumeration (Eq. 17) is an independent combinatorial formula, not fitted to the embedding results. The paper's self-citations (e.g., [14], [20]) appear only as background and are not load-bearing for the central claim. There are genuine validity gaps that a reviewer should flag as correctness risks: the theorem's projection direction (1_p ⊗ sigma) is never identified with the plotted singular vector v1 or v3, and Section 4.1.1 refers to 'v1 = 0' while Section 4.1.3 refers to 'v3 = 0'; moreover, Theorem 1 gives alpha*(rho-1)/(rho+1) with alpha approximately 6 for Lorenz, and the factor alpha is dropped in the stated f(rho) formula. These are unsupported steps in the argument, but they are not circular reductions of the claimed result to its inputs.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim depends mainly on the ergodic-theorem limit of time averages (which is standard) plus two data-dependent approximations: the symbol-to-time ratio equivalence for Lorenz and the completeness of the symbolic dynamics. No free parameters are fitted to produce the cluster separation; α and β are measured system properties.

assumptions (4)

domain assumption The Lorenz and Rössler symbolic dynamics are complete at the chosen parameter values, so every valid symbolic sequence is realized as a UPO.
Stated in Section 3.2 and 5, needed for the modified Redfield-Pólya count to equal actual UPO counts; likely false for Lorenz at standard parameters, which are known to have pruned symbolic dynamics.
domain assumption For the Lorenz attractor, the fraction of time a UPO spends in a lobe is approximately equal to the fraction of symbols in its symbolic name (|I_A|/t̃ ≈ M/(M+N)).
Used to define ρ = M/N for cluster sorting in Section 3.1; Table 2 shows errors up to 2.3% for ABBBBBB, and the approximation is not quantitatively linked to cluster positions.
domain assumption The ratio Δt/t̃ is irrational for each UPO, making the discrete sampling ergodic on the periodic orbit.
Required for the Birkhoff ergodic theorem in Theorem 1; not checked for individual orbits in the datasets.
domain assumption The range of the UPO Hankel matrix is contained in the range of the chaotic Hankel matrix for sufficiently large width (Lemma 1), so the projection V_proj = Σ_c^{-1}U_c^T H^(k) faithfully represents the UPO.
Justifies the embedding procedure; the proof uses approximation and closure arguments and does not rigorously establish exact inclusion.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Separation of periodic orbits in the delay embedded space of chaotic attractors." pith.science (2026). https://pith.science/paper/EEIYXHUG

@misc{pith2026241113103,
  author       = {Pith},
  title        = {Pith review of: Separation of periodic orbits in the delay embedded space of chaotic attractors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EEIYXHUG}},
  note         = {Machine review of arXiv:2411.13103}
}

read the original abstract

This work explores the intersection of time-delay embeddings, periodic orbit theory, and symbolic dynamics. Time-delay embeddings have been effectively applied to chaotic time series data, offering a principled method to reconstruct relevant information of the full attractor from partial time series observations. In this study, we investigate the structure of the unstable periodic orbits of an attractor using time-delay embeddings. First, we embed time-series data from a periodic orbit into a higher-dimensional space through the construction of a Hankel matrix, formed by arranging time-shifted copies of the data. We then examine the influence of the width and height of the Hankel matrix on the geometry of unstable periodic orbits in the delay-embedded space. The right singular vectors of the Hankel matrix provide a basis for embedding the periodic orbits. We observe that increasing the length of the delay (e.g., the height of the Hankel matrix) leads to a clear separation of the periodic orbits into distinct clusters within the embedded space. Our analysis characterizes these separated clusters and provides a mathematical framework to determine the relative position of individual unstable periodic orbits in the embedded space. Additionally, we present a modified formula to derive the symbolic representation of distinct periodic orbits for a specified sequence length, extending the Poly\'a-Redfield enumeration theorem.

Figures

Figures reproduced from arXiv: 2411.13103 by the authors.

**Figure 2.** Division of the phase space into two regions and symbolic notation assigned to each [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Construction of Hankel matrices for a single unstable periodic orbit and chaotic trajectory. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (9 more)

**Figure 4.** Figure 4: Schematic of the transformation of a UPO onto a unit circle and every time-step of the [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]

**Figure 5.** Figure 5: (a) Hierarchical clustering of the UPOs where repeated UPOs are eliminated from the tree. [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]

**Figure 6.** Figure 6: Lorenz attractor: Unfolding of the attractor for the unstable periodic orbits of the type [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]

**Figure 7.** Figure 7: Comparison between the proximity of the UPOs in the phase space and the embedded [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]

**Figure 8.** Figure 8: Unfolding of the attractor for the symmetric unstable periodic orbits of the type [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]

**Figure 9.** Figure 9: Lorenz attractor: Clustering of the separated unstable periodic orbits. We observe that [PITH_FULL_IMAGE:figures/full_fig_p019_9.png]

**Figure 10.** Figure 10: Lorenz attractor: Unfolding of the attractor for the symbolic sequences length less than 8. [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]

**Figure 11.** Figure 11: Rössler attractor: Unfolding of the attractor for UPOs of type [PITH_FULL_IMAGE:figures/full_fig_p021_11.png]

**Figure 12.** Figure 12: Rössler attractor: Unfolding of the attractor for UPOs of sequence length less than 8. [PITH_FULL_IMAGE:figures/full_fig_p022_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages

[64]

An equation for continuous chaos

Otto E Rössler. An equation for continuous chaos. Physics Letters A, 57(5):397–398, 1976

work page 1976
[65]

An equation for hyperchaos

Otto E Rossler. An equation for hyperchaos. Physics Letters A, 71(2-3):155–157, 1979

work page 1979
[66]

Caractérisation topologique et reconstruction d’attracteurs étranges

Christophe Letellier. Caractérisation topologique et reconstruction d’attracteurs étranges . PhD thesis, Université de Paris VII Denis Diderot, 1994

work page 1994
[67]

Caractérisation des attracteurs étranges par la population d’orbites périodiques

Pascal Dutertre. Caractérisation des attracteurs étranges par la population d’orbites périodiques. PhD thesis, Rouen, 1995

work page 1995
[68]

Principal component trajectories for modeling spectrally continuous dynamics as forced linear systems

Daniel Dylewsky, Eurika Kaiser, Steven L Brunton, and J Nathan Kutz. Principal component trajectories for modeling spectrally continuous dynamics as forced linear systems. Physical Review E, 105(1):015312, 2022

work page 2022
[69]

Space-time pod and the hankel matrix.Plos one, 18(8):e0289637, 2023

Peter Frame and Aaron Towne. Space-time pod and the hankel matrix.Plos one, 18(8):e0289637, 2023

work page 2023
[70]

Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series

Robert Vautard and Michael Ghil. Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series. Physica D: Nonlinear Phenomena, 35(3):395–424, 1989

work page 1989
[71]

Ergodic theory, volume 245

Isaac P Cornfeld, Sergej V Fomin, and Yakov Grigorevich Sinai. Ergodic theory, volume 245. Springer Science & Business Media, 2012

work page 2012

Show all 17 references

[72]

Kombinatorische anzahlbestimmungen für gruppen, graphen und chemische verbindungen

George Pólya. Kombinatorische anzahlbestimmungen für gruppen, graphen und chemische verbindungen. Acta Mathematica, 1937

1937
[73]

A survey of generalizations of pólya’s enumeration theorem

Nicolaas G de Bruijn. A survey of generalizations of pólya’s enumeration theorem. Nieuw Archief voor Wiskunde, 3(2):89–112, 1971

1971
[74]

The combinatorial significance of a theorem of pólya

John Riordan. The combinatorial significance of a theorem of pólya. Journal of the Society for Industrial and Applied Mathematics, 5(4):225–237, 1957

1957
[75]

The theory of group-reduced distributions.American Journal of Mathematics, 49(3):433–455, 1927

J Howard Redfield. The theory of group-reduced distributions.American Journal of Mathematics, 49(3):433–455, 1927

1927
[76]

Theory of groups of finite order

William Burnside. Theory of groups of finite order. University, 1911

1911
[77]

Symmetry decomposition of chaotic dynamics

Predrag Cvitanovi´ c and Bruno Eckhardt. Symmetry decomposition of chaotic dynamics. Nonlinearity, 6(2):277, 1993

1993
[78]

Coarse graining the state space of a turbulent flow using periodic orbits

Gökhan Yalnız, Björn Hof, and Nazmi Burak Budanur. Coarse graining the state space of a turbulent flow using periodic orbits. Physical Review Letters, 126(24):244502, 2021

2021
[79]

Decomposing the dynamics of the lorenz 1963 model using unstable periodic orbits: Averages, transitions, and quasi- invariant sets

Chiara Cecilia Maiocchi, Valerio Lucarini, and Andrey Gritsun. Decomposing the dynamics of the lorenz 1963 model using unstable periodic orbits: Averages, transitions, and quasi- invariant sets. Chaos: An Interdisciplinary Journal of Nonlinear Science, 32(3), 2022

1963
[80]

Separation of periodic orbits (data and codes)

Prerna Patil and Eurika Kaiser. Separation of periodic orbits (data and codes). https: //github.com/prernampatil/UnstablePeriodicOrbits.git, 2024. 27

2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [64]

An equation for continuous chaos

Otto E Rössler. An equation for continuous chaos. Physics Letters A, 57(5):397–398, 1976

work page 1976

[2] [65]

An equation for hyperchaos

Otto E Rossler. An equation for hyperchaos. Physics Letters A, 71(2-3):155–157, 1979

work page 1979

[3] [66]

Caractérisation topologique et reconstruction d’attracteurs étranges

Christophe Letellier. Caractérisation topologique et reconstruction d’attracteurs étranges . PhD thesis, Université de Paris VII Denis Diderot, 1994

work page 1994

[4] [67]

Caractérisation des attracteurs étranges par la population d’orbites périodiques

Pascal Dutertre. Caractérisation des attracteurs étranges par la population d’orbites périodiques. PhD thesis, Rouen, 1995

work page 1995

[5] [68]

Principal component trajectories for modeling spectrally continuous dynamics as forced linear systems

Daniel Dylewsky, Eurika Kaiser, Steven L Brunton, and J Nathan Kutz. Principal component trajectories for modeling spectrally continuous dynamics as forced linear systems. Physical Review E, 105(1):015312, 2022

work page 2022

[6] [69]

Space-time pod and the hankel matrix.Plos one, 18(8):e0289637, 2023

Peter Frame and Aaron Towne. Space-time pod and the hankel matrix.Plos one, 18(8):e0289637, 2023

work page 2023

[7] [70]

Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series

Robert Vautard and Michael Ghil. Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series. Physica D: Nonlinear Phenomena, 35(3):395–424, 1989

work page 1989

[8] [71]

Ergodic theory, volume 245

Isaac P Cornfeld, Sergej V Fomin, and Yakov Grigorevich Sinai. Ergodic theory, volume 245. Springer Science & Business Media, 2012

work page 2012