pith. sign in

arxiv: 2507.11714 · v2 · pith:IAQ6Z5CXnew · submitted 2025-07-15 · 🧮 math.DS

Reduced Order Modeling of Nonlinear Dynamical Systems Using Slow Manifolds

Dan Wilson This is my paper

Pith reviewed 2026-05-19 03:56 UTC · model grok-4.3

classification 🧮 math.DS
keywords slow manifoldsreduced order modelingisostable coordinatesnonlinear dynamical systemsHodgkin-Huxley neuronscircadian oscillationsmodel reductiontimescale separation
0
0 comments X

The pith

Slow manifolds formed by intersecting unstable and stable manifolds enable reduced-order modeling of nonlinear systems when transverse decay rates are large.

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

This paper shows that a low-dimensional slow manifold can be identified as the intersection of the unstable manifold of an unstable fixed point or periodic orbit and the stable manifold of a stable attractor. When perturbations decay rapidly away from this manifold, the dynamics restricted to it can be analyzed separately using isostable coordinates to produce a reduced-order model of the full system. The method is applied to a coupled Hodgkin-Huxley neuron network and a detailed circadian rhythm model, where the reduced descriptions support specific control tasks. A reader would care because many biological and physical systems display clear fast-slow separations that could make high-dimensional simulation and intervention far simpler if the transverse decay condition holds.

Core claim

The paper claims that when the decay rates of perturbations transverse to the unstable manifold are sufficiently large, the slow manifold defined by the intersection of an unstable manifold of an unstable fixed point or periodic orbit and the stable manifold of a stable attractor can be used for reduced modeling purposes by leveraging the isostable coordinate framework. Detailed examples are provided for two different highly nonlinear dynamical systems, the first being a coupled system of Hodgkin-Huxley neurons and the second being a biophysically detailed model of circadian oscillations. The resulting reduced order models are illustrated in two different biologically motivated control tasks

What carries the argument

The slow manifold obtained from the intersection of an unstable manifold and a stable manifold, together with isostable coordinates that track the decay of transverse perturbations.

If this is right

  • The reduced models accurately capture long-term behavior for control purposes in the example systems.
  • The same construction applies whether the unstable object is a fixed point or a periodic orbit.
  • Biological control objectives become feasible in lower-dimensional descriptions of neuron networks and circadian clocks.
  • Highly nonlinear models with clear timescale separation become more tractable without losing essential features.

Where Pith is reading between the lines

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

  • The approach might still work under moderate parameter changes that keep transverse decays fast but would require checking the boundary where the assumption fails.
  • It could be combined with other reduction techniques to handle even larger networks of oscillators.
  • Numerical tests on systems with multiple coexisting attractors would clarify how many such slow manifolds can be extracted simultaneously.

Load-bearing premise

The intersection of the unstable manifold and the stable manifold forms a slow manifold on which transverse decay rates are large enough to support reduced-order modeling.

What would settle it

A side-by-side simulation of the full high-dimensional system and the reduced model on the slow manifold that shows large divergence precisely when the transverse decay rates are moderate rather than large.

Figures

Figures reproduced from arXiv: 2507.11714 by Dan Wilson.

Figure 1
Figure 1. Figure 1: Panel A shows a toy system (4) with an unstable periodic orbit (dashed line), stable fixed [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Panel A of the toy system (4) with parameters chosen so that there is an unstable fixed [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: For the parameters used here, the model (63) is close to a subcritical Hopf bifurcation. [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Level sets of |ψ1| on the slow manifold of x0 are shown as colored lines in panels A-D. An initial condition in the basin of attraction of x0 quickly converges to the slow manifold (black line). For the same trajectory, panel E shows voltage traces for each neuron for this trajectory. The slow manifold of the fixed point x0 can be used for model reduction by considering the evolution of the 16-dimensional … view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of a control strategy for preventing action potentials in the model (63) as [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: A slow manifold of (B1) is given by the intersection of the unstable manifold of the [PITH_FULL_IMAGE:figures/full_fig_p018_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Applying the control strategy (69) to the circadian model (B1). In panel A, the con [PITH_FULL_IMAGE:figures/full_fig_p019_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of two alternative control strategies for driving the state from the stable [PITH_FULL_IMAGE:figures/full_fig_p021_8.png] view at source ↗
read the original abstract

Model order reduction in high-dimensional, nonlinear dynamical systems if often enabled through fast-slow timescale separation. One such approach involves identifying a low-dimensional slow manifold to which the state rapidly converges and subsequently studying the behavior on the slow manifold. This work investigates slow manifolds defined by the intersection of an unstable manifold of an unstable fixed point or periodic orbit and the stable manifold of a stable attractor. When the decay rates of perturbations transverse to the unstable manifold are sufficiently large, the resulting slow manifold can be used for reduced modeling purposes by leveraging the isostable coordinate framework. Detailed examples are provided for two different highly nonlinear dynamical systems, the first being a coupled system of Hodgkin-Huxley neurons and the second being a biophysically detailed model of circadian oscillations. The resulting reduced order models are illustrated in two different biologically motivated control objectives.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes defining slow manifolds in nonlinear dynamical systems as the intersection of the unstable manifold of an unstable fixed point or periodic orbit with the stable manifold of a stable attractor. It claims that when transverse decay rates to the unstable manifold are sufficiently large, the isostable coordinate framework can be applied to obtain reduced-order models, with demonstrations on a coupled Hodgkin-Huxley neuron system and a biophysically detailed circadian oscillator model for two biologically motivated control objectives.

Significance. If the transverse decay condition is verified, the approach extends isostable-based reduction techniques to systems featuring saddle structures, offering a systematic route to low-dimensional models in high-dimensional biological dynamics. The use of concrete, biophysically detailed examples strengthens potential applicability, though the absence of quantitative checks on the key decay-rate assumption limits immediate impact.

major comments (2)
  1. [§4 (Hodgkin-Huxley example)] The central claim that the intersection yields a usable slow manifold for isostable reduction rests on the unverified assumption that transverse decay rates are sufficiently large relative to motion along the manifold. No explicit computation of transverse Lyapunov exponents, decay times, or rate ratios is reported in the Hodgkin-Huxley example to confirm the condition holds.
  2. [§5 (circadian oscillations and control)] In the circadian model, the reduced-order controllers are presented without accompanying error metrics (e.g., L2 trajectory errors or phase-response curve comparisons) between the full-order system and the reduced model on the proposed slow manifold, leaving the practical accuracy of the reduction unquantified.
minor comments (2)
  1. [Introduction] Notation for isostable coordinates and manifold intersections should be introduced with a brief reminder of the cited framework to improve readability for readers unfamiliar with the prior literature.
  2. [Figures in §§4–5] Figure captions for the reduced-model trajectories should include the specific parameter values used in the control objectives to allow direct reproduction.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the presentation of our results. We respond to each major comment below and indicate the revisions we will make to the manuscript.

read point-by-point responses

  1. Referee: [§4 (Hodgkin-Huxley example)] The central claim that the intersection yields a usable slow manifold for isostable reduction rests on the unverified assumption that transverse decay rates are sufficiently large relative to motion along the manifold. No explicit computation of transverse Lyapunov exponents, decay times, or rate ratios is reported in the Hodgkin-Huxley example to confirm the condition holds.

    Authors: We agree that the manuscript would benefit from explicit verification of the transverse decay condition in the Hodgkin-Huxley example. While the theoretical development in the paper states the requirement that transverse decay rates be large relative to motion along the manifold, we did not include numerical checks of transverse Lyapunov exponents or rate ratios for this specific system. In the revised manuscript we will add these computations in §4, reporting the relevant exponents and confirming that the separation of timescales holds for the chosen parameters. revision: yes

  2. Referee: [§5 (circadian oscillations and control)] In the circadian model, the reduced-order controllers are presented without accompanying error metrics (e.g., L2 trajectory errors or phase-response curve comparisons) between the full-order system and the reduced model on the proposed slow manifold, leaving the practical accuracy of the reduction unquantified.

    Authors: We concur that quantitative error metrics would strengthen the demonstration of the reduced-order model's accuracy. The current manuscript illustrates the controllers on the circadian oscillator but does not report explicit L2 trajectory errors or phase-response curve comparisons. In the revision we will add these metrics in §5 for both control tasks, comparing the full-order trajectories and phase responses against those obtained from the reduced model on the slow manifold. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected; derivation is self-contained

full rationale

The paper defines the slow manifold explicitly as the intersection of the unstable manifold of an unstable fixed point or periodic orbit with the stable manifold of a stable attractor. It then states that when transverse decay rates to the unstable manifold are sufficiently large, this intersection can be leveraged for reduced-order modeling via the isostable coordinate framework, which is presented as a pre-existing method rather than derived or fitted from the current results. The Hodgkin-Huxley and circadian examples illustrate application but do not involve redefining fitted quantities as predictions or reducing the central claim to a self-citation chain. No self-definitional loops, ansatzes smuggled via citation, or uniqueness theorems imported from the same authors appear in the provided abstract and description; the load-bearing steps rest on external mathematical definitions and stated assumptions that remain independently verifiable.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract provides no explicit free parameters, axioms, or invented entities; the central claim rests on the geometric definition of the slow manifold and the assumption that transverse decay rates are large enough, both of which are stated without derivation or external verification in the given text.

pith-pipeline@v0.9.0 · 5659 in / 1330 out tokens · 26390 ms · 2026-05-19T03:56:26.548598+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    Ahmed, A

    T. Ahmed, A. Sadovnik, and D. Wilson. Data-driven inference of low-order isostable- coordinate-based dynamical models using neural networks. Nonlinear Dynamics, 111(3):2501– 2519, 2023

    work page 2023

  2. [2]

    S. L. Brunton, B. W. Brunton, J. L. Proctor, and J. N. Kutz. Koopman invariant subspaces and finite linear representations of nonlinear dynamical systems for control. PloS One, 11(2), 2016

    work page 2016

  3. [3]

    Cenedese, J

    M. Cenedese, J. Ax˚ as, B. B¨ auerlein, K. Avila, and G. Haller. Data-driven modeling and prediction of non-linearizable dynamics via spectral submanifolds. Nature Communications, 13(1):872, 2022

    work page 2022

  4. [4]

    Farjami, V

    S. Farjami, V. Kirk, and H. M. Osinga. Computing the stable manifold of a saddle slow manifold. SIAM Journal on Applied Dynamical Systems , 17(1):350–379, 2018

    work page 2018

  5. [5]

    Fenichel

    N. Fenichel. Geometric singular perturbation theory for ordinary differential equations.Journal of differential equations, 31(1):53–98, 1979

    work page 1979

  6. [6]

    Foias, M

    C. Foias, M. S. Jolly, I. G. Kevrekidis, G. R. Sell, and E. S. Titi. On the computation of inertial manifolds. Physics Letters A , 131(7):433–436, 1988

    work page 1988

  7. [7]

    Foias, G

    C. Foias, G. R. Sell, and R. Temam. Inertial manifolds for nonlinear evolutionary equations. Journal of Differential Equations , 73(2):309–353, 1988

    work page 1988

  8. [8]

    Guckenheimer

    J. Guckenheimer. Isochrons and phaseless sets. Journal of Mathematical Biology, 1(3):259–273, 1975

    work page 1975

  9. [9]

    Guckenheimer and C

    J. Guckenheimer and C. Kuehn. Computing slow manifolds of saddle type. SIAM Journal on Applied Dynamical Systems, 8(3):854–879, 2009

    work page 2009

  10. [10]

    Haller and S

    G. Haller and S. Ponsioen. Nonlinear normal modes and spectral submanifolds: existence, uniqueness and use in model reduction. Nonlinear Dynamics, 86:1493–1534, 2016

    work page 2016

  11. [11]

    A. L. Hodgkin and A. F. Huxley. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol., 117:500–44, 1952

    work page 1952

  12. [12]

    Johnston and S

    D. Johnston and S. M.-S. Wu. Foundations of Cellular Neurophysiology . MIT Press, Cam- bridge, MA, 1995

    work page 1995

  13. [13]

    Jordan and P

    D. Jordan and P. Smith. Nonlinear Ordinary Differential Equations: An Introduction for Scientists and Engineers , volume 10. Oxford University Press, Oxford, 2007

    work page 2007

  14. [14]

    Kaiser, J

    E. Kaiser, J. N. Kutz, and S. Brunton. Data-driven discovery of Koopman eigenfunctions for control. Machine Learning: Science and Technology , 2021. 25

    work page 2021

  15. [15]

    T. J. Kaper. Systems theory for singular perturbation problems. In Analyzing multiscale phenomena using singular perturbation methods , pages 85–131. AMS, 1999

    work page 1999

  16. [16]

    J. N. Kutz, S. L. Brunton, B. W. Brunton, and J. L. Proctor. Dynamic mode decomposition: data-driven modeling of complex systems . Society for Industrial and Applied Mathematics, Philadelphia, PA, 2016

    work page 2016

  17. [17]

    M. D. Kvalheim and S. Revzen. Existence and uniqueness of global Koopman eigenfunctions for stable fixed points and periodic orbits. Physica D: Nonlinear Phenomena , page 132959, 2021

    work page 2021

  18. [18]

    J. C. Leloup and A. Goldbeter. Toward a detailed computational model for the mammalian circadian clock. Proceedings of the National Academy of Sciences , 100(12):7051–7056, 2003

    work page 2003

  19. [19]

    Lusch, J

    B. Lusch, J. N. Kutz, and S. L. Brunton. Deep learning for universal linear embeddings of nonlinear dynamics. Nature Communications, 9(1):1–10, 2018

    work page 2018

  20. [20]

    Mauroy and I

    A. Mauroy and I. Mezi´ c. Global stability analysis using the eigenfunctions of the Koopman operator. IEEE Transactions on Automatic Control , 61(11):3356–3369, 2016

    work page 2016

  21. [21]

    Mauroy, I

    A. Mauroy, I. Mezi´ c, and J. Moehlis. Isostables, isochrons, and Koopman spectrum for the action–angle representation of stable fixed point dynamics. Physica D: Nonlinear Phenomena, 261:19–30, 2013

    work page 2013

  22. [22]

    I. Mezi´ c. Analysis of fluid flows via spectral properties of the Koopman operator. Annual Review of Fluid Mechanics , 45:357–378, 2013

    work page 2013

  23. [23]

    I. Mezi´ c. Spectrum of the Koopman operator, spectral expansions in functional spaces, and state-space geometry. Journal of Nonlinear Science , pages 1–55, 2019

    work page 2019

  24. [24]

    I. Mezi´ c. Spectrum of the Koopman operator, spectral expansions in functional spaces, and state-space geometry. Journal of Nonlinear Science , 30(5):2091–2145, 2020

    work page 2091

  25. [25]

    Monga and J

    B. Monga and J. Moehlis. Optimal phase control of biological oscillators using augmented phase reduction. Biological Cybernetics, 113(1-2):161–178, 2019

    work page 2019

  26. [26]

    H. M. Osinga and J. Moehlis. Continuation-based computation of global isochrons. SIAM Journal on Applied Dynamical Systems , 9(4):1201–1228, 2010

    work page 2010

  27. [27]

    Park and D

    Y. Park and D. Wilson. -body oscillator interactions of higher-order coupling functions. SIAM Journal on Applied Dynamical Systems , 23(2):1471–1503, 2024

    work page 2024

  28. [28]

    Ponsioen, S

    S. Ponsioen, S. Jain, and G. Haller. Model reduction to spectral submanifolds and forced- response calculation in high-dimensional mechanical systems. Journal of Sound and Vibration , 488:115640, 2020

    work page 2020

  29. [29]

    P. J. Schmid. Dynamic mode decomposition of numerical and experimental data. Journal of Fluid Mechanics, 656:5–28, 2010

    work page 2010

  30. [30]

    Sootla and A

    A. Sootla and A. Mauroy. Geometric properties of isostables and basins of attraction of monotone systems. IEEE Transactions on Automatic Control , 62(12):6183–6194, 2017. 26

    work page 2017

  31. [31]

    Takata, Y

    S. Takata, Y. Kato, and H. Nakao. Fast optimal entrainment of limit-cycle oscillators by strong periodic inputs via phase-amplitude reduction and Floquet theory.Chaos: An Interdisciplinary Journal of Nonlinear Science , 31(9), 2021

    work page 2021

  32. [32]

    S. Wiggins. Introduction to applied nonlinear dynamical systems and chaos, volume 2. Springer, 2003

    work page 2003

  33. [33]

    M. O. Williams, I. G. Kevrekidis, and C. W. Rowley. A data–driven approximation of the koopman operator: Extending dynamic mode decomposition. Journal of Nonlinear Science , 25(6):1307–1346, 2015

    work page 2015

  34. [34]

    D. Wilson. A data-driven phase and isostable reduced modeling framework for oscillatory dynamical systems. Chaos: An Interdisciplinary Journal of Nonlinear Science , 30(1):013121, 2020

    work page 2020

  35. [35]

    D. Wilson. Phase-amplitude reduction far beyond the weakly perturbed paradigm. Physical Review E, 101(2):022220, 2020

    work page 2020

  36. [36]

    D. Wilson. Identification and computation of slow manifolds using the isostable coordinate system. arXiv preprint:PENDING, 202025

    work page

  37. [37]

    D. Wilson. Analysis of input-induced oscillations using the isostable coordinate framework. Chaos: An Interdisciplinary Journal of Nonlinear Science , 31(2):023131, 2021

    work page 2021

  38. [38]

    D. Wilson. Data-driven inference of high-accuracy isostable-based dynamical models in response to external inputs. Chaos: An Interdisciplinary Journal of Nonlinear Science , 31(6):063137, 2021

    work page 2021

  39. [39]

    D. Wilson. An adaptive phase-amplitude reduction framework without O(ϵ) constraints on inputs. SIAM Journal on Applied Dynamical Systems , 21(1):204–230, 2022

    work page 2022

  40. [40]

    Wilson and B

    D. Wilson and B. Ermentrout. Greater accuracy and broadened applicability of phase reduction using isostable coordinates. Journal of Mathematical Biology , 76(1-2):37–66, 2018

    work page 2018

  41. [41]

    Wilson and J

    D. Wilson and J. Moehlis. Isostable reduction of periodic orbits. Physical Review E , 94(5):052213, 2016

    work page 2016

  42. [42]

    A. Winfree. The Geometry of Biological Time . Springer Verlag, New York, second edition, 2001. 27

    work page 2001