arxiv: 2605.06181 · v1 · submitted 2026-05-07 · 📡 eess.SY · cs.SY

Recognition: unknown

Synthesis of Limit Cycles and Reference Tracking via Switching Affine Systems

Nils Hanke , Zonglin Liu , Olaf Stursberg

Authors on Pith no claims yet

Pith reviewed 2026-05-08 06:43 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords limit cycle approximationswitching affine systemsnumerical optimizationglobal stabilityreference trackingLyapunov functionsnonlinear ODEs

0 comments

The pith

Switching affine systems with general partitions and external signals approximate globally stable limit cycles of nonlinear ODEs via constrained optimization from sampled data.

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

This paper develops a synthesis method to approximate limit cycles of nonlinear ODEs using switching affine dynamics. It extends beyond simple two-region or planar cases by allowing general partitions in higher-dimensional spaces plus external signals, then solves for a globally stable limit cycle through constrained numerical optimization that fits sampled data while enforcing Lyapunov stability. Once the periodic model is obtained, reference tracking is handled with multiple Lyapunov functions to reduce conservatism in convergence. A reader would care because the approach turns hard nonlinear periodic problems into optimizable piecewise affine ones that are easier to analyze and control numerically.

Core claim

Starting from sampled data of a nonlinear ODE, the synthesis minimizes approximation error to a limit cycle generated by a switching affine model while applying stability constraints to guarantee global stability; the optimization is solved over general state-space partitions augmented by external signals. For the resulting periodic switching system, reference tracking uses multiple Lyapunov functions to obtain less conservative convergence than a common Lyapunov function would allow.

What carries the argument

Constrained numerical optimization that trades off data-fitting error against Lyapunov stability constraints on a switching affine system, where the state space is divided into regions each following its own affine dynamics, possibly modulated by external signals.

If this is right

A feasible optimization solution produces a switching affine model whose trajectories converge globally to a limit cycle that approximates the sampled nonlinear behavior.
The common Lyapunov function enforces the global stability guarantee for the synthesized limit cycle.
Reference tracking for the periodic model converges with guarantees that are less conservative than single-Lyapunov results.
The method applies to higher-dimensional state spaces and more flexible partitions than prior two-region or planar approaches.

Where Pith is reading between the lines

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

The technique could support data-driven controller design for engineered periodic systems like mechanical oscillators or switched power converters without needing closed-form nonlinear analysis.
Coupling the partition choice with data-driven learning might scale the method to state spaces too large for manual design.
The same switching-affine template might serve as an intermediate model for stability certification of hybrid or sampled-data controllers.

Load-bearing premise

The nonlinear ODE can be sufficiently well approximated by a switching affine system under the chosen state-space partition and external signals, and the numerical optimizer will find a feasible solution satisfying the stability constraints.

What would settle it

For a nonlinear oscillator with a known stable limit cycle, such as the van der Pol system, running the synthesis and finding that the resulting switching model produces neither a matching periodic orbit nor global stability in simulation would show the claim does not hold.

Figures

Figures reproduced from arXiv: 2605.06181 by Nils Hanke, Olaf Stursberg, Zonglin Liu.

**Figure 1.** Figure 1: [12] Based on the set of samples F, the center point xs, and a set of selected points xˆ1, . . . , xˆnP ∈ F (each representing the first state of any subset F1, . . . , FnP along the limit cycle), the lines for partitioning the state space into regions Pi are determined. Case nx = 2: Let a center point xs of all points in F be determined by: xs,[q] = 1 2 max l∈{1,...,nF } x˜l,[q] − min l∈{1,...,nF } x˜l,… view at source ↗

**Figure 2.** Figure 2: [12] In the case nx = 3, the partition shown in view at source ↗

**Figure 3.** Figure 3: [12] Certain configurations of F are incompatible with the partitioning procedure. In example a), the condition is violated that Cix = di should contain only a sample point xˆi and Γ. In example b), the identification of A3, b3 ∈ P3 and A4, b4 ∈ P4 is not feasible. cycle that intersects itself. Although the proposed partitioning procedure may succeed in this scenario, it becomes clearly evident (e.g., in r… view at source ↗

**Figure 4.** Figure 4: [12] Sample points (black circles), switching bound view at source ↗

**Figure 5.** Figure 5: [12] Sample points (circles), switching boundaries view at source ↗

**Figure 6.** Figure 6: [12] Convergence of x1(t) and x2(t) for different initial points x(0) = view at source ↗

**Figure 7.** Figure 7: [12] Sample points (circles), green plane with minim view at source ↗

**Figure 8.** Figure 8: [12] Convergence of x1(t), x2(t), x3(t) for different initial points x(0) = view at source ↗

**Figure 9.** Figure 9: The state x˘(t) denotes the intersection point on the boundary Cix = di of the line segment connecting xr(t) and xc(t). Theorem 1. Given a system of type (3), a control law (24) and a periodic reference trajectory xr(t) governed by (4). Let the assumptions 1, 2 and 3, and the continuity condition (26) on the switching boundaries hold. If then positive scalars ρ > 0 ∈ R and σ > 0 ∈ R as well as a set of pos… view at source ↗

**Figure 10.** Figure 10: Evolution of the reference limit cycle (dashed gree view at source ↗

**Figure 11.** Figure 11: The evo view at source ↗

**Figure 11.** Figure 11: Evolution of the tracking error of the reference and view at source ↗

**Figure 12.** Figure 12: Evolution of xr,1(t) (green) and xc,1(t) (black). ence despite large deviations of the initialization, although the distance between xc(t) and xr(t) increases temporarily in the transient phase. 5 Conclusion This paper has introduced a method for approximating periodic behavior in nonlinear dynamical systems, extending beyond the planar case. By sampling the limit cycle of the nonlinear dynamics, switched… view at source ↗

**Figure 13.** Figure 13: Evolution of xr,2(t) (green) and xc,2(t) (black). conditions lead to a non-convex, nonlinear optimization problem, they simultaneously enforce smoothness of the limit cycle – a characteristic typically observed in real-world oscillating systems. While obtaining the global solution to the optimization problem cannot be assured a-priori, any feasible solution yields a convergent trajectory. The use of a c… view at source ↗

read the original abstract

This paper introduces a novel method to approximate limit cycles of nonlinear ODEs by use of switching affine dynamics in order to ease data-based modeling and analysis. Previous approaches to approximating limit cycles by switching systems have been largely confined to simple partitions into two-regions or low-dimensional (often planar) settings. In contrast, this study utilizes more general partitions in higher-dimensional state spaces, augmented by external signals, to develop a synthesis scheme that guarantees a globally stable limit cycle. The synthesis task is formulated and solved based on constrained numerical optimization. Starting from sampled data of the nonlinear dynamics, the method minimizes the error between the data and the limit cycle generated by the switching affine model, while employing stability constraints to ensure global stability. Based on the obtained model, the paper tackles the problem of reference tracking for switching affine systems with periodic behavior. While the approximation scheme is based on a common Lyapunov function, the reference tracking approach uses multiple Lyapunov functions to achieve less conservative convergence results. The principle and effectiveness of the proposed methods are illustrated through a set of examples.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note private letter to a colleague

This paper gives a direct optimization method to fit switching affine models to nonlinear data so the result has a globally stable limit cycle in higher dimensions, then relaxes to multiple Lyapunov functions for tracking.

read the letter

The core idea is to take sampled trajectories from a nonlinear ODE, partition the state space into several regions, introduce external signals, and solve a constrained optimization that minimizes the mismatch to a periodic orbit while enforcing a common Lyapunov function for global stability of the switched affine system. They then switch to multiple Lyapunov functions for the reference-tracking version to get less conservative convergence. This extends earlier switched-system approximations that were mostly limited to two regions or planar dynamics, so the generalization to arbitrary partitions and higher dimensions plus the tracking variant is the actual increment. The construction is internally consistent: the optimizer targets both fitting error and the Lyapunov conditions drawn from standard switched-systems results, with no circularity or unstated data requirements that jump out from the abstract. The examples are said to illustrate the principle, which is the right place to start for this kind of work. The main soft spot is that global stability still rests on the common Lyapunov function during synthesis, which is known to be conservative, and the quality of the approximation depends heavily on how well the chosen partition and external signals capture the original nonlinear flow. If the optimizer returns an infeasible problem or the cycle deviates noticeably, the method does not automatically diagnose why. Scalability with dimension or number of regions is not quantified in the summary, so it is unclear how far the approach reaches in practice. This is aimed at control engineers who already work with switched or hybrid models and need a data-driven route to periodic behavior, such as in mechanical or biological oscillators. A reader looking for a concrete optimization template rather than a fundamental new theory will find usable pieces. It deserves peer review because the claims are checkable against the examples and the extension is modest but well-motivated; the referees can focus on whether the numerical results hold up and how sensitive the partition choice is.

Referee Report

0 major / 3 minor

Summary. The paper proposes a data-driven synthesis method for globally stable limit cycles in nonlinear ODEs by approximating them with switching affine systems. It employs general state-space partitions augmented by external signals and formulates the task as a constrained numerical optimization that minimizes fitting error to sampled data while enforcing stability via a common Lyapunov function. The work then extends the model to reference tracking, replacing the common Lyapunov function with multiple Lyapunov functions for less conservative convergence guarantees, and demonstrates the approach on numerical examples.

Significance. If the central claims hold, the contribution lies in extending switched-system limit-cycle synthesis beyond low-dimensional or two-region partitions to higher-dimensional settings with external signals, while providing an optimization-based procedure that directly incorporates stability constraints. The shift to multiple Lyapunov functions for the tracking task is a standard relaxation that could improve practical applicability. The data-based nature and explicit handling of periodic behavior may aid control design for systems with limit cycles.

minor comments (3)

The abstract and introduction claim that the optimization 'guarantees a globally stable limit cycle,' but the dependence of feasibility on the choice of partition and external signals is not analyzed; a brief discussion of when the constrained problem is expected to admit solutions would strengthen the presentation.
In the reference-tracking section, the benefit of multiple Lyapunov functions over the common Lyapunov function used for synthesis is asserted but not quantified (e.g., via convergence-rate comparisons or conservatism metrics on the same example); adding such a comparison would clarify the practical gain.
Notation for the switching signal, the role of external signals, and the precise definition of the limit-cycle error metric should be introduced earlier and used consistently to improve readability for readers unfamiliar with switched affine systems.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary and significance assessment of our work on synthesis of limit cycles via switching affine systems. The recommendation for minor revision is noted, and we will incorporate any editorial or minor improvements in the revised version. No specific major comments were listed in the report.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper formulates limit-cycle synthesis as a constrained numerical optimization that fits switching affine dynamics to sampled data from the nonlinear ODE while enforcing global stability via a common Lyapunov function drawn from standard switched-systems theory. This does not reduce to a tautology or self-referential fit: the model is constructed to approximate external data and the stability certificate is an independent constraint rather than presupposed in the dynamics definition. The reference-tracking extension uses multiple Lyapunov functions as a recognized relaxation without circular reduction. No load-bearing self-citations, uniqueness theorems, or ansatzes imported from prior author work appear in the central derivation chain.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach rests on standard Lyapunov stability theory for switched systems and the assumption that a suitable partition and external signals exist to make the approximation feasible; no new entities are postulated.

axioms (2)

domain assumption A common Lyapunov function exists that certifies global stability of the synthesized limit cycle.
Invoked for the approximation scheme to guarantee global stability independent of initial conditions.
domain assumption Multiple Lyapunov functions can be constructed to obtain less conservative convergence for reference tracking.
Used to relax conservatism compared with a single common Lyapunov function.

pith-pipeline@v0.9.0 · 5481 in / 1347 out tokens · 65003 ms · 2026-05-08T06:43:16.652448+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

34 extracted references · 1 canonical work pages

[1]

International Journal of Control 31(6), 1041–1105 (1980)

Atherton, D., Dorrah, H.: A survey on non-linear oscillat ions. International Journal of Control 31(6), 1041–1105 (1980)

1980
[2]

Bernardo, M., Budd, C., Champneys, A.R., Kowalczyk, P.: P iecewise-smooth dy- namical systems: theory and applications, vol. 163. Spring er Science & Business Media (2008)

2008
[3]

Journal of mathematical analysis and appli cations 253(2), 671– 690 (2001)

Coll, B., Gasull, A., Prohens, R.: Degenerate hopf bifurc ations in discontinuous planar systems. Journal of mathematical analysis and appli cations 253(2), 671– 690 (2001)

2001
[4]

Demidovich, B.P.: Lectures on stability theory (1967)

1967
[5]

Automatica pp

Dörﬂer, F., Bullo, F.: Synchronization in complex networ ks of phase oscillators: A survey. Automatica pp. 1539–1564 (2014)

2014
[6]

International Journal o f Bifurcation and Chaos 8(11), 2073–2097 (1998)

Freire, E., Ponce, E., Rodrigo, F., Torres, F.: Bifurcati on sets of continuous piece- wise linear systems with two zones. International Journal o f Bifurcation and Chaos 8(11), 2073–2097 (1998)

2073
[7]

Nonlinear Analysis: Theory, Methods & Applications 74(18), 7532–7542 (2011)

Gaiko, V.A.: Multiple limit cycle bifurcations of the ﬁtz hugh–nagumo neuronal model. Nonlinear Analysis: Theory, Methods & Applications 74(18), 7532–7542 (2011)

2011
[8]

Acta Biotheoretica 69(4), 857–874 (2021)

Gonze, D., Ruoﬀ, P.: The goodwin oscillator and its legacy . Acta Biotheoretica 69(4), 857–874 (2021)

2021
[9]

Automatica 40(1), 21–35 (2004)

Habets, L., Van Schuppen, J.H.: A control problem for aﬃne dynamical systems on a full-dimensional polytope. Automatica 40(1), 21–35 (2004)

2004
[10]

In: European Control Conference

Hanke, N., Liu, Z., Stursberg, O.: Approximation of limi t cycles by using planar switching aﬃne systems with guarantees for uniqueness and s tability. In: European Control Conference. pp. 1460–1465 (2024)

2024
[11]

In: American Control Conference

Hanke, N., Liu, Z., Stursberg, O.: Approximation of plan ar periodic behavior from data with stability guarantees using switching aﬃne system s. In: American Control Conference. pp. 1944–1949 (2025)

1944
[12]

In: Proceedings of the 22nd International Conference on Informatics in Cont rol, Automation and Robotics - Volume 1: ICINCO

Hanke, N., Liu, Z., Stursberg, O.: On the synthesis of sta ble switching dy- namics to approximate limit cycles of nonlinear oscillator s. In: Proceedings of the 22nd International Conference on Informatics in Cont rol, Automation and Robotics - Volume 1: ICINCO. pp. 509–516. INSTICC, SciTe Press (2025). https://doi.org/10.5220/0013835000003982

work page doi:10.5220/0013835000003982 2025
[13]

In: European Control Conference

Hanke, N., Stursberg, O.: On the design of limit cycles of planar switching aﬃne systems. In: European Control Conference. pp. 2251–2256 (2 023)
[14]

SIAM review 51(4), 747–764 (2009)

Higham, N.J.: The scaling and squaring method for the mat rix exponential revis- ited. SIAM review 51(4), 747–764 (2009)

2009
[15]

In: European Control Conf

Johansson, M., Rantzer, A.: Computation of piecewise qu adratic lyapunov func- tions for hybrid systems. In: European Control Conf. pp. 200 5–2010. IEEE (1997)

2010
[16]

Johansson, M.K.J.: Piecewise linear control systems: a computational approach, vol. 284. Springer (2003)

2003
[17]

IF AC-PapersOnLine49(1), 320–325 (2016)

Joshi, S.K., Sen, S., Kar, I.N.: Synchronization of coup led oscillator dynamics. IF AC-PapersOnLine49(1), 320–325 (2016)

2016
[18]

Mathematical and Computer Modelli ng 55(3-4), 505–516 (2012) 26 N

Kai, T., Masuda, R.: Limit cycle synthesis of multi-moda l and 2-dimensional piecewise aﬃne systems. Mathematical and Computer Modelli ng 55(3-4), 505–516 (2012) 26 N. Hanke et al

2012
[19]

SIAM journal on matrix analysis and applications 19(3), 640–663 (1998)

Kenney, C.S., Laub, A.J.: A schur–fréchet algorithm for computing the logarithm and exponential of a matrix. SIAM journal on matrix analysis and applications 19(3), 640–663 (1998)

1998
[20]

Co mmunications in Nonlin- ear Science and Numerical Simulation 93, 105526 (2021)

Kudryashov, N.A.: The generalized duﬃng oscillator. Co mmunications in Nonlin- ear Science and Numerical Simulation 93, 105526 (2021)

2021
[21]

In: International symposium on mathematical problems in th eoretical physics

Kuramoto, Y.: Self-entrainment of a population of coupl ed non-linear oscillators. In: International symposium on mathematical problems in th eoretical physics. Ky- oto University, Japan. pp. 420–422. Springer (2005)

2005
[22]

Automatica 47(3), 608–613 (2011)

Lauer, F., Bloch, G., Vidal, R.: A continuous optimizati on framework for hybrid system identiﬁcation. Automatica 47(3), 608–613 (2011)

2011
[23]

Nonlinearity 21(9), 2121 (2008)

Llibre, J., Ponce, E., Torres, F.: On the existence and un iqueness of limit cycles in liénard diﬀerential equations allowing discontinuitie s. Nonlinearity 21(9), 2121 (2008)

2008
[24]

part i: Simplest case in R2

Lum, R., Chua, L.O.: Global properties of continuous pie cewise linear vector ﬁelds. part i: Simplest case in R2. International journal of circuit theory and applications 19(3), 251–307 (1991)

1991
[25]

Miljkovic, A., Hanke, N., Liu, Z., Stursberg, O.: Refere nce tracking and state obser- vation for switching aﬃne systems using piecewise quadrati c lyapunov functions. Proc. of IF AC Symposium on Nonlinear Control Systems p. TBD ( 2025)

2025
[26]

SIAM Journal on Applied Mathematics 50(6), 1645–1662 (1990)

Mirollo, R.E., Strogatz, S.H.: Synchronization of puls e-coupled biological oscilla- tors. SIAM Journal on Applied Mathematics 50(6), 1645–1662 (1990)

1990
[27]

European journal of control 13(2-3), 242–260 (2007)

Paoletti, S., Juloski, A.L., Ferrari-Trecate, G., Vida l, R.: Identiﬁcation of hybrid systems a tutorial. European journal of control 13(2-3), 242–260 (2007)

2007
[28]

: On convergence prop- erties of piecewise aﬃne systems

Pavlov, A., Pogromsky, A., Van De Wouw, N., Nijmeijer, H. : On convergence prop- erties of piecewise aﬃne systems. International Journal of Control 80(8), 1233–1247 (2007)

2007
[29]

IEEE Transactions on Powe r Electronics 18(1), 301–308 (2003)

Peterchev, A.V., Sanders, S.R.: Quantization resoluti on and limit cycling in digi- tally controlled pwm converters. IEEE Transactions on Powe r Electronics 18(1), 301–308 (2003)

2003
[30]

IEEE Transactions on Circuits and Systems II: Express Briefs 55(9), 882–886 (2008)

Teplinsky, A., Feely, O.: Limit cycles in a mems oscillat or. IEEE Transactions on Circuits and Systems II: Express Briefs 55(9), 882–886 (2008)

2008
[31]

PLoS One 15(8), e0235930 (2020)

Werckenthin, A., Huber, J., Arnold, T., Koziarek, S., Pl ath, M.J., Plath, J.A., Stursberg, O., Herzel, H., Stengl, M.: Neither per, nor tim1 , nor cry2 alone are es- sential components of the molecular circadian clockwork in the madeira cockroach. PLoS One 15(8), e0235930 (2020)

2020
[32]

Automatica 44(11), 2909–2915 (2008)

van de Wouw, N., Pavlov, A.: Tracking and synchronisatio n for a class of pwa systems. Automatica 44(11), 2909–2915 (2008)

2008
[33]

The European Ph ysical Journal Special Topics 228(9), 1839–1854 (2019)

Xu, Y., Luo, A.C.: Frequency-amplitude characteristic s of periodic motions in a periodically forced van der pol oscillator. The European Ph ysical Journal Special Topics 228(9), 1839–1854 (2019)

2019
[34]

IEEE Trans

Zhu, Y., Zheng, W.X.: Multiple lyapunov functions analy sis approach for discrete- time-switched piecewise-aﬃne systems under dwell-time co nstraints. IEEE Trans. on Automatic Control 65(5), 2177–2184 (2019)

2019