arxiv: 2604.01070 · v2 · submitted 2026-04-01 · 🧮 math.OC

Recognition: no theorem link

Stability, Contraction, and Controllers for Affine Systems

L. P. Wieringa , A. Padoan , F. Dorfler , J. Eising

Authors on Pith no claims yet

Pith reviewed 2026-05-13 21:56 UTC · model grok-4.3

classification 🧮 math.OC

keywords affine systemsbehavioral approachcontractionstabilityLyapunov theoremsfeedback controlimplementability

0 comments

The pith

Linear controllers suffice to implement contractive closed-loop behavior in affine systems, while affine controllers are required to place the equilibrium.

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

The paper establishes representation-independent necessary and sufficient conditions for contraction and stability in affine behaviors. It addresses converse Lyapunov theorems for input-output contraction, the implementability of prescribed contractive references, and the feedback type needed to realize them. A sympathetic reader would care because affine systems arise naturally from linearization of nonlinear dynamics yet retain enough structure for exact analysis, allowing behavioral methods to move beyond purely linear cases without committing to a specific state-space model.

Core claim

For affine behaviors the paper derives necessary and sufficient conditions for three problems: converse Lyapunov theorems establishing contraction of input-output systems, the existence and implementability of prescribed contractive references, and the question of whether linear or affine feedback suffices to realize those references. Linear controllers are shown to be sufficient for producing a contractive closed-loop behavior, while affine controllers are necessary when the equilibrium must also be placed at a desired location.

What carries the argument

The behavioral representation of an affine system, which treats the system as a set of trajectories rather than any particular state-space or input-output realization.

Load-bearing premise

The systems under consideration are affine and admit a behavioral representation independent of any particular state-space or input-output realization.

What would settle it

An explicit affine system together with a contractive reference for which no linear feedback achieves the closed-loop contraction property, or for which the derived implementability condition fails.

read the original abstract

Recent developments in data-driven control have revived interest in the behavioral approach to systems theory, where systems are defined as sets of trajectories rather than being described by a specific model or representation. However, most available results remain confined to linear systems, limiting the applicability of recent methods to complex behaviors. Affine systems form a natural intermediate class: they arise from linearization, capture essential nonlinear effects, and retain sufficient structure for analysis and design. This paper derives necessary and sufficient conditions independent of any particular representation for three fundamental stability problems for affine behaviors: (i) converse Lyapunov theorems for contraction of input-output systems; (ii) implementability and existence of prescribed contractive references; and (iii) whether these references can be implemented with linear or affine feedback control. For the latter, we show that linear controllers suffice for implementing contractive closed-loop, and and affine controllers are needed for equilibrium placement.

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

The paper gives representation-independent nec and suff conditions for contraction and implementability in affine behaviors, with a clean split on linear versus affine controllers that still needs explicit proof checks.

read the letter

The main thing to know is that this work pushes behavioral contraction theory from linear to affine systems and supplies necessary and sufficient conditions for three problems: converse Lyapunov results for contraction, implementability of contractive references, and the controller class required. The distinction that linear feedback handles contraction while affine feedback is needed for equilibrium placement is the practical takeaway and appears new relative to the linear literature. It does well by keeping the statements free of any fixed state-space or kernel representation, which aligns with the data-driven motivation and makes the results more portable. The framing of the three problems is direct and the abstract states the claims without hidden fitting or circular definitions. The soft spot is the controller separation itself. The stress-test note flags that the affine offset must decouple cleanly from the homogeneous contraction dynamics; if the proofs achieve this without selecting a particular representation that already encodes the constant term, the claim holds. If any step implicitly fixes the kernel, the representation-independence weakens. That is the one place where a referee should press for the full derivations. This paper is for control theorists working on behavioral methods or data-driven design who already know the linear case and want to handle the next layer of structure. It deserves a serious referee because the questions are well-posed, the extension is relevant, and the claims are specific enough to verify or correct in review. I would send it out.

Referee Report

3 major / 2 minor

Summary. The manuscript develops a behavioral-systems treatment of affine dynamical systems, deriving representation-independent necessary and sufficient conditions for three problems: (i) converse Lyapunov theorems establishing contraction of input-output trajectories, (ii) implementability of prescribed contractive reference trajectories, and (iii) the minimal controller class (linear versus affine) required to realize those references in closed loop. The central technical claim is that linear feedback is sufficient to enforce contraction of the closed-loop behavior while affine feedback is necessary to place the equilibrium at an arbitrary prescribed point.

Significance. If the stated conditions and controller distinctions hold, the work supplies a rigorous bridge between the well-developed linear behavioral theory and the practically relevant class of affine systems that arise from linearization or mild nonlinearities. The representation-free character of the results and the explicit separation of contraction (homogeneous) from equilibrium placement (particular solution) would be useful for data-driven controller synthesis on affine behaviors.

major comments (3)

[§4] §4 (Implementability of contractive references): The sufficiency argument for linear controllers achieving contraction appears to rest on decomposing the affine behavior into a homogeneous (difference) part and a particular (equilibrium) solution. The manuscript must explicitly verify that the contraction metric and the associated Lyapunov function are unaffected by the constant offset term; otherwise the claim that linear feedback suffices for contraction while affine feedback is required only for equilibrium placement is not yet load-bearing.
[Theorem 3.2] Theorem 3.2 (Converse Lyapunov for affine behaviors): The necessity direction invokes a behavioral representation that is independent of state-space realization. It is unclear whether the constructed Lyapunov function remains valid when the same trajectory set is expressed in a different kernel representation that encodes the affine constant differently; a short invariance argument under representation change is needed.
[§5.1] §5.1 (Closed-loop controller synthesis): The necessity proof that affine controllers are required for arbitrary equilibrium placement uses a specific input-output partitioning. If the partitioning is altered while preserving the same affine behavior, does the necessity still hold? The paper should state the precise invariance property that makes the linear-versus-affine distinction representation-independent.

minor comments (2)

[Abstract] Abstract, line 8: duplicated word 'and and affine' should be corrected.
[§2] Notation for the affine kernel representation (e.g., the constant vector in the behavioral equation) is introduced without an explicit comparison to the linear case; a short remark clarifying the difference would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough and constructive report. The comments highlight areas where explicit clarifications will strengthen the representation-independent claims. We address each major point below and will incorporate the suggested revisions.

read point-by-point responses

Referee: [§4] §4 (Implementability of contractive references): The sufficiency argument for linear controllers achieving contraction appears to rest on decomposing the affine behavior into a homogeneous (difference) part and a particular (equilibrium) solution. The manuscript must explicitly verify that the contraction metric and the associated Lyapunov function are unaffected by the constant offset term; otherwise the claim that linear feedback suffices for contraction while affine feedback is required only for equilibrium placement is not yet load-bearing.

Authors: We agree that an explicit verification strengthens the argument. The contraction metric and Lyapunov function are defined on trajectory differences, which cancel any constant offset by construction. In the revised version we will insert a short paragraph in §4 immediately after the decomposition, proving that both the metric and the Lyapunov function depend solely on the homogeneous component of the behavior and are therefore invariant to the choice of particular solution. revision: yes
Referee: [Theorem 3.2] Theorem 3.2 (Converse Lyapunov for affine behaviors): The necessity direction invokes a behavioral representation that is independent of state-space realization. It is unclear whether the constructed Lyapunov function remains valid when the same trajectory set is expressed in a different kernel representation that encodes the affine constant differently; a short invariance argument under representation change is needed.

Authors: The Lyapunov function is constructed directly from the set of trajectories belonging to the behavior, without reference to any particular kernel matrix. Because any two kernel representations describe exactly the same trajectory set, the function is automatically invariant. Nevertheless, we will add a concise invariance lemma (or remark) inside the proof of Theorem 3.2 that explicitly shows the construction is unchanged under reparameterization of the kernel representation. revision: yes
Referee: [§5.1] §5.1 (Closed-loop controller synthesis): The necessity proof that affine controllers are required for arbitrary equilibrium placement uses a specific input-output partitioning. If the partitioning is altered while preserving the same affine behavior, does the necessity still hold? The paper should state the precise invariance property that makes the linear-versus-affine distinction representation-independent.

Authors: The necessity argument relies on the intrinsic property that the affine behavior contains a nonzero constant trajectory. Any valid input-output partitioning that respects the system dimensions yields the same behavior; therefore the distinction between linear and affine controllers is preserved. In the revision we will add an explicit statement in §5.1 asserting that the linear-versus-affine controller classification is invariant under admissible changes of input-output partitioning, because it depends only on whether a constant shift can be absorbed without an additional constant input. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation rests on external behavioral axioms

full rationale

The paper states necessary and sufficient conditions for contraction and implementability of affine behaviors that are representation-independent. No equations or claims in the provided text reduce a prediction or theorem to a fitted parameter, self-definition, or self-citation chain by construction. The distinction between linear and affine controllers follows from the separation of homogeneous and particular solutions in affine trajectory sets, which is a standard structural property of the behavioral framework rather than an internal tautology. The central results therefore remain non-circular and externally grounded.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard behavioral-systems axioms for trajectory sets and the definition of affine behaviors; no free parameters or invented entities are introduced in the abstract.

axioms (2)

domain assumption Systems are defined as sets of trajectories rather than specific equations or state-space realizations.
Explicitly invoked in the abstract as the behavioral approach.
domain assumption Affine systems arise from linearization and capture essential nonlinear effects while retaining sufficient structure.
Stated directly as the motivation for studying this intermediate class.

pith-pipeline@v0.9.0 · 5457 in / 1238 out tokens · 30031 ms · 2026-05-13T21:56:25.498210+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

[1]

Linear tracking MPC for nonlinear systems-part II: The data-driven case,

J. Berberich, J. K ¨ohler, M. A. M ¨uller, and F. Allg ¨ower, “Linear tracking MPC for nonlinear systems-part II: The data-driven case,” IEEE Transactions on Automatic Control, vol. 67, no. 9, pp. 4406– 4421, 2022

work page 2022
[2]

An Overview of Systems-Theoretic Guarantees in Data-Driven Model Predictive Control,

J. Berberich and F. Allg ¨ower, “An Overview of Systems-Theoretic Guarantees in Data-Driven Model Predictive Control,”Annual Review of Control, Robot., and Auton. Syst., 2024

work page 2024
[3]

Choose wisely: Data- driven predictive control for nonlinear systems using online data selection,

J. N ¨af, K. Moffat, J. Eising, and F. D ¨orfler, “Choose wisely: Data- driven predictive control for nonlinear systems using online data selection,”https://arxiv.org/pdf/2503.18845, 2025

work page arXiv 2025
[4]

Less is more: Contextual sampling for nonlinear data-driven predictive control,

J. Beerwerth and B. Alrifaee, “Less is more: Contextual sampling for nonlinear data-driven predictive control,” https://arxiv.org/pdf/2503.23890, 2025

work page arXiv 2025
[5]

J. B. Rawlings, D. Q. Mayne, and M. Diehl,Model Predictive Control: Theory, Computation, and Design, 2nd ed. Nob Hill Publishing, 2017

work page 2017
[6]

J. W. Polderman and J. C. Willems,Introduction to Mathematical Systems Theory: A Behavioral Approach. Springer New York, 1998

work page 1998
[7]

A note on persistency of excitation,

J. C. Willems, P. Rapisarda, I. Markovsky, and B. L. De Moor, “A note on persistency of excitation,”Systems & Control Letters, vol. 54, no. 4, pp. 325–329, 2005

work page 2005
[8]

Data-driven optimal control of affine systems: A linear programming perspective,

A. Martinelli, M. Gargiani, M. Dra ˇskovi´c, and J. Lygeros, “Data-driven optimal control of affine systems: A linear programming perspective,” IEEE Control Systems Letters, vol. 6, pp. 3092–3097, 2022

work page 2022
[9]

Data-driven representations of conical, convex, and affine behaviors,

A. Padoan, F. D ˜A¶rfler, and J. Lygeros, “Data-driven representations of conical, convex, and affine behaviors,” inProceedings of the 62nd IEEE Conference on Decision and Control., 2023

work page 2023
[10]

From time series to affine systems,

A. Padoan, J. Eising, and I. Markovsky, “From time series to affine systems,”https://arxiv.org/abs/2510.22089, 2025

work page arXiv 2025
[11]

On quadratic differential forms,

J. C. Willems and H. L. Trentelman, “On quadratic differential forms,” SIAM Journal on Control and Optimization, vol. 36, no. 5, pp. 1703– 1749, 1998

work page 1998
[12]

A generalized Lyapunov stability theorem for discrete-time systems based on quadratic difference forms,

C. Kojima and K. Takaba, “A generalized Lyapunov stability theorem for discrete-time systems based on quadratic difference forms,”Trans- actions of the Society of Instrument and Control Engineers, vol. 42, pp. 2911 – 2916, 01 2006

work page 2006
[13]

On Lya- punov functions and data-driven dissipativity,

T. M. Maupong, J. C. Mayo-Maldonado, and P. Rapisarda, “On Lya- punov functions and data-driven dissipativity,”IFAC-PapersOnLine, vol. 50, no. 1, pp. 7783–7788, 2017

work page 2017
[14]

A behavioral approach to data-driven control with noisy input-output data,

H. J. van Waarde, J. Eising, M. K. Camlibel, and H. L. Trentelman, “A behavioral approach to data-driven control with noisy input-output data,”IEEE Transactions on Automatic Control, vol. 69, no. 2, pp. 813–827, 2024

work page 2024
[15]

From time se- ries to dissipativity of linear systems with dynamic supply rates,

H. J. van Waarde, J. Coulson, and A. Padoan, “From time se- ries to dissipativity of linear systems with dynamic supply rates,” https://arxiv.org/abs/2602.13654, 2026

work page arXiv 2026
[16]

Achievable behavior of general systems,

A. van der Schaft, “Achievable behavior of general systems,”Systems & Control Letters, vol. 49, no. 2, pp. 141–149, 2003

work page 2003
[17]

On interconnections, control, and feedback,

J. C. Willems, “On interconnections, control, and feedback,”IEEE Transactions on Automatic Control, vol. 42, no. 3, pp. 326–339, 1997

work page 1997
[18]

Regular imple- mentability and its application to stabilization of system behaviors,

M. N. Belur, H. L. Trentelman, and J. C. Willems, “Regular imple- mentability and its application to stabilization of system behaviors,” inProceedings of the 40th IEEE Conference on Decision and Control, vol. 4, 2001, pp. 3479–3484

work page 2001
[19]

Stabilization, pole placement, and regular implementability,

M. N. Belur and H. L. Trentelman, “Stabilization, pole placement, and regular implementability,”IEEE Transactions on Automatic Control, vol. 47, no. 5, pp. 735–744, 2002

work page 2002
[20]

Bullo,Contraction Theory for Dynamical Systems, 1.2 ed

F. Bullo,Contraction Theory for Dynamical Systems, 1.2 ed. Kindle Direct Publishing, 2024

work page 2024
[21]

On contraction analysis for non- linear systems,

W. Lohmiller and J.-J. E. Slotine, “On contraction analysis for non- linear systems,”Automatica, vol. 34, no. 6, pp. 683–696, 1998

work page 1998
[22]

How to represent and identify affine time-invariant systems?

I. Markovsky, J. Eising, and A. Padoan, “How to represent and identify affine time-invariant systems?”IEEE Control Systems Letters, vol. 9, pp. 1207–1212, 2025

work page 2025
[23]

An LMI condition for asymptotic stability of discrete-time system based on quadratic difference forms,

C. Kojima and K. Takaba, “An LMI condition for asymptotic stability of discrete-time system based on quadratic difference forms,”Pro- ceedings of the 2006 IEEE Conference on Computer Aided Control Systems Design, CACSD, pp. 1139 – 1143, 10 2006

work page 2006
[24]

Nonnegative realization of autonomous systems in the behavioral approach,

M. Valcher, “Nonnegative realization of autonomous systems in the behavioral approach,” inProceedings of the 41st IEEE Conference on Decision and Control, 2002., vol. 2, 2002, pp. 2131–2136

work page 2002
[25]

Synthesis of dissipative systems using quadratic differential forms: Part I,

J. C. Willems and H. L. Trentelman, “Synthesis of dissipative systems using quadratic differential forms: Part I,”IEEE Transactions on Automatic Control, vol. 47, no. 1, pp. 53–69, 2002

work page 2002
[26]

Towards a behavioral linear systems theory tailored to data-driven control,

J. Eising and F. D ˜A¶rfler, “Towards a behavioral linear systems theory tailored to data-driven control,” 2025, submitted

work page 2025
[27]

Paradigms and puzzles in the theory of dynamical systems,

J. C. Willems, “Paradigms and puzzles in the theory of dynamical systems,”IEEE Transactions on Automatic Control, vol. 36, no. 3, pp. 259–294, 1991. APPENDIX: PROOFS Proof of Lemma III.2.The equivalence of (i) and (iii) is clear by definition, and (ii) evidently implies (i). It remains to show (i)⇒(ii). Assume thatB ∈ A q is contractive. Then B= ker c R(σ...

work page 1991