pith. sign in

arxiv: 2607.00844 · v1 · pith:YFL27N5Snew · submitted 2026-07-01 · 🧮 math.OC · cs.SY· eess.SY

Experiment Design for Set-membership Identification: From Prior Knowledge to Universal Inputs

Amir Shakouri , Henk J. van Waarde , M. Kanat Camlibel This is my paper

Pith reviewed 2026-07-02 07:43 UTC · model grok-4.3

classification 🧮 math.OC cs.SYeess.SY
keywords set-membership identificationuniversal inputsexperiment designlinear time-invariant systemspersistently exciting inputsprior knowledgebounded noisesystem identification
0
0 comments X

The pith

Universal inputs can be designed from general prior knowledge to guarantee accurate identification data for any LTI system in the admissible set.

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

The paper develops methods to select finite-horizon input signals that produce data sufficient for set-membership identification of any linear time-invariant system consistent with a given prior on its parameters. This works in both noise-free cases and with bounded noise when the prior is of the right type. A sympathetic reader would care because the construction generalizes the persistently exciting inputs that follow from Willems' fundamental lemma, which require only controllability knowledge, and can require fewer samples for other priors. The result supplies explicit design procedures that ensure the collected trajectories separate or bound the possible systems to the desired accuracy.

Core claim

An input is universal for a prior-knowledge set if it yields data suitable for identification with prescribed accuracy when applied to every system whose parameters lie in that set. The authors supply constructive methods for finding such inputs, showing that persistently exciting inputs are universal only under controllability priors while other priors admit universal inputs with strictly better sample efficiency; certain priors even permit exact identification despite bounded noise.

What carries the argument

Universal inputs: input sequences that, for every system consistent with the prior-knowledge set, produce finite-horizon data enabling the target identification accuracy.

If this is right

  • Universal inputs exist and are computable for arbitrary admissible parameter sets, not only controllability sets.
  • For some priors the shortest universal input is shorter than any persistently exciting input that works for the same set.
  • Exact identification remains possible in the presence of bounded noise when the prior satisfies additional structural conditions.
  • The design reduces to checking rank or set-membership conditions on the data matrices generated by the candidate input.

Where Pith is reading between the lines

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

  • If the prior set itself can be updated from partial data, one could iteratively shorten the remaining experiment length.
  • The same universal-input construction might extend to switched or piecewise-linear systems whose mode sets play the role of the parameter set.
  • In applications where simulation or first-principles models already constrain parameters, the method could cut the length of required physical tests.

Load-bearing premise

The unknown system is linear time-invariant and the prior knowledge is a computable set of admissible parameter values for which the notion of universal inputs is well-defined.

What would settle it

Exhibit a specific prior-knowledge set and accuracy target such that every finite input sequence fails to produce data allowing the required identification accuracy for at least one system inside the set.

Figures

Figures reproduced from arXiv: 2607.00844 by Amir Shakouri, Henk J. van Waarde, M. Kanat Camlibel.

Figure 1
Figure 1. Figure 1: Sets Σ′ pk and Σ′′ pk, for Examples 23 and 25, shown by the red and blue areas, respectively. B. Experiment design for exact identification In this section, we consider universal experiment design for identification. We begin by defining the following notion: δpk := inf  [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
read the original abstract

We consider the problem of designing input signals for an unknown linear time-invariant system in such a way that the resulting data, within a finite horizon, is suitable for identification with a desired accuracy. We consider both noise-free and noisy settings with $\ell_\infty$--bounded noise models. We will take into account general prior knowledge of the system parameters. Central in our study is the concept of universal inputs. An input is called universal for identification if, when applied to any system complying with the prior knowledge, it yields data suitable for accurate identification. We provide new methods for designing such universal inputs. Our results generalize the experiment design approach based on Willems et al.'s fundamental lemma that relies on persistently exciting inputs, and that is limited to prior knowledge on controllability. It turns out that for other types of prior knowledge, there exist universal inputs that outperform the persistently exciting ones, e.g., in terms of sample efficiency. Moreover, we investigate types of prior knowledge that enable experiment design for exact identification in the presence of noise.

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

0 major / 2 minor

Summary. The paper develops a framework for designing input signals for set-membership identification of unknown linear time-invariant systems. It incorporates general prior knowledge on system parameters and defines universal inputs that guarantee data suitable for accurate identification (noise-free and with ℓ_∞-bounded noise) for every system consistent with the prior. The approach generalizes the persistently exciting input construction from Willems et al.'s fundamental lemma (restricted to controllability priors) and claims that other priors admit universal inputs with improved sample efficiency; it also identifies priors permitting exact identification despite noise.

Significance. If the constructions are correct and the universal inputs remain computable for nontrivial priors, the work would extend experiment design beyond the controllability case, potentially reducing data requirements and enabling exact recovery under bounded noise for selected priors. The explicit treatment of arbitrary admissible parameter sets as the basis for universality is a clear conceptual advance.

minor comments (2)
  1. The abstract states that universal inputs 'outperform the persistently exciting ones, e.g., in terms of sample efficiency,' but does not indicate whether this is shown by explicit construction, by a general theorem, or only by example; a concrete comparison (e.g., length of the shortest universal input versus the PE length) would strengthen the claim.
  2. Notation for the admissible parameter set and the precise definition of 'suitable for accurate identification' (e.g., the required rank or excitation condition) should be introduced early and used consistently throughout.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their summary of the manuscript and for recognizing the conceptual advance in generalizing persistently exciting inputs to universal inputs under arbitrary prior knowledge. The recommendation is listed as uncertain, but the report contains no specific major comments or requests for clarification. We therefore provide no point-by-point responses and remain available to address any questions that may arise during further review.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained via external reference

full rationale

The paper's central contribution generalizes experiment design for universal inputs from the external Willems et al. fundamental lemma (limited to controllability priors) to arbitrary admissible parameter sets for LTI systems. This extension is formulated directly from the set-membership prior knowledge without reducing any prediction or uniqueness claim to a self-citation, fitted input, or definitional loop. The abstract explicitly positions the outperforming inputs and noise-robust identification as consequences of the prior-knowledge formulation rather than internal re-derivations. No load-bearing self-citations or ansatz smuggling appear; the work remains independent of the authors' prior results.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard domain assumptions in linear systems theory and bounded-noise models; no free parameters or invented entities are visible from the abstract.

axioms (2)
  • domain assumption The plant is a linear time-invariant system.
    Explicitly stated in the abstract as the class of systems under consideration.
  • domain assumption Noise is bounded in the infinity norm.
    Stated as the noise model for the noisy setting.

pith-pipeline@v0.9.1-grok · 5725 in / 1335 out tokens · 25345 ms · 2026-07-02T07:43:13.531329+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages

  1. [1]

    G. C. Goodwin and R. L. Payne,Dynamic System Identification: Experiment Design and Data Analysis. Academic Press, 1977

    work page 1977

  2. [2]

    Ljung,System Identification: Theory for the User

    L. Ljung,System Identification: Theory for the User. Pearson Educa- tion, 1998

    work page 1998

  3. [3]

    van Overschee and B

    P. van Overschee and B. De Moor,Subspace Identification for Linear Systems: Theory—Implementation—Applications. Springer, 2012

    work page 2012

  4. [4]

    Milanese, J

    M. Milanese, J. Norton, H. Piet-Lahanier, and ´E. Walter,Bounding Approaches to System Identification. Springer, 2013

    work page 2013

  5. [5]

    Pintelon and J

    R. Pintelon and J. Schoukens,System Identification: A Frequency Domain Approach. John Wiley & Sons, 2012

    work page 2012

  6. [6]

    H. J. van Waarde, M. K. Camlibel, and H. L. Trentelman,Data-Based Linear Systems and Control Theory. Kindle Direct Publishing, 2025

    work page 2025

  7. [7]

    Set membership parameter estimation and design of experiments using homothety,

    S. Borchers, S. Rakovi ´c, and R. Findeisen, “Set membership parameter estimation and design of experiments using homothety,”IFAC Proceed- ings Volumes, vol. 44, no. 1, pp. 9035–9040, 2011

    work page 2011

  8. [8]

    Design of experiments for non- linear system identification: A set membership approach,

    M. Karimshoushtari and C. Novara, “Design of experiments for non- linear system identification: A set membership approach,”Automatica, vol. 119, p. 109036, 2020

    work page 2020

  9. [9]

    Beyond persistent excitation: Online experiment design for data-driven modeling and control,

    H. J. van Waarde, “Beyond persistent excitation: Online experiment design for data-driven modeling and control,”IEEE Control Systems Letters, vol. 6, pp. 319–324, 2021

    work page 2021

  10. [10]

    The shortest experiment for linear system identification,

    M. K. Camlibel, H. J. van Waarde, and P. Rapisarda, “The shortest experiment for linear system identification,”Systems & Control Letters, vol. 197, p. 106045, 2025

    work page 2025

  11. [11]

    From experiment design to closed-loop control,

    H. Hjalmarsson, “From experiment design to closed-loop control,” Automatica, vol. 41, no. 3, pp. 393–438, 2005

    work page 2005

  12. [12]

    Persistency of excitation, sufficient richness and parameter convergence in discrete time adaptive control,

    E.-W. Bai and S. S. Sastry, “Persistency of excitation, sufficient richness and parameter convergence in discrete time adaptive control,”Systems & Control letters, vol. 6, no. 3, pp. 153–163, 1985

    work page 1985

  13. [13]

    Persistence of excitation in linear systems,

    M. Green and J. B. Moore, “Persistence of excitation in linear systems,” Systems & Control Letters, vol. 7, no. 5, pp. 351–360, 1986

    work page 1986

  14. [14]

    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

  15. [15]

    A quantitative notion of persistency of excitation and the robust fundamental lemma,

    J. Coulson, H. J. van Waarde, J. Lygeros, and F. D ¨orfler, “A quantitative notion of persistency of excitation and the robust fundamental lemma,” IEEE Control Systems Letters, vol. 7, pp. 1243–1248, 2022

    work page 2022

  16. [16]

    A quantitative and constructive proof of Willems’ fun- damental lemma and its implications,

    J. Berberich, A. Iannelli, A. Padoan, J. Coulson, F. D ¨orfler, and F. Allg ¨ower, “A quantitative and constructive proof of Willems’ fun- damental lemma and its implications,” inAmerican Control Conference (ACC). IEEE, 2023, pp. 4155–4160

    work page 2023

  17. [17]

    Robust optimal experiment design for system identification,

    C. R. Rojas, J. S. Welsh, G. C. Goodwin, and A. Feuer, “Robust optimal experiment design for system identification,”Automatica, vol. 43, no. 6, pp. 993–1008, 2007

    work page 2007

  18. [18]

    Optimal experiment designs with respect to the intended model application,

    M. Gevers and L. Ljung, “Optimal experiment designs with respect to the intended model application,”Automatica, vol. 22, no. 5, pp. 543– 554, 1986

    work page 1986

  19. [19]

    Consistency and relative efficiency of subspace methods,

    M. Deistler, K. Peternell, and W. Scherrer, “Consistency and relative efficiency of subspace methods,”Automatica, vol. 31, no. 12, pp. 1865– 1875, 1995

    work page 1995

  20. [20]

    Finite sample properties of linear model identification,

    E. Weyer, R. C. Williamson, and I. M. Mareels, “Finite sample properties of linear model identification,”IEEE Transactions on Automatic Control, vol. 44, no. 7, pp. 1370–1383, 2002

    work page 2002

  21. [21]

    Finite sample properties of system iden- tification methods,

    M. C. Campi and E. Weyer, “Finite sample properties of system iden- tification methods,”IEEE Transactions on Automatic Control, vol. 47, no. 8, pp. 1329–1334, 2002

    work page 2002

  22. [22]

    Beyond asymp- totics: Targeted exploration with finite-sample guarantees,

    J. Venkatasubramanian, J. K ¨ohler, and F. Allg ¨ower, “Beyond asymp- totics: Targeted exploration with finite-sample guarantees,” inConfer- ence on Decision and Control (CDC). IEEE, 2025, pp. 75–81

    work page 2025

  23. [23]

    Experiment design in a bounded-error context: comparison with D-optimality,

    L. Pronzato and E. Walter, “Experiment design in a bounded-error context: comparison with D-optimality,”Automatica, vol. 25, no. 3, pp. 383–391, 1989

    work page 1989

  24. [24]

    Design of experiments for guaranteed parameter estimation in membership setting,

    S. Borchers and R. Findeisen, “Design of experiments for guaranteed parameter estimation in membership setting,” inConference on Decision and Control (CDC) and European Control Conference (ECC). IEEE, 2011, pp. 2602–2607

    work page 2011

  25. [25]

    The sample complexity of worst-case identification of FIR linear systems,

    M. A. Dahleh, T. V . Theodosopoulos, and J. N. Tsitsiklis, “The sample complexity of worst-case identification of FIR linear systems,”Systems & Control Letters, vol. 20, no. 3, pp. 157–166, 1993

    work page 1993

  26. [26]

    On the time complexity of worst-case system identification,

    K. Poolla and A. Tikku, “On the time complexity of worst-case system identification,”IEEE Transactions on Automatic Control, vol. 39, no. 5, pp. 944–950, 1994

    work page 1994

  27. [27]

    For differential equations withrparameters,2r+ 1 experiments are enough for identification,

    E. Sontag, “For differential equations withrparameters,2r+ 1 experiments are enough for identification,”Journal of Nonlinear Science, vol. 12, no. 6, pp. 553–583, 2003

    work page 2003

  28. [28]

    Comparison of closed-loop system identification techniques to quantify multi-joint human balance control,

    D. Engelhart, T. A. Boonstra, R. G. Aarts, A. C. Schouten, and H. van der Kooij, “Comparison of closed-loop system identification techniques to quantify multi-joint human balance control,”Annual Reviews in Control, vol. 41, pp. 58–70, 2016

    work page 2016

  29. [29]

    Statistical learning theory for control: A finite-sample perspective,

    A. Tsiamis, I. Ziemann, N. Matni, and G. J. Pappas, “Statistical learning theory for control: A finite-sample perspective,”IEEE Control Systems Magazine, vol. 43, no. 6, pp. 67–97, 2023

    work page 2023

  30. [30]

    Optimal two- and three-impulse fixed-time rendezvous in the vicinity of a circular orbit,

    J. E. Prussing, “Optimal two- and three-impulse fixed-time rendezvous in the vicinity of a circular orbit,”AIAA Journal, vol. 8, no. 7, pp. 1221–1228, 1970

    work page 1970

  31. [31]

    Optimal estimation theory for dynamic systems with set membership uncertainty: An overview,

    M. Milanese and A. Vicino, “Optimal estimation theory for dynamic systems with set membership uncertainty: An overview,”Automatica, vol. 27, no. 6, pp. 997–1009, 1991

    work page 1991

  32. [32]

    Learning the uncertainty sets of linear control systems via set membership: A non- asymptotic analysis,

    Y . Li, J. Yu, L. Conger, T. Kargin, and A. Wierman, “Learning the uncertainty sets of linear control systems via set membership: A non- asymptotic analysis,” inInternational Conference on Machine Learning (ICML), 2024

    work page 2024

  33. [33]

    Chebyshev centers and radii for sets induced by quadratic matrix inequalities,

    A. Shakouri, H. J. van Waarde, and M. K. Camlibel, “Chebyshev centers and radii for sets induced by quadratic matrix inequalities,”Mathematics of Control, Signals, and Systems, vol. 37, pp. 1007–1034, 2025

    work page 2025

  34. [34]

    Set membership identification of linear systems with guaranteed simulation accuracy,

    M. Lauricella and L. Fagiano, “Set membership identification of linear systems with guaranteed simulation accuracy,”IEEE Transactions on Automatic Control, vol. 65, no. 12, pp. 5189–5204, 2020

    work page 2020

  35. [35]

    Subspace model identification part 1. the output-error state-space model identification class of algorithms,

    M. Verhaegen and P. Dewilde, “Subspace model identification part 1. the output-error state-space model identification class of algorithms,” International Journal of Control, vol. 56, no. 5, pp. 1187–1210, 1992

    work page 1992

  36. [36]

    Identifica- tion of stable models in subspace identification by using regularization,

    T. van Gestel, J. A. Suykens, P. van Dooren, and B. De Moor, “Identifica- tion of stable models in subspace identification by using regularization,” IEEE Transactions on Automatic Control, vol. 46, no. 9, pp. 1416–1420, 2002

    work page 2002

  37. [37]

    Subspace identification with guar- anteed stability using constrained optimization,

    S. L. Lacy and D. S. Bernstein, “Subspace identification with guar- anteed stability using constrained optimization,”IEEE Transactions on Automatic Control, vol. 48, no. 7, pp. 1259–1263, 2003

    work page 2003

  38. [38]

    Subspace identification with eigenvalue constraints,

    D. N. Miller and R. A. De Callafon, “Subspace identification with eigenvalue constraints,”Automatica, vol. 49, no. 8, pp. 2468–2473, 2013

    work page 2013

  39. [39]

    Identification of positive linear systems with poisson output transformation,

    A. De Santis and L. Farina, “Identification of positive linear systems with poisson output transformation,”Automatica, vol. 38, no. 5, pp. 861–868, 2002

    work page 2002

  40. [40]

    Identification of positive real models in subspace identification by using regularization,

    I. Goethals, T. van Gestel, J. Suykens, P. van Dooren, and B. De Moor, “Identification of positive real models in subspace identification by using regularization,”IEEE Transactions on Automatic Control, vol. 48, no. 10, pp. 1843–1847, 2003

    work page 2003

  41. [41]

    A novel subspace identification approach with passivity enforcement,

    L. F. Rodrigues, L. P. Ihlenfeld, and G. H. da Costa Oliveira, “A novel subspace identification approach with passivity enforcement,” Automatica, vol. 132, p. 109798, 2021

    work page 2021

  42. [42]

    Subspace identification with moment matching,

    M. Inoue, “Subspace identification with moment matching,”Automatica, vol. 99, pp. 22–32, 2019

    work page 2019

  43. [43]

    Kernel-based identification with fre- quency domain side-information,

    M. Khosravi and R. S. Smith, “Kernel-based identification with fre- quency domain side-information,”Automatica, vol. 150, p. 110813, 2023

    work page 2023

  44. [44]

    Experiment design using prior knowledge on controllability and stabilizability,

    A. Shakouri, H. J. van Waarde, and M. K. Camlibel, “Experiment design using prior knowledge on controllability and stabilizability,”arXiv preprint arXiv:2512.01876 (to appear in 23rd IFAC World Congress), 2025

    work page arXiv 2025

  45. [45]

    A new perspective on Willems’ fundamental lemma: Universality of persistently exciting inputs,

    ——, “A new perspective on Willems’ fundamental lemma: Universality of persistently exciting inputs,”IEEE Control Systems Letters, vol. 9, pp. 583–588, 2025

    work page 2025

  46. [46]

    On controllability and persistency of excitation in data- driven control: Extensions of Willems’ fundamental lemma,

    Y . Yu, S. Talebi, H. J. van Waarde, U. Topcu, M. Mesbahi, and B. Ac ¸ıkmes,e, “On controllability and persistency of excitation in data- driven control: Extensions of Willems’ fundamental lemma,” inIEEE Conference on Decision and Control (CDC). IEEE, 2021, pp. 6485– 6490

    work page 2021

  47. [47]

    Chebyshev center of the intersection of balls: Complexity, relaxation and approximation,

    Y . Xia, M. Yang, and S. Wang, “Chebyshev center of the intersection of balls: Complexity, relaxation and approximation,”Mathematical Pro- gramming, vol. 187, no. 1, pp. 287–315, 2021

    work page 2021

  48. [48]

    Regularization in regression with bounded noise: A Chebyshev center approach,

    A. Beck and Y . C. Eldar, “Regularization in regression with bounded noise: A Chebyshev center approach,”SIAM Journal on Matrix Analysis and Applications, vol. 29, no. 2, pp. 606–625, 2007

    work page 2007

  49. [49]

    Beer,Topologies on Closed and Closed Convex Sets

    G. Beer,Topologies on Closed and Closed Convex Sets. Springer, 1993, vol. 268

    work page 1993

  50. [50]

    Bernstein,Scalar, Vector, and Matrix Mathematics: Theory, Facts, and Formulas-Revised and Expanded Edition

    D. Bernstein,Scalar, Vector, and Matrix Mathematics: Theory, Facts, and Formulas-Revised and Expanded Edition. Princeton University Press, 2018

    work page 2018

  51. [51]

    Comprehensive survey and assessment of spacecraft relative motion dynamics models,

    J. Sullivan, S. Grimberg, and S. D’Amico, “Comprehensive survey and assessment of spacecraft relative motion dynamics models,”Journal of Guidance, Control, and Dynamics, vol. 40, no. 8, pp. 1837–1859, 2017

    work page 2017

  52. [52]

    Quadratic matrix inequalities with applications to data-based control,

    H. J. van Waarde, M. K. Camlibel, J. Eising, and H. L. Trentelman, “Quadratic matrix inequalities with applications to data-based control,” SIAM Journal on Control and Optimization, vol. 61, no. 4, pp. 2251– 2281, 2023

    work page 2023

  53. [53]

    R. A. Horn and C. R. Johnson,Topics in Matrix Analysis. Cambridge University Press, 1994

    work page 1994

  54. [54]

    A matrix lower bound,

    J. F. Grcar, “A matrix lower bound,”Linear Algebra and its Applications, vol. 433, no. 1, pp. 203–220, 2010. Amir Shakouriis currently pursuing the Ph.D. degree in applied mathematics with the Sys- tems, Control, and Optimization Group, Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Gronin- gen, Groningen,...

    work page 2010

  55. [55]

    His research interests include learning and data-driven control, sys- tem identification and identifiability, networks of dynamical systems, and robust and optimal control

    He was also a visiting researcher at the University of Washington, Seattle in 2019-2020. His research interests include learning and data-driven control, sys- tem identification and identifiability, networks of dynamical systems, and robust and optimal control. Dr. van Waarde is the recipient of the 2025 SIAM Activity Group on Control and Systems Theory P...

    work page 2019