Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Data-Efficient Control of Polynomial Systems via Physics-Guided Quadratic Constraints

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a physics-guided quadratic constraint, added to data from a single short noisy trajectory, is enough to synthesize a robust control barrier certificate and its safety controller for an unknown discrete-time…

desk verdict Promising physics-guided data-driven safety synthesis with a clean Theorem 1, but the central SOS theorem has an unproven dilation equivalence and the key uncertainty bound is calibrated post-hoc, so the claimed data savings are provisional. read the letter →

arxiv 2508.01315 v2 pith:6WGCBVVW submitted 2025-08-02 eess.SY cs.SY

MohammadHossein Ashoori , Ali Aminzadeh , Amy Nejati , Abolfazl Lavaei This is my paper
classification eess.SYcs.SY MSC 93C5593D3090C2293C10
keywords robustcontrolbarriercertificatesdata-drivensafetysynthesisphysics-guidedquadraticconstraintssum-of-squaresoptimizationdiscrete-timepolynomialsystemsunknown-but-boundeddisturbancesampleefficiencycontrollers
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

The paper claims that for a discrete-time input-affine polynomial system whose matrices are unknown and whose output is corrupted by bounded noise, safety can be certified directly from a single short input-state trajectory, provided one also has a rough nominal model obtained from physical first principles. The certificate is a quadratic robust control barrier function $\beta(x)=x^{\top}P x$, and the controller is a state-dependent linear-in-$x$ policy $u=K(x)x$; both are produced jointly by one sum-of-squares optimization program. If that program is feasible, the barrier and controller are guaranteed to work for every system matrix consistent with the observed data and with the physics-guided uncertainty bound, so no explicit system identification step is needed. In the three benchmarks, adding the physics constraint cuts the number of data samples needed for an infinite-horizon guarantee by more than an order of magnitude compared with the purely data-driven version.

What carries the argument

The load-bearing object is the robust control barrier certificate in the inverse variable: rather than searching for a positive-definite $P$ directly, the SOS program searches for $\tilde P=P^{-1}$ and a polynomial $\tilde K(x)$, so that the decrease condition $\beta(x^+)\le \lambda\beta(x)+\delta$ becomes a linear matrix inequality. The decrease condition is enforced not for one model but for every $\Xi$ in the set cut out by the data-conformity inequalities (13) and the physics-guided quadratic constraint (16); the S-procedure pulls those quadratic inequalities into the SOS condition with polynomial multipliers $\kappa_i(x)$. Young's inequality $\mu J^{\top}P J + (1+\mu)^{-1}\omega^{\top}P\omega$ bounds the disturbance term, and the Schur complement turns the weighted disturbance bound into the LMI (23a). The physics-guided constraint (16) is what shrinks the admissible set enough to make a short trajectory sufficient.

What would settle it

Choose a nominal model, fix $\epsilon_a$ smaller than the actual perturbation between the true matrices and the nominal ones, collect a trajectory, run Algorithm 1 to feasibility, and simulate the closed loop from many initial conditions in $X_0$: if any trajectory enters $X_u$, the uniform guarantee stated in Theorem 2 is false, since the true $\Xi$ lies outside the physics-guided set (16).

Watch

Extended reading notes

Core claim

The central claim is Theorem 2: if the SOS program (23a)–(23d) is feasible, then $\beta(x)=x^{\top}P x$ is a robust control barrier certificate for the unknown discrete-time input-affine polynomial system and $u=K(x)x$ is its robust safety controller, with $\gamma_1=\tilde\gamma_1^2$, $\gamma_2=\tilde\gamma_2^2$, and $\delta=\tilde\delta^2$. The certificate is uniform: the inequalities (4a)–(4c) hold for every system matrix $\Xi=[A\;B]$ that satisfies the data-conformity inequalities (13) and the physics-guided quadratic constraint (16). Theorem 1 then converts the certificate into a safety guarantee: if $\delta \leq \gamma_1(1-\lambda)$, all trajectories starting in the initial set stay out of the unsafe set for all time; for larger $\delta$ up to the bound in (5), the same avoidance is guaranteed over a finite horizon. The proof uses the Schur complement to rewrite the weighted disturbance bound, Young's inequality to absorb the disturbance cross term, the S-procedure to enforce the barrier decrease for all admissible $\Xi$, and a dilation argument to linearize the bilinear coupling between $P^{-1}$ and the controller.

Load-bearing premise

Everything rests on the prior that the true system matrix lies in the weighted spectral-norm ball of radius $\epsilon_a$ around the first-principles nominal matrix; the experiments set that radius after generating the perturbation, so no data actually verifies it.

Editorial extensions

If this is right

  • If the paper is right, an infinite-horizon robust safety certificate for a nonlinear polynomial system can be obtained from a single short trajectory without identifying the matrices $A$ and $B$.
  • Three benchmarks (a Lorenz system, a rotating rigid spacecraft, and a degree-3 polynomial extension) require only 9, 27, and 17 data samples respectively for infinite-horizon guarantees, versus 130, 80, and 50 for the purely data-driven version.
  • The R-CBC notion is less conservative than robust-invariant-set methods because only the initial set, not the whole safe set, must lie inside a barrier level set.
  • If the physics prior is absent or the bound $\epsilon_a$ is so loose that the constraint is uninformative, the optimization degrades to the purely data-driven case and needs more data for the same guarantee.

Reading between the lines

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

  • Editorial inference: the reported sample counts are meaningful only if $\epsilon_a$ is fixed before seeing the data; in the benchmarks it is chosen large enough to cover perturbations generated in advance, so a practitioner still needs a principled first-principles error bound for real deployment.
  • Editorial inference: because the barrier level sets are ellipsoids, the method is likely to become infeasible or conservative when initial and unsafe regions are strongly non-convex; the paper's own suggested direction of composite barrier functions is the natural test of the method's ceiling.
  • Editorial inference: Theorem 1's finite-time regime can be read as a design knob—lowering $\lambda$ shortens the guaranteed horizon but relaxes the decrease condition (4c), which may let the SOS program remain feasible in cases where an infinite-horizon guarantee is impossible.
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.

Referee Report

3 major / 5 minor

Summary. The paper develops a physics-guided, direct data-driven framework for synthesizing robust control barrier certificates (R-CBCs) and robust safety controllers for discrete-time input-affine nonlinear polynomial systems with unknown-but-bounded disturbances. The certificate is restricted to the quadratic form B(x)=x^T P x, and the synthesis is cast as an SOS/LMI program that combines noisy input-state trajectory data with a prior constraint on the spectral-norm distance between the unknown system matrix Ξ=[A B] and a nominal first-principles value Ξ0. The main theoretical result, Theorem 2, claims that feasibility of the SOS program (23a)-(23d) implies the robust decrease condition (4c) uniformly for all Ξ consistent with the data-conformity and physics-guided constraints, without an identification step. Three benchmarks (Lorenz, rotating rigid spacecraft, and a degree-3 extension) are used to demonstrate that the physics-guided constraint reduces the number of required data samples relative to a purely data-driven baseline.

Significance. If Theorem 2 is fully established, the contribution is significant: it offers a single-trajectory, SOS-based safety-certificate synthesis for nonlinear polynomial systems that explicitly exploits physics-based side information to reduce data requirements, and the infinite/finite-time split in Theorem 1 is a useful refinement of standard barrier-certificate conditions. The paper is also honest about the conservatism of the quadratic barrier and about the sufficient-only nature of the SOS conditions. However, the central theorem currently rests on an unproved and dimensionally delicate equivalence between a robust condition over all admissible Ξ and the tractable matrix inequality (23d); until that equivalence is supplied, the main safety guarantee is not established. The empirical claims in Table I and Figures 1-3 are consistent with the intended theory but do not compensate for the missing proof step.

major comments (3)
  1. [§IV-A, Eq. (19) and the proof of Theorem 2] The key step from the robust barrier condition to the SOS program is not proved. In (19) the closed-loop dynamics are written as x^+ = Z(x)x + ω with Z(x)=[φ(x); ψ(x)K(x)], but this omits the unknown matrix Ξ=[A B]; the correct expression is x^+ = Ξ [Ψ(x); ψ(x)K(x)] x + ω, with Ψ(x) from (3). As printed, (19) is dimensionally inconsistent because Z(x)x has dimension d_φ+d_ψ while x^+ has dimension n. Later, in (28), Ξ reappears in a quadratic form, so the derivation silently repairs the omission. More importantly, the only bridge from the Ξ-dependent condition over all admissible systems to the tractable condition is the single sentence in §IV-A: 'one can use dilation [49] and show that inequality (31) is equivalent to (23d)'. No derivation is given, and this is not a routine congruence: (13) and (16) are quadratic matrix inequalities in Ξ, whereas (28) is an n×n matrix inequality depending on Ξ, so the equivalence requires an explicit elimination of Ξ with careful block-dimension accounting. Because the benchmarks have d_φ much larger than n (e.g., Lorenz has d_φ=10 and n=3), the equivalence must hold for tall dictionaries; a hidden d=n assumption would invalidate the published solver outputs. Please provide a complete proof of (31)⇔(23d), including the dimensions of all blocks and the treatment of the multipliers κ_0(x), κ_i(x).
  2. [§IV-A, S-procedure step (31)] The S-procedure application that produces (31) is not justified in its printed form. The data-conformity constraints (13) and the physics-guided constraint (16) are quadratic matrix inequalities in Ξ, while (28) is a matrix inequality after multiplication by [Ξ^T I] or a similar congruence. The multipliers κ(x) in (31) are scalar polynomial functions, but a matrix-valued uncertainty generally requires either matrix multipliers or a Schur-complement reduction to a largest-singular-value condition. The paper should state the exact form of the quadratic constraints being dualized, the signs and degrees of the multipliers, and the direction of the implication that is actually used. Without this, it is unclear whether (23d) is merely sufficient or also necessary for (31), and the statement 'equivalence' is unsupported.
  3. [§V-A, §V-B, §V-C and Table I] The empirical data-efficiency claim depends on an a priori valid bound ε_a in (14), but in all three benchmarks ε_a is set after the perturbation of Ξ has been generated, as 'sufficiently large to accommodate these perturbations'. The same applies to ε_ω. Nothing in the data can certify that the true Ξ lies in the physics-guided set (16); if ε_a is underestimated, Theorem 2 provides no safety guarantee. The paper should explain how ε_a and ε_ω are selected before solving the SOS program, or provide a sensitivity analysis showing how the required trajectory length T varies with conservative overestimates of ε_a. Without such an explanation, the 'data-efficient' comparison in Table I is not supported by a valid guarantee for the exact system used in the simulations.
minor comments (5)
  1. [§II-C, Definition 2] The quantifier structure in condition (4c) is ambiguous: the text reads 'for all x∈X, u∈U such that ω∈W', but it should clarify whether (4c) must hold for all control inputs u or only for the synthesized feedback u=K(x)x. Since Theorem 2 designs the controller, the intended condition is likely closed-loop, but Definition 2 as written suggests a stronger requirement.
  2. [§III-A, Eq. (12)-(13)] The derivation of (13) should state the dimensions of the data matrices explicitly: Ξ is n×(d_φ+d_ψ), the regressor [φ(x_j); ψ(x_j)u_j] is (d_φ+d_ψ)×1, and the T matrix inequalities are obtained by applying (10) to each residual. As printed, the indexing is hard to follow because the same symbols φ and ψ are used for both the vector dictionary and the matrix Ψ(x) from (3).
  3. [§IV-A, Eq. (23d)] The displayed inequality (23d) is garbled in terms of notation: it mixes ar P, P, ar K(x), K(x), and the multipliers κ(x) without indicating which variables are design variables and which are fixed. Please rewrite (23d) with a clear block structure and define all dimensions in a table or remark.
  4. [§IV-A, proof of Theorem 2, inequalities (23b)-(23c)] The passage from (23b)-(23c) to (4a)-(4b) uses a complement argument and then replaces a negative-semidefinite condition by a conservative '≤' inequality. The exact relationship between γ_1, γ_2 and the barred quantities γbar_1, γbar_2, including the claimed squares, should be stated explicitly, because the direction of the level-set inclusion depends on whether ar P is P^{-1} or P.
  5. [§V and Table I] Table I is difficult to read because the numerical entries are formatted as garbled expressions (e.g., '����×��'). Please typeset the table with explicit decimal values for T_PGDD, T_DD, γ_1, γ_2, δ, and the runtime, and state whether the reported T values are the smallest values that were actually tested or a lower bound from a bisection search.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: Theorem 2 synthesizes the certificate from SOS feasibility; self-citations are contextual, and the dilation gap and post-hoc epsilon_a choice are correctness/assumption risks, not circular reductions.

full rationale

Walking the derivation chain from Definition 2 through Theorem 2, I find no circular reduction. The robust decrease condition (4c) is not assumed inside the SOS program; it is derived from feasibility of (23a)-(23d) via Young's inequality, Schur complements, the S-procedure, and a dilation claimed equivalent to (31). The target property (unsafe avoidance) enters only as constraints (23b)-(23c) on the synthesized P, so this is standard certificate synthesis rather than self-definitional fitting. The uncertainty sets (13) and (16) define the quantification in Theorem 2; the theorem is conditional on the prior bound epsilon_a, but the guarantee is for all Xi in that set, not identical to the choice of epsilon_a. Self-citations ([7], [27], [39], [57]) are contextual comparisons and future-work pointers; none carries a load-bearing step of Theorem 2, and the cited technical tools (Schur complement [47], S-procedure [33], dilation [49]) are external. Two caveats should be weighed as correctness risks, not circularity: (i) the sentence in Section IV-A, 'one can use dilation [49] and show that inequality (31) is equivalent to (23d),' is an omitted proof of a load-bearing equivalence; if it fails, SOS feasibility may not imply robust safety. (ii) In the benchmarks (Sections V-A to V-C), epsilon_a is set after the perturbation is generated, as 'sufficiently large to accommodate these perturbations', so the empirical data-efficiency claim rests on an unverified prior; this weakens the demonstration but does not make the derivation circular.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The method's guarantees are conditional on the user-supplied bounds epsilon_omega and epsilon_a, the weight matrix W, the tuning parameters mu and lambda, and the assumption that the dictionary spans the true dynamics. In the experiments these are chosen with knowledge of the generated perturbations, which is the main practical weakness of the evaluation. No new physical entities are introduced.

free parameters (5)
  • epsilon_omega (disturbance bound) = not reported; set 'sufficiently large' in all benchmarks
    Defines the data-conformity set via (8). In simulation it is chosen to cover the generated disturbance, which is a post-hoc calibration rather than a genuine prior.
  • epsilon_a (model mismatch bound) = not reported; set 'sufficiently large' in all benchmarks
    Defines the physics-guided set via (14). In simulation it is chosen after generating the perturbation, so the reported data savings depend on this calibration.
  • W (weight matrix) = not reported; identity in the special case
    Used in both (8) and (14); the benchmarks do not state the concrete matrix W.
  • mu (Young's inequality parameter) = 0.001 (Lorenz), 0.001 (spacecraft), 0.0001 (higher-degree)
    Chosen by hand to balance the disturbance bound in (26); affects feasibility and tightness.
  • lambda (decay rate) = initialized at 0.1 and increased until a valid solution is found
    Remark 8; the final lambda determines whether the guarantee is infinite or finite horizon.
assumptions (6)
  • domain assumption The true dynamics are representable as x+ = A zeta(x) + B eta(x) u + omega with a known rich dictionary (zeta, eta).
    Invoked in Definition 1 and Section III-A; if the dictionary is not rich enough, the data-conformity set (13) cannot represent the true system and Theorem 2 does not apply.
  • domain assumption The disturbance omega satisfies the known weighted bound omega^T W^T W omega <= epsilon_omega^2 of (8).
    Used to derive the data-conformity set (13) and Lemma 1; a violation invalidates the safety guarantee.
  • domain assumption The true system matrix Xi = [A B] lies within the weighted norm ball ||Xi - Xi0||_W <= epsilon_a of (14).
    Central to the physics-guided set (16); the data-efficiency claim depends on this prior being valid.
  • domain assumption The state set X is a basic semi-algebraic set defined by known polynomial inequalities.
    Needed for the SOS relaxation (34) via Positivstellensatz.
  • standard math Standard results: Schur complement, S-procedure, Cauchy-Schwarz, Young's inequality, and the dilation lemma of [49].
    Applied in the proof of Theorem 2; the dilation equivalence (31)<->(23d) is cited without proof.
  • ad hoc to paper The barrier certificate is restricted to the quadratic form B(x)=x^T P x.
    Stated in (18) and acknowledged in Section IV-D as a trade-off for SOS tractability; it limits which initial and unsafe set geometries can be certified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Data-Efficient Control of Polynomial Systems via Physics-Guided Quadratic Constraints." pith.science (2026). https://pith.science/paper/6WGCBVVW

@misc{pith2026250801315,
  author       = {Pith},
  title        = {Pith review of: Data-Efficient Control of Polynomial Systems via Physics-Guided Quadratic Constraints},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6WGCBVVW}},
  note         = {Machine review of arXiv:2508.01315}
}
read the original abstract

This work addresses the critical challenge of guaranteeing safety for complex dynamical systems where precise mathematical models are uncertain and data measurements are corrupted by noise. We develop a physics-guided, direct data-driven framework for synthesizing robust safety controllers for discrete-time nonlinear polynomial systems that are subject to unknown-but-bounded disturbances. To do so, we introduce a notion of safety through robust control barrier certificates, which ensure avoidance of unsafe regions, offering a less conservative alternative to existing methods based on robust invariant sets. To achieve data efficiency, we further integrate physical information, formulated as quadratic constraints on system and control matrices, with observed noisy data. This integration drastically reduces data requirements, enabling robust safety analysis with significantly shorter trajectories compared to purely data-driven methods. The proposed synthesis procedure is formulated as a sum-of-squares optimization program that systematically designs the barrier and its associated controller by leveraging both collected data and underlying physical laws. The efficacy of our framework is demonstrated on three benchmark systems, confirming its ability to offer robust safety guarantees with reduced data demands.

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Data-Driven Stochastic Control: Foundations and Guarantees

    eess.SY 2025-07 conditional novelty 6.0 of 10

    A data-driven framework synthesizes stochastic control barrier certificates and safety controllers for unknown polynomial stochastic systems, with probabilistic safety guarantees and certified confidence from multiple...

  2. Data-Driven Formal Methods for Complex Dynamical Systems: A Survey

    eess.SY 2026-07 accept novelty 2.0 of 10

    A taxonomy and survey of data-driven formal verification and controller synthesis, organized around abstraction-based, functional-certificate, and compositional methods with PAC, Lipschitz, and structural-property guarantees.

Reference graph

Works this paper leans on

59 extracted references · 48 canonical work pages · cited by 2 Pith papers

  1. [37]

    Trade-offs in learning controllers from noisy data,

    A. Bisoffi, C. D. Persis, and P. Tesi, “Trade-offs in learning controllers from noisy data,”Systems & Control Letters, vol. 154, 2021

    work page 2021

  2. [49]

    Zhang,The Schur Complement and Its Applications

    F. Zhang,The Schur Complement and Its Applications. Springer Science & Business Media, 2006, vol. 4

    2006

  3. [1]

    Analysis and design of safety- critical, cyber-physical systems,

    J. McGregor, D. Gluch, and P. Feiler, “Analysis and design of safety- critical, cyber-physical systems,”ACM SIGAda Ada Letters, vol. 36, no. 2, pp. 31–38, 2017

    work page 2017

  4. [2]

    Safety verification of hybrid systems using barrier certificates,

    S. Prajna and A. Jadbabaie, “Safety verification of hybrid systems using barrier certificates,” inInternational Workshop on Hybrid Systems: Computation and Control, 2004, pp. 477–492

    work page 2004

  5. [3]

    Constructive safety using control barrier functions,

    P. Wieland and F. Allg ¨ower, “Constructive safety using control barrier functions,”IFAC Proceedings Volumes, vol. 40, no. 12, pp. 462–467, 2007

    work page 2007

  6. [4]

    Control barrier functions: Theory and applications,

    A. D. Ames, S. Coogan, M. Egerstedt, G. Notomista, K. Sreenath, and P. Tabuada, “Control barrier functions: Theory and applications,” in18th European Control Conference (ECC), 2019, pp. 3420–3431

    work page 2019

  7. [5]

    A barrier function approach to finite-time stochastic system verification and control,

    C. Santoyo, M. Dutreix, and S. Coogan, “A barrier function approach to finite-time stochastic system verification and control,”Automatica, vol. 125, 2021

    work page 2021

  8. [6]

    W. Xiao, C. G. Cassandras, and C. Belta,Safe Autonomy with Control Barrier Functions: Theory and Applications, 2023

    work page 2023

Show all 59 references
  1. [7]

    PRoTECT: Parallelized con- struction of safety barrier certificates for nonlinear polynomial systems,

    B. Wooding, V . Horbanov, and A. Lavaei, “PRoTECT: Parallelized con- struction of safety barrier certificates for nonlinear polynomial systems,” arXiv: 2404.14804, 2024

    2024 arXiv

  2. [8]

    Safety verification for probabilistic hybrid systems,

    L. Zhang, Z. She, S. Ratschan, H. Hermanns, and E. M. Hahn, “Safety verification for probabilistic hybrid systems,”European Journal of Control, vol. 18, no. 6, pp. 572–587, 2012

    2012

  3. [9]

    Control barrier functions for complete and incomplete in- formation stochastic systems,

    A. Clark, “Control barrier functions for complete and incomplete in- formation stochastic systems,” inAmerican Control Conference (ACC), 2019, pp. 2928–2935

    2019

  4. [10]

    Multi-robot collision avoidance under uncertainty with probabilistic safety barrier certificates,

    W. Luo, W. Sun, and A. Kapoor, “Multi-robot collision avoidance under uncertainty with probabilistic safety barrier certificates,” inAdvances in Neural Information Processing Systems, vol. 33, 2020, pp. 372–383

    2020

  5. [11]

    Scalable synthesis of safety barrier certifi- cates for networks of stochastic switched systems,

    A. Lavaei and E. Frazzoli, “Scalable synthesis of safety barrier certifi- cates for networks of stochastic switched systems,”IEEE Transactions on Automatic Control, vol. 11, no. 69, pp. 7294–7309, 2024

    2024

  6. [12]

    Context-triggered games for reactive synthesis over stochastic systems via control barrier certificates,

    A. Nejati, S. Prakash Nayak, and A.-K. Schmuck, “Context-triggered games for reactive synthesis over stochastic systems via control barrier certificates,” inProceedings of the 27th ACM International Conference on Hybrid Systems: Computation and Control, 2024, pp. 1–12. 12

    2024

  7. [13]

    Automated verifica- tion and synthesis of stochastic hybrid systems: A survey,

    A. Lavaei, S. Soudjani, A. Abate, and M. Zamani, “Automated verifica- tion and synthesis of stochastic hybrid systems: A survey,”Automatica, vol. 146, 2022

    2022

  8. [14]

    K. Zhou, J. C. Doyle, and K. Glover,Robust and Optimal Control. Prentice Hall, 1996

    1996

  9. [15]

    An indirect data-driven method for trajectory tracking control of a class of nonlinear discrete-time systems,

    Z. Wang, R. Lu, F. Gao, and D. Liu, “An indirect data-driven method for trajectory tracking control of a class of nonlinear discrete-time systems,” IEEE Transactions on Industrial Electronics, vol. 64, no. 5, pp. 4121– 4129, 2016

    2016

  10. [16]

    Bridging direct and indirect data-driven control formulations via regularizations and relaxations,

    F. D ¨orfler, J. Coulson, and I. Markovsky, “Bridging direct and indirect data-driven control formulations via regularizations and relaxations,” IEEE Transactions on Automatic Control, vol. 68, no. 2, pp. 883–897, 2022

    2022

  11. [17]

    On direct vs indirect data-driven predictive control,

    V . Krishnan and F. Pasqualetti, “On direct vs indirect data-driven predictive control,” in60th IEEE Conference on Decision and Control (CDC), 2021, pp. 736–741

    2021

  12. [18]

    Formulas for data-driven control: Stabilization, optimality, and robustness,

    C. De Persis and P. Tesi, “Formulas for data-driven control: Stabilization, optimality, and robustness,”IEEE Transactions on Automatic Control, vol. 65, no. 3, pp. 909–924, 2019

    2019

  13. [19]

    Guarantees for data-driven control of nonlinear systems using semidefinite programming: A survey,

    T. Martin, T. B. Sch ¨on, and F. Allg ¨ower, “Guarantees for data-driven control of nonlinear systems using semidefinite programming: A survey,” Annual Reviews in Control, vol. 56, 2023

    2023

  14. [20]

    The scenario approach to robust control design,

    G. C. Calafiore and M. C. Campi, “The scenario approach to robust control design,”IEEE Transactions on automatic control, vol. 51, no. 5, pp. 742–753, 2006

    2006

  15. [21]

    The scenario approach for systems and control design,

    M. C. Campi, S. Garatti, and M. Prandini, “The scenario approach for systems and control design,”Annual Reviews in Control, vol. 33, no. 2, pp. 149–157, 2009

    2009

  16. [22]

    Performance bounds for the scenario approach and an extension to a class of non-convex programs,

    P. Mohajerin Esfahani, T. Sutter, and J. Lygeros, “Performance bounds for the scenario approach and an extension to a class of non-convex programs,”IEEE Transactions on Automatic Control, vol. 60, no. 1, pp. 46–58, 2014

    2014

  17. [23]

    On the road between robust optimization and the scenario approach for chance constrained optimiza- tion problems,

    K. Margellos, P. Goulart, and J. Lygeros, “On the road between robust optimization and the scenario approach for chance constrained optimiza- tion problems,”IEEE Transactions on Automatic Control, vol. 59, no. 8, pp. 2258–2263, 2014

    2014

  18. [24]

    From noisy data to feedback controllers: Nonconservative design via a matrix S-lemma,

    H. J. Van Waarde, M. K. Camlibel, and M. Mesbahi, “From noisy data to feedback controllers: Nonconservative design via a matrix S-lemma,” IEEE Transactions on Automatic Control, vol. 67, no. 1, pp. 162–175, 2020

    2020

  19. [25]

    purely data-driven method

    by allowing the initial region to be a subset of the safe set, rather than requiring strict invariance of the entire safe set. It also enables direct synthesis via a single-shot SOS optimization, which avoids the potentially conservative overapproximation of system and control...

  20. [26]

    Meta results on data-driven control of nonlinear systems,

    N. Monshizadeh, C. De Persis, and P. Tesi, “Meta results on data-driven control of nonlinear systems,”arXiv:2405.10064, 2024

    2024 arXiv

  21. [27]

    Data-driven design of safe control for polynomial systems,

    A. Luppi, A. Bisoffi, C. De Persis, and P. Tesi, “Data-driven design of safe control for polynomial systems,”European Journal of Control, vol. 75, 2024

    2024

  22. [28]

    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

    2005

  23. [29]

    From a single trajectory to safety controller synthesis of discrete-time nonlinear poly- nomial systems,

    B. Samari, O. Akbarzadeh, M. Zaker, and A. Lavaei, “From a single trajectory to safety controller synthesis of discrete-time nonlinear poly- nomial systems,”IEEE Control Systems Letters, vol. 8, pp. 3123–3128, 2024

    2024

  24. [30]

    Physics-informed data-driven safe and optimal control design,

    N. Niknejad and H. Modares, “Physics-informed data-driven safe and optimal control design,”IEEE Control Systems Letters, vol. 8, pp. 285– 290, 2023

    2023

  25. [31]

    Willems’ fundamental lemma for state-space systems and its extension to multiple datasets,

    H. J. Van Waarde, C. De Persis, M. K. Camlibel, and P. Tesi, “Willems’ fundamental lemma for state-space systems and its extension to multiple datasets,”IEEE Control Systems Letters, vol. 4, no. 3, pp. 602–607, 2020

    2020

  26. [32]

    A stabilization algorithm for a class of uncertain linear systems,

    I. R. Petersen, “A stabilization algorithm for a class of uncertain linear systems,”Systems & control letters, vol. 8, no. 4, pp. 351–357, 1987

    1987

  27. [33]

    Learning dynamical systems with side information,

    A. A. Ahmadi and B. El Khadir, “Learning dynamical systems with side information,”SIAM Review, vol. 65, no. 1, pp. 183–223, 2023

    2023

  28. [34]

    Data-driven control via petersen’s lemma,

    A. Bisoffi, C. De Persis, and P. Tesi, “Data-driven control via petersen’s lemma,”Automatica, vol. 145, 2022

    2022

  29. [35]

    From noisy data to feedback controllers: Nonconservative design via a matrix s-lemma,

    H. J. van Waarde, M. K. Camlibel, and M. Mesbahi, “From noisy data to feedback controllers: Nonconservative design via a matrix s-lemma,” IEEE Transactions on Automatic Control, vol. 67, no. 1, pp. 162–175, 2022

    2022

  30. [36]

    Data-driven stabilization of nonlinear systems via taylor’s expansion,

    M. Guo, C. De Persis, and P. Tesi, “Data-driven stabilization of nonlinear systems via taylor’s expansion,” inHybrid and Networked Dynamical Systems: Modeling, Analysis and Control. Springer, 2024, pp. 273–299

    2024

  31. [38]

    Formal verification of unknown discrete-and continuous-time systems: A data- driven approach,

    A. Nejati, A. Lavaei, P. Jagtap, S. Soudjani, and M. Zamani, “Formal verification of unknown discrete-and continuous-time systems: A data- driven approach,”IEEE Transactions on Automatic Control, vol. 68, no. 5, pp. 3011–3024, 2023

    2023

  32. [39]

    Data-based transfer stabilization in linear systems,

    L. Li, C. De Persis, P. Tesi, and N. Monshizadeh, “Data-based transfer stabilization in linear systems,”IEEE Transactions on Automatic Con- trol, vol. 69, no. 3, pp. 1866–1873, 2023

    2023

  33. [40]

    Application of robust control barrier function with stochastic disturbance model for discrete time systems,

    R. Takano, H. Oyama, and M. Yamakita, “Application of robust control barrier function with stochastic disturbance model for discrete time systems,”IFAC-PapersOnLine, vol. 51, no. 31, pp. 46–51, 2018

    2018

  34. [41]

    A physics- informed scenario approach with data mitigation for safety verification of nonlinear systems,

    A. Aminzadeh, M. H. Ashoori, A. Nejati, and A. Lavaei, “A physics- informed scenario approach with data mitigation for safety verification of nonlinear systems,”arXiv:2412.03932, 2024

    2024 arXiv

  35. [42]

    Robust safety under stochastic uncertainty with discrete-time control barrier functions,

    R. K. Cosner, P. Culbertson, A. J. Taylor, and A. D. Ames, “Robust safety under stochastic uncertainty with discrete-time control barrier functions,”arXiv:2302.07469, 2023

    2023 arXiv

  36. [43]

    Robustness of control barrier functions for safety critical control,

    X. Xu, P. Tabuada, J. W. Grizzle, and A. D. Ames, “Robustness of control barrier functions for safety critical control,”IFAC-PapersOnLine, vol. 48, no. 27, pp. 54–61, 2015

    2015

  37. [44]

    Verification and synthesis of robust control barrier functions: Multilevel polynomial optimization and semidefinite relaxation,

    S. Kang, Y . Chen, H. Yang, and M. Pavone, “Verification and synthesis of robust control barrier functions: Multilevel polynomial optimization and semidefinite relaxation,” in62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 8215–8222

    2023

  38. [45]

    Robust control barrier functions for constrained stabiliza- tion of nonlinear systems,

    M. Jankovic, “Robust control barrier functions for constrained stabiliza- tion of nonlinear systems,”Automatica, vol. 96, pp. 359–367, 2018

    2018

  39. [46]

    Data- driven feedback linearization with complete dictionaries,

    C. De Persis, D. Gadginmath, F. Pasqualetti, and P. Tesi, “Data- driven feedback linearization with complete dictionaries,” in62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 3037–3042

    2023

  40. [47]

    Learning robust output control barrier functions from safe expert demonstrations,

    L. Lindemann, A. Robey, L. Jiang, S. Das, S. Tu, and N. Matni, “Learning robust output control barrier functions from safe expert demonstrations,”IEEE Open Journal of Control Systems, 2024

    2024

  41. [48]

    On classes of summable functions and their Fourier series,

    W. H. Young, “On classes of summable functions and their Fourier series,”Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, vol. 87, no. 594, pp. 225–229, 1912

    1912

  42. [50]

    SOS- TOOLS: Control applications and new developments,

    S. Prajna, A. Papachristodoulou, P. Seiler, and P. A. Parrilo, “SOS- TOOLS: Control applications and new developments,” inProceedings of IEEE International Conference on Robotics and Automation, 2004, pp. 315–320

    2004

  43. [51]

    LMI properties and applications in systems, stability, and control theory,

    R. J. Caverly and J. R. Forbes, “LMI properties and applications in systems, stability, and control theory,”arXiv:1903.08599, 2019

    1903 arXiv

  44. [52]

    H. K. Khalil and J. W. Grizzle,Nonlinear Systems. Prentice Hall Upper Saddle River, NJ, 2002

    2002

  45. [53]

    Synchronization of complex networks of identical and nonidentical chaotic systems via model-matching control,

    D. L ´opez-Mancilla, G. L ´opez-Cahuich, C. Posadas-Castillo, C. Casta ˜neda, J. Garc ´ıa-L´opez, J. V ´azquez-Guti´errez, and E. Tlelo- Cuautle, “Synchronization of complex networks of identical and nonidentical chaotic systems via model-matching control,”Plos one, vol. 14, n...

    2019

  46. [54]

    Deterministic nonperiodic flow,

    E. N. Lorenz, “Deterministic nonperiodic flow,”Journal of Atmospheric Sciences, vol. 20, no. 2, pp. 130–141, 1963

    1963

  47. [55]

    Chaotic synchronization and secure communication based on descriptor observer,

    H. Wang, Z. Han, W. Zhang, and Q. Xie, “Chaotic synchronization and secure communication based on descriptor observer,”Nonlinear Dynamics, vol. 57, no. 1, pp. 69–73, 2009

    2009

  48. [56]

    S. H. Strogatz,Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering. Chapman and Hall/CRC, 2024

    2024

  49. [57]

    to address the scalability limitations of monolithic SOS programs for high-dimensional systems by decomposing the global synthesis problem into smaller, tractable subproblems. Furthermore, extending the framework tostochasticcontrol barrier certificates for systems affected by...

  50. [58]

    J. C. Sprott,Elegant Chaos: Algebraically Simple Chaotic Flows. World Scientific, 2010

    2010

  51. [59]

    From formal meth- ods to data-driven safety certificates of unknown large-scale networks,

    O. Akbarzadeh, B. Samari, A. Nejati, and A. Lavaei, “From formal meth- ods to data-driven safety certificates of unknown large-scale networks,” IEEE Transactions on Automatic Control, to appear, 2026. DATA-EFFICIENT CONTROL OF POL YNOMIAL SYSTEMS VIA PHYSICS-GUIDED QUADRATIC C...

    2026

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.