pith. sign in

arxiv: 2604.28108 · v2 · submitted 2026-04-30 · 📡 eess.SY · cs.SY

Hierarchical Control for Continuous-time Systems via General Approximate Alternating Simulation Relations

Zhiyuan Huang , Shuo Li , Murat Arcak , Majid Zamani , Bingzhuo Zhong This is my paper

Pith reviewed 2026-05-07 06:40 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords approximate alternating simulationhierarchical controlcontinuous-time systemscontrol refinementmodel abstractionsimulation relationsstability preservation
0
0 comments X

The pith

A general approximate alternating simulation relation enables hierarchical control of continuous-time systems by tolerating larger model mismatches.

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

This paper introduces the ε-gAAS relation as a relaxed version of alternating simulation for continuous-time systems. The relation allows abstract models to differ more substantially from concrete dynamics than traditional methods permit. The authors examine the properties of this relation and create a control refinement technique to transfer abstract controllers to the real system. This framework supports hierarchical control where high-level decisions use simplified models. Such an approach could make abstraction-based design feasible for a broader range of complex continuous systems.

Core claim

The central discovery is the definition of the ε-gAAS relation for continuous-time systems, which relaxes existing simulation relations to tolerate larger mismatches between abstract and concrete models. Its properties are investigated, and a control refinement method is developed to enable hierarchical control while preserving closed-loop properties. Case studies illustrate the effectiveness and advantages over existing methods.

What carries the argument

The ε-gAAS relation, which provides conditions for the existence of a control refinement map that accounts for approximation errors in continuous-time dynamics.

Load-bearing premise

That the defined ε-gAAS relation allows a practical control refinement that maintains the desired closed-loop properties in the concrete system despite the permitted mismatches, and that such relations can be found for systems of interest.

What would settle it

Demonstrating a concrete system and abstract model pair where the ε-gAAS relation holds but the refined controller does not achieve the guaranteed stability or safety in the actual continuous-time dynamics.

Figures

Figures reproduced from arXiv: 2604.28108 by Bingzhuo Zhong, Majid Zamani, Murat Arcak, Shuo Li, Zhiyuan Huang.

Figure 1
Figure 1. Figure 1: Hierarchical control system architecture. view at source ↗
Figure 3
Figure 3. Figure 3: The simulation result of the proposed approach. view at source ↗
Figure 4
Figure 4. Figure 4: The simulation result by implementing control view at source ↗
Figure 2
Figure 2. Figure 2: The simulation result of the proposed approach. view at source ↗
read the original abstract

This paper introduces a general approximate alternating simulation relation (\emph{$\varepsilon$-gAAS relation}) for continuous-time systems, which relaxes existing simulation relations to tolerate larger mismatches between abstract and concrete models. The definition of gAAS for continuous-time systems is first proposed, and its properties are investigated. Then, a control refinement method is developed to enable hierarchical control for the gAAS relation. Finally, case studies demonstrate the effectiveness of the proposed approach, highlighting its advantages over existing methods.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 3 minor

Summary. The manuscript introduces the ε-gAAS relation as a relaxation of existing approximate alternating simulation relations for continuous-time systems. It first defines the gAAS relation for CT dynamics, investigates its properties (including transitivity and compositionality), develops a control refinement procedure that maps abstract controllers to concrete ones while ensuring the mismatch remains bounded by ε, and validates the approach via case studies on linear and nonlinear systems that demonstrate advantages in terms of allowable abstraction error and computational efficiency over prior methods.

Significance. If the central claims hold, the work provides a useful generalization of simulation relations that tolerates larger mismatches while still enabling property-preserving hierarchical control for continuous-time systems. The explicit construction of the refinement map and the property proofs are strengths; the case studies supply concrete evidence of practical utility. This could meaningfully extend the toolkit for abstraction-based synthesis in cyber-physical systems, particularly where strict bisimulations are too conservative.

major comments (2)
  1. [§4.1, Definition 3] §4.1, Definition 3: The ε-gAAS relation relaxes the standard alternating simulation condition by allowing an ε-ball around the abstract trajectory; however, the error propagation argument in the subsequent theorem relies on a uniform Lipschitz constant for the vector field. This assumption is load-bearing for the closed-loop stability guarantee and should be stated explicitly as a hypothesis rather than left implicit in the trajectory-matching clause.
  2. [§5.2, Theorem 4] §5.2, Theorem 4: The control refinement procedure claims that any controller synthesized on the abstract system yields a concrete closed-loop trajectory whose distance to the abstract one is at most ε. The proof sketch uses a Gronwall-type inequality, but the constant in the bound appears to depend on the time horizon; the manuscript should clarify whether the result is uniform in time or only holds for finite horizons, as this directly affects safety specifications.
minor comments (3)
  1. [Introduction] The notation for the relation (ε-gAAS) is introduced in the abstract but first formally defined only in §3; a brief reminder of the symbol in the introduction would improve readability.
  2. [Case Studies] Figure 3 (nonlinear case study): the y-axis label on the error plot is missing the unit; adding it would make the quantitative comparison with the baseline method clearer.
  3. [Related Work] Several references to prior work on approximate bisimulations (e.g., Girard et al.) are cited but the precise difference in the relaxation parameter is not tabulated; a short comparison table would help readers assess the generalization.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive assessment and the detailed comments on our manuscript. We have carefully considered each point and provide point-by-point responses below. Revisions will be made to enhance the clarity of the assumptions and the scope of the theoretical guarantees.

read point-by-point responses

  1. Referee: [§4.1, Definition 3] §4.1, Definition 3: The ε-gAAS relation relaxes the standard alternating simulation condition by allowing an ε-ball around the abstract trajectory; however, the error propagation argument in the subsequent theorem relies on a uniform Lipschitz constant for the vector field. This assumption is load-bearing for the closed-loop stability guarantee and should be stated explicitly as a hypothesis rather than left implicit in the trajectory-matching clause.

    Authors: We appreciate the referee's observation. The uniform Lipschitz continuity of the vector field is indeed a key assumption underlying the error propagation in the theorem following Definition 3. While it was implicitly used in the trajectory-matching condition, we agree that making it explicit strengthens the presentation. In the revised manuscript, we will introduce it as a standing Assumption in §4.1, before Definition 3, and reference it clearly in the theorem statement. This change will be reflected in the updated version. revision: yes

  2. Referee: [§5.2, Theorem 4] §5.2, Theorem 4: The control refinement procedure claims that any controller synthesized on the abstract system yields a concrete closed-loop trajectory whose distance to the abstract one is at most ε. The proof sketch uses a Gronwall-type inequality, but the constant in the bound appears to depend on the time horizon; the manuscript should clarify whether the result is uniform in time or only holds for finite horizons, as this directly affects safety specifications.

    Authors: Thank you for highlighting this important aspect. The application of the Gronwall inequality in the proof of Theorem 4 does produce a bound whose multiplicative constant depends on the time horizon T, typically of the form involving e^{LT} for Lipschitz constant L. Consequently, the guarantee that the distance remains at most ε holds for finite time horizons, with the effective bound scaling with T. We will revise the manuscript to explicitly state that Theorem 4 provides a finite-horizon guarantee. For infinite-horizon safety specifications, we note that the result can be applied over finite intervals or under additional assumptions ensuring non-expansive error propagation (e.g., when the closed-loop dynamics are contracting). The revised text will include a remark discussing these implications for safety properties to avoid any ambiguity. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper introduces a novel definition of the ε-gAAS relation for continuous-time systems, derives its properties via direct mathematical analysis from that definition, and constructs a control refinement procedure that maps abstract controllers to concrete ones while bounding mismatch by ε. These steps rely on standard simulation relation theory and explicit proofs rather than reducing to self-citations, fitted parameters renamed as predictions, or ansatzes smuggled via prior work. Case studies provide empirical validation without the central claims being forced by construction from the inputs. The derivation chain is self-contained against external benchmarks in control theory.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Review based on abstract only; full text unavailable for detailed audit. The primary new element is the ε-gAAS relation itself as a conceptual entity. No free parameters, standard mathematical axioms, or other invented entities are identifiable from the given information.

invented entities (1)
  • ε-gAAS relation no independent evidence
    purpose: Relaxed approximate alternating simulation allowing larger mismatches for hierarchical control in continuous-time systems
    Newly introduced in the paper per the abstract; no independent evidence or falsifiable handle provided in abstract.

pith-pipeline@v0.9.0 · 5384 in / 1211 out tokens · 62042 ms · 2026-05-07T06:40:47.823422+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

17 extracted references · 1 canonical work pages

  1. [1]

    Farhat, H. (2020). Control of nondeterministic systems for bisimulation equivalence under partial information. IEEE Transactions on Automatic Control, 65(12), 5437--5443

    2020

  2. [2]

    and Pappas, G.J

    Girard, A. and Pappas, G.J. (2006). Hierarchical control using approximate simulation relations. In Proceedings of the 45th IEEE Conference on Decision and Control, 264--269. IEEE

    2006

  3. [3]

    and Pappas, G.J

    Girard, A. and Pappas, G.J. (2007). Approximation metrics for discrete and continuous systems. IEEE Transactions on Automatic Control, 52(5), 782--798

    2007

  4. [4]

    and Pappas, G.J

    Girard, A. and Pappas, G.J. (2009). Hierarchical control system design using approximate simulation. Automatica, 45(2), 566--571

    2009

  5. [5]

    and Pappas, G.J

    Girard, A. and Pappas, G.J. (2011). Approximate bisimulation: A bridge between computer science and control theory. European Journal of Control, 17(5-6), 568--578

    2011

  6. [6]

    and Grizzle, J.W

    Khalil, H.K. and Grizzle, J.W. (2002). Nonlinear systems, volume 3. Prentice hall Upper Saddle River, NJ

    2002

  7. [7]

    Nadali, A., Zhong, B., Trivedi, A., and Zamani, M. (2024). Transfer learning for control systems via neural simulation relations. arXiv preprint arXiv:2412.01783

    work page arXiv 2024

  8. [8]

    Pappas, G.J. (2003). Bisimilar linear systems. Automatica, 39(12), 2035--2047

    2003

  9. [9]

    Reissig, G., Weber, A., and Rungger, M. (2016). Feedback refinement relations for the synthesis of symbolic controllers. IEEE Transactions on Automatic Control, 62(4), 1781--1796

    2016

  10. [10]

    Smith, S.W., Yin, H., and Arcak, M. (2019). Continuous abstraction of nonlinear systems using sum-of-squares programming. In 2019 IEEE 58th Conference on Decision and Control (CDC), 8093--8098. IEEE

    2019

  11. [11]

    Tabuada, P. (2009). Verification and control of hybrid systems: a symbolic approach. Springer Science & Business Media

    2009

  12. [12]

    Van der Schaft, A. (2004). Equivalence of dynamical systems by bisimulation. IEEE transactions on automatic control, 49(12), 2160--2172

    2004

  13. [13]

    and Zamani, M

    Zhang, K. and Zamani, M. (2017). Infinite-step opacity of nondeterministic finite transition systems: A bisimulation relation approach. In 2017 IEEE 56th Annual Conference on Decision and Control (CDC), 5615--5619. IEEE

    2017

  14. [14]

    Zhong, B., Arcak, M., and Zamani, M. (2024). Hierarchical control for cyber-physical systems via general approximate alternating simulation relations. IFAC-PapersOnLine, 58(11), 13--18

    2024

  15. [15]

    Zhong, B., Lavaei, A., Zamani, M., and Caccamo, M. (2023). Automata-based controller synthesis for stochastic systems: A game framework via approximate probabilistic relations. Automatica, 147, 110696

    2023

  16. [16]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1 'mid.sent...

  17. [17]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...