pith. sign in

arxiv: 2106.01813 · v4 · submitted 2021-06-03 · 📡 eess.SY · cs.SY

Identification of diffusively coupled linear networks through structured polynomial models

E.M.M. (Lizan) Kivits , Paul M.J. Van den Hof This is my paper

Pith reviewed 2026-05-24 12:30 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords diffusively coupled networksprediction error identificationpolynomial modelsinterconnection structurelinear time-invariant systemsconvex optimizationundirected graphs
0
0 comments X

The pith

Prediction error methods for polynomial models can consistently identify parameters and interconnection structures in diffusively coupled networks.

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

The paper shows that existing prediction error identification techniques for linear time-invariant systems can be set up to recover both the numerical values and the connection pattern in networks whose links obey diffusive coupling rules. These rules produce symmetric interactions that fit undirected graphs, as is typical for many physical component networks. The work also supplies a multi-step least-squares procedure that turns the resulting nonconvex fitting task into a sequence of convex problems. A reader would care because accurate structure recovery matters for simulation, monitoring, and control of interconnected physical systems.

Core claim

Prediction error identification methods developed for linear time-invariant systems in polynomial form can be configured to consistently identify the parameters and the interconnection structure of diffusively coupled networks, with a multi-step least squares convex optimization algorithm developed to solve the nonconvex optimization problem that results.

What carries the argument

Structured polynomial models that embed the symmetry constraints of diffusive couplings to allow joint estimation of parameters and undirected interconnection structure.

If this is right

  • Both network parameters and the undirected interconnection graph can be recovered consistently from input-output data.
  • The nonconvex identification problem admits a solution via successive convex least-squares steps.
  • Existing polynomial-form prediction error methods extend to the diffusively coupled case without requiring a priori knowledge of the topology.

Where Pith is reading between the lines

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

  • The same structuring idea might allow topology learning in other symmetric coupling settings such as certain electrical or thermal networks.
  • Online versions could support adaptive monitoring where the interconnection pattern changes over time.
  • Simulation studies on networks with known ground-truth structure would provide direct numerical checks of the consistency property.

Load-bearing premise

Physical dynamic networks consist of diffusive couplings that produce symmetric cause-effect relationships representable by undirected graphs.

What would settle it

Running the configured method on data from a network whose couplings are known to be asymmetric and finding that the recovered structure or parameters are inconsistent would show the claim does not hold.

Figures

Figures reproduced from arXiv: 2106.01813 by E.M.M. (Lizan) Kivits, Paul M.J. Van den Hof.

Figure 1
Figure 1. Figure 1: A network of masses (Mj0), dampers (Djk) and springs (Kjk). wj (t), respectively. The diffusive type of coupling induces the symmetry constraints Djk = Dkj and Kjk = Kkj ∀j, k. An obvious physical example of such a network is the mass￾spring-damper system shown in [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A physical network as defined in Definition 1, with nodes [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The continuous-time network model with interconnection dynamics [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Boxplot of the relative estimation errors of the parameters of [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: The true parameter values (blue) and the mean of the estimated [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: A module representation of a diffusively coupled network with three [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
read the original abstract

Physical dynamic networks most commonly consist of interconnections of physical components that can be described by diffusive couplings. These diffusive couplings imply that the cause-effect relationships in the interconnections are symmetric and therefore physical dynamic networks can be represented by undirected graphs. This paper shows how prediction error identification methods developed for linear time-invariant systems in polynomial form can be configured to consistently identify the parameters and the interconnection structure of diffusively coupled networks. Further, a multi-step least squares convex optimization algorithm is developed to solve the nonconvex optimization problem that results from the identification method.

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 / 3 minor

Summary. The manuscript claims that prediction error identification methods for linear time-invariant systems expressed in polynomial form can be configured to achieve consistent identification of both parameters and interconnection structure for diffusively coupled networks (represented via undirected graphs due to symmetric cause-effect relations). It further develops a multi-step least-squares convex optimization algorithm to address the nonconvex problem that arises from the structured identification criterion.

Significance. If the consistency result and algorithm hold, the contribution lies in extending established prediction-error theory to structured network identification without introducing new identifiability conditions or ad-hoc entities. The diffusive-coupling premise is taken as modeling foundation rather than derived, and the method preserves the external consistency guarantees of the underlying framework while providing a practical convex procedure.

minor comments (3)
  1. §2, the transition from the general polynomial LTI predictor to the diffusive-coupling constraint (symmetric adjacency) should include an explicit statement of how the parameter vector is reparameterized to enforce the undirected-graph structure; the current wording leaves the mapping implicit.
  2. Algorithm 1, step 3: the least-squares subproblem is stated to be convex, but the paper should clarify whether the weighting matrix is fixed from a prior step or updated iteratively, as this affects the overall convergence argument.
  3. The numerical example section would benefit from a table reporting both parameter error and structure recovery rate (e.g., false-positive edges) across Monte-Carlo runs, rather than a single realization.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive evaluation of the manuscript, the accurate summary of its contributions, and the recommendation for minor revision. No specific major comments were provided in the report.

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper applies established external prediction-error identification theory for LTI systems in polynomial form to the class of diffusively coupled networks. The diffusive-coupling premise is introduced as a standard modeling assumption for physical networks rather than derived from the identification procedure itself. The consistency claim for parameter and structure recovery is presented as a configuration of prior methods, without any reduction of the target result to a fitted quantity defined by the method or to a self-citation chain. No load-bearing self-citations, self-definitional steps, or ansatzes smuggled via citation are indicated.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review; ledger populated from explicit statements in the provided abstract. The central claim rests on the diffusive-coupling symmetry assumption and standard LTI polynomial identification theory.

axioms (1)
  • domain assumption Diffusive couplings imply symmetric cause-effect relationships that allow representation by undirected graphs
    Stated in the first two sentences of the abstract as the modeling premise for physical dynamic networks.

pith-pipeline@v0.9.0 · 5618 in / 1149 out tokens · 30394 ms · 2026-05-24T12:30:56.712620+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    Reduction of second- order network systems with structure preservation,

    X. Cheng, Y . Kawano, and J. M. A. Scherpen, “Reduction of second- order network systems with structure preservation,” IEEE Transactions on Automatic Control , vol. 62, no. 10, pp. 5026–5038, 2017

    work page 2017

  2. [2]

    Kron reduction of graphs with applications to electrical networks,

    F. D ¨orfler and F. Bullo, “Kron reduction of graphs with applications to electrical networks,” IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 60, no. 1, pp. 150–163, 2013

    work page 2013

  3. [3]

    Electrical networks and algebraic graph theory: Models, properties, and applications,

    F. D ¨orfler, J. W. Simpson-Porco, and F. Bullo, “Electrical networks and algebraic graph theory: Models, properties, and applications,” Proceed- ings of the IEEE , vol. 106, no. 5, pp. 977–1005, 2018

    work page 2018

  4. [4]

    Ljung, System identification: theory for the user

    L. Ljung, System identification: theory for the user . Englewood Cliffs, NJ: Prentice-Hall, 1999

    work page 1999

  5. [5]

    Extracting second order systems from state space representations,

    M. Friswell, S. Garvey, and J. Penny, “Extracting second order systems from state space representations,” AIAA Journal, vol. 37, no. 1, pp. 132– 135, 1999

    work page 1999

  6. [6]

    Deriving mechanical structures in physical coor- dinates from data-driven state-space realizations,

    P. Lopes dos Santos, J. A. Ramos, T. Azevedo-Perdico ´ulis, and J. L. Martins de Carvalho, “Deriving mechanical structures in physical coor- dinates from data-driven state-space realizations,” in Proceedings of the 2015 American Control Conference (ACC) , 2015, pp. 1107–1112

    work page 2015

  7. [7]

    Identifying second-order mod- els of mechanical structures in physical coordinates: an orthogonal complement approach,

    J. Ramos, G. Merc `ere, and O. Prot, “Identifying second-order mod- els of mechanical structures in physical coordinates: an orthogonal complement approach,” in Proceedings of the 12th European Control Conference (ECC), 2013, pp. 3973–3978

    work page 2013

  8. [8]

    Identification of mass, damping, and stiffness matrices of mechanical systems,

    C.-P. Fritzen, “Identification of mass, damping, and stiffness matrices of mechanical systems,” Journal of Vibration, Acoustics, Stress, and Reliability in Design , vol. 108, no. 1, pp. 9–16, 1986

    work page 1986

  9. [9]

    Constructing second-order models of mechanical systems from identified state space realizations. part i: theoretical discussions,

    H. Lus ¸, M. D. Angelis, R. Betti, and R. W. Longman, “Constructing second-order models of mechanical systems from identified state space realizations. part i: theoretical discussions,” Journal of Engineering Mechanics, vol. 129, no. 5, pp. 477–488, 2003

    work page 2003

  10. [10]

    The Mathworks, Inc., Matlab’s System Identification Toolbox , 2021

    work page 2021

  11. [11]

    Model reduction methods for complex network systems,

    X. Cheng and J. M. A. Scherpen, “Model reduction methods for complex network systems,” Annual Reviews of Control, Robotics, and Autonomous Systems, vol. 4, pp. 425–453, 2021

    work page 2021

  12. [12]

    Revealing network connectivity from response dynamics,

    M. Timme, “Revealing network connectivity from response dynamics,” Physical Review Letters , vol. 98, 2007

    work page 2007

  13. [13]

    Network identification via node knockout,

    M. Nabi-Abdolyousefi and M. Mesbahi, “Network identification via node knockout,” IEEE Transactions on Automatic Control , vol. 57, no. 12, pp. 3214–3219, 2012

    work page 2012

  14. [14]

    Subspace identification of large-scale interconnected systems,

    A. Haber and M. Verhaegen, “Subspace identification of large-scale interconnected systems,” IEEE Transactions on Automatic Control , vol. 59, no. 10, pp. 2754–2759, 2014

    work page 2014

  15. [15]

    Dynamical structure functions for the reverse engineering of lti networks,

    J. Gonc ¸alves, R. Howes, and S. Warnick, “Dynamical structure functions for the reverse engineering of lti networks,” in Proceedings of the 46th IEEE Conference on Decision and Control (CDC), 2007, pp. 1516–1522

    work page 2007

  16. [16]

    On the problem of reconstructing an unknown topology via locality properties of the wiener filter,

    D. Materassi and M. V . Salapaka, “On the problem of reconstructing an unknown topology via locality properties of the wiener filter,” IEEE Transactions on Automatic Control, vol. 57, no. 7, pp. 1765–1777, 2012

    work page 2012

  17. [17]

    Topology identification of heterogeneous networks: Identifiability and reconstruction,

    H. J. van Waarde, P. Tesi, and M. K. Camlibel, “Topology identification of heterogeneous networks: Identifiability and reconstruction,” Automat- ica, vol. 123, p. 109331, 2021

    work page 2021

  18. [18]

    Identification of dynamic models in complex networks with prediction error methods-basic methods for consistent module estimates,

    P. M. J. Van den Hof, A. Dankers, P. S. C. Heuberger, and X. Bombois, “Identification of dynamic models in complex networks with prediction error methods-basic methods for consistent module estimates,” Automat- ica, vol. 49, no. 10, pp. 2994–3006, 2013

    work page 2013

  19. [19]

    Parametric identification using weighted null-space fitting,

    M. Galrinho, C. R. Rojas, and H. Hjalmarsson, “Parametric identification using weighted null-space fitting,” IEEE Transactions on Automatic Control, vol. 64, no. 7, pp. 2798–2813, 2019

    work page 2019

  20. [20]

    A dynamic network approach to identification of physical systems,

    E. M. M. Kivits and P. M. J. Van den Hof, “A dynamic network approach to identification of physical systems,” in Proceedings of the 58th IEEE Conference on Decision and Control (CDC) , 2019, pp. 4533–4538

    work page 2019

  21. [21]

    Identification of dynamic models in complex networks with prediction error methods: Predictor input selection,

    A. G. Dankers, P. M. J. Van den Hof, X. Bombois, and P. S. C. Heuberger, “Identification of dynamic models in complex networks with prediction error methods: Predictor input selection,” IEEE Transactions on Automatic Control , vol. 61, no. 4, pp. 937–952, 2016

    work page 2016

  22. [22]

    Mesbahi and M

    M. Mesbahi and M. Egerstedt, Graph theoretic methods in multiagent networks. Princeton University Press, 2010

    work page 2010

  23. [23]

    A class of discrete time models for a continuous time system,

    T. Mori, P. Nikiforuk, M. Gupta, and N. Hori, “A class of discrete time models for a continuous time system,” in Proceedings of the 1987 American Control Conference (ACC) , 1987, pp. 953–957

    work page 1987

  24. [24]

    E. J. Hannan and M. Deistler, The Statistical Theory of Linear Systems . SIAM, 2012

    work page 2012

  25. [25]

    Identifica- tion of dynamic networks operating in the presence of algebraic loops,

    H. H. M. Weerts, P. M. J. Van den Hof, and A. G. Dankers, “Identifica- tion of dynamic networks operating in the presence of algebraic loops,” in Proceedings of the 55th IEEE Conference on Decision and Control (CDC), 2016, pp. 4606–4611

    work page 2016

  26. [26]

    On the factorization of rational matrices,

    D. Youla, “On the factorization of rational matrices,” IRE Transactions on Information Theory , vol. 7, no. 3, pp. 172–189, 1961

    work page 1961

  27. [27]

    Identifi- ability of linear dynamic networks,

    H. H. M. Weerts, P. M. J. Van den Hof, and A. G. Dankers, “Identifi- ability of linear dynamic networks,” Automatica, vol. 89, pp. 247–258, 2018

    work page 2018

  28. [28]

    Kailath, Linear systems

    T. Kailath, Linear systems. Englewood Cliffs, NJ: Prentice-Hall, 1980

    work page 1980

  29. [29]

    The properties of the parameterization of ARMAX systems and their relevance for structural estimation and dynamic specification,

    M. Deistler, “The properties of the parameterization of ARMAX systems and their relevance for structural estimation and dynamic specification,” Econometrica, vol. 51, no. 4, pp. 1187–1207, 1983

    work page 1983

  30. [30]

    Identifiability of dy- namical networks with partial node measurements,

    J. M. Hendrickx, M. Gevers, and A. S. Bazanella, “Identifiability of dy- namical networks with partial node measurements,” IEEE Transactions on Automatic Control , vol. 64, no. 6, pp. 2240–2253, 2019

    work page 2019

  31. [31]

    Allocation of excitation signals for generic identifiability of linear dynamic networks,

    X. Cheng, S. Shi, and P. M. J. Van den Hof, “Allocation of excitation signals for generic identifiability of linear dynamic networks,” IEEE Transactions on Automatic Control , vol. 67, no. 2, pp. 692–705, 2022

    work page 2022

  32. [32]

    A practical method for the consistent identification of a module in a dynamical network,

    M. Gevers, A. S. Bazanella, and G. V . da Silva, “A practical method for the consistent identification of a module in a dynamical network,” IFAC-PapersOnLine, vol. 51, no. 15, pp. 862–867, 2018

    work page 2018

  33. [33]

    A sequential least squares algorithm for ARMAX dynamic network identification,

    H. H. M. Weerts, M. Galrinho, G. Bottegal, H. Hjalmarsson, and P. M. J. Van den Hof, “A sequential least squares algorithm for ARMAX dynamic network identification,” IFAC-PapersOnLine, vol. 51, no. 15, pp. 844–849, 2018

    work page 2018

  34. [34]

    A scalable multi-step least squares method for network identification with unknown disturbance topology,

    S. J. M. Fonken, K. R. Ramaswamy, and P. M. J. Van den Hof, “A scalable multi-step least squares method for network identification with unknown disturbance topology,”Automatica, vol. 141, no. 110295, 2022

    work page 2022

  35. [35]

    Asymptotic properties of the least-squares method for estimating transfer functions and disturbance spectra,

    L. Ljung and B. Wahlberg, “Asymptotic properties of the least-squares method for estimating transfer functions and disturbance spectra,” Ad- vances in Applied Probabilities , vol. 24, no. 2, pp. 412–440, 1992

    work page 1992

  36. [36]

    Least squares methods for system identification of struc- tured models,

    M. Galrinho, “Least squares methods for system identification of struc- tured models,” Ph.D. dissertation, KTH School of Electrical Engineering, 2016

    work page 2016

  37. [37]

    E. K. P. Chong and S. H. ˙Zak, An introduction to optimization , 3rd ed. Wiley Interscience, 2008

    work page 2008

  38. [38]

    Network identifi- cation with partial excitation and measurement,

    A. S. Bazanella, M. Gevers, and J. M. Hendrickx, “Network identifi- cation with partial excitation and measurement,” in Proceedings of the 58th IEEE Conference on Decision and Control (CDC), 2019, pp. 5500– 5506. Lizan Kivits received her bachelor’s degree in Electrical Engineering and her master’s degree in Systems and Control (with great appreciation) fr...

    work page 2019