pith. sign in

arxiv: 2602.11921 · v4 · pith:VZCLPQBZnew · submitted 2026-02-12 · 🧮 math.OC

Relationship Between Controllability Scoring and Optimal Experimental Design

Kazuhiro Sato This is my paper

Pith reviewed 2026-05-21 13:30 UTC · model grok-4.3

classification 🧮 math.OC
keywords controllability scoringoptimal experimental designGramian decompositionD-optimalityA-optimalitynetworked dynamical systemsnode centrality
0
0 comments X

The pith

The finite-time controllability Gramian decomposes additively across nodes, linking controllability scores directly to D-optimality and A-optimality in experimental design.

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

The paper establishes a connection between controllability scoring for nodes in networked dynamical systems and optimal experimental design. The finite-time controllability Gramian breaks down additively by each state node. This creates an affine matrix structure that matches the information matrix used in OED. Consequently, volumetric controllability scores correspond to D-optimality and average energy scores to A-optimality. The analysis also shows controllability scoring usually has unique solutions and a distinctive long-horizon downweighting effect for certain nodes.

Core claim

The finite-time controllability Gramian decomposes additively across nodes, yielding an affine matrix model of the same form as the information-matrix model in OED. This yields a direct correspondence between the volumetric controllability score (VCS) and D-optimality, and between the average energy controllability score (AECS) and A-optimality, implying that the classical D/A invariance gap has a direct analogue in controllability scoring. By contrast, controllability scoring generically admits a unique optimizer, unlike approximate-OED formulations. Finally, source-like state nodes without a negative self-loop can be increasingly downweighted by AECS as the horizon grows.

What carries the argument

Additive decomposition of the finite-time controllability Gramian across state nodes, producing an affine matrix equivalent to the OED information matrix.

If this is right

  • The D/A optimality gap from experimental design applies to controllability measures.
  • Controllability scoring problems have unique optimizers in contrast to OED.
  • Average energy scoring downweights source nodes over long horizons without an OED equivalent.
  • OED methods can be repurposed for selecting important nodes in control systems.

Where Pith is reading between the lines

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

  • This link may enable using OED optimization techniques to solve controllability problems more efficiently.
  • It opens the possibility of joint design where experiments serve both estimation and control objectives.
  • The long-horizon phenomenon could influence how control strategies are planned for extended time periods.

Load-bearing premise

The finite-time controllability Gramian decomposes additively across nodes in the linear networked dynamical system.

What would settle it

Finding a linear system where the controllability Gramian does not add up across nodes or where the scores fail to match the corresponding optimality criteria in computation.

Figures

Figures reproduced from arXiv: 2602.11921 by Kazuhiro Sato.

Figure 1
Figure 1. Figure 1: Network employed in the numerical experiments. [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
read the original abstract

Controllability scores provide control-theoretic centrality measures that quantify the relative importance of state nodes in networked dynamical systems. We establish a structural connection between finite-time controllability scoring and approximate optimal experimental design (OED): the finite-time controllability Gramian decomposes additively across nodes, yielding an affine matrix model of the same form as the information-matrix model in OED. This yields a direct correspondence between the volumetric controllability score (VCS) and D-optimality, and between the average energy controllability score (AECS) and A-optimality, implying that the classical D/A invariance gap has a direct analogue in controllability scoring. By contrast, we point out that controllability scoring generically admits a unique optimizer, unlike approximate-OED formulations. Finally, we uncover a long-horizon phenomenon with no OED counterpart: source-like state nodes without a negative self-loop can be increasingly downweighted by AECS as the horizon grows. Two numerical examples corroborate this long-horizon downweighting behavior.

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 a direct structural correspondence between finite-time controllability scoring and approximate optimal experimental design (OED) for linear networked dynamical systems. The finite-time controllability Gramian decomposes additively across nodes as W(T) = sum_i W_i(T), producing an affine matrix model identical in form to the OED information matrix. This implies that the volumetric controllability score (VCS) corresponds to D-optimality and the average energy controllability score (AECS) to A-optimality. The paper further notes that controllability scoring admits a unique optimizer (unlike approximate OED) and identifies a long-horizon downweighting effect for source-like nodes without negative self-loops, with two numerical examples provided as corroboration.

Significance. If the claimed equivalence holds, the work provides a clean bridge between control-theoretic centrality measures and classical OED criteria, potentially enabling transfer of optimality results and algorithms between the fields. The additive Gramian decomposition is a standard consequence of linearity and the structure of B, lending rigor to the central claim. The long-horizon phenomenon has no direct OED analogue and is a genuine contribution; the numerical examples supply initial empirical support for it.

minor comments (3)
  1. The long-horizon downweighting claim (abstract and final section) is supported only by two numerical examples; a brief analytic argument or explicit condition on the eigenvalues of A would strengthen the result without altering scope.
  2. Notation for the controllability Gramian W(T) and the OED information matrix should be aligned more explicitly when the affine equivalence is stated, to avoid any reader confusion between the two matrix families.
  3. The manuscript would benefit from a short remark on whether the unique-optimizer property survives when the controllability scores are replaced by their OED-optimal counterparts under the same affine model.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their accurate summary of the manuscript, positive significance assessment, and recommendation for minor revision. No specific major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper's central derivation establishes a structural correspondence by showing that the finite-time controllability Gramian decomposes additively as W(T) = sum_i W_i(T) due to linearity of the integral and the form BB^T = sum e_i e_i^T under standard basis inputs. This produces an affine matrix in the node selection vector that matches the information-matrix form in OED, directly mapping VCS to D-optimality and AECS to A-optimality. The steps rely on standard linear-system identities and matrix algebra rather than fitted parameters, self-definitions, or load-bearing self-citations; the paper further contrasts the models by noting uniqueness properties and long-horizon effects absent in OED. The derivation is therefore self-contained against external mathematical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claim rests on standard linear system assumptions for defining the controllability Gramian and its decomposition; no free parameters or new entities are introduced beyond established definitions in control theory and OED.

axioms (1)
  • domain assumption The networked system is linear and time-invariant, allowing definition of the finite-time controllability Gramian.
    Invoked to enable the additive decomposition across nodes as the basis for the OED equivalence.

pith-pipeline@v0.9.0 · 5697 in / 1431 out tokens · 61133 ms · 2026-05-21T13:30:21.720844+00:00 · methodology

Review history (2 revisions) →

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

16 extracted references · 16 canonical work pages · 1 internal anchor

  1. [1]

    On the role of network centrality in the controllability of complex networks,

    N. Bof, G. Baggio, and S. Zampieri, “On the role of network centrality in the controllability of complex networks,”IEEE Transactions on Control of Network Systems, vol. 4, no. 3, pp. 643–653, 2017

    work page 2017

  2. [2]

    Controllability of complex networks,

    Y .-Y . Liu, J.-J. Slotine, and A.-L. Barab´asi, “Controllability of complex networks,”Nature, vol. 473, no. 7346, pp. 167–173, 2011

    work page 2011

  3. [3]

    Controllability metrics, limitations and algorithms for complex networks,

    F. Pasqualetti, S. Zampieri, and F. Bullo, “Controllability metrics, limitations and algorithms for complex networks,”IEEE Transactions on Control of Network Systems, vol. 1, no. 1, pp. 40–52, 2014

    work page 2014

  4. [4]

    On submodularity and controllability in complex dynamical networks,

    T. H. Summers, F. L. Cortesi, and J. Lygeros, “On submodularity and controllability in complex dynamical networks,”IEEE Transactions on Control of Network Systems, vol. 3, no. 1, pp. 91–101, 2016

    work page 2016

  5. [5]

    Controllability of large-scale networks: The control energy exponents,

    G. Baggio and S. Zampieri, “Controllability of large-scale networks: The control energy exponents,”IEEE Transactions on Control of Network Systems, vol. 11, no. 2, pp. 808–820, 2024

    work page 2024

  6. [6]

    Controllability scores for selecting control nodes of large-scale network systems,

    K. Sato and S. Terasaki, “Controllability scores for selecting control nodes of large-scale network systems,”IEEE Transactions on Automatic Control, vol. 69, no. 7, pp. 4673–4680, 2024

    work page 2024

  7. [7]

    Uniqueness Analysis of Controllability Scores and Their Application to Brain Networks,

    K. Sato and R. Kawamura, “Uniqueness Analysis of Controllability Scores and Their Application to Brain Networks,”IEEE Transactions on Control of Network Systems, vol. 12, no. 4, pp. 2568–2580, 2025

    work page 2025

  8. [8]

    Target Controllability Scores for Actuation-Constrained Network Intervention

    K. Sato, “Target controllability score,” 2025. [Online]. Available: https://arxiv.org/abs/2510.13354

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  9. [9]

    The polytope of optimal ap- proximate designs: extending the selection of informative experiments,

    R. Harman, L. Filov ´a, and S. Rosa, “The polytope of optimal ap- proximate designs: extending the selection of informative experiments,” Statistics and Computing, vol. 34, no. 6, p. 211, 2024

    work page 2024

  10. [10]

    Optimal experimental design: Formulations and computations,

    X. Huan, J. Jagalur, and Y . Marzouk, “Optimal experimental design: Formulations and computations,”Acta Numerica, vol. 33, pp. 715–840, 2024

    work page 2024

  11. [11]

    Pukelsheim,Optimal design of experiments

    F. Pukelsheim,Optimal design of experiments. SIAM, 2006

    work page 2006

  12. [12]

    A-optimal versus d-optimal de- sign of screening experiments,

    B. Jones, K. Allen-Moyer, and P. Goos, “A-optimal versus d-optimal de- sign of screening experiments,”Journal of Quality Technology, vol. 53, no. 4, pp. 369–382, 2021

    work page 2021

  13. [13]

    Optimum allocation in linear regression theory,

    G. Elfving, “Optimum allocation in linear regression theory,”The Annals of Mathematical Statistics, pp. 255–262, 1952

    work page 1952

  14. [14]

    Optimum experimental designs,

    J. Kiefer, “Optimum experimental designs,”Journal of the Royal Sta- tistical Society: Series B (Methodological), vol. 21, no. 2, pp. 272–304, 1959

    work page 1959

  15. [15]

    Infinite-horizon controllability scores for linear time-invariant systems,

    K. Umezu and K. Sato, “Infinite-horizon controllability scores for linear time-invariant systems,”arXiv preprint arXiv:2601.10260, 2026

    work page arXiv 2026

  16. [16]

    Zhang,The Schur complement and its applications

    F. Zhang,The Schur complement and its applications. Springer Science & Business Media, 2006, vol. 4

    work page 2006