Pith. sign in

REVIEW 2 major objections 5 minor 13 references

Energy-stable Port-Hamiltonian Systems

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper introduces energy-stable port-Hamiltonian (es-pH) systems, a reformulation of port-Hamiltonian systems in which Galerkin projection and Petrov-Galerkin time discretization automatically preserve the system's power balance and…

desk verdict A clean but narrow extension of energy-stable systems to port-Hamiltonian ones; the main theorem is correct, but the time-discretization proof has a genuine gap and the class does not cover most standard pH systems. read the letter →

arxiv 2506.06471 v1 pith:K34TZQUK submitted 2025-06-06 math.NA cs.NA

Patrick Buchfink , Silke Glas , Hans Zwart This is my paper
classification math.NAcs.NA MSC 37M1565P10
keywords energy-stablesystemsport-Hamiltonianstructure-preservingdiscretizationmodelreductionDiracstructuredissipationinequalitypowerbalanceGalerkinprojection
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 tries to establish a new formalism, energy-stable port-Hamiltonian (es-pH) systems, that merges the energy-stable framework with port-Hamiltonian (pH) systems by adding an input-output port. The payoff is structural: a Galerkin projection in space or a Petrov-Galerkin projection in time of an es-pH system is again an es-pH system, and the power balance inequality $\frac{d}{dt}H(x)\le\langle y,u\rangle$ is inherited automatically. The paper proves this inequality for the infinite-dimensional formulation and shows that every es-pH system is a port-Hamiltonian system via a Dirac structure. If the framework covers the pH systems one wants to simulate, it would make structure-preserving discretization and model reduction essentially free, since no special integrator or reduction algorithm is needed beyond projection.

What carries the argument

The central object is the es-pH system — a port-Hamiltonian system whose structure and dissipation matrices act on $(\dot{x},u)$ rather than on $(\partial H/\partial x,u)$ — in its variational form (2.5): for every test pair $(v,v_{IO})$, $\langle [-\partial H/\partial x; y] + (\lambda-\phi)[\dot{x};u], [v;v_{IO}]\rangle = 0$. The mechanism is to test this equation with $v=\dot{x}$ and $v_{IO}=u$; the skew-symmetric $\lambda$ term then vanishes by symmetry, the positive-semidefinite $\phi$ term contributes a nonpositive dissipation, and the chain rule turns the left side into $\frac{d}{dt}H(x)$, giving the power balance and dissipation inequality. Since the entire structure is captured by this bilinear form, restricting to subspaces or temporal test spaces preserves the form, so any Galerkin or compatible Petrov-Galerkin projection is automatically structure-preserving.

What would settle it

Take a standard input-state-output port-Hamiltonian system whose output depends on $\partial H/\partial x$ and $u$, for example $y = G^\top\partial H/\partial x + S u$, and check whether there exist coefficient blocks $\omega,\rho,\gamma,\pi,\mu,\sigma$ that cast it into the es-pH form (1.2); a concrete system for which this is provably impossible would refute the claim that the es-pH formulation offers natural structure-preserving discretization for the pH systems one wants to simulate.

Watch

Extended reading notes

Core claim

The central discovery is a new class of open, dissipative systems, the energy-stable port-Hamiltonian (es-pH) systems, defined by pairing the dynamics $(-\omega(x)+\rho(x))\dot{x} = -\partial H/\partial x + (\gamma(x)-\pi(x))u$ with the output equation $y = (\gamma(x)^\top+\pi(x)^\top)\dot{x} + (-\mu(x)+\sigma(x))u$, where the combined block matrices $\lambda$ and $\phi$ are pointwise skew-symmetric and symmetric positive semidefinite, respectively. Theorem 2.1 shows that every classical solution obeys the power balance identity $\frac{d}{dt}H(x) = -\langle \phi[\dot{x};u],[\dot{x};u]\rangle + \langle y,u\rangle$ and hence the dissipation inequality $\frac{d}{dt}H(x) \le \langle y,u\rangle$. Because the system is encoded as a single variational equation in $(\dot{x},u)$, a Galerkin projection onto closed subspaces of the state and input-output spaces produces another es-pH system, and the same structure is shown to survive a Petrov-Galerkin time discretization when the temporal subspaces satisfy a compatibility condition. Section 3.3 reformulates the es-pH system on a Dirac structure, proving it is a port-Hamiltonian system in the established sense.

Load-bearing premise

The load-bearing premise is that the port-Hamiltonian systems one actually wants to simulate can be written in the es-pH form; the paper shows every es-pH system is a port-Hamiltonian system, but it does not show that every port-Hamiltonian system admits an es-pH representation.

Editorial extensions

If this is right

  • Spatial discretization of an es-pH system by a Galerkin projection onto closed subspaces yields a reduced model that is again es-pH, so the discrete model inherits the power balance and dissipation inequality without adding stabilization or correction terms.
  • Temporal discretization by a Petrov-Galerkin method, under the subspace condition $\frac{dx_N}{dt}\in W_N$ and $y_N\in W_{IO,N}$, preserves a discrete energy balance, extending discrete-gradient and average-vector-field ideas to systems with an input-output port.
  • Because the es-pH class is a subclass of port-Hamiltonian systems via the Dirac-structure construction, all existing pH analysis applies while the new form adds projection-based structure preservation.
  • The Banach-space formulation covers PDEs, ODEs, and DAEs, so the same structure-preserving projection framework applies to wave and elasticity equations, shallow water equations, and magneto-quasistatics.
  • Model reduction of parametric es-pH systems is obtained by the same Galerkin projection, as noted in the paper, although the parametric details are left for future work.

Reading between the lines

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

  • The paper proves only that every es-pH system is a pH system; whether every standard pH system can be rewritten in es-pH form is left open, so the practical reach of the framework depends on the size of this subclass.
  • The time-discretization result relies on a compatibility condition between temporal subspaces that the paper assumes rather than verifies; a concrete map from standard time-stepping methods (discrete gradients, average vector field, collocation methods) to subspaces satisfying it would turn the framework into an algorithm.
  • Because the output in es-pH form is driven by $(\dot{x},u)$ rather than by $(\partial H/\partial x,u)$, applying the framework to a mechanical or electrical system may require reinterpreting what the output port measures; this modeling step is not addressed in the paper.
  • A natural next step is to test the framework numerically on one of the listed applications, such as shallow water or magneto-quasistatics, and compare the discrete energy balance of the projected es-pH scheme against a standard pH discretization.
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

2 major / 5 minor

Summary. The manuscript introduces energy-stable port-Hamiltonian (es-pH) systems as an extension of energy-stable systems by an input–output port. It gives three formulations: a finite-dimensional matrix form (1.2), a Banach-space variational form (2.2)–(2.5), and a finite-dimensional Dirac-structure form (3.6). Theorem 2.1 establishes a power balance equation and a dissipation inequality for classical solutions. The paper then claims that Galerkin projection in space, model-order reduction, and a Petrov–Galerkin time discretization all preserve the es-pH structure, and that the construction is a genuine pH system via a Dirac structure.

Significance. If the main structural claims hold, the es-pH formulation is attractive because structure preservation would follow from a plain (Petrov-)Galerkin projection, without additional correction terms or fitting. The proof of Theorem 2.1 is concise and correct, and the Dirac-structure embedding in Section 3.3 is a faithful mapping rather than a circular restatement of the desired result. The paper contains no fitted parameters and the central algebraic conditions are stated in a checkable way. The main reservation is that the temporal-discretization claim advertised in the abstract and discussed in Section 2.5 is not established as written; until that is repaired, the claimed advantage over existing pH discretization techniques is fully supported only for the spatial-discretization and MOR settings.

major comments (2)
  1. [§2.5 (Eq. (2.8))] The discrete energy balance does not follow from the stated hypotheses. Reproducing the proof of Theorem 2.1 requires inserting the test pair (v,v_IO) = (d_t x_N, u_N) into (2.8). The paper assumes d_t x_N ∈ W_N, but u_N is only assumed to belong to V_IO,N; the stated condition y_N ∈ W_IO,N does not make u_N an admissible test function in the output test space. Consequently the displayed EBE and integrated DI after (2.8) are unproven as stated. The admissibility condition should be corrected (for example, by requiring u_N ∈ W_IO_N, or by proving an appropriate compatibility condition), and a concrete example of temporal spaces satisfying it should be supplied; none of the methods cited from [3] is shown to satisfy the condition in the port setting.
  2. [Abstract and §2.5] The abstract promises structure preservation 'in space and time' and in model reduction. The spatial and MOR parts are rigorously supported by Theorem 2.1 and the Galerkin argument in §2.4. The temporal part, however, depends entirely on the unverified admissibility condition discussed in the previous comment. Because no concrete temporal spaces are constructed or verified, the paper does not currently deliver on the time-discretization portion of the headline claim. The authors should either supply the missing construction or explicitly restrict the claim to spatial discretization and model reduction.
minor comments (5)
  1. [§2.4] The output y is consistently an element of the dual space V'_IO, see Eq. (2.3); the statement after (2.5) that the Galerkin output satisfies ~y ∈ C^1(It, ~V_IO) should instead read ~y ∈ C^1(It, ~V'_IO), with the usual identification in the finite-dimensional setting.
  2. [§2.5 (displayed EBE)] In the discrete energy balance displayed after (2.8), the integrand uses x(t), u(t), and φ|_{x(t)} rather than the discrete curves x_N(t), u_N(t); this should be corrected so that the discrete and continuous arguments are not confused.
  3. [§3.3 (Eq. (3.6))] It should be stated explicitly that L|x is skew-symmetric with respect to the duality pairing (3.2), which is the property that makes the graph D_es a Dirac structure; the current presentation leaves this essential verification implicit.
  4. [§3.2 and §3.3] The port-variable convention is reversed between the iso-pH and es-pH Dirac formulations: in §3.2 one sets f_FIO = y and e_FIO = u, while in §3.3 one sets f_FIO = u and e_FIO = y. The authors should warn the reader explicitly, since this switch affects the signs of the feedthrough terms and the interpretation of passivity.
  5. [Conclusion] The paper states that a numerical study is future work. For a methods paper this is acceptable, but a small illustrative example verifying the temporal admissibility condition would substantially increase confidence in the claimed structure preservation in time.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the PBE/DI follow from explicit defining structural assumptions, the Dirac-structure embedding is faithful, and the Galerkin closure is a direct computation; the only flagged gap (Section 2.5 temporal EBE) is an omitted proof condition, not a circular reduction.

full rationale

The paper's derivation chain is self-contained. Theorem 2.1 derives the PBE/DI (2.6) from the defining properties of an es-pH system: evaluating the variational form (2.5) at (dx/dt, u), the lambda term vanishes by the skew-symmetry built into the definition in Section 2.3 and the phi term is bounded above by zero by the defining positive (semi-)definiteness. This is a legitimate derivation from explicitly stated structural assumptions, not a circular reduction: the DI is not part of the definition of es-pH. Section 3.3 shows es-pH systems are pH systems by constructing a Dirac structure D_es = {(f,e) : e = L f} with L skew-symmetric; such a graph satisfies (3.1)(i) by skew-symmetry and (3.1)(ii) by the graph-dimension theorem, and the resistive relation f_FR = -phi-hat e_FR reproduces (2.2)-(2.3) verbatim. That is a faithful embedding of one formalism into another, not a renaming that presupposes the conclusion. The Galerkin closure claim (1.3)/(Section 2.4) is a direct algebraic computation: congruence with V preserves the block structure (omega, rho, gamma, pi, mu, sigma) and the skew-symmetry/PSD of lambda and phi, so the reduced system is again es-pH by construction. No parameter fitting occurs anywhere, no external benchmark is invoked, and the only self-citation is [11] (an author appears among eleven authors of an application paper cited only for 'the applications listed therein'), which is purely motivational and not load-bearing. One genuine caveat, flagged for completeness: Section 2.5 claims an EBE holds under 'If d/dt x_N(.) in W_N and y(.) in W_IO,N,' but reproducing the proof of Theorem 2.1 requires testing (2.8) with v_IO = u_N, and only u_N in V_IO,N is assumed; the stated condition y in W_IO,N (with y valued in V'_IO, not V_IO) does not supply the needed inclusion u_N in W_IO,N. This is an unproven conditional - an omitted proof that Section 2.5 itself marks with 'If' - not a circular step, since the discrete EBE is not assumed among its own hypotheses. The paper also states upfront that existence and uniqueness are out of scope. Verdict: no significant circularity; the temporal-discretization concern is a correctness gap, not a circularity.

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

The es-pH system is a novel formalism, but introduces no fitted constants or physical entities. The derivations rely on standard functional analysis and the assumption that target systems can be represented in this form; the temporal discretization result further requires a discrete-inclusion condition that is stated but not constructed. These are the costs the reader pays beyond the abstract definitions.

assumptions (5)
  • domain assumption Energy H is C^1 and solution curves are C^1 so the chain rule applies in time.
    Theorem 2.1 uses d/dt H(x(t)) = <∂H/∂x, dx/dt>; C^1 regularity is assumed in Section 2.2-2.3 and again for temporal discretization.
  • domain assumption For any system of interest there exist pointwise bounded operators ω,ρ,γ,π,μ,σ such that λ is skew-symmetric and φ is symmetric positive semidefinite.
    This is the definition of an es-pH system (Section 2.3). The paper does not prove that arbitrary pH systems admit such a representation.
  • domain assumption The projected operators and reduced Hamiltonian retain the required symmetry/definiteness and regularity after Galerkin projection.
    Section 2.4 assumes the reduced model (1.3) is again es-pH; the congruence argument preserves algebraic properties but nonlinear state-dependence and regularity are inherited without proof.
  • ad hoc to paper In the temporal Petrov-Galerkin setting, the discrete derivative lies in the test space and the discrete output lies in the test output space.
    Section 2.5 obtains the discrete EBE only under dx_N/dt ∈ W_N and y_N ∈ W_IO,N; no construction of spaces satisfying this is provided.
  • standard math The graph of a skew-symmetric operator on a finite-dimensional space is a Dirac structure (power-conserving and maximal dimension).
    Used in Section 3.1-3.3 to justify that the constructed relation is a Dirac structure.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Energy-stable Port-Hamiltonian Systems." pith.science (2026). https://pith.science/paper/K34TZQUK

@misc{pith2026250606471,
  author       = {Pith},
  title        = {Pith review of: Energy-stable Port-Hamiltonian Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K34TZQUK}},
  note         = {Machine review of arXiv:2506.06471}
}
read the original abstract

We combine energy-stable and port-Hamiltonian (pH) systems to obtain energy-stable port-Hamiltonian (espH) systems. The idea is to extend the known energy-stable systems with an input-output port, which results in a pH formulation. One advantage of the new espH formulation is that it naturally preserves its espH structure throughout discretization (in space and time) and model reduction.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

13 extracted references · 12 canonical work pages

  1. [3]

    Egger, O

    H. Egger, O. Habrich, and V. Shashkov . On the energy stable approximation of Hamiltonian and gradient systems.Computational Methods in Applied Mathematics, 21(2):335–349, 2021

    work page 2021

  2. [1]

    T. J. Bridges and S. Reich . Numerical methods for Hamiltonian PDEs.Journal of Physics A: Mathematical and General, 39(19):5287–5320, apr 2006

    work page 2006

  3. [2]

    Chaturantabut, C

    S. Chaturantabut, C. Beattie, and S. Gugercin . Structure-preserving model reduction for nonlinear port-Hamiltonian systems.SIAM Journal on Scientific Computing, 38(5):B837–B865, 2016. 10 P. BUCHFINK, S. GLAS, AND H. ZW ART

    work page 2016

  4. [4]

    Giesselmann, A

    J. Giesselmann, A. Karsai, and T. Tscherpel . Energy-consistent Petrov–Galerkin time discretiza- tion of port-Hamiltonian systems, 2024

    work page 2024

  5. [5]

    P. L. Kinon, T. Thoma, P. Betsch, and P. Kotyczka . Port-Hamiltonian formulation and structure-preserving discretization of hyperelastic strings.arXiv preprint arXiv:2304.10957, 2023

    arXiv 2023

  6. [6]

    Liljegren-Sailer and N

    B. Liljegren-Sailer and N. Marheineke . Structure-preserving Galerkin approximation for a class of nonlinear port-Hamiltonian partial differential equations on networks.PAMM, 19(1):e201900399, 2019

    work page 2019

  7. [7]

    C. Lubich. From Quantum to Classical Molecular Dynamics: Reduced Models and Numerical Analysis. Zurich lectures in advanced mathematics. European Mathematical Society, 2008

    work page 2008

  8. [8]

    J. E. Marsden, A. Weinstein, T. S. Ratiu, R. Schmid, and R. Spencer . Hamiltonian systems with symmetry, coadjoint orbits and plasma physics. InProceedings of the IUTAM-ISIMM symposium on modern developments in analytical mechanics, volume 117, pages 289–340, 1983

    work page 1983

Show all 13 references
  1. [9]

    Mehrmann and B

    V. Mehrmann and B. Unger . Control of port-Hamiltonian differential-algebraic systems and applications. Acta Numerica, 32:395–515, 2023

    2023

  2. [10]

    R. V. Polyuga and A. J. van der Schaft . Effort-and flow-constraint reduction methods for structure preserving model reduction of port-Hamiltonian systems.Systems & Control Letters, 61(3):412–421, 2012

    2012

  3. [11]

    Rettberg, D

    J. Rettberg, D. Wittwar, P. Buchfink, A. Brauchler, P. Ziegler, J. Fehr, and B. Haasdonk . Port-Hamiltonian fluid-structure interaction modelling and structure-preserving model order reduction of a classical guitar.Mathematical and Computer Modelling of Dynamical Systems, 29(1...

    2023

  4. [12]

    Schwerdtner, T

    P. Schwerdtner, T. Moser, V. Mehrmann, and M. Voigt . Optimization-based model order reduction of port-Hamiltonian descriptor systems.Systems & Control Letters, 182:105655, 2023

    2023

  5. [13]

    A. J. van der Schaft and D. Jeltsema . Port-Hamiltonian systems theory: An introductory overview. Foundations and Trends® in Systems and Control, 1(2-3):173–378, 2014. † Department of Applied Mathematics, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands Em...

    2014

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.