Pith. sign in

REVIEW 4 major objections 4 minor 33 references

Regularization and passivity-preserving model reduction of quasilinear magneto-quasistatic coupled problems

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The singular components of FEM-discretized quasilinear magneto-quasistatic systems can be projected out, POD model reduction then preserves passivity, and a DEIM-error-sized output perturbation restores passivity when DEIM breaks it.

desk verdict A sound, genuinely useful paper on passivity-preserving MOR for quasilinear MQS systems; the main theorems hold, but the X2 rank hypothesis and a zero-initial-state condition need explicit fixing before acceptance. read the letter →

arxiv 2411.14226 v1 pith:XAY6KFXO submitted 2024-11-21 math.NA cs.NA

Johanna Kerler-Back , Timo Reis , Tatjana Stykel This is my paper · ORCID
classification math.NAcs.NA MSC 12H2015A2234A0937L0578A3093A1593C10
keywords magneto-quasistaticsystemsdifferential-algebraicequationspassivityregularizationmodelorderreductionproperorthogonaldecompositiondiscreteempiricalinterpolationenforcement
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

Low-frequency electromagnetic devices coupled to electrical circuits are modelled by quasilinear magneto-quasistatic (MQS) equations; after finite element (FEM) discretization on a 3D domain, the semidiscrete system is a singular differential-algebraic equation (DAE). The paper shows that a constant coordinate transformation can identify and project out the singular state components, turning the DAE into a regular index-one system and then into an ordinary differential equation without changing its input-output behaviour. The paper then proves that proper orthogonal decomposition (POD) model reduction of this system preserves passivity, i.e. the reduced model never generates energy on its own, which is exactly what is needed for safe coupling to circuit simulators. Because the discrete empirical interpolation method (DEIM) used to accelerate the nonlinear evaluation destroys the symmetric structure behind passivity, the paper restores input-output passivity by perturbing the output by an amount controlled by a computable DEIM error bound.

What carries the argument

The central object is the condensed form of the matrix pencil $\lambda E_r - A_r(x_r)$ under a constant coordinate transformation $W$ applied by congruence, which yields blocks $E_{11}$, $I$, $0$ on the energy side and $A_{11}(x_r)$, $0$, $I$ on the operator side; this splitting separates the regular dynamics from the zero- and infinite-eigenvalue parts, and projecting out the infinite part is what regularizes the DAE. Passivity is carried by the magnetic-energy storage function built from the reluctivity integral $\vartheta(\xi,s)=\int_0^{\sqrt{s}}\nu(\xi,\zeta)\zeta\,d\zeta$, whose dissipation rate is nonnegative because the matrices $M_{11}$ and $R$ are positive definite. For the DEIM correction, the load-bearing estimate is the bound $\|\varepsilon(t)\|\le \theta(t)=\Delta_{\mathrm{DEIM}}\mu^{-1}(e^{\mu t/\lambda_{\min}(E)}-1)$, obtained from a logarithmic Lipschitz constant and an exponential comparison inequality for differential inequalities; $\theta(t)$ sizes the output perturbation.

What would settle it

Construct a 3D MQS finite element model with two windings placed in the same non-conducting region so that the coupling matrix $X_2$ has linearly dependent columns, and recompute the condensed form of Theorem 6 and the index-one criterion of Theorem 5; if the pencil remains regular and the criterion holds, the full-rank condition is dispensable, while if either fails, the regularization guarantee collapses. Alternatively, in the paper's own 2D transformer, search for an admissible input where $\int_0^t u(\tau)^T y_\delta(\tau)\,d\tau<0$ for some $t$ despite the bound (56), which would invalidate the passivity-enforcement claim.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes a chain of structural guarantees for the quasilinear MQS system discretized by edge elements. The singular components of the FEM DAE can be removed by a constant coordinate transformation: Theorem 5 shows the condensed system has tractability index one and Theorem 6 gives a simultaneous block-diagonalization of the pencil separating regular, zero, and infinite parts. The original FEM model, the regularized DAE, and the equivalent ODE are all passive with the magnetic energy as storage function. POD reduction applied to the structured ODE preserves passivity (Theorem 11). When DEIM is used to approximate the nonlinearity, the symmetric structure is lost, so passivity can fail; the paper proves that adding an output perturbation $\delta(t) u$ with $\delta(t)=\|C\|\theta(t)/\|u(t)\|$ restores input-output passivity, where $\theta(t)$ is an explicitly computable DEIM state-error bound (Theorem 15).

Load-bearing premise

The whole chain of guarantees assumes that the winding currents couple to the non-conducting part of the finite element mesh in linearly independent ways, i.e. the matrix $X_2$ has full column rank; the proofs use this condition, but the stated assumptions never list it.

Editorial extensions

If this is right

  • Reduced models of low-frequency electromagnetic devices can be coupled to circuit simulators with a guaranteed nonnegative energy balance, not just with good approximation accuracy.
  • The projection-based regularization gives a systematic route from a singular FEM DAE to an ODE, so standard ODE model reduction techniques apply without tree-cotree or grad-div gauging.
  • POD-reduced MQS models are passive by construction and need no post-processing or passivity enforcement.
  • For POD-DEIM models, passivity enforcement adds only a small, bounded perturbation to the output; in the paper's 2D transformer experiment the perturbed output stays close to the POD-DEIM output.
  • The DEIM state-error bound $\theta(t)$ is computed from reduced matrices only, so the passivity guarantee can be evaluated without re-solving the full FEM model.

Reading between the lines

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

  • The same output-perturbation mechanism should work for any approximation of the nonlinearity that comes with a computable state-error bound, not only for DEIM; any such error bound could be plugged into the $\delta(t)$ formula.
  • If the winding-coupling matrix $X_2$ is rank-deficient in an industrial device, the paper's chain of guarantees would need a fallback, such as a rank-revealing preprocessing or a generalized-inverse regularization.
  • The paper proves input-output passivity of the perturbed DEIM system but does not construct a storage function for it, so a state-space passivity certificate would need an additional reachability or storage-function argument.
  • Because $\theta(t)$ grows exponentially in $\mu t/\lambda_{\min}(E)$, the perturbation may become large over long time horizons; a tighter, problem-adapted bound or a different enforcement strategy would be needed for long-time simulation.
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

4 major / 4 minor

Summary. The manuscript studies the quasilinear magneto-quasistatic FEM system (8) with circuit coupling. It proposes a regularization that projects out singular state components via a condensed form, transforms the regularized DAE into an ODE, proves passivity of the FEM, regularized, and POD-reduced models, and develops a perturbation-based passivity enforcement for POD-DEIM-reduced models driven by a computable state-error bound. Numerical experiments on a 2D transformer illustrate the error bounds and the small output perturbation.

Significance. If the claims hold, the paper delivers a constructive route from a singular 3D FEM MQS DAE to a passive reduced ODE model with a computable a priori bound for DEIM-induced loss of passivity. The storage-function arguments in Theorems 7, 9, and 11 are direct, and the bound in Theorem 15 is explicit in terms of reduced quantities only. These are useful, concrete results for circuit-coupled electromagnetic simulation. The main gaps are missing hypotheses rather than errors in the derivations themselves.

major comments (4)
  1. [Section 3.1, after Eq. (10)] The sentence "due to Assumption 1 e)(iii) the matrix X2 has full column rank" refers to an item that does not exist; Assumption 1 e) lists only (i) and (ii). The full-column-rank condition is used in the proof of Theorem 5 to conclude that Yhat_{C2}^T X2 has full column rank, in Theorem 6 to ensure the condensed form, and in Eqs. (26) and (31) where (X2^T Yhat_{C2} Yhat_{C2}^T X2)^{-1} appears. If X2 is rank deficient, e.g., when a winding is supported in the conducting subdomain, the regularized index-one DAE and the subsequent ODE transformation can fail. This condition should be added as an explicit hypothesis in Assumption 1, or an equivalent condition should be proved from the mesh and winding assumptions.
  2. [Section 5, inequality (45)] The conclusion that the integral of u^T y_POD over [0,t] is nonnegative uses io-passivity of the POD-reduced model, which by Definition 2 and Remark 3 requires zero initial state E x(0)=0. The text does not state x_POD,0=0 before this inequality, and in the general setup of Sections 2 and 3 the initial field A0 is arbitrary. For a nonzero initial state, the free-response contribution to y_POD can make the integral negative, so the perturbation rule (47) need not enforce io-passivity. Add a zero-initial-state hypothesis, or an estimate involving S_POD(x_POD,0), to the passivity enforcement statement.
  3. [Theorem 13, Eq. (54) and following claim] The claim that mu = min(mu1, mu2) < 0 is not established. It requires U^T C_d^T M_f C_d U and U_{a1}^T C_1^T M_{f,1} C_1 U_{a1} to be positive definite, but only positive semidefiniteness follows from the given information; in 3D, C_d has a nontrivial kernel, so the reduced matrices can be singular, and A_l is only shown negative semidefinite, so lambda_max(A_l)=0 is possible. If mu=0, the formula in Theorem 15 must be interpreted as a limit and the statement that theta(t) remains bounded needs separate justification. Please state an explicit positive-definiteness condition for the reduced curl Gramians or prove it from the construction of U.
  4. [Theorem 12, proof around Eq. (50)] The derivation of Delta_DEIM contains two inequalities written with 'lesssim', but the theorem states (49) as a deterministic bound without quantifying the constants. Since the perturbation in (47) is chosen from theta(t), an unquantified constant in Delta_DEIM weakens the guarantee that the perturbed system is io-passive. Please either prove the bound with explicit constants under stated assumptions on the snapshot richness, or state the theorem with the same 'lesssim' convention and discuss the validity of the passivity guarantee.
minor comments (4)
  1. [Section 4, first paragraph] The phrase 'Thought for general systems' should read 'Although for general systems'.
  2. [Theorem 13, display (54)] The word 'entrees' should be 'entries' in the definition of the matrices M_f and M_{f,1}.
  3. [Theorems 12 and 15] The displayed error bounds divide by L2[f] and by mu; the paper should state the limiting interpretation when these quantities vanish, since zero is not excluded by the current assumptions.
  4. [Section 5 and Conclusion] The text sometimes says 'passivity enforcement', but the proven property is io-passivity with zero initial state; the terminology should be aligned to avoid overstating the result.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: regularization, passivity, and DEIM error bounds are derived from stated structural and monotonicity assumptions, not from fitted values or self-referential definitions.

full rationale

The paper's derivation chain is self-contained. The regularization in Section 3.2 is obtained by a concrete linear coordinate transformation that exposes the common kernel of E and K(a), and the index-one property of Theorem 5 and the condensed form of Theorem 6 are proved directly from positive definiteness of M11 and Mν(Cd a) and the full column rank of X2. Passivity of the FEM, regularized, and POD-reduced models (Theorems 7, 9, 11) is established by explicit storage functions built from the magnetic energy and by direct differentiation of the discrete equations; no step defines a quantity in terms of the prediction being made. The POD-DEIM passivity enforcement in Section 5 is a construction rather than a prediction: the output perturbation δ(t)u(t) is chosen in (47) precisely to make inequality (45) nonnegative, and the size is controlled by the DEIM state error bound θ(t) of Theorem 15. That bound is derived from a standard DEIM snapshot estimate (the ΔDEIM term from [30]) and a logarithmic Lipschitz bound proved in Theorem 13 using the monotonicity lemma from [4]. The constants in Table 1 are computed from the reduced system matrices and snapshot singular values, not fitted to the reported output errors. The self-citations [4,5,17,18] are prior results used as framework or lemmas; they are not the only support for the new claims, and the load-bearing steps are re-proven or extended here. One rigor gap is flagged: the full column rank of X2 is load-bearing for Theorems 5 and 6 and the ODE transformation (31), but the text in Section 3.1 justifies it by a nonexistent 'Assumption 1 e)(iii)'. This is an incomplete hypothesis or missing proof, not a circular step, so it does not raise the circularity score.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central results rest on the standard MQS assumptions, on two technical conditions that are not stated as assumptions (full column rank of X2 and positive definiteness of the reduced curl Gramians), and on a zero-initial-state caveat for io-passivity enforcement. No new physical entities are introduced.

free parameters (3)
  • POD truncation order r = 35
    Chosen from POD snapshot singular values (sigma_36/sigma_1 = 1.14e-7); algorithmic hyperparameter, not fit to enforce passivity.
  • DEIM truncation order l = 9
    Chosen from DEIM snapshot singular values (sigma_10/sigma_1 = 1.40e-3); algorithmic hyperparameter.
  • DEIM index set K = 9 greedy-selected indices
    Standard DEIM greedy selection; affects Delta_DEIM and the size of the passivity-enforcing output perturbation.
assumptions (6)
  • domain assumption Assumption 1: bounded simply connected Lipschitz domain, conducting/non-conducting decomposition, sigma piecewise constant positive on conductors, nu strongly monotone and Lipschitz, R symmetric positive definite, winding functions divergence-free with disjoint supports.
    Used throughout; guarantees unique solvability of the continuous problem via [4, Thm. 9] and passivity via [5].
  • domain assumption The non-conducting winding coupling block X2 has full column rank.
    Never stated in Assumption 1; Section 3.1 cites nonexistent item e)(iii). Load-bearing for the condensed form (Theorem 6), index-one property (Theorem 5), and ODE transformation.
  • domain assumption Initial stored energy is zero for the io-passivity argument.
    Section 5 uses nonnegativity of the integral of u^T y_POD, which is only guaranteed from io-passivity when the initial state, or initial storage, is zero.
  • ad hoc to paper Reduced curl Gramians U^T C_d^T M_f C_d U and U^T C_1^T M_{f,1} C_1 U are positive definite, giving mu < 0.
    Theorem 13 requires a negative logarithmic Lipschitz bound; this positive definiteness is asserted without proof and could fail if the reduced basis contains gradient components.
  • standard math DEIM error estimate of Chaturantabut and Sorensen (2010).
    Invoked in Theorem 12 to define Delta_DEIM and the state error bound.
  • standard math Gronwall's inequality and norm equivalence.
    Used in Theorems 12 and 15 to convert the differential inequality into an exponential state error bound.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Regularization and passivity-preserving model reduction of quasilinear magneto-quasistatic coupled problems." pith.science (2026). https://pith.science/paper/XAY6KFXO

@misc{pith2026241114226,
  author       = {Pith},
  title        = {Pith review of: Regularization and passivity-preserving model reduction of quasilinear magneto-quasistatic coupled problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XAY6KFXO}},
  note         = {Machine review of arXiv:2411.14226}
}
read the original abstract

We consider the quasilinear magneto-quasistatic field equations that arise in the simulation of low-frequency electromagnetic devices coupled to electrical circuits. Spatial discretization of these equations on 3D domains using the finite element method results in a singular system of differential-algebraic equations (DAEs). First, we analyze the structural properties of this system and present a novel regularization approach based on projecting out the singular state components. Next, we explore the passivity of the variational magneto-quasistatic problem and its discretization by defining suitable storage functions. For model reduction of the magneto-quasistatic system, we employ the proper orthogonal decomposition (POD) technique combined with the discrete empirical interpolation method (DEIM), to facilitate efficient evaluation of the system's nonlinearities. Our model reduction approach involves the transformation of the regularized DAE into a system of ordinary differential equations, leveraging a special block structure inherent in the problem, followed by applying standard model reduction techniques to the transformed system. We prove that the POD-reduced model preserves passivity, and for the POD-DEIM-reduced model, we propose to enforce passivity by perturbing the output in a way that accounts for DEIM errors. Numerical experiments illustrate the effectiveness of the presented model reduction methods and the passivity enforcement technique.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 22 canonical work pages

  1. [1]

    Academic Press, Boston (1998)

    Bossavit, A.: Computational Electromagnetism: Variational Formulations, Com- plementarity, Edge Elements. Academic Press, Boston (1998)

    work page 1998

  2. [2]

    Cambridge University Press, Cambridge (2017)

    Griffiths, D.J.: Introduction to Electrodynamics. Cambridge University Press, Cambridge (2017). https://doi.org/10.1017/9781108333511

    doi:10.1017/9781108333511 2017

  3. [3]

    John Wiley & Sons, New York (1999) 30

    Jackson, J.D.: Classical Electrodynamics. John Wiley & Sons, New York (1999) 30

    work page 1999

  4. [4]

    Chill, R., Reis, T., Stykel, T.: Analysis of a quasilinear coupled magneto- quasistatic model: solvability and regularity of solutions. J. Math. Anal. Appl. 523(2), 127033 (2023) https://doi.org/10.1016/j.jmaa.2023.127033

    arXiv 2023

  5. [5]

    Reis, T., Stykel, T.: Passivity, port-hamiltonian formulation and solution esti- mates for a quasilinear coupled magneto-quasistatic system. Evol. Equ. Control Theory 12(4), 1208–1232 (2023) https://doi.org/10.3934/eect.2023008

    doi:10.3934/eect.2023008 2023

  6. [6]

    Bartel, A., Baumanns, S., Sch¨ ops, S.: Structural analysis of electrical circuits including magnetoquasistatic devices. Appl. Numer. Math. 61(12), 1257–1270 (2011) https://doi.org/10.1016/j.apnum.2011.08.004

    work page doi:10.1016/j.apnum.2011.08.004 2011

  7. [7]

    Cortes Garcia, I., Gersem, H., Sch¨ ops, S.: A structural analysis of field/circuit coupled problems based on a generalised circuit element. Numer. Algor. 83, 373– 394 (2020) https://doi.org/10.1007/s11075-019-00686-x

    work page doi:10.1007/s11075-019-00686-x 2020

  8. [8]

    Springer, Heidelberg (2010)

    Alonso Rodr ´ ıguez, A., Valli, A.: Eddy Current Approximation of Maxwell Equations: Modeling, Simulation and Applications. Springer, Heidelberg (2010). https://doi.org/10.1007/978-88-470-1506-7

    doi:10.1007/978-88-470-1506-7 2010

Show all 33 references
  1. [9]

    IEEE Trans

    Bossavit, A.: ”Stiff” problems in eddy-current theory and the regularization of Maxwell’s equations. IEEE Trans. Magn. 37(5), 3542–3545 (2001) https://doi. org/10.1109/20.952657

    2001 doi

  2. [10]

    Clemens, M., Weiland, T.: Discrete electromagnetism with the finite integration technique. Prog. Electromagn. Res. 32, 65–87 (2001) https://doi.org/10.2528/ PIER00080103

    2001

  3. [11]

    Sch¨ ops, S.: Multiscale modeling and multirate time-integration of field/circuit coupled problems. Ph.D. thesis, Bergische Universit¨ at Wuppertal (2011)

    2011

  4. [12]

    IEEE Trans

    Manges, J.B., Cendes, Z.J.: A generalized tree-cotree gauge for magnetic field computation. IEEE Trans. Magn. 31(3), 1342–1347 (1995) https://doi.org/10. 1109/20.376275

    1995

  5. [13]

    Computing 64(2), 97–122 (2000) https://doi.org/10.1007/s006070050005

    Hiptmair, R.: Multilevel gauging for edge elements. Computing 64(2), 97–122 (2000) https://doi.org/10.1007/s006070050005

    2000 doi

  6. [14]

    IEEE Trans

    Clemens, M., Weiland, T.: Regularization of eddy-current formulations using discrete grad-div operators. IEEE Trans. Magn. 38(2), 569–572 (2002) https: //doi.org/10.1109/20.996149

    2002 doi

  7. [15]

    Schoenmaker, W., Magnus, W., Meuris, P.: Ghost fields in classical gauge the- ories. Phys. Rev. Lett. 88, 181602 (2002) https://doi.org/10.1103/PhysRevLett. 88.181602

    2002 doi

  8. [16]

    (eds.): Model Order Reduction

    Benner, P., Grivet-Talocia, S., Quarteroni, A., Rozza, G., Schilders, W., Silveira, 31 L.M. (eds.): Model Order Reduction. Volume 1: System- and Data-Driven Meth- ods and Algorithms; Volume 2: Snapshot-Based Methods and Algorithms; Volume 3: Applications. De Gruyter, Berlin, B...

    2021

  9. [17]

    Kerler-Back, J., Stykel, T.: Model order reduction for linear and nonlinear magneto-quasistatic equations. Int. J. Numer. Meth. Engng 111(13), 1274–1299 (2017) https://doi.org/10.1016/j.ifacol.2015.05.126

    2017 doi

  10. [18]

    In: Beattie, C., Benner, P., Embree, M., Gugercin, S., Lefteriu, S

    Kerler-Back, J., Stykel, T.: Balanced truncation model reduction for 3D lin- ear magneto-quasistatic field problems. In: Beattie, C., Benner, P., Embree, M., Gugercin, S., Lefteriu, S. (eds.) Realization and Model Reduction of Dynamical Systems - A Festschrift in Honor of the ...

    2022 doi

  11. [19]

    IEEE Trans

    Montier, L., Pierquin, A., Henneron, T., Cl´ enet, S.: Structure preserving model reduction of low frequency electromagnetic problem based on POD and DEIM. IEEE Trans. Magn. 53(6), 1–4 (2017) https://doi.org/10.1109/TMAG.2017. 2663761

    2017 doi

  12. [20]

    Sato, M. Y. Clemens, Igarashi, H.: Adaptive subdomain model order reduction with discrete empirical interpolation method for nonlinear magneto-quasi-static problems. IEEE Trans. Magn. 52(3), 1–4 (2016) https://doi.org/10.1109/TMAG. 2015.2489264

    2016

  13. [21]

    COMPEL 32(6), 2063–2083 (2013) https://doi.org/10.1108/COMPEL-01-2013-0004

    Sch¨ ops, S., De Gersem, H., Weiland, T.: Winding functions in transient magneto- quasistatic field-circuit coupled simulations. COMPEL 32(6), 2063–2083 (2013) https://doi.org/10.1108/COMPEL-01-2013-0004

    2013 doi

  14. [22]

    Springer Series in Computational Math- ematics, vol

    Girault, V., Raviart, P.-A.: Finite Element Methods for the Navier–Stokes Equations - Theory and Algorithms. Springer Series in Computational Math- ematics, vol. 5. Springer, Berlin Heidelberg (1986). https://doi.org/10.1007/ 978-3-642-61623-5

    1986

  15. [23]

    Prentice Hall, Englewood Cliffs, NJ (1973)

    Anderson, B.D.O., Vongpanitlerd, S.: Network Analysis and Synthesis. Prentice Hall, Englewood Cliffs, NJ (1973)

    1973

  16. [24]

    Willems, J.C.: Dissipative dynamical systems part I: General theory. Arch. Ratio- nal Mech. Anal. 45(5), 321–351 (1972) https://doi.org/10.1007/BF00276493

    1972 doi

  17. [25]

    PhD thesis, Technische Universit¨ at Berlin (2010)

    Br¨ ull, T.: Dissipativity of Linear Quadratic Systems. PhD thesis, Technische Universit¨ at Berlin (2010)

    2010

  18. [26]

    Hill, D.J., Moylan, P.J.: Dissipative dynamical systems: Basic input-output and state properties. J. Franklin Inst. 309(5), 327–357 (1980) https://doi.org/10. 1016/0016-0032(80)90026-5

    1980

  19. [27]

    N´ ed´ elec, J.C.: Mixed finite elements inR3. Numer. Math. 35, 315–341 (1980) 32 https://doi.org/10.1007/BF01396415

    1980 doi

  20. [28]

    COMPEL 30(6), 1701–1714 (2011) https://doi.org/10.1108/03321641111168039

    Clemens, M., Sch¨ ops, S., De Gersem, H., Bartel, A.: Decomposition and regu- larization of nonlinear anisotropic curl-curl DAEs. COMPEL 30(6), 1701–1714 (2011) https://doi.org/10.1108/03321641111168039

    2011 doi

  21. [29]

    Differential-Algebraic Equations Forum

    Lamour, R., M¨ arz, R., Tischendorf, C.: Differential-Algebraic Equations: A Pro- jector Based Analysis. Differential-Algebraic Equations Forum. Springer, Berlin, Heidelberg (2013). https://doi.org/10.1007/978-3-642-27555-5

    2013 doi

  22. [30]

    Chaturantabut, S., Sorensen, D.C.: Nonlinear model reduction via discrete empir- ical interpolation. SIAM J. Sci. Comput. 32(5), 2737–2764 (2010) https://doi. org/10.1137/090766498

    2010 doi

  23. [31]

    Transactions of the Royal Institute of Technology, vol

    Dahlquist, G.: Stability and Error Bounds in the Numerical Integration of Ordi- nary Differential Equations. Transactions of the Royal Institute of Technology, vol. 130. Stockholm, Sweden (1959)

    1959

  24. [32]

    History and modern theory

    S¨ oderlind, G.: The logarithmic norm. History and modern theory. BIT 46, 631– 652 (2006) https://doi.org/10.1007/s10543-006-0069-9

    2006 doi

  25. [33]

    Academic Press, San Diego (1998) 33

    Pachpatte, B.G.: Inequalities for Differential and Integral Equations. Academic Press, San Diego (1998) 33

    1998

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.