Pith. sign in

REVIEW 2 major objections 5 minor 38 references

Convergence analysis of GMRES applied to Helmholtz problems near resonances

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 ↗ Review JSON

Pith's one-line read Each plateau in GMRES residual histories near a Helmholtz resonance is the Krylov subspace slowly resolving a small eigenvalue, and deflating that eigenvalue's eigenvector removes the plateau.

desk verdict Subset-based harmonic Ritz bound with a clean plateau diagnosis; the non-normal scattering test needs a conditioning check, but the mathematics holds and the paper deserves serious review. read the letter →

arxiv 2505.16345 v2 pith:XIOLWQJP submitted 2025-05-22 math.NA cs.NA

Victorita Dolean , Pierre Marchand , Axel Modave , Timothée Raynaud This is my paper · ORCID
classification math.NAcs.NA MSC 65F1065N2265N30
keywords GMRESharmonicRitzvaluesHelmholtzequationresonancequasi-resonancedeflationcomplexshiftedLaplacianpreconditionerconvergencebounds
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

Near a resonance or quasi-resonance, the finite-element Helmholtz system matrix has a small eigenvalue, and GMRES convergence slows dramatically. The paper argues that the slowdown is not a uniform spectral phenomenon: each plateau in the residual history is the Krylov subspace slowly resolving one small eigenvalue through a harmonic Ritz value, and the plateau ends when that resolution succeeds. Building on an earlier single-eigenvalue bound, it proves a subset version: once harmonic Ritz values approximate a group of eigenvalues, the factor in the convergence bound involving those eigenvalues tends to one, so the remaining eigenvalues alone govern subsequent convergence. Numerical experiments on a resonant cavity and on a quasi-resonant scattering problem confirm that deflating the corresponding eigenvectors removes the plateaus, and that combining deflation with a complex shifted Laplacian preconditioner is the most robust acceleration strategy.

What carries the argument

The machinery is the harmonic Ritz value, the root of the GMRES minimizing polynomial at iteration $l$; because the polynomial is normalized to one at zero, the residual can be factored as $p_l(z)=\prod_j(1-z/\nu_j^{(l)})$. The paper's central object is the rational factor $s_l^J(z)$ obtained by replacing some of those roots by the corresponding eigenvalues; the numerator cancels the denominator so the factor is well defined, and its size on the remaining spectrum controls progress after iteration $l$. A spectral-projector identity shows $s_l^J(A)=P_J^c s_l^J(A)$, which powers the pseudospectral bound, while deflation operators $P_{\mathrm{def}}=I-AQ$ and $Q_{\mathrm{def}}=I-QA$ built from approximate eigenvectors remove the captured modes explicitly.

What would settle it

Run GMRES on a nearly defective matrix, for example a Helmholtz-like matrix with a Jordan block at the smallest eigenvalue, track harmonic Ritz values, and check whether the residual drops as soon as a harmonic Ritz value matches that eigenvalue; if the plateau persists after the match, or if deflating the corresponding eigenvector fails to remove it, the diagonalizable-based bound is not the operative mechanism.

Watch

Extended reading notes

Core claim

The central discovery is a convergence bound that tracks GMRES between any two iterations using harmonic Ritz values. For diagonalizable $A$, if at iteration $l$ the harmonic Ritz values in a set $\mathcal{N}_J^{(l)}$ approximate the eigenvalues in $\Lambda_J$, then $$\frac{\|r_{l+m}\|_2}{\|r_l\|_2} \le \Big(\sum_{\lambda_i\in\Lambda_J^c} \kappa(\lambda_i)\Big) \max_{\lambda_i\in\Lambda_J^c} |s_l^J(\lambda_i)| \min_{q_m\in\mathcal{P}$_m^{1}$}\max_{\lambda_i\in\Lambda_J^c}|q_m(\lambda_i)|,$$ where $s_l^J(z)=\prod_{\lambda_j\in\Lambda_J}(1-z/\lambda_j)/\prod_{\nu_j\in\mathcal{N}_J^{(l)}}(1-z/\nu_j)$. When the harmonic Ritz values closely match the eigenvalues in $\Lambda_J$, the factor $\max_{\lambda_i\in\Lambda_J^c}|s_l^J(\lambda_i)|$ approaches one, meaning the approximated eigenvalues cease to hinder convergence and GMRES behaves as if they had been deflated. A pseudospectral analogue replaces the eigenvalue sum by a contour integral of the resolvent, covering cases where non-normality or near-defectiveness makes the eigenbasis story unreliable.

Load-bearing premise

Everything spectral in the main bound assumes the system matrix can be diagonalized with a full set of eigenvectors; if the finite-element Helmholtz matrix were defective or nearly defective, the simple eigenvalue-tracking diagnosis of plateaus would not follow in the stated form.

Editorial extensions

If this is right

  • Each plateau in a GMRES residual history near a resonance corresponds to one small eigenvalue being resolved by a harmonic Ritz value; the sharp drop begins once the approximation is good enough for the factor $|s_l^J(\lambda_i)|$ to approach one on the remaining spectrum.
  • Deflating the eigenvector of the resonant mode, even an approximate mode from a simplified closed-cavity problem, removes the corresponding plateau and reduces iteration counts across an interval of wavenumbers, not only at the resonant value.
  • Restarted GMRES near a resonance can fail because each restart discards the Krylov basis before the small eigenvalues are captured; deflation restores convergence, and in the hardest quasi-resonant case only deflation combined with ILU-preconditioned complex shifted Laplacian succeeds.
  • Because the bound is stage-by-stage, it predicts nonlinear residual histories rather than a single asymptotic rate, explaining why linear-rate bounds are poor descriptive tools for these problems.

Reading between the lines

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

  • Beyond the paper, harmonic Ritz values could be monitored during a solve to trigger deflation dynamically, adding a deflation vector only when a value has converged to a problematic eigenvalue; this would remove the need for a priori spectral information.
  • The plateau-capture mechanism suggests a practical estimator: the ratio of consecutive residual norms should spike precisely while a harmonic Ritz value approaches a small eigenvalue, which could be used to decide when to enlarge the deflation space.
  • The bound's mechanism is not tied to the complex shifted Laplacian specifically; the same plateau structure should appear with other preconditioners whenever the preconditioned matrix is diagonalizable, which is a testable prediction for sweeping or domain-decomposition preconditioners.
  • In 3D problems where exact eigenvectors are unavailable, the scattering benchmark suggests that physically motivated approximate modes, such as closed-cavity eigenmodes extended by zero, may be enough to remove the plateaus.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 paper studies convergence of GMRES for finite-element Helmholtz systems near resonant wavenumbers. It proves a bound (Theorem 2.6) in which, once harmonic Ritz values at iteration l approximate a subset of eigenvalues Lambda_J, the factor max_{lambda_i in Lambda_c^J} |s_l^J(lambda_i)| approaches 1, so that subsequent convergence is controlled by the remaining eigenvalues; a pseudospectral version (Theorem 2.9) is also given. The paper then discusses deflation of eigenvectors associated with small eigenvalues and its combination with a complex-shifted Laplacian preconditioner with ILU approximation. Numerical experiments on a closed cavity (normal symmetric matrix) and an open-cavity scattering problem with PMLs (complex non-normal matrix) show plateaus in the GMRES residual histories that end when an HR value approaches a small eigenvalue, and demonstrate that deflation removes the associated plateaus.

Significance. The algebraic core of the paper is sound and useful: Lemma 2.5, Theorem 2.6, and Theorem 2.9 are proved by elementary arguments, the bounds are parameter-free in the sense that no quantity is fitted, and the cavity experiments directly connect the end of a plateau to an HR value approaching an eigenvalue. This is a genuinely useful diagnostic for Helmholtz solvers near resonances. The main weakness is that the non-normal scattering benchmark invokes Theorem 2.6 without verifying its diagonalizability assumption or quantifying the eigenvalue condition numbers in the bound; the pseudospectral theorem is not used for the diagnosis. With that gap repaired, the paper would be a solid contribution to the numerical analysis of GMRES for resonant Helmholtz problems.

major comments (2)
  1. [Section 5, paragraph 'Influence of the wavenumber on the residual history'; Assumption 2.4; Theorem 2.6] The plateau in the scattering residual history is explained by 'HR values and Theorem 2.6', but the scattering matrix is explicitly complex and non-normal, and the paper does not verify that Assumption 2.4 (diagonalizability) holds or report the quantities kappa(lambda_i) = ||v_i|| ||w_i|| that appear in the bound. If the eigenvector matrix is ill-conditioned, the sum sum_{lambda_i in Lambda_c^J} kappa(lambda_i) can be arbitrarily large, making the bound vacuous; the observed correlation would then be consistent with the theorem but not explained by it. The authors should either (i) verify diagonalizability and show that the condition-number sum is moderate for the matrices used, (ii) use Theorem 2.9 with a concrete estimate of L(Gamma_c^J)/(2 pi epsilon_c^J), or (iii) explicitly label the Section 5 explanation as heuristic.
  2. [Section 5, Figure 5.3] The diagnosis that an HR value 'approaches the smallest eigenvalue' is only qualitative. For a complex spectrum, 'smallest' is ambiguous, and the figure alone does not show the distance between the HR value and the chosen eigenvalue. Reporting the numerical values of the eigenvalue and the HR value at iterations 850 and 1100, together with the corresponding factor max|s_l^J(lambda_i)|, would make the claimed mechanism quantitative and checkable.
minor comments (5)
  1. [Section 2.4, Lemma 2.5] The sentence 'Suppose that Theorem 2.4 holds' refers to a non-existent Theorem 2.4; it should say 'Assumption 2.4'.
  2. [Section 2.4, proof of Lemma 2.5] The expression q_m(A) s_l^J(A) p_l(A) is formally undefined if an HR value nu_j^(l) coincides exactly with an eigenvalue of A, since I - A/nu_j is then singular. Please state that the rational expression is understood before cancellation, or handle the equality case by continuity.
  3. [Throughout] There are several typos: 'interessed' (Section 1), 'mentionned' (Section 3.3), 'slighlty' (Section 6), and 'the bound in (2.3) less predictive' (Section 2.2) should read 'is less predictive'.
  4. [Section 5, Figure 5.3] The left panel of Figure 5.3 is hard to read at the scale shown; marking the specific eigenvalue and HR value used in the diagnosis would improve the presentation.
  5. [Section 3.3, equation (3.7)] It would help to clarify that the system (P_def A A_epsilon^{-1}) tilde{x} = P_def b is solved for tilde{x} and that the reconstruction u = Q b + Q_def A_epsilon^{-1} tilde{x} produces the solution of the preconditioned deflated problem; currently this is only implicit.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the HR-value convergence bound is derived from GMRES optimality and standard spectral/pseudospectral estimates; self-citations are background motivation, not load-bearing.

full rationale

The paper's central theoretical result (Theorem 2.6) is a conditional bound: if harmonic Ritz values approximate a subset of eigenvalues, then the GMRES residual ratio is bounded by a factor depending only on the remaining eigenvalues. This is proven directly from the defining property that HR values are roots of the GMRES minimizing polynomial (Definition 2.1 and (2.4)), not from any fitted quantity. The cancellation in Lemma 2.5 and the definition of s_l^J in (2.5) are algebraic identities, so the claim that well-approximated eigenvalues cease to hinder convergence is a theorem, not an input. No parameter is fitted to the experiments; the numerical sections use computed HR values and spectra only to illustrate the bound. The pseudospectral bound (Theorem 2.9) is likewise derived from standard spectral-projector and resolvent estimates. Self-citations [21,24] are used for background facts about quasimodes and prior numerical observations; these facts are not used to prove the new bounds, and Section 5 independently reports the small eigenvalue and HR-value trajectory. The skeptical concern that Theorem 2.6 is applied in Section 5 to a non-normal matrix without verifying diagonalizability or bounding kappa(lambda_i) is a correctness and validation gap, not circularity. Overall, the derivation chain is self-contained and no step reduces to its own conclusion.

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

The theory introduces no fitted constants. All parameters in the experiments are standard algorithmic choices (mesh size, tolerance, restart length, CSL shift epsilon=k taken from [22]). The main assumptions are mathematical regularity (diagonalizability, resolvent bound) and the physical modeling assumption that closed-cavity modes approximate open-cavity quasimodes.

assumptions (6)
  • standard math GMRES residual norm is the minimum over polynomials of degree l with q(0)=1, and the roots of the optimal residual polynomial are the harmonic Ritz values (Eq. 2.2, Definition 2.1).
    Foundational for all convergence bounds; taken from the standard GMRES literature (Saad and Schultz).
  • domain assumption Assumption 2.4: A is non-singular and diagonalizable.
    Used to write A = V D V^{-1} and to justify the eigenvalue-based bound in Theorem 2.6; the paper notes this may fail for general non-normal Helmholtz matrices and offers Theorem 2.9 for the non-normal case.
  • standard math Spectral projectors satisfy the standard properties listed in Lemma 2.8 (idempotence, completeness, orthogonality, commutativity with A, kernel property).
    Used in the proof of Lemma 2.8 and Theorem 2.9.
  • domain assumption There exists a contour Gamma_c^J on which the resolvent norm ||(A-zI)^{-1}|| is bounded by 1/epsilon_c^J (Theorem 2.9).
    A regularity assumption on the contour for the pseudospectral bound; the paper notes that epsilon may be very small for highly non-normal matrices.
  • domain assumption For the continuous and discrete Helmholtz operators, quasimodes correspond to small eigenvalues (from [21]) and cause GMRES stagnation.
    Links the PDE phenomenon to the spectral properties on which the convergence interpretation relies; this is a cited prior result, not proved here.
  • domain assumption Eigenvectors of the closed cavity problem extended by zero are good approximations of the open-cavity quasimodes (Section 5).
    The deflation experiments rely on this approximation; the paper supports it numerically by showing the deflated spectrum has the quasimode eigenvalue moved to zero.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Convergence analysis of GMRES applied to Helmholtz problems near resonances." pith.science (2026). https://pith.science/paper/XIOLWQJP

@misc{pith2026250516345,
  author       = {Pith},
  title        = {Pith review of: Convergence analysis of GMRES applied to Helmholtz problems near resonances},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XIOLWQJP}},
  note         = {Machine review of arXiv:2505.16345}
}
read the original abstract

The finite element solution of Helmholtz problems near resonant or quasi-resonant frequencies poses significant challenges, as iterative solvers typically suffer from severely degraded convergence. We analyze the convergence behavior of GMRES applied to linear systems arising from such configurations. Theoretical convergence estimates are derived based on harmonic Ritz values, highlighting their proximity to small eigenvalues as a key determining factor. We further examine deflation strategies and their interplay with preconditioning techniques, using the Complex Shifted Laplacian preconditioner as a case study. Numerical experiments on resonant and quasi-resonant test cases validate the theoretical framework and demonstrate the effectiveness of deflation strategies. This study provides new insights and practical guidance for analyzing and improving iterative solvers for time-harmonic problems near resonances.

Figures

Figures reproduced from arXiv: 2505.16345 by the authors.

Figure 2.1
Figure 2.1. GMRES convergence history for a cavity problem close to a resonance. [PITH_FULL_IMAGE:figures/full_fig_p003_2_1.png] view at source ↗
Figure 4.1
Figure 4.1. Cavity benchmark. Relative L 2 -error and number of GMRES iterations to reach the tolerance 10−6 on the relative residual as a function of k. The vertical lines correspond to resonance wavenumbers. with the solution given by u = (A−1 ε + Q)x. This approach has the same additional cost per iteration of (3.7), but it does not require the initial computation of AZ and Z ∗A. However, since the numerical results obtained… view at source ↗
Figure 4
Figure 4. [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figures from the paper (13 more)
Figure 4.2
Figure 4.2. Figure 4.2: Cavity benchmark. Convergence analysis for [PITH_FULL_IMAGE:figures/full_fig_p013_4_2.png]
Figure 4
Figure 4. Figure 4: a [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 4
Figure 4. Figure 4: a [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 4.3
Figure 4.3. Figure 4.3: Cavity benchmark. Mode (3,3) of the Laplacian. [PITH_FULL_IMAGE:figures/full_fig_p015_4_3.png]
Figure 4
Figure 4. Figure 4: a [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 4.4
Figure 4.4. Figure 4.4: Cavity benchmark. Number of GMRES iteration to reach the relative residual [PITH_FULL_IMAGE:figures/full_fig_p016_4_4.png]
Figure 5
Figure 5. Figure 5 [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 5.1
Figure 5.1. Figure 5.1: Scattering benchmark. Computational domain with PML. [PITH_FULL_IMAGE:figures/full_fig_p018_5_1.png]
Figure 5
Figure 5. Figure 5: a [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 5.2
Figure 5.2. Figure 5.2: Scattering benchmark. Number of GMRES iteration to reach the relative residual [PITH_FULL_IMAGE:figures/full_fig_p019_5_2.png]
Figure 5.3
Figure 5.3. Figure 5.3: Scattering benchmark. HR values and deflated spectrum. [PITH_FULL_IMAGE:figures/full_fig_p020_5_3.png]
Figure 5
Figure 5. Figure 5: a [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]
Figure 5
Figure 5. Figure 5: c [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 20 canonical work pages

  1. [1]

    I. M. Babuška and S. A. Sauter , Is the Pollution Effect of the FEM Avoidable for the Helmholtz Equation Considering High Wave Numbers? , SIAM Journal on Numerical Analysis, 34 (1997), pp. 2392--2423, https://doi.org/10.1137/s0036142994269186

    doi:10.1137/s0036142994269186 1997

  2. [2]

    Bastian, G

    P. Bastian, G. Wittum, and W. Hackbusch , Additive and multiplicative multi-grid — A comparison , Computing, 60 (1998), pp. 345--364, https://doi.org/10.1007/bf02684380

    work page doi:10.1007/bf02684380 1998

  3. [3]

    Beckermann, S

    B. Beckermann, S. A. Goreinov, and E. E. Tyrtyshnikov , Some Remarks on the Elman Estimate for GMRES , SIAM Journal on Matrix Analysis and Applications, 27 (2005), pp. 772--778, https://doi.org/10.1137/040618849

    work page doi:10.1137/040618849 2005

  4. [4]

    Berenger , A perfectly matched layer for the absorption of electromagnetic waves , Journal of Computational Physics, 114 (1994), pp

    J.-P. Berenger , A perfectly matched layer for the absorption of electromagnetic waves , Journal of Computational Physics, 114 (1994), pp. 185--200, https://doi.org/10.1006/jcph.1994.1159

    arXiv 1994

  5. [5]

    B \' e riot and A

    H. B \' e riot and A. Modave , An automatic perfectly matched layer for acoustic finite element simulations in convex domains of general shape , International Journal for Numerical Methods in Engineering, (2020), https://doi.org/10.1002/nme.6560

    doi:10.1002/nme.6560 2020

  6. [6]

    Berm \' u dez, L

    A. Berm \' u dez, L. Hervella-Nieto, A. Prieto, and R. Rodr \' guez , An exact bounded PML for the Helmholtz equation , Comptes Rendus Mathematique, 339 (2004), pp. 803--808, https://doi.org/10.1016/j.crma.2004.10.006

    doi:10.1016/j.crma.2004.10.006 2004

  7. [7]

    Berm \' u dez, L

    A. Berm \' u dez, L. Hervella-Nieto, A. Prieto, and R. Rodr \' guez , An optimal perfectly matched layer with unbounded absorbing function for time-harmonic acoustic scattering problems , Journal of Computational Physics, 223 (2007), pp. 469--488, https://doi.org/10.1016/j.jcp.2006.09.018

    doi:10.1016/j.jcp.2006.09.018 2007

  8. [8]

    Bécache, A

    E. Bécache, A. S. Bonnet-Ben Dhia, and G. Legendre , Perfectly Matched Layers for Time-Harmonic Acoustics in the Presence of a Uniform Flow , SIAM Journal on Numerical Analysis, 44 (2006), pp. 1191--1217, https://doi.org/10.1137/040617741

    work page doi:10.1137/040617741 2006

Show all 38 references
  1. [9]

    S. L. Campbell, I. C. F. Ipsen, C. T. Kelley, and C. D. Meyer , GMRES and the minimal polynomial , BIT Numerical Mathematics, 36 (1996), pp. 664--675, https://doi.org/10.1007/bf01733786

    1996 doi

  2. [10]

    Cao , A note on the convergence behavior of GMRES , Applied Numerical Mathematics, 25 (1997), pp

    Z.-H. Cao , A note on the convergence behavior of GMRES , Applied Numerical Mathematics, 25 (1997), pp. 13--20, https://doi.org/10.1016/s0168-9274(97)00039-1

    1997 doi

  3. [11]

    S. N. Chandler-Wilde, I. G. Graham, S. Langdon, and M. Lindner , Condition number estimates for combined potential boundary integral operators in acoustic scattering , Journal of Integral Equations and Applications, 21 (2009), https://doi.org/10.1216/jie-2009-21-2-229

    2009 doi

  4. [12]

    Crouzeix and C

    M. Crouzeix and C. Palencia , The Numerical Range is a (1+ 2 ) -Spectral Set , SIAM Journal on Matrix Analysis and Applications, 38 (2017), pp. 649--655, https://doi.org/10.1137/17m1116672

    2017 doi

  5. [13]

    H. V. der Vorst and C. Vuik , The superlinear convergence behaviour of GMRES , Journal of Computational and Applied Mathematics, 48 (1993), pp. 327--341, https://doi.org/10.1016/0377-0427(93)90028-a

    1993 doi

  6. [14]

    Dolean, P

    V. Dolean, P. Jolivet, and F. Nataf , An Introduction to Domain Decomposition Methods: Algorithms, Theory, and Parallel Implementation , Society for Industrial and Applied Mathematics, Nov. 2015, https://doi.org/10.1137/1.9781611974065

    2015 doi

  7. [15]

    Embree , How Descriptive are GMRES Convergence Bounds? , 2022, https://doi.org/10.48550/ARXIV.2209.01231

    M. Embree , How Descriptive are GMRES Convergence Bounds? , 2022, https://doi.org/10.48550/ARXIV.2209.01231

    2022 doi arXiv

  8. [16]

    Embree , Extending Elman's Bound for GMRES , 2023, https://doi.org/10.48550/ARXIV.2312.15022

    M. Embree , Extending Elman's Bound for GMRES , 2023, https://doi.org/10.48550/ARXIV.2312.15022

    2023 doi arXiv

  9. [17]

    Engquist and A

    B. Engquist and A. Majda , Absorbing boundary conditions for numerical simulation of waves , Proceedings of the National Academy of Sciences, 74 (1977), pp. 1765--1766, https://doi.org/10.1073/pnas.74.5.1765

    1977 doi

  10. [18]

    Y. A. Erlangga , Advances in Iterative Methods and Preconditioners for the Helmholtz Equation , Archives of Computational Methods in Engineering, 15 (2007), pp. 37--66, https://doi.org/10.1007/s11831-007-9013-7

    2007 doi

  11. [19]

    Y. A. Erlangga and R. Nabben , Deflation and Balancing Preconditioners for Krylov Subspace Methods Applied to Nonsymmetric Matrices , SIAM Journal on Matrix Analysis and Applications, 30 (2008), pp. 684--699, https://doi.org/10.1137/060678257

    2008 doi

  12. [20]

    O. G. Ernst and M. J. Gander , Why it is Difficult to Solve Helmholtz Problems with Classical Iterative Methods , Springer Berlin Heidelberg, Aug. 2012, pp. 325--363, https://doi.org/10.1007/978-3-642-22061-6_10

    2012 doi

  13. [21]

    Galkowski, P

    J. Galkowski, P. Marchand, and E. A. Spence , Eigenvalues of the Truncated Helmholtz Solution Operator under Strong Trapping , SIAM Journal on Mathematical Analysis, 53 (2021), pp. 6724--6770, https://doi.org/10.1137/21m1399658

    2021 doi

  14. [22]

    M. J. Gander, I. G. Graham, and E. A. Spence , Applying GMRES to the Helmholtz equation with shifted Laplacian preconditioning: what is the largest shift for which wavenumber-independent convergence is guaranteed? , Numerische Mathematik, 131 (2015), pp. 567--614, https://doi....

    2015 doi

  15. [23]

    García Ramos, R

    L. García Ramos, R. Kehl, and R. Nabben , Projections, Deflation, and Multigrid for Nonsymmetric Matrices , SIAM Journal on Matrix Analysis and Applications, 41 (2020), pp. 83--105, https://doi.org/10.1137/18m1180268

    2020 doi

  16. [24]

    Marchand, J

    P. Marchand, J. Galkowski, E. A. Spence, and A. Spence , Applying GMRES to the Helmholtz equation with strong trapping: how does the number of iterations depend on the frequency? , Advances in Computational Mathematics, 48 (2022), https://doi.org/10.1007/s10444-022-09931-9

    2022 doi

  17. [25]

    Marchner, H

    P. Marchner, H. Bériot, X. Antoine, and C. Geuzaine , Stable Perfectly Matched Layers with Lorentz transformation for the convected Helmholtz equation , Journal of Computational Physics, 433 (2021), p. 110180, https://doi.org/10.1016/j.jcp.2021.110180

    2021

  18. [26]

    G. A. Meurant , Computer solution of large linear systems , vol. 28 of Studies in mathematics and its applications, Elsevier, 1999

    1999

  19. [27]

    Modave, C

    A. Modave, C. Geuzaine, and X. Antoine , Corner treatments for high-order local absorbing boundary conditions in high-frequency acoustic scattering , Journal of Computational Physics, 401 (2020), p. 109029, https://doi.org/10.1016/j.jcp.2019.109029

    2020

  20. [28]

    L. G. Ramos and R. Nabben , A two-level shifted Laplace preconditioner for Helmholtz problems: Field-of-values analysis and wavenumber-independent convergence , 2020, https://doi.org/10.48550/ARXIV.2006.08750

    2020 doi arXiv

  21. [29]

    L. G. Ramos, O. Sète, and R. Nabben , Preconditioning the Helmholtz equation with the shifted Laplacian and Faber polynomials , ETNA - Electronic Transactions on Numerical Analysis, 54 (2021), pp. 534--557, https://doi.org/10.1553/etna_vol54s534

    2021 doi

  22. [30]

    Saad , Iterative Methods for Sparse Linear Systems , Society for Industrial and Applied Mathematics, 2003, https://doi.org/10.1137/1.9780898718003

    Y. Saad , Iterative Methods for Sparse Linear Systems , Society for Industrial and Applied Mathematics, 2003, https://doi.org/10.1137/1.9780898718003

    2003 doi

  23. [31]

    Saad , Numerical Methods for Large Eigenvalue Problems , Society for Industrial and Applied Mathematics, 2011, https://doi.org/10.1137/1.9781611970739

    Y. Saad , Numerical Methods for Large Eigenvalue Problems , Society for Industrial and Applied Mathematics, 2011, https://doi.org/10.1137/1.9781611970739

    2011 doi

  24. [32]

    Saad and M

    Y. Saad and M. H. Schultz , GMRES : A Generalized Minimal Residual Algorithm for Solving Nonsymmetric Linear Systems , SIAM Journal on Scientific and Statistical Computing, 7 (1986), pp. 856--869, https://doi.org/10.1137/0907058

    1986 doi

  25. [33]

    Spillane and D

    N. Spillane and D. B. Szyld , New Convergence Analysis of GMRES with Weighted Norms, Preconditioning and Deflation, Leading to a New Deflation Space , 2023, https://doi.org/10.48550/ARXIV.2312.13625

    2023 doi arXiv

  26. [34]

    L. N. Trefethen and M. Embree , Spectra and Pseudospectra: The Behavior of Nonnormal Matrices and Operators , Princeton University Press, 2005, https://doi.org/10.1515/9780691213101

    2005 doi

  27. [35]

    Turkel and A

    E. Turkel and A. Yefet , Absorbing PML boundary layers for wave-like equations , Applied Numerical Mathematics, 27 (1998), pp. 533--557, https://doi.org/10.1016/s0168-9274(98)00026-9

    1998 doi

  28. [36]

    van der Sluis and H

    A. van der Sluis and H. A. van der Vorst , The rate of convergence of Conjugate Gradients , Numerische Mathematik, 48 (1986), pp. 543--560, https://doi.org/10.1007/bf01389450

    1986 doi

  29. [37]

    Vuik , Krylov Subspace Solvers and Preconditioners , ESAIM: Proceedings and Surveys, 63 (2018), pp

    C. Vuik , Krylov Subspace Solvers and Preconditioners , ESAIM: Proceedings and Surveys, 63 (2018), pp. 1--43, https://doi.org/10.1051/proc/201863001

    2018

  30. [38]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1 'skip if FUNCTION new.block.check...

Pith tools

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