pith. sign in

arxiv: 2606.12577 · v1 · pith:7IC5QM3Znew · submitted 2026-06-10 · 🧮 math.NA · cs.NA

Cascading Smoothers for Multigrid

Robert I. Saye This is my paper

Pith reviewed 2026-06-27 08:49 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords multigrid methodssmoothersFrobenius norm minimizationPoisson equationStokes systemfinite elementsdiscontinuous Galerkinerror propagators
0
0 comments X

The pith

Cascading smoothers, optimized sequentially by Frobenius norm minimization, match or exceed classical smoothers in multigrid V-cycles across many problems without tuning.

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

The paper introduces cascading smoothers as sequences of block-diagonal smoothers where each step is chosen to minimize the Frobenius norm of the error propagator from the preceding steps. It presents both additive and multiplicative versions and demonstrates their use inside standard multigrid V-cycles. These smoothers are shown to work effectively on finite difference, finite element, and discontinuous Galerkin discretizations of Poisson, elliptic interface, and Stokes problems, including multiphase cases. They perform at least as well as optimally damped classical smoothers yet avoid the need for problem-specific damping parameters. This matters because effective smoothers are essential for the linear scaling of multigrid, and a general method that works across discretizations could simplify solver design for complex applications.

Core claim

Cascading smoothers consist of an ordered sequence of single-step block-diagonal smoothers, with each level in the cascade optimised to maximally damp the output of prior steps via a Frobenius norm minimisation of the corresponding error propagators. The additive formulation is analogous to Jacobi and the multiplicative to Gauss-Seidel. When used within a standard multigrid V-cycle, they deliver convergence that closely matches or significantly outperforms optimally-damped classical counterparts on finite difference, finite element, and discontinuous Galerkin discretisations of Poisson, elliptic interface, and Stokes systems as well as multiphase variants, while requiring no parameter tuning

What carries the argument

Cascading smoothers defined by sequential Frobenius-norm minimization of error propagators in additive or multiplicative block-diagonal form.

Load-bearing premise

Sequential Frobenius-norm minimization at each cascade level produces stable and effective damping for arbitrary discretizations and operators without introducing new instabilities.

What would settle it

Finding a discretization or operator where the cascading smoothers diverge or perform substantially worse than their optimally damped classical counterparts on a standard test problem would falsify the effectiveness.

Figures

Figures reproduced from arXiv: 2606.12577 by Robert I. Saye.

Figure 1
Figure 1. Figure 1: Optimal damping parameters for block-Jacobi and block-Gauss-Seidel. Circles and blue/green shades correspond to 2D applications; diamonds and red/orange shades correspond to 3D applications. Within each group of three, the left-most point corresponds to J, the middle point to GS , and right-most to GS ; bars indicate ranges of near-optimal ω, i.e., damping parameters that achieve speeds η within 10% of the… view at source ↗
Figure 2
Figure 2. Figure 2: Multigrid solver performance for the 2D (top) and 3D (bottom) finite difference Poisson problem considered in §3.4.1; see §3.3 for a visualisation guide. method, except that each colour sweep has its own level-dependent, automatically-determined damping coefficient λℓ,R/B [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Multigrid solver performance for the 2D (top) and 3D (bottom) continuous Galerkin nodal finite element method considered in §3.4.2. blocking, defined by collecting the lower-left ℘ d nodes into each cell’s block. For example, for bilinear/trilinear elements, each block contains just one node; for biquadratic (resp., triquadratic) elements, a block contains 4 (resp., 8) nodes. Meanwhile, for multicoloured s… view at source ↗
Figure 4
Figure 4. Figure 4: Multigrid solver performance for the 2D (top) and 3D (bottom) uniform Cartesian grid, discontinuous Galerkin problem considered in §3.4.3 [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Examples of the nonuniform quadtree (left) and octree (right) used in the adaptive mesh refinement problems. Each example corresponds to a base-level grid size of n = 4, further refined up to four levels around a circular arc or spherical shell of the same radius. regard to numerical fluxes, penalty parameters, amalgamation of quadrature rules, etc.; see [14, 20] for details. We present two cases here, eac… view at source ↗
Figure 6
Figure 6. Figure 6: Multigrid solver performance for the 2D quadtree mesh (top) and 3D octree mesh (bottom) discontinuous Galerkin problem considered in §3.4.3. compiles the results.10 Overall, in terms of both η and the ν-weighted solver cost Crel, we see that a multicoloured smoother yields the fastest solvers. However, this neglects the complications of large-scale colouring: the graph colouring algorithm itself is a nontr… view at source ↗
Figure 7
Figure 7. Figure 7: Multigrid solver performance for the 2D (top) and 3D (bottom) high-contrast multiphase elliptic interface problem considered in §3.5. where Ω is a domain in R d divided into one or more subdomains/phases Ωi , Γij := ∂Ωi ∩ ∂Ωj is the interface between phase i and j, and ΓD and ΓN denote the parts of ∂Ω upon which Dirichlet or Neumann boundary conditions are prescribed, respectively. Here, [[·]] denotes the … view at source ↗
Figure 8
Figure 8. Figure 8: Blocking methods for Stokes problems. (left) Finite difference staggered grid: pressure nodes are cell-centered (•), velocity nodes are face-centered ( and for x- and y-components, respectively); dashed boxes depict the non-overlapping cell-wise blocking approach. (middle) Vanka patch for the Taylor-Hood element: each patch includes the central pressure node (◦) and every velocity node (•) from adjoining c… view at source ↗
Figure 9
Figure 9. Figure 9: Multigrid solver performance for the 2D (top) and 3D (bottom) staggered grid finite difference Stokes problem considered in §3.6.1. In contrast, when designing smoothers for Stokes problems, the prevailing consensus is that non-overlapping block approaches lead to ineffective multigrid solvers [10]. Roughly speaking, for saddle-point systems that contain a constraint equation (in this case, velocity diverg… view at source ↗
Figure 10
Figure 10. Figure 10: Multigrid solver performance for the 2D (top) and 3D (bottom) Taylor-Hood finite element Stokes problem considered in §3.6.2. With this setup, [PITH_FULL_IMAGE:figures/full_fig_p019_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Corresponding to a ℘ = 2 mixed-degree LDG method, multigrid solver performance for a 2D (top) and 3D (bottom) single-phase, uniform viscosity, steady-state, stress-form Stokes problem on a uniform Cartesian grid with periodic boundary conditions. of quadrature rules, etc.; see [14, 20] in the scalar elliptic case, whose generalisation to the Stokes case is straightforward [21]. In the setting of a DG meth… view at source ↗
Figure 12
Figure 12. Figure 12: Corresponding to a ℘ = 2 mixed-degree LDG method, multigrid solver performance for a 2D (top) and 3D (bottom) high-contrast multiphase, steady-state, stress-form Stokes problem on a uniform Cartesian grid with Dirichlet boundary conditions [PITH_FULL_IMAGE:figures/full_fig_p021_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Corresponding to a ℘ = 2 mixed-degree LDG method, multigrid solver performance for a 2D (left) and 3D (right) single-phase, unsteady stress-form Stokes problem on a uniform Cartesian grid with periodic boundary conditions. cascading smoothers do not have any fine-scaled parameters to tune, in practice we may still need to test which forward/reverse application yields the best results. Our final applicatio… view at source ↗
read the original abstract

Multigrid methods are among the most effective frameworks for solving large-scale sparse systems. However, achieving their hallmark linear scaling and rapid convergence crucially depends on an effective smoother algorithm, whose design is often highly problem-dependent. This paper develops a new approach, referred to as \textit{cascading smoothers} due to their operation as an ordered sequence of single-step block-diagonal smoothers. Each level in the cascade is optimised to maximally damp the output of prior steps via a Frobenius norm minimisation of the corresponding error propagators. In particular, we develop an additive (resp., multiplicative) formulation analogous to Jacobi (resp., Gauss-Seidel). Applied within a standard multigrid V-cycle, we show they are remarkably effective across a wide array of problems, including finite difference, finite element, and discontinuous Galerkin discretisations applied to Poisson, elliptic interface, and Stokes systems as well as multiphase variants. In every case, cascading smoothers closely match or significantly outperform their optimally-damped classical counterparts, yet require no parameter tuning apart from a few discrete solver choices. Additionally, the approach is highly parallelisable and robust to geometric and operator complexities such as unstructured meshes and high-contrast coefficients.

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

2 major / 0 minor

Summary. The manuscript introduces cascading smoothers for multigrid, consisting of ordered sequences of single-step block-diagonal smoothers (additive or multiplicative) in which each stage is chosen by direct Frobenius-norm minimization of the error propagator that remains after the preceding stages. These smoothers are inserted into standard V-cycles and are asserted to be effective, without parameter tuning beyond a few discrete choices, for finite-difference, finite-element and discontinuous-Galerkin discretizations of Poisson, elliptic-interface, Stokes and multiphase problems on both structured and unstructured meshes.

Significance. If the reported numerical performance is reproducible, the construction supplies a largely parameter-free route to effective smoothers for a range of operators that are otherwise difficult to treat with classical damped Jacobi or Gauss-Seidel methods. The emphasis on sequential norm minimization and the claimed robustness to high-contrast coefficients and geometric complexity constitute a potentially useful addition to the multigrid toolbox.

major comments (2)
  1. The central construction minimizes the Frobenius norm of each successive error propagator, yet the manuscript supplies no analysis showing that this procedure guarantees that the spectral radius of the composite smoother remains strictly less than one. Because the Frobenius norm only bounds the 2-norm and does not control the spectral radius for non-normal matrices, the procedure can in principle produce an unstable smoother for the Stokes or high-contrast DG operators that are included in the claimed test suite.
  2. The abstract asserts that the cascading smoothers 'closely match or significantly outperform their optimally-damped classical counterparts' across every listed discretization and PDE; however, the provided text contains no quantitative tables, iteration counts, or convergence-rate comparisons that would allow an independent assessment of this performance claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments. We respond point by point to the major comments below.

read point-by-point responses

  1. Referee: The central construction minimizes the Frobenius norm of each successive error propagator, yet the manuscript supplies no analysis showing that this procedure guarantees that the spectral radius of the composite smoother remains strictly less than one. Because the Frobenius norm only bounds the 2-norm and does not control the spectral radius for non-normal matrices, the procedure can in principle produce an unstable smoother for the Stokes or high-contrast DG operators that are included in the claimed test suite.

    Authors: We agree that the manuscript provides no theoretical analysis guaranteeing that the spectral radius of the composite smoother is strictly less than one. The Frobenius-norm minimization reduces the Frobenius norm (hence bounds the 2-norm) of each successive propagator, but this does not control the spectral radius for non-normal matrices. The paper relies on numerical evidence of stable convergence for the Stokes and high-contrast DG cases. In revision we will add an explicit discussion of this limitation together with the observed empirical stability. revision: yes

  2. Referee: The abstract asserts that the cascading smoothers 'closely match or significantly outperform their optimally-damped classical counterparts' across every listed discretization and PDE; however, the provided text contains no quantitative tables, iteration counts, or convergence-rate comparisons that would allow an independent assessment of this performance claim.

    Authors: The full manuscript contains numerical results with iteration counts and convergence comparisons. To make the abstract claim directly verifiable, we will add a compact summary table of key metrics and insert explicit references from the abstract and introduction to the relevant tables and figures. revision: yes

Circularity Check

0 steps flagged

No circularity: cascading smoothers defined by explicit independent Frobenius minimization

full rationale

The paper defines cascading smoothers directly via sequential Frobenius-norm minimization of error propagators at each cascade level (additive or multiplicative). This is an optimization procedure applied to the constructed propagators rather than a fit to solution data or a self-referential definition. Effectiveness is shown via numerical experiments on multiple discretizations and operators; no load-bearing self-citation, uniqueness theorem, or renaming of known results is present in the provided text. The central construction does not reduce to its inputs by definition.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review; the method rests on the domain assumption that Frobenius-norm minimization of successive error propagators yields effective smoothers, but no explicit free parameters, invented entities, or additional axioms are stated.

axioms (1)
  • domain assumption Error propagators of block-diagonal smoothers can be minimized level-by-level via Frobenius norm to produce an effective cascade.
    This is the core optimization step described in the abstract.

pith-pipeline@v0.9.1-grok · 5731 in / 1179 out tokens · 25945 ms · 2026-06-27T08:49:34.216688+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

30 extracted references · 23 canonical work pages · 1 internal anchor

  1. [1]

    Adams, M

    M. Adams, M. Brezina, J. Hu, and R. Tuminaro,Parallel multigrid smoothing: polynomial versus Gauss-Seidel, Journal of Computational Physics, 188 (2003), pp. 593–610, https://doi.org/10.1016/ S0021-9991(03)00194-3

    2003

  2. [2]

    Adsuara, I

    J. Adsuara, I. Cordero-Carri ´on, P. Cerd ´a-Dur´an, and M. Aloy,Scheduled Relaxation Jacobi method: Improvements and applications, Journal of Computational Physics, 321 (2016), pp. 369–413, https://doi.org/10.1016/j.jcp.2016.05.053

    work page doi:10.1016/j.jcp.2016.05.053 2016

  3. [3]

    Adsuara, I

    J. Adsuara, I. Cordero-Carri ´on, P. Cerd´a-Dur´an, V. Mewes, and M. Aloy,On the equivalence between the Scheduled Relaxation Jacobi method and Richardson’s non-stationary method, Journal of Computational Physics, 332 (2017), pp. 446–460, https://doi.org/10.1016/j.jcp.2016.12.020

    work page doi:10.1016/j.jcp.2016.12.020 2017

  4. [4]

    A. H. Baker, R. D. Falgout, T. V. Kolev, and U. M. Yang,Multigrid smoothers for ultraparallel 23 computing, SIAM Journal on Scientific Computing, 33 (2011), pp. 2864–2887, https://doi.org/10. 1137/100798806

    2011

  5. [5]

    Birken, J

    P. Birken, J. Bull, and A. Jameson,A study of multigrid smoothers used in compressible CFD based on the convection diffusion equation, in 7th European Congress on Computational Methods in Applied Sciences and Engineering, ECCOMAS Congress 2016, National Technical University of Athens, 2016, pp. 2648–2663

    2016

  6. [6]

    W. L. Briggs, V. E. Henson, and S. F. McCormick,A Multigrid Tutorial, Second Edition, Society for Industrial and Applied Mathematics, second ed., 2000, https://doi.org/10.1137/1.9780898719505

    work page doi:10.1137/1.9780898719505 2000

  7. [7]

    Br¨oker and M

    O. Br¨oker and M. J. Grote,Sparse approximate inverse smoothers for geometric and algebraic multi- grid, Applied Numerical Mathematics, 41 (2002), pp. 61–80, https://doi.org/10.1016/S0168-9274(01) 00110-6

    work page doi:10.1016/s0168-9274(01 2002

  8. [8]

    Br¨oker, M

    O. Br¨oker, M. J. Grote, C. Mayer, and A. Reusken,Robust parallel smoothing for multigrid via sparse approximate inverses, SIAM Journal on Scientific Computing, 23 (2001), pp. 1396–1417, https://doi.org/10.1137/S1064827500380623

    work page doi:10.1137/s1064827500380623 2001

  9. [9]

    Brown, Y

    J. Brown, Y. He, S. MacLachlan, M. Menickelly, and S. M. Wild,Tuning multigrid methods with robust optimization and local Fourier analysis, SIAM Journal on Scientific Computing, 43 (2021), pp. A109–A138, https://doi.org/10.1137/19M1308669

    work page doi:10.1137/19m1308669 2021

  10. [10]

    Claus and M

    L. Claus and M. Bolten,Nonoverlapping block smoothers for the Stokes equations, Numerical Linear Algebra with Applications, 28 (2021), p. e2389, https://doi.org/10.1002/nla.2389

    work page doi:10.1002/nla.2389 2021

  11. [11]

    Cockburn, G

    B. Cockburn, G. Kanschat, D. Sch ¨otzau, and C. Schwab,Local discontinuous Galerkin methods for the Stokes system, SIAM Journal on Numerical Analysis, 40 (2002), pp. 319–343, https: //doi.org/10.1137/S0036142900380121

    work page doi:10.1137/s0036142900380121 2002

  12. [12]

    J. W. Demmel,Applied Numerical Linear Algebra, Society for Industrial and Applied Mathematics, 1997, https://doi.org/10.1137/1.9781611971446

    work page doi:10.1137/1.9781611971446 1997

  13. [13]

    P. E. Farrell, Y. He, and S. P. MacLachlan,A local Fourier analysis of additive Vanka relaxation for the Stokes equations, Numerical Linear Algebra with Applications, 28 (2021), p. e2306, https: //doi.org/10.1002/nla.2306

    work page doi:10.1002/nla.2306 2021

  14. [14]

    Fortunato, C

    D. Fortunato, C. H. Rycroft, and R. I. Saye,Efficient operator-coarsening multigrid schemes for local discontinuous Galerkin methods, SIAM Journal on Scientific Computing, 41 (2019), pp. A3913– A3937, https://doi.org/10.1137/18M1206357

    work page doi:10.1137/18m1206357 2019

  15. [15]

    M. J. Grote and T. Huckle,Parallel preconditioning with sparse approximate inverses, SIAM Journal on Scientific Computing, 18 (1997), pp. 838–853, https://doi.org/10.1137/S1064827594276552

    work page doi:10.1137/s1064827594276552 1997

  16. [16]

    F. H. Harlow and J. E. Welch,Numerical Calculation of Time-Dependent Viscous Incompressible Flow of Fluid with Free Surface, The Physics of Fluids, 8 (1965), pp. 2182–2189, https://doi.org/10. 1063/1.1761178

    1965

  17. [17]

    Huckle and M

    T. Huckle and M. Sedlacek,Smoothing and regularization with modified sparse approximate inverses, Journal of Electrical and Computer Engineering, 2010 (2010), p. 930218, https://doi.org/doi.org/10. 1155/2010/930218

    2010

  18. [18]

    Lottes,Optimal polynomial smoothers for multigrid V-cycles, Numerical Linear Algebra with Applications, 30 (2023), p

    J. Lottes,Optimal polynomial smoothers for multigrid V-cycles, Numerical Linear Algebra with Applications, 30 (2023), p. e2518, https://doi.org/10.1002/nla.2518

    work page doi:10.1002/nla.2518 2023

  19. [19]

    2003.Iterative Methods for Sparse Linear Systems(second ed.)

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

    work page doi:10.1137/1.9780898718003 2003

  20. [20]

    R. I. Saye,Efficient multigrid solution of elliptic interface problems using viscosity-upwinded local discontinuous Galerkin methods, Communications in Applied Mathematics and Computational Science, 14 (2019), pp. 247–283, https://doi.org/10.2140/camcos.2019.14.247

    work page doi:10.2140/camcos.2019.14.247 2019

  21. [21]

    R. I. Saye,Fast multigrid solution of high-order accurate multiphase Stokes problems, Communications in Applied Mathematics and Computational Science, 15 (2020), pp. 147–196, https://doi.org/10. 2140/camcos.2020.15.33

    2020

  22. [22]

    R. I. Saye,A comparative study of efficient multigrid solvers for high-order local discontinuous Galerkin methods: Poisson, elliptic interface, and multiphase Stokes problems, Communications in Applied Mathematics and Computational Science, 20 (2025), pp. 253–291, https://doi.org/10.2140/camcos. 2025.20.253

    work page doi:10.2140/camcos 2025

  23. [23]

    R. I. Saye,Efficient multigrid solvers for mixed-degree local discontinuous Galerkin multiphase Stokes problems, 2025, https://doi.org/10.48550/arXiv.2506.09933, https://arxiv.org/abs/2506.09933

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2506.09933 2025

  24. [24]

    St¨uben,A review of algebraic multigrid, Journal of Computational and Applied Mathematics, 128 (2001), pp

    K. St¨uben,A review of algebraic multigrid, Journal of Computational and Applied Mathematics, 128 (2001), pp. 281–309, https://doi.org/10.1016/S0377-0427(00)00516-1. Numerical Analysis 2000. Vol. VII: Partial Differential Equations

    work page doi:10.1016/s0377-0427(00)00516-1 2001

  25. [25]

    Tang and W

    W.-P. Tang and W. L. Wan,Sparse approximate inverse smoother for multigrid, SIAM Jour- nal on Matrix Analysis and Applications, 21 (2000), pp. 1236–1252, https://doi.org/10.1137/ S0895479899339342. [26]U. Trottenberg, C. Oosterlee, and A. Sch ¨uller,Multigrid, Academic Press, 2001. 24

    2000

  26. [26]

    van Leer, C.-H

    B. van Leer, C.-H. Tai, and K. G. Powell,Design of optimally smoothing multi-stage schemes for the Euler equations, in 9th Computational Fluid Dynamics Conference, 1989, pp. 40–59, https://doi.org/10.2514/6.1989-1933

    work page doi:10.2514/6.1989-1933 1989

  27. [27]

    S. P. Vanka,Block-implicit multigrid solution of Navier-Stokes equations in primitive variables, Journal of Computational Physics, 65 (1986), pp. 138–158, https://doi.org/10.1016/0021-9991(86)90008-2

    work page doi:10.1016/0021-9991(86)90008-2 1986

  28. [28]

    S. P. Vanka,A calculation procedure for three-dimensional steady recirculating flows using multigrid methods, Computer Methods in Applied Mechanics and Engineering, 55 (1986), pp. 321–338, https://doi.org/10.1016/0045-7825(86)90058-7. [30]P. Wesseling,An Introduction to Multigrid Methods, John Wiley & Sons, 1992. [31]R. Wienands and W. Joppich,Practical F...

    work page doi:10.1016/0045-7825(86)90058-7 1986

  29. [29]

    X. I. Yang and R. Mittal,Acceleration of the Jacobi iterative method by factors exceeding 100 using scheduled relaxation, Journal of Computational Physics, 274 (2014), pp. 695–708, https: //doi.org/10.1016/j.jcp.2014.06.010

    work page doi:10.1016/j.jcp.2014.06.010 2014

  30. [30]

    bubble” surrounded by gas, and a gas “bubble

    I. Yavneh,On Red-Black SOR Smoothing in Multigrid, SIAM Journal on Scientific Computing, 17 (1996), pp. 180–192, https://doi.org/10.1137/0917013. 25 CASCADING SMOOTHERS FOR MULTIGRID SUPPLEMENTARY MATERIAL Robert I. Saye Lawrence Berkeley National Laboratory, Berkeley, California, USA|rsaye@lbl.gov|June 5, 2026 SM5. Multicoloured Cascading Smoothers.As no...

    work page doi:10.1137/0917013 1996