pith. sign in

arxiv: 2604.19230 · v1 · submitted 2026-04-21 · 🧮 math.NA · cs.NA

Preconditioners for the Onsager-Stefan-Maxwell equations for multicomponent diffusion

Kars Knook , Aaron Baier-Reinio , Patrick E. Farrell This is my paper

Pith reviewed 2026-05-10 02:22 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords Onsager-Stefan-Maxwell equationsaugmented Lagrangian preconditionermulticomponent diffusiongeometric multigridfinite element methodsPicard linearizationNewton linearizationrobust solvers
0
0 comments X

The pith

An augmented Lagrangian preconditioner is proven discretization-robust for the stationary Onsager-Stefan-Maxwell equations under ideal gaseous conditions.

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

The paper develops solvers for the Onsager-Stefan-Maxwell equations that describe coupled diffusive fluxes among multiple chemical species. It introduces an augmented Lagrangian preconditioner and proves that this preconditioner yields iteration counts independent of mesh size and polynomial degree for a Picard linearization in the isobaric, isothermal, ideal gas case. For the Newton linearization the same preconditioner serves as a block-diagonal smoother inside a monolithic geometric multigrid iteration that also incorporates vertex-star Schwarz methods. Numerical experiments then demonstrate that the overall strategy remains effective or shows only mild parameter dependence when the model is extended to cross-diffusion, nonideal mixing, temperature gradients, convection, and electrochemical driving forces.

Core claim

The central claim is that an augmented Lagrangian preconditioner for the Picard linearization of the stationary Onsager-Stefan-Maxwell equations is discretization-robust in the isobaric, isothermal, ideal gaseous setting. For the Newton linearization the preconditioner is placed as a block-diagonal smoother inside a monolithic geometric multigrid iteration combined with vertex-star Schwarz methods. The resulting solver is shown through experiments to handle a range of additional physical effects while retaining robustness or only mild dependence on mesh refinement and polynomial degree.

What carries the argument

The augmented Lagrangian preconditioner, which augments the discrete system to control the coupling between species fluxes and is used either directly or as a smoother inside geometric multigrid.

If this is right

  • The preconditioner removes the need to retune solver parameters when the mesh is refined for ideal-gas multicomponent diffusion.
  • The multigrid-Schwarz combination extends the same robustness property to the fully nonlinear Newton linearization.
  • The strategy remains applicable when the model includes temperature gradients, nonideal mixing, convection, and electrochemical effects.
  • Concrete applications such as airway cross-diffusion, gas separation, and electrolytic plating become computationally tractable at high resolution.

Where Pith is reading between the lines

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

  • If the observed numerical robustness persists beyond the tested regimes, the same framework could support routine high-resolution simulations of industrial multicomponent separation processes.
  • The combination of augmented Lagrangian augmentation with monolithic multigrid may transfer to other strongly coupled transport systems in fluid mechanics.
  • Testing the identical preconditioner on time-dependent or three-dimensional OSM problems would provide a direct check on broader applicability.

Load-bearing premise

The proof that iteration counts remain independent of discretization parameters is stated only for the isobaric, isothermal, ideal gaseous setting.

What would settle it

A sequence of refined meshes or increasing polynomial degrees in the ideal gaseous Picard case where the number of preconditioner iterations grows substantially would falsify the robustness claim.

Figures

Figures reproduced from arXiv: 2604.19230 by Aaron Baier-Reinio, Kars Knook, Patrick E. Farrell.

Figure 1
Figure 1. Figure 1: Solver diagram for the monolithic GMG preconditioner with AL smoother, [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Simplified mesh, original mesh, and mole fraction of oxygen in the human [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Carbon dioxide mole fraction on a cross section of a bronchiole and streamlines [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Schematic of the gas separation chamber and (b) helium mole fraction. [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Benzene mole fraction on a cross section of the microfluidic device and stream [PITH_FULL_IMAGE:figures/full_fig_p022_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Magnitude of the current density in the steady-state Hull cell. Current flows [PITH_FULL_IMAGE:figures/full_fig_p024_6.png] view at source ↗
read the original abstract

The Onsager-Stefan-Maxwell (OSM) equations are an important model of mass transport in multicomponent flows with multiple chemical species. They describe the coupling of diffusive fluxes between species, accounting for their interactions through frictional and thermodynamic driving forces. In this work we propose an augmented Lagrangian preconditioner and prove its discretization-robustness for a Picard linearization of the stationary OSM equations in the isobaric, isothermal, ideal gaseous setting. For the Newton linearization we employ the augmented Lagrangian preconditioner as a block diagonal smoother inside a monolithic geometric multigrid iteration and combine with vertex star Schwarz methods. This strategy is shown to be applicable in a wide variety of settings which incorporate cross-diffusion, nonideal mixing, thermal, pressure, convective, and electrochemical effects. We demonstrate robustness or mild dependence with respect to mesh refinement and polynomial degree of the proposed monolithic preconditioning strategy for different types of multicomponent flows in several applications: cross-diffusion in the human airways, separation of gases under a temperature gradient, nonideal mixing of benzene and cyclohexane, and electrolytic transport in a Hull cell undergoing electroplating.

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

1 major / 3 minor

Summary. The manuscript develops preconditioning strategies for the Onsager-Stefan-Maxwell equations governing multicomponent diffusion. It proves discretization-robustness of an augmented Lagrangian preconditioner for the Picard linearization of the stationary equations under isobaric, isothermal, ideal-gas assumptions. For the Newton linearization in more general settings (incorporating cross-diffusion, nonideal mixing, thermal, pressure, convective, and electrochemical effects), the same preconditioner is deployed as a block-diagonal smoother within a monolithic geometric multigrid iteration that is further combined with vertex-star Schwarz methods. Robustness or mild parameter dependence with respect to mesh size and polynomial degree is demonstrated numerically on four applications: cross-diffusion in human airways, gas separation under a temperature gradient, nonideal benzene-cyclohexane mixing, and electrolytic transport in a Hull cell.

Significance. If the analysis and experiments hold, the work supplies a practical and partially analyzed route to robust solvers for a class of coupled diffusion problems that arise in chemical engineering, biology, and electrochemistry. The explicit separation of the proven ideal-gas Picard case from the numerically supported general Newton case, together with the use of established multigrid and domain-decomposition components, is a constructive contribution to the literature on preconditioners for multicomponent transport.

major comments (1)
  1. [Section 3 (proof) and Section 5 (numerical results)] The discretization-robustness proof is stated only for the isobaric, isothermal, ideal-gas Picard linearization. While the manuscript correctly limits the analytic claim and supports broader applicability by numerical experiment, the absence of any a-priori bound on the augmented-Lagrangian parameter with respect to the number of species or the magnitude of the thermodynamic driving forces leaves open whether the observed robustness in the nonideal cases is accidental or structural.
minor comments (3)
  1. [Abstract and §1] The abstract and introduction would benefit from a single sentence that explicitly separates the proven ideal-gas Picard result from the numerically demonstrated general Newton result.
  2. [§2 and §4] Notation for the species mass fluxes, velocities, and chemical potentials is introduced in §2 but is not always restated when the linearized operators are written in §4; a short table of symbols would improve readability.
  3. [Figure captions in §6] The captions of the numerical figures (e.g., those showing iteration counts versus h and p) should list the precise values of the augmented-Lagrangian parameter and the number of Schwarz iterations used in each experiment.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful review and the recommendation for minor revision. We address the major comment below.

read point-by-point responses

  1. Referee: [Section 3 (proof) and Section 5 (numerical results)] The discretization-robustness proof is stated only for the isobaric, isothermal, ideal-gas Picard linearization. While the manuscript correctly limits the analytic claim and supports broader applicability by numerical experiment, the absence of any a-priori bound on the augmented-Lagrangian parameter with respect to the number of species or the magnitude of the thermodynamic driving forces leaves open whether the observed robustness in the nonideal cases is accidental or structural.

    Authors: We agree that the discretization-robustness proof is confined to the isobaric, isothermal, ideal-gas Picard linearization, as stated in the manuscript. Extending the analysis to the general Newton case with nonideal mixing and other effects would require bounding the augmented Lagrangian parameter in terms of the number of species and the thermodynamic driving forces, which is a nontrivial task due to the nonlinear and coupled nature of the thermodynamic factors. We do not currently have such an a priori bound. However, the numerical experiments in Section 5 demonstrate robustness across a range of applications with varying numbers of species and driving forces, suggesting that the observed behavior is structural rather than accidental. We will revise the manuscript to include a brief discussion in the conclusions highlighting this limitation and the supporting numerical evidence. revision: partial

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper derives an augmented Lagrangian preconditioner directly from the structure of the Picard-linearized stationary OSM equations under the stated isobaric/isothermal/ideal-gas assumptions and proves its discretization-robustness by direct analysis of the resulting saddle-point system. For the Newton linearization in broader regimes the same operator is deployed only as a smoother inside an existing monolithic geometric multigrid + vertex-star Schwarz framework, with performance assessed solely by numerical experiments on four concrete applications; the authors make no analytic claim for those cases. No equation reduces to a self-definition, no fitted parameter is relabeled as a prediction, and no load-bearing step rests on a self-citation whose content is itself unverified or circular. The central claims therefore stand on independent analytic and numerical content.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work relies on standard numerical analysis techniques for PDEs; no free parameters or invented entities are mentioned in the abstract.

axioms (2)
  • domain assumption The stationary Onsager-Stefan-Maxwell equations admit Picard and Newton linearizations.
    Used as the basis for designing and applying the preconditioners.
  • standard math Finite element discretization is employed for the spatial approximation of the equations.
    Standard assumption for such PDE problems in numerical analysis.

pith-pipeline@v0.9.0 · 5501 in / 1412 out tokens · 64471 ms · 2026-05-10T02:22:31.226457+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

14 extracted references · 14 canonical work pages

  1. [1]

    [1]P. R. Amestoy, I. S. Duff, J.-Y. L’Excellent, and J. Koster,A fully asynchronous multi- frontal solver using distributed dynamic scheduling, SIAM J. Matrix Anal. Appl., 23 (2001), pp. 15–41. [2]D. N. Arnold, R. S. F alk, and R. Winther,Multigrid inH(div)andH(curl), Numer. Math., 85 (2000), pp. 197–217. [3]F. R. Aznaran, P. E. F arrell, C. W. Monroe, an...

    work page 2001

  2. [2]

    Baier-Reinio and P

    [4]A. Baier-Reinio and P. E. F arrell,High-order finite element methods for three-dimensional multicomponent convection-diffusion, SIAM J. Sci. Comput., 48 (2025), pp. A540–A567. [5]A. Baier-Reinio, P. E. F arrell, and C. W. Monroe,Finite element methods for electroneu- tral multicomponent electrolyte flows, arXiv preprint arXiv:2510.14923, (2025). [6]K. ...

    work page arXiv 2025

  3. [3]

    Benzi and M

    [8]M. Benzi and M. A. Olshanskii,An augmented Lagrangian-based approach to the Oseen problem, SIAM J. Sci. Comput., 28 (2006), pp. 2095–2113. [9]J. Betteridge, P. E. F arrell, M. Hochsteger, C. Lackner, J. Sch ¨oberl, S. Zampini, and U. Zerbinati,ngsPETSc: a coupling between NETGEN/NGSolve and PETSc, J. Open Source Softw., 9 (2024), p

    work page 2006

  4. [4]

    PRECONDITIONERS FOR MULTICOMPONENT DIFFUSION25 [10]R. B. Bird,Transport phenomena, Appl. Mech. Rev., 55 (2002), pp. R1–R4. [11]D. Boffi, F. Brezzi, M. Fortin, et al.,Mixed Finite Element Methods and Applications, vol. 44 of Springer Ser. Comput. Math., Springer Berlin Heidelberg,

    work page 2002

  5. [5]

    Braukhoff, I

    [13]M. Braukhoff, I. Perugia, and P. Stocker,An entropy structure preserving space-time formulation for cross-diffusion systems: analysis and Galerkin discretization, SIAM J. Numer. Anal., 60 (2022), pp. 364–395. [14]P. D. Brubeck and P. E. F arrell,Multigrid solvers for the de Rham complex with optimal complexity in polynomial degree, SIAM J. Sci. Comput...

    work page 2022

  6. [6]

    [16]L. D. Dalcin, R. R. Paz, P. A. Kler, and A. Cosimo,Parallel distributed computing using Python, Adv. Water Resour., 34 (2011), pp. 1124–1139. [17]S. C. Eisenstat and H. F. W alker,Choosing the forcing terms in an inexact Newton method, SIAM J. Sci. Comput., 17 (1996), pp. 16–32. [18]A. Ern and V. Giovangigli,Multicomponent Transport Algorithms, vol. 2...

    work page 2011

  7. [7]

    [19]P. E. F arrell, M. G. Knepley, L. Mitchell, and F. Wechsung,PCPATCH: software for the topological construction of multigrid relaxation methods, ACM Trans. Math. Softw., 47 (2021), p. 1–22. [20]A. Fick, ¨Uber diffusion, Anal. Phys., 170 (1855), pp. 59–86. [21]C. Geuzaine and J.-F. Remacle,Gmsh: A 3-D finite element mesh generator with built- in pre- an...

    work page 2021

  8. [8]

    Hafskjold, T

    [25]B. Hafskjold, T. Ikeshoji, and S. K. Ratkje,On the molecular mechanism of thermal diffusion in liquids, Mol. Phys., 80 (1993), pp. 1389–1412. [26]W. W. Hager,Updating the inverse of a matrix, SIAM Rev., 31 (1989), pp. 221–239. [27]D. A. Ham, P. H. J. Kelly, L. Mitchell, C. J. Cotter, R. C. Kirby, K. Sagiyama, N. Bouziani, S. Vorderwuelbecke, T. J. Gre...

    work page 1993

  9. [9]

    Helfand,On inversion of the linear laws of irreversible thermodynamics, J

    [28]E. Helfand,On inversion of the linear laws of irreversible thermodynamics, J. Chem. Phys., 33 (1960), pp. 319–322. [29]X. Huo, A. J ¨ungel, and A. E. Tzavaras,Existence and weak–strong uniqueness for Maxwell– Stefan–Cahn–Hilliard systems, Ann. Inst. Henri Poincar´ e C, 41 (2023), pp. 797–852. [30]R. C. Kirby and L. Mitchell,Solver composition across t...

    work page 1960

  10. [10]

    Onsager,Theories and problems of liquid diffusion, Ann

    [38]L. Onsager,Theories and problems of liquid diffusion, Ann. N. Y. Acad. Sci., 46 (1945), pp. 241–265. [39]A. M. Ostrowski,A quantitative formulation of Sylvester’s law of inertia, Proc. Natl. Acad. 26K. KNOOK, A. BAIER-REINIO, P. E. FARRELL Sci. USA, 45 (1959), pp. 740–744. [40]J. H. Perry, ed.,Chemical Engineers’ Handbook, McGraw-Hill Book Company, In...

    work page 1945

  11. [11]

    Rathgeber, D

    [41]F. Rathgeber, D. A. Ham, L. Mitchell, M. Lange, F. Luporini, A. T. McRae, G.-T. Bercea, G. R. Markall, and P. H. Kelly,Firedrake: automating the finite element method by composing abstractions, ACM Trans. Math. Softw., 43 (2016), pp. 1–27. [42]G. W. Richardson, J. M. Foster, R. Ranom, C. P. Please, and A. M. Ramos,Charge transport modelling of Lithium...

    work page 2016

  12. [12]

    Saad and M

    [43]Y. Saad and M. H. Schultz,GMRES: A generalized minimal residual algorithm for solving nonsymmetric linear systems, SIAM J. Sci. Stat. Comput., 7 (1986), pp. 856–869. [44]J. Sch ¨oberl,NETGEN: an advancing front 2D/3D-mesh generator based on abstract rules, Comput. Vis. Sci., 1 (1997), pp. 41–52. [45]J. Sch ¨oberl,Multigrid methods for a parameter depe...

    work page 1986

  13. [13]

    Udoetok,Thermal conductivity of binary mixtures of gases, Front

    [49]E. Udoetok,Thermal conductivity of binary mixtures of gases, Front. Heat Mass Transf., 4 (2013). [50]A. V an-Brunt, P. E. F arrell, and C. W. Monroe,Augmented saddle-point formulation of the steady-state Stefan–Maxwell diffusion problem, IMA J. Appl. Math., 42 (2022), pp. 3272–3305. [51]A. V an-Brunt, P. E. F arrell, and C. W. Monroe,Consolidated theo...

    work page 2013

  14. [14]

    [56]Software used in ‘Preconditioners for the Onsager–Stefan–Maxwell equations for multicompo- nent diffusion’, 2026, https://doi.org/10.5281/zenodo.19592785

    work page doi:10.5281/zenodo.19592785 2026