Partitioned Conservative, Variable Step, Second-Order Method for Magneto-hydrodynamics In Els\"asser Variables

Catalin Trenchea; Wenlong Pei; Zhen Yao

REVIEW 4 major objections 4 minor 42 references

Partitioned Conservative, Variable Step, Second-Order Method for Magneto-hydrodynamics In Els\"asser Variables

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper proposes a partitioned iterative method for the MHD system in Elsässer variables and proves it converges linearly to the fully coupled implicit midpoint solution while conserving energy, cross-helicity, and magnetic helicity.

desk verdict Solid and useful analysis of a practical partitioned midpoint MHD solver, with a real gap between the conservation claim and the convergence theory. read the letter →

arxiv 2507.12700 v1 pith:Z3AVLJBH submitted 2025-07-17 math.NA cs.NA

Zhen Yao , Catalin Trenchea , Wenlong Pei This is my paper

classification math.NAcs.NA MSC 76W0565M1265M6035Q3035Q61

keywords magnetohydrodynamicsElsässervariablespartitionedtime-steppingimplicitmidpointmethodlinearconvergencequadraticinvariantsvariablestepsizetimeadaptivity

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

This paper proposes an iterative partitioned solver for the magnetohydrodynamic equations written in Elsässer variables and tries to establish that it matches the accuracy and conservation of the fully coupled implicit midpoint rule at lower cost. The key claim is that the partitioned iterations converge linearly to the midpoint solution whenever the step is small relative to a curvature measure of the exact solution, and that the resulting variable-step method is second-order in time and conserves energy, cross-helicity, and magnetic helicity. A sympathetic reader would care because the scheme splits each time step into two half-size subproblems that can be solved in parallel, so the same physical fidelity costs less per step, with a local-truncation-error controller to choose steps. Numerical tests on a travelling-wave solution and Hartmann flows support the accuracy and conservation statements, and a stiff Lindberg-type test shows the adaptive version beats constant stepping at equal step count.

What carries the argument

The central mechanism is the partition-and-refactorization: the midpoint rule (1.4) is refactored into a backward-Euler solve over the first half-interval followed by the linear post-process $z^\pm_{n+1} = 2z^\pm_{n+1/2} - z^\pm_n$, and the backward-Euler solve is replaced by the fixed-point iteration (1.6), where each Elsässer component is updated using the other component from the previous iterate. This makes two half-size linear subproblems that can run in parallel. The convergence proof is carried by a weighted $H^1(\Omega)$ error norm $G(k)$ that absorbs the mixed diffusion term through a polarization identity and controls the cross-coupling convection terms through the Ladyzhenskaya inequality; the time-step restriction (3.1) is exactly the condition that keeps the contraction factor positive, with rate $1 - 2\nu\nu_m/(\nu^2+\nu\nu_m+\nu_m^2)$ from (3.2). Closure of the space-time error analysis uses the Stokes projection and a discrete Grönwall inequality, with a condition (3.19) of the same flavor.

What would settle it

Take the 2D travelling-wave test with $B_\circ=(0,0)$ and $\nu=\nu_m=2.5\times 10^{-4}$, and for each step compute $\gamma_n$ from the known exact solution (5.2). If $\tau_n$ is chosen to violate (3.1) yet the measured contraction factor still obeys (3.2) on a long run, the theorem's condition is not necessary; if instead the iterate errors stop contracting at the predicted rate on a single such run, that confirms the restriction is doing real work.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the fully coupled, symplectic, second-order implicit midpoint discretization of the Elsässer-variable MHD system can be solved by a partitioned fixed-point iteration that decouples the two Elsässer components, and that this iteration inherits the midpoint method's guarantees. Specifically, Theorem 3.1 proves the iterates converge linearly, in a weighted $H^1$ norm, to the half-step solution of the fully coupled scheme under a time-step restriction depending on the gradient of the exact solution at midpoints; Theorem 3.2 proves unconditional stability; Theorem 3.3 proves exact conservation of the three quadratic invariants in the ideal inviscid, unforced case; and Theorem 3.5 proves second-order space-time error estimates in $L^2$ and $H^1$. The paper presents the refactorization of midpoint into a backward-Euler half-step plus linear postprocess as the device that turns the monolithic solve into two half-size problems, and uses an AB2-like local truncation error estimator to drive step-size selection.

Load-bearing premise

The convergence and error proofs require the time step to be small enough relative to the slope of the exact solution at each half-step, but that slope is not known before the solution is computed and the implemented algorithm never checks the resulting bound.

Editorial extensions

If this is right

If (3.1) holds, each time step's iteration cost is bounded by a fixed-point contraction with rate independent of $h$, so the partition does not degrade the nonlinear solve.
The method is second-order accurate on arbitrary time grids, because it inherits the midpoint rule's variable-step consistency.
Energy, cross-helicity, and magnetic helicity are conserved at the discrete level in the ideal case (zero forcing, zero viscosity and diffusivity), which makes the method suited to long-time MHD runs.
The AB2-like local-truncation-error estimator gives a cheap way to adapt the step; the stiff Lindberg-type numerical test indicates the adaptive version achieves smaller end-time error than constant stepping with the same number of steps.
Compared with the BDF2-AB2 partitioned method used as a baseline, the midpoint-based scheme keeps second-order accuracy even for a vanishing external field, where the baseline loses order on the travelling-wave test.

Reading between the lines

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

Beyond the paper: the contraction proof's rate depends only on $\nu$ and $\nu_m$, not on $h$, so the same decoupling idea should transfer to other quadratic-conservative systems written as two coupled transport equations, such as other two-field plasma or wave models.
Beyond the paper: since condition (3.1) uses the exact solution's half-step gradient, a practical implementation could estimate $\gamma_n$ from the current iterate and either enforce the bound by shrinking $\tau_n$ or accept it as a heuristic; the numerical tests suggest the heuristic regime is benign, but that is an empirical question the paper leaves open.
Beyond the paper: the conserved quantities could serve as post-hoc indicators, since a run that drifts in $E_n$, $HC_n$, or $HM_n$ after many steps would signal that the solver left the conservative regime, which is a cheap diagnostic the paper does not discuss.
Beyond the paper: the AB2-like LTE estimator is essentially free because it reuses previous solutions, so the same controller could be attached to other one-leg or symplectic integrators for MHD, provided their step-change stability is verified.

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.

Desk Editor's Note

read the letter

Here's my take. The paper is a serious numerical analysis contribution, not a breakthrough. What is actually new: a proof of linear convergence for a partitioned iteration that decouples the two Elsässer equations inside the refactorized midpoint method, and a variable-step finite element error analysis for the resulting scheme. The adaptive LTE controller is standard but nicely implemented and tested. The numerics are credible: second-order convergence is demonstrated, the comparison with BDF2-AB2 is informative, and the stiff Lindberg test shows the adaptive mechanism doing real work. No free parameters; the stability proof is clean; the conservation proof for the ideal case is correct.

The soft spots are real but not fatal. First, the abstract and conclusion say the algorithm 'unconditionally conserves' energy, cross-helicity, and magnetic helicity. Theorem 3.3 proves conservation only for f=0 and ν=ν_m=0. That gap is not just wording: the stress-test note is on target. In the ideal regime, the right-hand side of the step restriction (3.1) is zero and the contraction factor in (3.2) becomes 1, so Theorem 3.1 gives no linear convergence guarantee for the exact-conservation case. The method may still work in practice, but the theory does not cover the intersection of the paper's two headline properties. The authors should either prove a separate convergence result for the inviscid case, or explicitly state the limitation.

Second, the convergence and error theorems depend on step restrictions involving γ_n and Z_n, i.e., norms of the exact half-step solution. This is common in a priori error estimates, but the adaptive controller never checks these conditions, so the practical algorithm is not guaranteed to stay in the regime the theory covers. A short discussion of this limitation is needed.

Third, the relationship to ref. [39], which has the same title as this paper, is never clarified. The reader needs to know whether this is an extended version and what the new results are.

Minor: magnetic helicity conservation is claimed but never tested numerically; only energy and cross-helicity plots appear.

Who should read it: numerical PDE researchers working on partitioned and adaptive methods for MHD. It deserves serious peer review; I'd send it out. I'd ask for revisions that qualify the conservation claims and address the ideal-case convergence gap.

Referee Report

4 major / 4 minor

Summary. The paper proposes a partitioned iterative solver for the fully implicit midpoint discretization of the incompressible MHD equations in Elsässer variables. At each half-step, the nonlinear backward Euler system is solved by a fixed-point iteration that decouples the z+ and z− equations into two half-size problems; after convergence, the next value is obtained by linear extrapolation. The authors prove linear H1 convergence of the iteration under time-step condition (3.1), unconditional stability of the monolithic midpoint method, exact conservation of energy, cross-helicity and magnetic helicity in the ideal dissipation-free case, and second-order space-time error estimates under condition (3.19). An AB2-like local truncation error estimator and an adaptive step controller are described and tested on a travelling-wave problem, a Hartmann flow, and a stiff Lindberg-type flow.

Significance. If the results hold, the main contribution is an efficient way to make the conservative midpoint method computationally feasible for MHD by decoupling the two Elsässer components while retaining the favorable properties of the fully coupled method. The proofs are detailed and self-contained, the error analysis is carried out at the fully discrete finite element level, and the numerical experiments cover convergence, long-time behavior, and adaptivity. The main reservations concern the mismatch between the advertised claims (unconditional conservation, proven convergence) and the actual theorems, which are conditional and restricted to the ideal case. The core numerical analysis appears sound, and the issues are fixable by careful restatement and additional discussion.

major comments (4)

[Abstract and §6] The abstract and the conclusion state that the algorithm 'unconditionally conserves' the energy, cross-helicity and magnetic helicity, but Theorem 3.3 proves conservation only in the ideal case f=0 and ν=ν_m=0, and only for the spatially continuous formulation (1.4)/(3.9). The fully discrete finite element method (3.15), which is what the numerical tests actually implement, is not covered by that theorem. Since Section 5 uses ν=ν_m>0 and finite elements, the reported conservation errors in Figures 5.1 and 5.2 are nonzero, so the unconditional conservation claim is not supported. The claims should be restated to match the hypotheses of Theorem 3.3.
[Theorem 3.1 / §3.1] There is a load-bearing gap between the convergence theorem and the conservation theorem in the regime the paper advertises as its main advantage. In the ideal case ν=ν_m=0, the right-hand side of the time-step restriction (3.1) vanishes because of the factor (2νν_m/(d(ν+ν_m)))^{d/(4-d)}, and the contraction factor in (3.2) degenerates to 1. Thus Theorem 3.1 does not establish convergence of the partitioned iteration (1.6) exactly when Theorem 3.3 proves exact conservation. The paper should either prove a degenerate-case convergence result, or explicitly state that the convergence theorem requires positive dissipation and explain how exact conservation is obtained in practice in the ideal case.
[Theorems 3.1 and 3.5 / §3.1 and §3.4] The convergence and error theorems are conditional on time-step restrictions (3.1) and (3.19) that depend on unknown exact-solution quantities: γ_n in (3.1) and Z_n in (3.19). The adaptive controller in Section 4 (Algorithm 1) never checks or enforces these restrictions, so the proven bounds are not verified for the implemented algorithm. The authors should either provide a computable a posteriori criterion, or report in the numerical tests whether the restrictions are actually satisfied for the chosen adaptive and constant steps, or clearly state that the theorems are conditional regularity/step-size results that the implementation does not monitor.
[Theorem 3.5 / §3.4] The error estimates in Theorem 3.5 are for the fully coupled midpoint solution (the limit of the iteration), not for the finite k-th iterate z±_{(k)} produced by (1.6) after a stopping criterion is applied. The paper uses a tolerance on successive differences in Section 5, but no combined error bound in terms of τ, h, and the iteration count is provided. A statement such as ||z±_{(k)} - z±(t_n)|| ≤ C(τ^2 + h^r + ρ^k) would make the overall second-order accuracy claim for the partitioned method rigorous.

minor comments (4)

[Eq. (3.6)] In the display after testing with 2z±_{n+1/2}, the mixed dissipation term is written as 2ν−(z∓_{n+1/2}, z±_{n+1/2}); it should be the inner product of gradients, 2ν−(∇z∓_{n+1/2}, ∇z±_{n+1/2}), so that the subsequent polarization identity is dimensionally consistent.
[Theorem 3.5 / §3.4] The regularity assumptions use the space (H^ℓ)^d without specifying ℓ (presumably ℓ = r+1 or a sufficiently large value). The value of ℓ should be stated explicitly so that the norms in (3.18) are unambiguous.
[Section 5.2] The reported convergence rates for the Hartmann flow are called 'super convergence' in the text; since the exact solution is steady, this elevated rate is likely problem-specific and should not be presented as a general property of the method.
[Algorithm 1 / §4] The 'else' branch of Algorithm 1 says the current step is adjusted and the solution recomputed, but the pseudo-code does not show the loop back to the computation step; making this control flow explicit would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's claims are proved from the midpoint scheme by explicit estimates, not by construction or by load-bearing self-citations.

full rationale

The derivation chain is self-contained. The monolithic midpoint discretization (1.4) and its BE refactorization (1.5) are the starting input; the partitioned iteration (1.6) is a fixed-point iteration, and Theorem 3.1 proves contraction in the weighted H1 norm G(k) via energy estimates and the Ladyzhenskaya inequality, under the explicit condition (3.1). The contraction factor in (3.2) is an algebraically derived constant, not an assumed rate. Theorem 3.3 proves conservation of energy, cross-helicity, and magnetic helicity directly from the u-B form (3.9) using skew-symmetry and boundary conditions; it does not assume the conclusion. Theorem 3.5 derives the second-order error bound from the consistency estimates (3.16)-(3.17), Stokes projection bounds, and a discrete Gronwall argument; no parameter is fitted to the target error to manufacture the rate. The only load-bearing citations are to the known refactorization of the midpoint rule [8] and standard adaptive controllers [22], and these are auxiliary identities rather than unverified uniqueness claims: they do not carry the convergence or conservation conclusions. The step restrictions (3.1)/(3.19) depend on exact-solution gradients and are not enforced by Algorithm 1; that is a verifiability/coverage limitation (and the nu=nu_m=0 degeneration is a real gap in the theory), but it is not circular: the theorems simply fail to apply in that limit rather than deriving their conclusions from the same assumptions. No equation in the paper reduces to an earlier equation by definition, and no fitted quantity is relabeled as a prediction.

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

The analysis relies on standard numerical analysis tools (inf-sup condition, Sobolev inequalities, regularity assumptions) and the known midpoint refactorization. No new physical entities or fitted parameters are introduced. The main additional assumption is the unverifiable time step restriction depending on the exact solution gradient.

assumptions (6)

standard math The discrete inf-sup (LBB) condition holds for the chosen finite element spaces, e.g., Taylor-Hood P2-P1.
Used in the Stokes projection estimate (2.6) and throughout the error analysis. Assumed in Section 2, equation (2.4).
standard math The Ladyzhenskaya and Sobolev embedding inequalities (2.9) and (2.10) hold in the bounded domain.
Invoked in the convergence analysis of subiterates (Section 3.1) and in the error estimates (Section 3.4).
domain assumption The exact solution satisfies the regularity assumptions (3.18), including z± in L∞(H^{r+1}), z±_tt in L^2(H^{r+1}), and p in L^2(H^{s+1}).
Required for the error estimates in Theorem 3.5. These are standard but not always guaranteed for MHD solutions with rough data.
domain assumption The time step tau_n satisfies the smallness conditions (3.1) and (3.19), which depend on gamma_n, the H1 norm of the exact solution at the half-step.
The linear convergence and error estimates hold only under these restrictions. The restrictions are not verifiable a priori and are not enforced by the implemented algorithm.
standard math The refactorization of the midpoint rule into a backward Euler solve plus a linear extrapolation (Step 1 and Step 2 in (1.5)) is valid.
This is a known result from Burkardt and Trenchea [8], cited in the introduction and used as the basis for the partitioned iteration.
standard math For divergence-free B with B = curl A, a vector potential A exists with A=0 on the boundary.
Used in the magnetic helicity conservation proof in Theorem 3.3. Standard result from Girault and Raviart [20].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Partitioned Conservative, Variable Step, Second-Order Method for Magneto-hydrodynamics In Els\"asser Variables." pith.science (2026). https://pith.science/paper/Z3AVLJBH

@misc{pith2026250712700,
  author       = {Pith},
  title        = {Pith review of: Partitioned Conservative, Variable Step, Second-Order Method for Magneto-hydrodynamics In Els\"asser Variables},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z3AVLJBH}},
  note         = {Machine review of arXiv:2507.12700}
}

abstract

Magnetohydrodynamics (MHD) describes the interaction between electrically conducting fluids and electromagnetic fields. We propose and analyze a symplectic, second-order algorithm for the evolutionary MHD system in Els\"asser variables. We reduce the computational cost of the iterative non-linear solver, at each time step, by partitioning the coupled system into two subproblems of half size, solved in parallel. We prove that the iterations converge linearly, under a time step restriction similar to the one required in the full space-time error analysis. The variable step algorithm unconditionally conserves the energy, cross-helicity and magnetic helicity, and numerical solutions are second-order accurate in the $L^{2}$ and $H^{1}$-norms. The time adaptive mechanism, based on a local truncation error criterion, helps the variable step algorithm balance accuracy and time efficiency. Several numerical tests support the theoretical findings and verify the advantage of time adaptivity.

Figures

Figures reproduced from arXiv: 2507.12700 by the authors.

**Figure 5.1.** The performance of the BDF2-AB2 algorithm is a slig [PITH_FULL_IMAGE:figures/full_fig_p024_5_1.png] view at source ↗

**Figure 5.2.** The error in the cross helicity HCn behaves similarly to the errors in the kinetic energy, for both algorithms. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_5_2.png] view at source ↗

**Figure 5.3.** Time component function proposed by Lindberg [PITH_FULL_IMAGE:figures/full_fig_p026_5_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 5.4.** Figure 5.4: Both algorithms obtain the true pattern in kinetic [PITH_FULL_IMAGE:figures/full_fig_p027_5_4.png]

**Figure 5.5.** Figure 5.5: The adaptive algorithm outperforms the correspon [PITH_FULL_IMAGE:figures/full_fig_p027_5_5.png]

**Figure 5.6.** Figure 5.6: The adaptive algorithm (Algorithm (1)) balances the efficiency and accuracy: the estimator of LTE Tbn+1 at most time steps is close to, but below the required tolerance Tol = 1.e−4, so that τn at most time steps is above τmin. The adaptive algorithm (Algorithm (1)) …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

42 extracted references · 38 canonical work pages

[39]

Trenchea

C. Trenchea. Partitioned conservative, variable step , second-order method for magneto- hydrodynamics in Els¨ asser variables. ROMAI J., 15(2):117–137, 2019

work page 2019
[1]

Alfv´ en

H. Alfv´ en. Existence of electromagnetic-hydrodynami c waves. Nature, 150:405, 1942

work page 1942
[2]

U. M. Ascher and S. Reich. The midpoint scheme and variant s for Hamiltonian systems: advantages and pitfalls. SIAM J. Sci. Comput., 21(3):1045–1065, 1999

work page 1999
[3]

Barleon, V

L. Barleon, V. Casal, and L. Lenhart. MHD ﬂow in liquid-me tal-cooled blankets. Fusion Eng. Des., 14:401 – 412, 1991

work page 1991
[4]

J. D. Barrow, R. Maartens, and C. G. Tsagas. Cosmology wit h inhomogeneous magnetic ﬁelds. Phys. Rep., 449(6):131–171, 2007

work page 2007
[5]

P. B. Bochev and C. Scovel. On quadratic invariants and sy mplectic structure. BIT, 34(3):337– 345, 1994

work page 1994
[6]

S. C. Brenner and L. R. Scott. The mathematical theory of ﬁnite element methods , volume 15 of Texts in Applied Mathematics. Springer-Verlag, New York, 1994

work page 1994
[7]

Burkardt, W

J. Burkardt, W. Pei, and C. Trenchea. A stress test for the midpoint time-stepping method. Int. J. Numer. Anal. Model., 19(2-3):299–314, 2022

work page 2022

Show all 42 references

[8]

Burkardt and C

J. Burkardt and C. Trenchea. Refactorization of the midp oint rule. Appl. Math. Lett., 107:106438, 2020

2020
[9]

Capuano, B

F. Capuano, B. Sanderse, E. De Angelis, and G. Coppola. A m inimum-dissipation time- integration strategy for large-eddy simulation of incompr essible turbulent ﬂows. In AIMETA 2017 Proceedings of the XXIII Conference of the Italian Association of Theoretical and Applied Mecha...

2017
[10]

P. G. Ciarlet. The ﬁnite element method for elliptic problems, volume 40 of Classics in Applied Mathematics. Society for Industrial and Applied Mathemati cs (SIAM), Philadelphia, PA,
[11]

G. G. Dahlquist, W. Liniger, and O. Nevanlinna. Stabili ty of two-step methods for variable integration steps. SIAM J. Numer. Anal., 20(5):1071–1085, 1983

1983
[12]

P. A. Davidson. An introduction to magnetohydrodynamics. Cambridge Texts in Applied Mathematics. Cambridge University Press, Cambridge, 2001

2001
[13]

Dobrowolny, A

M. Dobrowolny, A. Mangeney, and P. Veltri. Fully develo ped anisotropic hydromagnetic tur- bulence in interplanetary space. Phys. Rev. Lett., 45(2):144–147, 1980

1980
[14]

Dormy and M

E. Dormy and M. N´ u˜ nez. Introduction [Special issue: M agnetohydrodynamics in astrophysics and geophysics]. Geophys. Astrophys. Fluid Dyn., 101(3-4):169, 2007

2007
[15]

W. M. Els¨ asser. The hydromagnetic equations. Phys. Rev., 79:183–183, Jul 1950

1950
[16]

J. A. Font. General relativistic hydrodynamics and mag netohydrodynamics: hyperbolic systems in relativistic astrophysics. In Hyperbolic problems: theory, numerics, applications , pages 3–17. Springer, Berlin, 2008

2008
[17]

G. P. Galdi. An introduction to the mathematical theory of the Navier-Stokes equations . Springer Monographs in Mathematics. Springer, New York, se cond edition, 2011. Steady- state problems

2011
[18]

Galtier, S

S. Galtier, S. V. Nazarenko, A. C. Newell, and A. Pouquet . A weak turbulence theory for incompressible magnetohydrodynamics. J. Plasma Phys., 63:447–488, 6 2000. 29

2000
[19]

Gerbeau, C

J.-F. Gerbeau, C. Le Bris, and T. Leli` evre. Mathematical methods for the magnetohydrodynamics of liquid metals. Numerical Mathematics and Scientiﬁc Computa- tion. Oxford University Press, Oxford, 2006

2006
[20]

Girault and P.-A

V. Girault and P.-A. Raviart. Finite element methods for Navier-Stokes equations , volume 5 of Springer Series in Computational Mathematics. Springer-Verlag, Berlin, 1986. Theory and algorithms

1986
[21]

Goldreich and S

P. Goldreich and S. Sridhar. Toward a theory of interste llar turbulence. II: Strong Alfv´ enic turbulence. ApJ, 438:763–775, 1995

1995
[22]

Hairer, S

E. Hairer, S. P. Nørsett, and G. W anner. Solving ordinary diﬀerential equations. I, volume 8 of Springer Series in Computational Mathematics. Springer-Verlag, Berlin, second edition,
[23]

Hashizume

H. Hashizume. Numerical and experimental research to s olve MHD problem in liquid blanket system. Fusion Eng. Des., 81:1431 – 1438, 2006

2006
[24]

J. G. Heywood and R. Rannacher. Finite-element approxi mation of the nonstationary Navier- Stokes problem. IV. Error analysis for second-order time di scretization. SIAM J. Numer. Anal., 27(2):353–384, 1990

1990
[25]

V. John. Finite element methods for incompressible ﬂow problems, volume 51 of Springer Series in Computational Mathematics. Springer, Cham, 2016

2016
[26]

R. H. Kraichnan. Inertial-range spectrum of hydromagn etic turbulence. Phys. Fluids, 8(7):1385–1387, 1965

1965
[27]

Layton, W

W. Layton, W. Pei, Y. Qin, and C. Trenchea. Analysis of th e variable step method of Dahlquist, Liniger and Nevanlinna for ﬂuid ﬂow. Numer. Methods Partial Diﬀerential Equations, 38(6):1713–1737, 2022

2022
[28]

Layton, W

W. Layton, W. Pei, and C. Trenchea. Refactorization of a variable step, unconditionally stable method of Dahlquist, Liniger and Nevanlinna. Appl. Math. Lett., 125:Paper No. 107789, 2022

2022
[29]

Layton, W

W. Layton, W. Pei, and C. Trenchea. Time step adaptivity in the method of Dahlquist, Liniger and Nevanlinna. Advances in Computational Science and Engineering, 1(3):320–350, 2023

2023
[30]

Li and C

Y. Li and C. Trenchea. Partitioned second order method f or magnetohydrodynamics in Els¨ asser variables. Discrete Contin. Dyn. Syst. Ser. B, 23(7):2803–2823, 2018

2018
[31]

T. Lin, J. Gilbert, R. Kossowsky, and P. S. U. S. COLLEGE. Sea-W aterMagnetohydrodynamic Propulsion for Next-Generation Undersea Vehicles. Defense Technical Information Center, 1990

1990
[32]

Lindberg

B. Lindberg. On a dangerous property of methods for stiﬀ diﬀerential equations. Nordisk Tidskr. Informationsbehandling (BIT), 14:430–436, 1974

1974
[33]

W. Pei. The semi-implicit DLN algorithm for the Navier- Stokes equations. Numer. Algorithms, 97(4):1673–1713, 2024

2024
[34]

W. Pei. The variable time-stepping DLN-ensemble algor ithms for incompressible Navier-Stokes equations. arXiv 2407.19101, 2024

2024 arXiv
[35]

B. Punsly. Black hole gravitohydromagnetics, volume 355 of Astrophysics and Space Science Library. Springer-Verlag, Berlin, second edition, 2008

2008
[36]

Shehzad, T

S. Shehzad, T. Hayat, and A. Alsaedi. Inﬂuence of convec tive heat and mass conditions in MHD ﬂow of nanoﬂuid. Bull. Pol. Acad. Sci. Tech. Sci., 63(2):465–474, 2015

2015
[37]

Siddiqua and W

F. Siddiqua and W. Pei. Variable time step method of Dahl quist, Liniger and Nevanlinna (DLN) for a corrected Smagorinsky model. Int. J. Numer. Anal. Model., 21(6):879–909, 2024

2024
[38]

Trenchea

C. Trenchea. Unconditional stability of a partitioned IMEX method for magnetohydrodynamic ﬂows. Appl. Math. Lett., 27:97–100, 2014

2014
[40]

Wilson, A

N. Wilson, A. Labovsky, and C. Trenchea. High Accuracy M ethod for Magnetohydrodynamics System in Els¨ asser Variables. Comput. Methods Appl. Math., 15(1):97–110, 2015

2015
[41]

Yuksel and R

G. Yuksel and R. Ingram. Numerical analysis of a ﬁnite el ement, Crank-Nicolson discretization for MHD ﬂows at small magnetic Reynolds numbers. Int. J. Numer. Anal. Model., 10(1):74– 98, 2013. 30

2013
[2002]

Reprint of the 1978 original [North-Holland, Amsterd am]

1978

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [39]

Trenchea

C. Trenchea. Partitioned conservative, variable step , second-order method for magneto- hydrodynamics in Els¨ asser variables. ROMAI J., 15(2):117–137, 2019

work page 2019

[2] [1]

Alfv´ en

H. Alfv´ en. Existence of electromagnetic-hydrodynami c waves. Nature, 150:405, 1942

work page 1942

[3] [2]

U. M. Ascher and S. Reich. The midpoint scheme and variant s for Hamiltonian systems: advantages and pitfalls. SIAM J. Sci. Comput., 21(3):1045–1065, 1999

work page 1999

[4] [3]

Barleon, V

L. Barleon, V. Casal, and L. Lenhart. MHD ﬂow in liquid-me tal-cooled blankets. Fusion Eng. Des., 14:401 – 412, 1991

work page 1991

[5] [4]

J. D. Barrow, R. Maartens, and C. G. Tsagas. Cosmology wit h inhomogeneous magnetic ﬁelds. Phys. Rep., 449(6):131–171, 2007

work page 2007

[6] [5]

P. B. Bochev and C. Scovel. On quadratic invariants and sy mplectic structure. BIT, 34(3):337– 345, 1994

work page 1994

[7] [6]

S. C. Brenner and L. R. Scott. The mathematical theory of ﬁnite element methods , volume 15 of Texts in Applied Mathematics. Springer-Verlag, New York, 1994

work page 1994

[8] [7]

Burkardt, W

J. Burkardt, W. Pei, and C. Trenchea. A stress test for the midpoint time-stepping method. Int. J. Numer. Anal. Model., 19(2-3):299–314, 2022

work page 2022