Pith. sign in

REVIEW 1 major objections 7 minor 18 references

Simulating the photospheric to coronal plasma using magnetohydrodyanamic characteristics II: reflections on non-reflecting boundary conditions

T0 review · 1 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper shows that two common MHD non-reflecting boundary conditions, 'Fixed' (incoming characteristics set to zero) and 'Cancellation' (incoming set to cancel transverse terms), both fail at sub-Alfvénic advection of a spheromak: one…

desk verdict Careful warning: standard NRBCs can badly misbehave for sub-Alfvenic advection of magnetic structures; the main caveat is real but minor. read the letter →

arxiv 2501.17995 v1 pith:MYHDI5TV submitted 2025-01-29 astro-ph.SR physics.plasm-ph

N. Dylan Kee , Lucas A. Tarr , Peter W. Schuck , Mark G. Linton , James E. Leake This is my paper · ORCID
classification astro-ph.SRphysics.plasm-ph
keywords non-reflectingboundaryconditionscharacteristics-basedMHDLaRe3Dspheromakadvectionmagnetohydrodynamicsimulationssolarcoronaincomingcharacteristicderivativesforcebalanceatboundaries
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 argues that two standard implementations of non-reflecting boundary conditions (NRBCs) in magnetohydrodynamic (MHD) simulations do not do what users typically hope: when a self-contained, force-balanced magnetic structure such as a spheromak is advected through the boundary at sub-Alfvénic speed, a 'Fixed' NRBC makes the structure bounce off the wall and a 'Cancellation' NRBC accelerates it out of the box, while the same structure in a larger ground-truth simulation simply passes through at constant velocity. The authors implement both NRBCs in the LaRe3D code by coupling it to a characteristics-based solver, and they show the deviations are the boundary condition working as mathematically designed: the incoming characteristic derivatives, which would encode the external universe's response, are either zeroed out or set only from transverse interior information, which cannot preserve the force balance of an advecting magnetic structure. If true, the common practice of trusting a small box with an NRBC to reproduce a larger simulation is unsafe for magnetically structured outflows, and the errors are not numerical artifacts but intrinsic to the boundary-condition design.

What carries the argument

The central object is the characteristics-based decomposition of ideal MHD in the direction normal to the boundary. The flux Jacobian in that direction is diagonalized to obtain eight boundary-normal characteristic derivatives $L_\sigma$ (entropy, Alfvén, slow magnetosonic, and fast magnetosonic modes) with propagation speeds $v_z$, $v_z\pm c_a$, $v_z\pm c_s$, and $v_z\pm c_f$. Outgoing derivatives are computed from the simulation interior; incoming derivatives must be prescribed by the boundary condition. The paper implements two prescriptions: the Fixed NRBC sets all incoming $L_{\sigma,I}=0$, and the Cancellation NRBC sets $L_{\sigma,I}=-\sum_\zeta S^{-1}_{\sigma,\zeta} C_\zeta$, where $C_\zeta$ collects transverse and inhomogeneous terms. The argument identifies a force-balance conflict: preserving $\partial_t v=0$ for an advected force-free structure requires incoming characteristics to balance outgoing ones in a time-varying way, and neither prescription can do this once part of the structure has left the domain.

What would settle it

For the same spheromak advection case, set the incoming characteristic derivatives at the NRBC plane each timestep to the values read from the ground-truth simulation at that plane; if the spheromak then passes through at constant velocity matching the ground truth, the bounce and suction are caused by the loss of incoming information as claimed, whereas if substantial reflections remain, the coupling itself is non-negligibly reflective.

Watch

Extended reading notes

Core claim

The central claim is that both standard implementations of non-reflecting boundary conditions yield physically real but unintended consequences when complex MHD features cross the boundary. In the paper's spheromak advection tests, a Fixed NRBC, with incoming characteristic derivatives set to zero ($L_{\sigma,I}=0$), causes part of the spheromak to bounce off the boundary, while a Cancellation NRBC, with $L_{\sigma,I}=-\sum_\zeta S^{-1}_{\sigma,\zeta} C_\zeta$, pulls the spheromak through faster than the advection speed. These behaviors disappear when the same initial condition is run on an extended grid that keeps the spheromak in the volume; the differences grow as the advection speed drops and more incoming characteristics appear, with the largest jump between super-Alfvénic and sub-Alfvénic outflow. The paper concludes that the loss of incoming information is the root cause: a non-reflecting boundary removes the impact of outgoing characteristics on incoming ones, and that loss is exactly what the boundary condition is designed to produce, even though it is not what users want when a structured magnetic field leaves the domain.

Load-bearing premise

The paper's attribution of the bounce and suction to the boundary-condition design presupposes that numerical reflections from the coupling between the main code and the characteristics-based boundary layer are negligible, a conclusion inferred from the earlier data-driving tests rather than measured directly in these nonlinear spheromak runs.

Editorial extensions

If this is right

  • Users cannot assume a smaller NRBC-bounded domain reproduces a larger simulation once a magnetically structured, force-balanced feature crosses the boundary at sub-Alfvénic speeds; differences propagate inward at all characteristic speeds.
  • The mismatch grows with the number of incoming characteristic derivatives: simulations with super-Alfvénic outflow closely match the ground truth, while sub-Alfvénic cases show order-unity differences in a weighted mean-squared-error metric.
  • These failures are not numerical instabilities; the boundary does exactly what it is designed to do, so numerical tuning of the coupling cannot fix them.
  • For simulations of coronal mass ejections, a Fixed NRBC biases eruptions to be harder to trigger and a Cancellation NRBC biases them to be easier, potentially confounding studies of eruption onset.
  • A 'minimum impact' boundary condition must allow departed structures to influence the interior through a time-dependent, nonlocal prescription of incoming characteristics; the paper identifies this as future work.

Reading between the lines

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

  • A direct testable extension would advect the same spheromak through an NRBC that stores the departed state in a buffer zone or prescribes incoming characteristics from the ground-truth solution at the boundary plane; if the bounce and suction disappear, the loss-of-information interpretation is confirmed.
  • The same argument implies that contamination from a mis-specified data-driven boundary propagates inward with speed at most $v_\perp + c_f$, so results near a driven boundary can be trusted only for a limited time after an inconsistency arises.
  • The paper's pairwise cancellation analysis suggests that no locally constructed boundary condition based only on transverse gradients can preserve force balance for an advected magnetic structure; a nonlocal or time-history term appears unavoidable.
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

1 major / 7 minor

Summary. This paper implements two standard flavors of characteristics-based non-reflecting boundary conditions (NRBCs) in the Lagrangian-remap MHD code LaRe3D. The boundary region is handled by a separate characteristic solver, CHAR, which populates the LaRe3D ghost cells; the approach follows the authors' Paper I (Tarr et al. 2024). After validating the implementation on a hot-sphere expansion in an angled magnetic field (§3) and on 1D Alfvén-wave, Sod, and Brio-Wu problems (Appendix A), the paper advects a force-balanced spheromak through the boundary at four speeds chosen so that the number of incoming characteristic derivatives in the spheromak interior is 0, 1, 2, or 3 (Cases 1-4). Compared against ground-truth simulations on an extended grid from the same initial conditions, the NRBC simulations at the two sub-Alfvénic advection speeds (Cases 3 and 4) diverge severely: the 'Fixed' NRBC (incoming characteristics L_σ,I set to zero) causes the spheromak to bounce off the boundary, while the 'Cancellation' NRBC (L_σ,I = -Σ_ζ S^-1_σζ C_ζ) accelerates the spheromak through the boundary. The paper attributes these behaviors to the mathematical design of the NRBCs, namely the irretrievable loss of incoming information representing the external universe's reaction, and warns that NRBCs of these types should not be used as proxies for a larger simulation volume when self-contained magnetic structures cross the boundary at sub-Alfvénic speeds.

Significance. If the attribution holds, this is an important and transferable caution for computational solar and space physics. The paper shows quantitatively (wMSD rising from ~10^-4 to order unity as the number of incoming characteristics increases) that a 'non-reflecting' boundary is not a minimal-impact boundary, and that the two common recipes bias the dynamics in opposite directions (bounce versus suck-through), which the authors argue would respectively suppress or enhance CME eruptions in coronal simulations (§6). Strengths include purpose-built extended-grid ground truths; honest error metrics (wMSD with its velocity-omission caveat, per-variable ϖ_99); a clean experimental control of the number of incoming characteristics; a concrete mechanism in §5 (force balance of the advecting structure requires incoming characteristics that carry information about the portion of the structure that has already left the domain, so the boundary condition effectively makes the external universe exert a force); and falsifiable, mutually opposite predictions for the two NRBC classes. The 1D test suite in Appendix A and the Mathematica verification of the eigensystem are useful supporting material.

major comments (1)
  1. [§4.2, §4.2.2, and §6 (with Appendix A)] The paper's central interpretation, that the spheromak bounce (Fixed NRBC) and suck-through (Cancellation NRBC) are the boundary condition 'done exactly what it is constructed to do' (§6), rests on the assertion in §4.2 that numerical reflections from the CHAR/LaRe3D coupling are negligible. The support offered is Paper I's data-driven test, in which L_σ,I was prescribed from the full ground-truth U, plus the Case 1 result in §4.2.2 (wMSD < 10^-4 with #L_σ,I = 0), which does directly demonstrate nearly reflection-free 3D transmission of a spheromak through the boundary. However, neither test covers the operating mode in which the anomalies occur: Cases 3 and 4, in which L_σ,I is actively prescribed (to 0 or to -Σ_ζ S^-1_σζ C_ζ) over long times (t up to 50 for Case 4) while the strong transverse gradients of the spheromak surface sit at the boundary, and in which the paper itself documents a positive feedback loop that raises #L_σ,I beyond its tuned value. The 1D problems in Appendix A show residual discrepancies of 0.1-1% (up to a few percent for the reversed Sod shock and the Brio-Wu compound wave), and the paper does not separate the design-level contribution from the coupling contribution in those tests either. I recommend adding a direct oracle-boundary test in the 3D spheromak geometry: prescribe L_σ,I at every step from the ground-truth simulation in the Case 4 setup (the 'blue-line' mode of Paper I's Figure 8) and verify that the NRBC simulation then reproduces the ground truth within the Case 1 tolerance. If it does, the design-level attribution in §6 is established by measurement rather than by inference; if it does not, a portion of the observed bounce and acceleration must be re-attributed to the coupling, and the conclusions should be qualified accordingly.
minor comments (7)
  1. [Title] The word 'magnetohydrodyanamic' in the title is misspelled; it should read 'magnetohydrodynamic.'
  2. [§3, Eq. (3)] In the definition of the wMSD, the vector N^T is given as (ρ, ϵ, vy, vy, vz, Bx, By, Bz); the second entry should be vx.
  3. [§3, Figure 8 caption] The caption states that 'Panels c) and e) on the right' show the absolute difference between the ground truth and the NRBC simulation; the right column actually contains panels c) and f), while panel e) is in the middle column.
  4. [Appendix A] The sentence 'All figures plot an absolute error in each MHD property with a gradient in the z − direction' is garbled; presumably the intent is 'All figures plot the absolute error in each MHD property that has a gradient in the z-direction.'
  5. [§2.1.2 / §5 (footnote 1)] The characterization of the Bifrost boundary conditions as L_σ,I = -Σ_ζ S^-1_σζ C_ζ evaluated at v = 0 is presented as a factual description of another group's code and is used in §5 as a stepping stone toward the 'minimum impact' boundary condition; please verify this description against Gudiksen et al. (2011), since an inaccuracy would be visible to a large user community.
  6. [§4.2.2, Figure 5 and Eq. (3)] The covariance normalization K is recomputed for each case at its own τ_ad = 1/v_ad, so the wMSD values are not strictly comparable across the four panels of Figure 5; the text's cross-case comparison ('the largest jump in wMSD occurs between Case 2 and Case 3') should carry this caveat, and the per-variable ϖ_99 curves in Figure 7, which are also case-normalized, should be cited as the safer cross-case diagnostics.
  7. [Software / data availability] The paper would benefit from an explicit code and data availability statement covering access to CHAR, the LaRe3D version and configuration used for each run, and the data underlying Figures 2, 5, and 7.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the spheromak results are independent simulation outputs, and the Paper I self-citation for negligible numerical reflections is supporting evidence rather than a circular premise.

full rationale

The paper's central claim—that Fixed and Cancellation NRBCs cause an advected spheromak to bounce or accelerate instead of passing at constant velocity—is not derived from the boundary conditions' definitions by construction; it is obtained by running LaRe3D/CHAR and comparing against an extended-grid ground truth simulation. The two NRBC prescriptions (L_sigma,I = 0 and L_sigma,I = -sum S^{-1}_sigma,zeta C_zeta) are inputs, and the observed behavior is a nontrivial simulation outcome. No parameter is fitted to the target data and no predicted quantity is re-labeled from a fit. The characteristic decomposition is restated from Paper I (Eqs. P1.14 and P1.18) and is independently verifiable; the spheromak is the standard Rosenbluth-Bussac analytic solution, not an author-defined ansatz. The only load-bearing self-citation is the Section 4.2 assertion that numerical reflections from the CHAR/LaRe3D coupling are negligible, justified by Paper I's data-driving test with full ground-truth U. That is an independent empirical test of the same coupling, not an assumption of the present conclusion, so it is not circular. It is, however, an evidentiary gap for the 3D sub-Alfvenic case: Paper I prescribed L_sigma,I from the full ground truth, whereas here L_sigma,I is set to 0 or to the cancellation expression, and Appendix A shows residual 0.1%-1% differences even in 1D tests. That weakens the attribution of the bounce/acceleration entirely to NRBC design, but it is a correctness/support limitation, not a circularity. Accordingly the circularity score is 1.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central demonstration depends on a set of hand-chosen test parameters (advection speeds, plasma beta, relaxation time) that select the incoming-characteristic regime, and on two background assumptions about the numerical method: that the CHAR/LaRe3D coupling has negligible numerical reflections (inferred from Paper I) and that the extended-grid ground truth is undisturbed by its far boundary. No parameters are fitted to data, and no new physical entities are introduced.

free parameters (5)
  • advection speeds vad for Cases 1-4 = -3.5, -2.3, -1.0, -0.1
    Hand-chosen to place the spheromak advection in different rows of Table 1, controlling the number of incoming characteristics. They are experimental design parameters, not fitted, but the demonstration of the paper's central claim depends on these choices.
  • plasma beta inside the spheromak = ~0.1 (B0 = 2.0, rho = 1, epsilon = 1)
    Chosen so that the fast, Alfven, and slow characteristic speeds separate enough to select distinct incoming-characteristic regimes. The required pre-relaxation expansion of the spheromak scales with this choice.
  • hot sphere perturbation amplitude and radius = 10% energy perturbation, radius 0.1*sqrt(2)
    Test parameters for the validation case; not central to the spheromak conclusions.
  • spheromak relaxation time before advection = several cs crossing times
    The simulation is allowed to relax to a force-balanced state before advection is switched on; the outward expansion and gas-pressure buildup depend on this duration.
  • surface smoothing width for spheromak tangential discontinuity = a few grid cells
    Numerical smoothing of the surface current to avoid grid-scale discontinuities; a standard numerical choice.
assumptions (5)
  • standard math Ideal MHD equations closed with the ideal gas law P = (gamma-1)*rho*epsilon
    Invoked in Section 2.1, Eq. P1.1, as the model to be solved.
  • standard math Eigen-decomposition of the z-direction flux matrix following Roe and Balsara (1996), with characteristic speeds as defined in Eq. P1.7
    Provides the characteristic derivatives L_sigma used to separate incoming and outgoing information at the boundary.
  • domain assumption The extended-grid ground truth simulation with a zero-gradient far boundary is an accurate reference for the true MHD evolution up to the time analysis is halted
    The authors halt before the spheromak's leading edge reaches the far boundary, so the far-boundary treatment should not influence the comparison window.
  • ad hoc to paper Numerical reflections from the CHAR/LaRe3D coupling are negligible
    Inferred from Paper I's data-driving test using the same CHAR code, not measured directly in the present nonlinear runs; the paper's central interpretation relies on this.
  • domain assumption Findings from homogeneous, dimensionless test cases transfer to stratified solar atmosphere simulations
    The paper motivates the work with solar applications but presents only uniform-background tests; applying the conclusions to a stratified photosphere-to-corona model is an extrapolation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Simulating the photospheric to coronal plasma using magnetohydrodyanamic characteristics II: reflections on non-reflecting boundary conditions." pith.science (2026). https://pith.science/paper/MYHDI5TV

@misc{pith2026250117995,
  author       = {Pith},
  title        = {Pith review of: Simulating the photospheric to coronal plasma using magnetohydrodyanamic characteristics II: reflections on non-reflecting boundary conditions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MYHDI5TV}},
  note         = {Machine review of arXiv:2501.17995}
}
read the original abstract

We present our implementation of non-reflecting boundary conditions in the magnetohydrodynamics (MHD) code LaRe3D. This implementation couples a characteristics-based boundary condition with a Lagrangian remap code, demonstrating the generality and flexibility of such non-reflecting boundary conditions for use with arbitrary grid-based MHD schemes. To test this implementation for perturbations on a background state, we present simulations of a hot sphere in an angled magnetic field. We then examine a series of simulations where we advect a spheromak through a non-reflecting boundary condition at four speeds related to the fast and slow magnetosonic speeds and the Alfven speed. We compare the behavior of these simulations to ground truth simulations run from the same initial condition on an extended grid that keeps the spheromak in the simulation volume at all times. We find that the non-reflecting boundary condition can lead to severe, physical differences developing between a simulation using a non-reflecting boundary and a ground truth simulation using a larger simulation volume. We conclude by discussing the origins of these differences.

Figures

Figures reproduced from arXiv: 2501.17995 by the authors.

Figure 1
Figure 1. Snapshots of the hot sphere with angled B test of the non-reflecting boundary conditions. Each panel shows an isocontour of internal energy 1% above the background and a map of the vertical velocity at the layer of the non-reflecting boundary. The results from a simulation with a Fixed NRBC (here Lσ,I = 0) are in the top row, and those from a simulation with a Cancellation NRBC (Lσ,I = − P ζ S −1 σ,ζCζ ) are in the … view at source ↗
Figure 2
Figure 2. Mean squared difference between the ground truth hot sphere simulation and the simulations with non￾reflecting boundary conditions. The solid black curve de￾notes the simulation with a Fixed NRBC (here Lσ,I = 0) and the dashed red curve the simulation with a Cancellation NRBC (Lσ,I = − P ζ S −1 σ,ζCζ ). do show subtle differences from one another, here most evident in the interference pattern of non-zero vertical ve… view at source ↗
Figure 3
Figure 3. Contours of the characteristic velocities cf (red), ca (blue), and cs (yellow) for each of the four selected values of vad. In each panel, |vad| is less than the characteristic velocity inside the contour. The four cases are Case 1: vad < −cf (top left), Case 2: −cf < vad < −ca (top right), Case 3: −ca < vad < −cs (bottom left), and Case 4: −cs < vad < 0 (bottom right). For Case 3 the contour of cf is outside the bo… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Magnetic field lines in the volume, and the normal component of the magnetic field, Bz, at the location of the non-reflecting boundary for the simulations with −ca < vad (= −1.0) < −cs. The simulation with a Fixed NRBC (here Lσ,I = 0) is on the top and with a Cancellat…
Figure 5
Figure 5. Figure 5: Lines plot the wMSD for each simulation (left axis), and symbols plot the maximum number of incoming charac￾teristics, #Lσ,I , within the boundary layer (right axis), as a function of time, for Case 1 (a), Case 2 (b), Case 3 (c), and Case 4 (d). Solid black curves and …
Figure 6
Figure 6. Figure 6: Magnetic field lines in the volume, and the normal component of the magnetic field, Bz, at the location of the non-reflecting boundary for the simulations with −cs < vad (= −0.1) < 0 (Case 4). The simulation with a Fixed NRBC (here Lσ,I = 0) is on the left and with a C…
Figure 7
Figure 7. Figure 7: Threshold ϖ99 for density (a), energy (b), vx (c), vy (e), vz (g), Bx (d), By (f), and Bz (h). In each panel, black curves denote simulations with a Fixed NRBC (here Lσ,I = 0) and red curves simulations with a Cancellation NRBC (Lσ,I = − P ζ S −1 σ,ζCζ ). The four case…
Figure 8
Figure 8. Figure 8: 1D test case of an Alfv´en wave with perturbations in the x-direction propagating in the negative z-direction. On the left, panel a) shows vx and panel d) shows Bx for a time immediately before the wave reaches the non-reflecting boundary at z = 0. Panels b) and e) res…
Figure 9
Figure 9. Figure 9: The 1D Sod shock tube problem. The left column shows the conditions after the solution has had time to develop from the initial conditions and the middle column shows a time after the left propagating sonic rarefaction wave has begun interacting with the non-reflecting…
Figure 9
Figure 9. Figure 9: Finally, we present the reversed Brio-Wu shock tube. To fully compare the ground truth solution and the simulation with a non-reflecting boundary condition, the results of the reversed Brio-Wu shock tube are split over Figures 12 and 13. For completeness, we show the r…
Figure 12
Figure 12. Figure 12 [PITH_FULL_IMAGE:figures/full_fig_p031_12.png]
Figure 13
Figure 13. Figure 13: As [PITH_FULL_IMAGE:figures/full_fig_p032_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 7 canonical work pages

  1. [1]

    D., Longbottom, A

    Arber, T. D., Longbottom, A. W., Gerrard, C. L., & Milne, A. M. 2001, Journal of Computational Physics, 171, 151, doi: 10.1006/jcph.2001.6780

    arXiv 2001

  2. [2]

    Brio, M., & Wu, C. C. 1988, Journal of Computational Physics, 75, 400, doi: 10.1016/0021-9991(88)90120-9

    doi:10.1016/0021-9991(88)90120-9 1988

  3. [3]

    2012, in High Performance Visualization–Enabling Extreme-Scale Scientific Insight, 357–372, doi: 10.1201/b12985

    Childs, H., Brugger, E., Whitlock, B., et al. 2012, in High Performance Visualization–Enabling Extreme-Scale Scientific Insight, 357–372, doi: 10.1201/b12985

    doi:10.1201/b12985 2012

  4. [4]

    2008, A&A, 490, 353, doi: 10.1051/0004-6361:200810386

    Grappin, R., Aulanier, G., & Pinto, R. 2008, A&A, 490, 353, doi: 10.1051/0004-6361:200810386

    work page doi:10.1051/0004-6361:200810386 2008

  5. [5]

    2000, A&A, 362, 342

    Grappin, R., L´ eorat, J., & Buttighoffer, A. 2000, A&A, 362, 342

    work page 2000

  6. [6]

    V., Carlsson, M., Hansteen, V

    Gudiksen, B. V., Carlsson, M., Hansteen, V. H., et al. 2011, A&A, 531, A154, doi: 10.1051/0004-6361/201116520

    doi:10.1051/0004-6361/201116520 2011

  7. [7]

    R., Millman, K

    Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2

    doi:10.1038/s41586-020-2649-2 2020

  8. [8]

    D., & van Leer, B

    Harten, A., Lax, P. D., & van Leer, B. 1983, SIAM Review, 25

    work page 1983

Show all 18 references
  1. [9]

    Hedstrom, G. W. 1979, Journal of Computational Physics, 30, 222, doi: 10.1016/0021-9991(79)90100-1

    1979 doi

  2. [10]

    Hunter, J. D. 2007, Computing in Science & Engineering, 9, 90, doi: 10.1109/MCSE.2007.55

    2007 doi

  3. [11]

    2011, ApJ, 727, 101, doi: 10.1088/0004-637X/727/2/101

    Jiang, C., Feng, X., Fan, Y., & Xiang, C. 2011, ApJ, 727, 101, doi: 10.1088/0004-637X/727/2/101

    2011 doi

  4. [12]

    2005, ApJ, 624, 392, doi: 10.1086/428822

    Landi, S., Velli, M., & Einaudi, G. 2005, ApJ, 624, 392, doi: 10.1086/428822

    2005 doi

  5. [13]

    A., et al

    Lionello, R., Downs, C., Linker, J. A., et al. 2013, ApJ, 777, 76, doi: 10.1088/0004-637X/777/1/76

    2013 doi

  6. [14]

    L., & Balsara, D

    Roe, P. L., & Balsara, D. S. 1996, SIAM J. of Applied Mathematics, 56, 57, doi: 10.1137/S003613999427084X

    1996 doi

  7. [15]

    N., & Bussac, M

    Rosenbluth, M. N., & Bussac, M. N. 1979, Nuclear Fusion, 19, 489

    1979

  8. [16]

    Sod, G. A. 1978, Journal of Computational Physics, 27, 1, doi: 10.1016/0021-9991(78)90023-2

    1978 doi

  9. [17]

    Leake, J. E. 2024, ApJS, 270, 30, doi: 10.3847/1538-4365/ad0e0c

    2024 doi

  10. [18]

    Thompson, K. W. 1987, Journal of Computational Physics, 68, 1, doi: 10.1016/0021-9991(87)90041-6 Wolfram Research, I. 2019, Mathematica, Version 12.0,

    1987 doi

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.