Suppressing spurious oscillations and particle noise in particle-in-cell simulations

Gabor Toth; Hongyang Zhou; Yuxi Chen

REVIEW 3 major objections 7 minor 24 references

Suppressing spurious oscillations and particle noise in particle-in-cell simulations

T0 review · 3 major / 7 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Flux-limited diffusion in the Maxwell solver and comoving-frame current calculation suppress spurious oscillations and particle noise in particle-in-cell shock simulations.

desk verdict Two practical fixes for semi-implicit PIC, but the comoving-frame current truncation likely breaks charge conservation and the paper never checks it. read the letter →

arxiv 2506.11320 v1 pith:R3UK6P2O submitted 2025-06-12 physics.comp-ph physics.plasm-phphysics.space-ph

Yuxi Chen , Hongyang Zhou , Gabor Toth This is my paper · ORCID

classification physics.comp-phphysics.plasm-phphysics.space-ph PACS 52.65.Rr

keywords particle-in-cellsimulationsemi-implicitPICGL-ECSIMLax-Friedrichsdiffusionfluxlimitercomoving-framecurrentstatisticalnoisecollisionlessshock

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 specific numerical choices cure the two dominant failure modes of semi-implicit particle-in-cell simulations of collisionless shocks. The first is a Lax-Friedrichs-type diffusion term, switched on by a flux limiter only where the solution is non-smooth, added to the Maxwell solver; it removes the spurious oscillations that the underlying implicit solver inherits from a Lax-Wendroff-style discretization. The second is computing the current density in the frame comoving with the plasma, which deletes the noise-dominated contribution from bulk transport of net charge and makes fast-flow simulations stable at moderate particle counts. If the argument holds, the FLEKS code can simulate quasi-parallel shocks, quasi-perpendicular shocks, and magnetic reconnection without numerical ringing or excessive particle noise, and the same fixes transfer to explicit PIC codes.

What carries the argument

Two mechanisms carry the argument. (1) Limited Lax-Friedrichs diffusion: the magnetic-field update is rewritten in flux form, $B_z^{n+1}=B_z^n-(\Delta t/\Delta x)(f_{B,i+1/2}-f_{B,i-1/2})$, with $f_{B,i+1/2}=cE_y^{n+\theta}-(1-\phi)a(B_{i+1}-B_i)/2$, and a corresponding implicit diffusion term enters the electric-field equation; the monotonized-central limiter $\phi(r,\beta)$ selects the first-order diffusive flux near discontinuities and the original second-order flux in smooth regions, while $a$ is taken as the local bulk speed. A hyperbolic/parabolic divergence-cleaning step repairs $\nabla\cdot\mathbf{B}=0$ broken by the added diffusion. (2) Comoving-frame current: the grid current is decomposed as $J_g=\frac{1}{\Delta V}\sum q(v-u)W+\frac{u}{\Delta V}\sum qW$; because the second, net-charge-transport term is noise-dominated when the Debye length is unresolved, the paper keeps only the first term and applies it inside the semi-implicit predictor through a mass matrix. These two pieces are what let the shock simulations run stably and quietly.

What would settle it

Set up a uniform plasma with a net charge density moving at bulk velocity $u$, with grid resolution fine enough to resolve the Debye length, and compare the computed current and electric-field evolution with and without the dropped net-charge-transport term; if the comoving-frame current fails to reproduce the physical $\rho u$ current and the associated field response, the scale-separation assumption is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that both defects—spurious oscillations at discontinuities and particle-noise instability in fast flows—are symptoms of identifiable discretization choices and can be removed without changing the underlying semi-implicit scheme. The implicit Maxwell solver rings at discontinuities because its update has the same structure as the second-order Lax-Wendroff advection scheme; adding a Lax-Friedrichs-type diffusion flux with a monotonized-central limiter to the $\mathbf{B}$ and $\mathbf{E}$ updates makes the scheme essentially TVD near sharp features while retaining second-order accuracy in smooth regions, with the diffusion speed set to the local bulk flow speed rather than the speed of light. On the noise side, when the bulk velocity $u$ greatly exceeds the thermal spread, the dominant current-noise term is the transport of net charge; the paper drops that term and keeps only the comoving-frame current, an approximation justified by the separation between the Debye length and the simulation grid scale. The numerical tests then show that the combined method stabilizes uniform fast flows and 1D quasi-parallel shocks, leaves quasi-perpendicular shock and reconnection solutions essentially unaltered, and slightly improves electron-pressure preservation in reconnection.

Load-bearing premise

The load-bearing premise is that the current produced by the bulk motion of net charge can be safely dropped, which holds only because physical charge separation happens far below the grid scale and noise swamps the physical net charge; in simulations that resolve the Debye length or treat boundary sheaths, the omitted term carries real physics.

Editorial extensions

If this is right

FLEKS can run 1D quasi-parallel shock simulations that were unstable without the fixes, and the solutions converge with grid spacing of $0.4 d_i$ or finer while resolving physical upstream fast-magnetosonic waves and shocklets.
Quasi-perpendicular shock simulations generate and propagate upstream whistler waves even at coarse resolution ($\Delta x=4 d_e$), and the wave amplitudes converge when the cell size is $d_e$ or smaller.
In a uniform solar-wind-like fast flow, total particle momentum is conserved with only one pass of current smoothing when the comoving-frame current is used; without it, momentum is lost regardless of smoothing.
In 2D asymmetric magnetic reconnection, the new techniques do not degrade the solution relative to a high-particle-count baseline, and smoothing the current reduces numerical electron cooling in the exhaust.
The paper states that the two techniques are applicable to explicit PIC codes as well as to the semi-implicit GL-ECSIM scheme.

Reading between the lines

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

[Inference] The noise reduction rests on a scale-separation bet: it will likely fail in regimes that resolve the Debye length, such as sheath or double-layer problems, where the dropped net-charge-transport current is physical; a hybrid scheme that restores that term when local resolution demands it would be a natural extension.
[Inference] The choice of the local bulk speed as the diffusion speed is empirical, so a more systematic criterion tied to the local CFL number or characteristic speeds would be needed before the limiter is trusted in relativistic or very high-Mach regimes.
[Inference] The comoving-frame current plus digital smoothing acts like a variance-reduction filter on the current, so one could quantify its effective noise spectrum and compare it directly with delta-f or other low-noise PIC approaches.
[Inference] A clean portability test would be to implement the same limited Lax-Friedrichs flux in an explicit Yee/FDTD PIC code and check that a discontinuous light wave remains oscillation-free at CFL numbers below one.

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

Colleague,

The short version: this paper has two genuinely useful techniques — a flux-limited Lax-Friedrichs diffusion added to the implicit Maxwell update, and a comoving-frame current density that subtracts the net-charge transport term. Both target real problems (oscillations at shocks, noise in fast flows), and the tests show they do the job. The paper is worth reading and worth a serious referee. But there's a soft spot that needs attention before I'd trust the accuracy claims.

The comoving-frame current works by dropping the second term in Eq. 28, i.e., u times the net charge density. The authors justify this by Debye-scale separation. That's an honest caveat, but they never check the consequence. The modified current no longer satisfies the discrete continuity equation with the actual particle charge. In a semi-implicit solver like GL-ECSIM, Gauss's law is enforced through that consistency. By dropping the transport term, the electric field becomes consistent with a charge that doesn't match the real particles. The noise-induced net charge is roughly 10% of electron charge density at ppc=100, with grid-scale gradients; u is large in fast flows, so ∇·(u ρ_net) doesn't have to be small. None of the Section 3 tests report ∇·E or charge-conservation error. The uniform-flow test measures momentum, not Gauss's law. So the shock and reconnection accuracy conclusions rest on shakier ground than the paper claims. This is the single biggest issue.

The rest is more routine. The identity α[u + βE_M] = u is stated without proof; that's a gap in a numerical methods paper. The diffusion speed a was chosen by trial, and the flux limiter β and smoothing passes are empirical; some sensitivity checks are given, but there's no systematic parameter study. Convergence is by visual comparison, not error norms. No code or data are released, which makes reproducibility harder.

That said, the paper is honest about its limitations, compares against standard filters, and the companion paper (Zhou et al. 2025) adds external validation. If the authors can show that ∇·E stays bounded and charge is conserved to an acceptable level in these test problems, or explain why the missing term is harmless, I'd be happy to see this published. As it stands, it's a conditional accept with a specific technical request.

Referee Report

3 major / 7 minor

Summary. The paper presents two numerical techniques implemented in the semi-implicit particle-in-cell code FLEKS/GL-ECSIM: (1) a Lax-Friedrichs-type diffusion term with a van Leer-type flux limiter added to the Maxwell solver to suppress spurious oscillations near discontinuities, and (2) a current-density formulation computed in the comoving frame, obtained by dropping the net-charge transport term in the current, to reduce particle noise in fast-flow plasmas. The methods are tested on vacuum light-wave propagation with discontinuities, a uniform fast flow, 1D quasi-parallel and quasi-perpendicular shocks, and 2D asymmetric magnetic reconnection. The authors report that the limiter removes oscillations in the vacuum test, that the comoving-frame current stabilizes the uniform fast flow and conserves momentum, and that the shock and reconnection simulations remain stable and qualitatively faithful.

Significance. If the claims hold, the proposed techniques would be practically useful for kinetic simulations of collisionless shocks and reconnection in space plasmas, where both numerical oscillations and particle noise are common obstacles. The paper is clearly written and the qualitative tests support the oscillation-suppression and noise-reduction mechanisms: the vacuum discontinuity test demonstrates the limiter's effect, the uniform-flow test shows momentum conservation with the comoving current, and the shock tests display plausible physical structures. The authors also correctly identify that their methods are not restricted to FLEKS and could be adapted to explicit PIC codes. However, the central accuracy claim is not fully supported because the modified current is not shown to preserve discrete Gauss's law or charge conservation, and several key parameters are chosen empirically without sensitivity analysis; these issues need to be addressed before the method can be considered validated.

major comments (3)

[Section 2.1, Eqs. (28), (32), (35)] The comoving-frame current Jbar_C is obtained by dropping the net-charge transport term (the second term in Eq. 28), so Jbar_C is not the current produced by the actual particle motion. As a result, the discrete continuity equation and hence discrete Gauss's law are modified: the truncation error involves the divergence of u times the net charge density. The paper states that noise-induced net charge can be about 10% of the electron charge density at ppc=100, and that grid-scale gradients of this noise need not be small. None of the tests in Section 3 report the resulting div(E) error or the charge-conservation residual. Because GL-ECSIM's advertised Gauss's-law preservation is a central property, and because the shock and reconnection accuracy claims rely on this modified current, the paper should quantify max|div(E) - 4*pi*rho| (or the equivalent discrete residual) over time in the uniform-flow, shock, and reconnection tests, and compare the results with and without the comoving-frame approximation. This is a load-bearing verification gap.
[Section 1.2, Eq. (21) and the choice a = u] The diffusion speed is set to the local plasma bulk speed u based on 'numerical experiments' and 'after some trials', rather than on a stability or accuracy analysis. The manuscript also does not report a sensitivity study for this choice, despite it governing the balance between oscillation suppression and numerical diffusion in all shock and reconnection tests. Since the method is advertised as generally applicable to explicit and semi-implicit PIC codes, the empirical choice of a = u should be supported by a parameter study (for example a = c, a = u/2, a = 2u, and a spatially varying u) with quantitative error measures in at least the vacuum-wave and quasi-parallel shock tests. Without this, the generality of the method is not established.
[Section 3.1, Figure 3] The paper reports that at CFL=2 the Lax-Friedrichs diffusion 'does not entirely eliminate oscillations' but only reduces their amplitude, and then dismisses this with the statement that practical discontinuities propagate at plasma speeds and are well resolved. This is a limitation that should be stated explicitly and tested in a relevant plasma problem, because semi-implicit PIC codes often operate at electromagnetic CFL numbers larger than one. The claim that the methods are robust for shock simulations would be stronger if the authors either demonstrate adequate oscillation control in a CFL>1 shock test or clearly delimit the CFL range for which the limiter is effective.

minor comments (7)

[Section 1.2, Eqs. (19)-(20)] The index notation is confusing: Eq. (19) uses d_{i+1} - d_i with d_i defined at cell centers, while Eq. (20) refers to phi_{E,i+1} and a_{E,i+1} without defining these shifted quantities clearly. Please clarify whether the electric-field diffusion flux is evaluated at cell centers or cell faces.
[Section 1.2, Eq. (21)] The text says beta in [1,2] for the MC limiter, then states that beta = 0 results in the 'unlimited first order scheme'. This is inconsistent because Eq. (21) with beta = 0 gives the first-order LF flux, not an unlimited scheme. Use a consistent definition and explain the beta = 0 case separately.
[Section 1.2, Eqs. (22)-(23)] The smoothness indicator r has a denominator that can vanish in regions where the field is locally constant; the manuscript does not explain how this division by zero is handled in the code. Please state the safe evaluation used.
[Section 2.1, Eq. (31)-(32)] The identity alpha_p [u(x_{s,p}) + beta_s E_M(x_{s,p})] = u(x_{s,p}) is stated as 'after some calculations' but no derivation is given. Given that this identity is essential to the simplified form of the comoving current, a short derivation or a reference would be helpful.
[Section 3.2, Figure 4] The uniform-flow test measures particle-momentum conservation only. It would be useful to also report the divergence error and the field-energy evolution, since the comoving-current truncation affects the field-particle coupling directly.
[Section 3.3 and Section 3.5] The convergence statements ('close agreement', 'negligible effect') are based on visual inspection rather than quantitative error norms or physical diagnostics such as reconnection rate. Adding quantitative measures would strengthen the validation.
[Section 1.4, Eq. (26)] The symbol phi is reused for the digital-filter weight in Eq. (26) and for the flux limiter in Eq. (21); using different symbols would avoid ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the new numerical techniques are constructed from explicit algebra and standard numerical-analysis ideas, and the validation rests on independent benchmark problems rather than on inputs that already contain the claimed outcomes.

full rationale

The paper does not derive its conclusions from the quantities it is trying to predict. The Lax-Friedrichs-type diffusion term is introduced explicitly as a numerical flux with a standard van Leer limiter (Eqs. 17-23), and the choice of local diffusion speed is disclosed as an empirical finding ('After some trials, we found that using the local plasma bulk speed for a = u provides an optimal balance'), not as a derived prediction. The comoving-frame current is obtained by an algebraic split of the particle current into a velocity-difference term and a net-charge transport term (Eq. 28), followed by a stated physical scale-separation assumption; the identity alpha[u + beta E_M] = u is a direct algebraic consequence of the rotation operator, not an imported result. Validation uses independent problems: vacuum discontinuity advection, uniform fast flow, quasi-parallel and quasi-perpendicular shocks, and magnetic reconnection, with baseline comparisons and grid-convergence checks. Self-citations to the underlying GL-ECSIM/FLEKS solvers and to the companion shock-study paper provide context and the base method, but the central techniques are defined and tested within this paper rather than being forced by those citations. The main substantive concerns about the comoving-frame approximation, such as possible violation of discrete Gauss's law or charge conservation, are correctness risks outside the circularity definition, since no test result is shown to be equivalent by construction to the model's assumptions.

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

The central claim rests on five tuned or assumed ingredients: the diffusion speed a equals the local bulk velocity (chosen empirically), the flux limiter parameter beta, the divergence cleaning coefficients, and the smoothing parameters for the velocity and current. The key physical assumption is that the net-charge transport current is negligible; the key algebraic assumption is the Boris identity alpha[u + beta E_M] = u, stated without proof. No new physical entities are introduced.

free parameters (5)

LF diffusion speed a = a = u (local plasma bulk velocity)
Chosen after numerical trials in Section 1.2 as an optimal balance between accuracy and stability; no derivation from first principles.
Flux limiter parameter beta = beta in [1,2]; tests use 1 and 1.5
Tunable parameter in the van Leer MC limiter, Eq. 21; affects diffusion and oscillation amplitude in Sections 3.3 and 3.4.
Divergence cleaning coefficients = c_h = 0.8 * dx/dt; dt * c_h^2 / c_p^2 = 0.1
Empirical settings for the hyperbolic/parabolic cleaning added because the LF diffusion breaks div(B) = 0, Section 1.3.
Number of smoothing passes for current = nsj = 1 to 3
Digital filter passes on the comoving-frame current; the paper states 1 to 3 passes are sufficient, Sections 2.2 and 3.5.
Smoothing parameters for comoving velocity = 5 passes, phi = 0.5, alternating d = 1 and d = 2
Used to obtain a smooth bulk velocity field u(x) before computing the comoving current, Section 2.1.

assumptions (5)

standard math Yee/FDTD central-difference schemes produce spurious oscillations near discontinuities
Cited from [1] and used to motivate the LF diffusion; not proven in the paper.
domain assumption The implicit theta-method Maxwell solver (Eq. 7) is stable and second-order accurate at theta = 0.5
Taken from the authors' prior GL-ECSIM papers [8,9]; is the base solver being modified.
ad hoc to paper The identity alpha[u + beta_s E_M] = u holds for the Boris rotation matrix alpha of Eq. 30
Stated in Section 2.1 as 'After some calculations' with no derivation or citation; used to simplify the comoving current equation.
domain assumption The current from transport of net charge is negligible compared with the current from ion-electron velocity differences at resolved scales
Central approximation of the comoving-frame method, justified by Debye scale being unresolved in Section 2.1; acknowledged as not universally valid.
ad hoc to paper Using the local plasma bulk speed u as the LF diffusion speed keeps the scheme stable for the tested shock and reconnection problems
Empirical finding in Section 1.2; not derived from a stability analysis of the coupled Maxwell-PIC system.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Suppressing spurious oscillations and particle noise in particle-in-cell simulations." pith.science (2026). https://pith.science/paper/R3UK6P2O

@misc{pith2026250611320,
  author       = {Pith},
  title        = {Pith review of: Suppressing spurious oscillations and particle noise in particle-in-cell simulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R3UK6P2O}},
  note         = {Machine review of arXiv:2506.11320}
}

read the original abstract

Particle-in-cell (PIC) simulations are essential for studying kinetic plasma processes, but they often suffer from statistical noise, especially in plasmas with fast flows. We have also found that the typical central difference scheme used in PIC codes to solve Maxwell's equations produces spurious oscillations near discontinuities, which can lead to unphysical solutions. In this work, we present numerical techniques to address these challenges within the semi-implicit PIC code FLEKS, which is based on the Gauss's Law-satisfying Energy-Conserving Semi-Implicit Particle-in-Cell method (GL-ECSIM). First, we introduce a Lax-Friedrichs-type diffusion term with a flux limiter into the Maxwell solver to suppress unphysical oscillations near discontinuities. Second, we propose a novel approach for calculating the current density in the comoving frame, which significantly reduces particle noise in simulations with fast plasma flows. Numerical tests are presented to demonstrate the effectiveness of these methods in mitigating spurious oscillations and noise in shock and magnetic reconnection simulations.

Figures

Figures reproduced from arXiv: 2506.11320 by the authors.

**Figure 2.** Light wave simulations with smoothing the electric field using a digital filter for [PITH_FULL_IMAGE:figures/full_fig_p015_2.png] view at source ↗

**Figure 3.** Light wave simulations with CFL=2 at t = 0.1. 3.2. Uniform fast flow This 1D test evaluates the stability of the algorithm for a uniform fast plasma flow. The initial conditions are representative of the typical solar wind at 1 AU: density n = 5 cm−3 , velocity u = [400, 0, 0] km/s, magnetic field B = [0, 0, 5] nT, ion temperature Ti = 72, 000 K, and electron temperature Te = 2Ti . The ion-electron mass ratio is set… view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Time evolution of the normalized total particle momentum in uniform fast flow [PITH_FULL_IMAGE:figures/full_fig_p018_4.png]

**Figure 5.** Figure 5: 1D quasi-parallel shock simulation results at [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]

**Figure 6.** Figure 6: 1D quasi-perpendicular shock simulation results at [PITH_FULL_IMAGE:figures/full_fig_p024_6.png]

**Figure 7.** Figure 7: Magnetic reconnection simulation results for simulations listed in Table 1. [PITH_FULL_IMAGE:figures/full_fig_p025_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 15 canonical work pages

[1]

Taflove, S

A. Taflove, S. C. Hagness, M. Piket-May, Computational electromagnet- ics: the finite-difference time-domain method, The Electrical Engineering Handbook 3 (2005) 15

work page 2005
[2]

C. K. Birdsall, A. B. Langdon, Plasma physics via computer simulation, CRC press, 2018

2018
[3]

Fi´ uza, M

F. Fi´ uza, M. Marti, R. Fonseca, L. Silva, J. Tonge, J. May, W. B. Mori, Efficient modeling of laser–plasma interactions in high energy density scenarios, Plasma Physics and Controlled Fusion 53 (2011) 074004

work page 2011
[4]

T. D. Arber, K. Bennett, C. S. Brady, A. Lawrence-Douglas, M. G. Ramsay, N. J. Sircombe, P. Gillies, R. G. Evans, H. Schmitz, A. R. Bell, C. P. Ridgers, Contemporary particle-in-cell approach to laser-plasma modelling, Plasma Physics and Controlled Fusion 57 (2015) 113001. doi:10.1088/0741-3335/57/11/113001

doi:10.1088/0741-3335/57/11/113001 2015
[5]

Exactly Energy Conserving Semi-Implicit Particle in Cell Formulation

G. Lapenta, Exactly energy conserving semi-implicit particle in cell formulation, Journal of Computational Physics 334 (2017) 349–366. URL: http://dx.doi.org/10.1016/j.jcp.2017.01.002. doi:10.1016/j.jcp.2017.01.002, arXiv: 1602.06326 Publisher: Elsevier Inc

work page Pith review arXiv 2017
[6]

J.-L. Vay, C. Geddes, E. Cormier-Michel, D. Grote, Numer- ical methods for instability mitigation in the modeling of laser wakefield accelerators in a Lorentz-boosted frame, Jour- nal of Computational Physics 230 (2011) 5908–5929. URL: https://linkinghub.elsevier.com/retrieve/pii/S0021999111002270. doi:10.1016/j.jcp.2011.04.003

doi:10.1016/j.jcp.2011.04.003 2011
[7]

G. R. Werner, L. C. Adams, J. R. Cary, Suppressing grid instability and noise in particle-in-cell simulation by smoothing, arXiv preprint arXiv:2503.05123 (2025)

arXiv 2025
[8]

Y. Chen, G. T´ oth, H. Zhou, X. Wang, FLEKS: A flex- ible particle-in-cell code for multi-scale plasma simulations, 27 Computer Physics Communications 287 (2023) 108714. URL: https://www.sciencedirect.com/science/article/pii/S0010465523000590. doi:10.1016/j.cpc.2023.108714

arXiv 2023

Show all 24 references

[9]

Y. Chen, G. T´ oth, Gauss’s Law satisfying Energy-Conserving Semi- Implicit Particle-in-Cell method, Journal of Computational Physics 386 (2019) 632–652. URL: https://doi.org/10.1016/j.jcp.2019.02.032. doi:10.1016/j.jcp.2019.02.032, arXiv: 1808.05745 Publisher: Elsevier Inc

2019 arXiv
[10]

Hirsch, Numerical Computation of Internal and External Flows, Volume 1, Fundamentals of Numerical Discretization, John Wiley & Sons, Toronto, 1989

C. Hirsch, Numerical Computation of Internal and External Flows, Volume 1, Fundamentals of Numerical Discretization, John Wiley & Sons, Toronto, 1989

1989
[11]

Harten, G

A. Harten, G. Zwas, Self-adjusting hybrid schemes for shock computation, J. Comput. Phys. 9 (1972) 568–583

1972
[12]

van Leer, Towards the ultimate conservative difference scheme

B. van Leer, Towards the ultimate conservative difference scheme. V. A second-order sequel to Godunov’s method, J. Comput. Phys. 32 (1979) 101–136

1979
[13]

R. J. LeVeque, Finite volume methods for hyperbolic problems, volume 31, Cambridge university press, 2002

2002
[14]

Yee, Numerical solution of initial boundary value problems involving maxwell’s equations in isotropic media, IEEE Transactions on antennas and propagation 14 (1966) 302–307

K. Yee, Numerical solution of initial boundary value problems involving maxwell’s equations in isotropic media, IEEE Transactions on antennas and propagation 14 (1966) 302–307

1966
[15]

Lapenta, Exactly energy conserving semi-implicit particle in cell formu- lation, J

G. Lapenta, Exactly energy conserving semi-implicit particle in cell formu- lation, J. Comput. Phys. 334 (2017) 349. doi:10.1016/j.jcp.2017.01.002

2017 doi
[16]

Dedner, F

A. Dedner, F. Kemm, D. Kr¨ oner, C. Munz, T. Schnitzer, M. Wesenberg, Hyperbolic divergence cleaning for the MHD equations, J. Comput. Phys. 175 (2003) 645–673

2003
[17]

T´ oth, B

G. T´ oth, B. van der Holst, I. V. Sokolov, D. L. De Zeeuw, T. I. Gombosi, F. Fang, W. B. Manchester, X. Meng, D. Najib, K. G. Powell, Q. F. Stout, A. Glocer, Y. J. Ma, M. Opher, Adaptive numerical algorithms in space weather modeling, Journal of Computational Physics 231 (201...

2012 doi
[18]

Y. Chen, G. T´ oth, P. Cassak, X. Jia, T. I. Gombosi, J. A. Slavin, S. Markidis, I. B. Peng, V. K. Jordanova, M. G. Henderson, Global Three-Dimensional Simulation of Earth’s Dayside Reconnection Using a Two-Way Coupled Magnetohydrodynamics With Embedded Particle- in-Cell Model...

2017 doi
[19]

Y. Chen, G. T´ oth, H. Hietala, S. K. Vines, Y. Zou, Y. Nishimura, M. V. Silveira, Z. Guo, Y. Lin, S. Markidis, Magnetohydrodynamic with embedded particle-in-cell simulation of the Geospace Environ- ment Modeling dayside kinetic processes challenge event, arXiv (2020). doi:10....

2020 doi
[20]

Y. Chen, G. T´ oth, X. Jia, J. A. Slavin, W. Sun, S. Markidis, T. I. Gombosi, J. M. Raines, Studying Dawn-Dusk Asymmetries of Mercury’s Magnetotail Using MHD-EPIC Simulations, Journal of Geophysical Re- search: Space Physics 124 (2019) 8954–8973. doi:10.1029/2019JA026840, arXi...

2019 arXiv
[21]

J.-L. Vay, P. Colella, P. McCORQUODALE, B. Van Straalen, A. Fried- man, D. Grote, Mesh refinement for particle-in-cell plasma simulations: Applications to and benefits for heavy ion fusion, Laser and Particle Beams 20 (2002) 569–575. doi:10.1017/S0263034602204139

2002 doi
[22]

Lapenta, Particle simulations of space weather, Journal of Computational Physics 231 (2012) 795–821

G. Lapenta, Particle simulations of space weather, Journal of Computational Physics 231 (2012) 795–821. URL: http://dx.doi.org/10.1016/j.jcp.2011.03.035. doi:10.1016/j.jcp.2011.03.035, publisher: Elsevier Inc

2012 doi
[23]

H. Zhou, Y. Chen, C. Dong, L. Wang, Y. Zou, B. Walsh, G. T´ oth, Planar collisionless shock simulations with semi-implicit particle-in-cell model fleks, 2025. arXiv:2506.08384

2025
[24]

Y. Zou, B. M. Walsh, Y. Chen, H. Zhou, S. Raptis, Control of solar wind on magnetic field fluctuations in the subsolar magnetosheath, Journal of Geophysical Research: Space Physics 130 (2025) e2025JA033856. doi:https://doi.org/10.1029/2025JA033856. 29

2025 doi

Pith tools

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

[1] [1]

Taflove, S

A. Taflove, S. C. Hagness, M. Piket-May, Computational electromagnet- ics: the finite-difference time-domain method, The Electrical Engineering Handbook 3 (2005) 15

work page 2005

[2] [2]

C. K. Birdsall, A. B. Langdon, Plasma physics via computer simulation, CRC press, 2018

2018

[3] [3]

Fi´ uza, M

F. Fi´ uza, M. Marti, R. Fonseca, L. Silva, J. Tonge, J. May, W. B. Mori, Efficient modeling of laser–plasma interactions in high energy density scenarios, Plasma Physics and Controlled Fusion 53 (2011) 074004

work page 2011

[4] [4]

T. D. Arber, K. Bennett, C. S. Brady, A. Lawrence-Douglas, M. G. Ramsay, N. J. Sircombe, P. Gillies, R. G. Evans, H. Schmitz, A. R. Bell, C. P. Ridgers, Contemporary particle-in-cell approach to laser-plasma modelling, Plasma Physics and Controlled Fusion 57 (2015) 113001. doi:10.1088/0741-3335/57/11/113001

doi:10.1088/0741-3335/57/11/113001 2015

[5] [5]

Exactly Energy Conserving Semi-Implicit Particle in Cell Formulation

G. Lapenta, Exactly energy conserving semi-implicit particle in cell formulation, Journal of Computational Physics 334 (2017) 349–366. URL: http://dx.doi.org/10.1016/j.jcp.2017.01.002. doi:10.1016/j.jcp.2017.01.002, arXiv: 1602.06326 Publisher: Elsevier Inc

work page Pith review arXiv 2017

[6] [6]

J.-L. Vay, C. Geddes, E. Cormier-Michel, D. Grote, Numer- ical methods for instability mitigation in the modeling of laser wakefield accelerators in a Lorentz-boosted frame, Jour- nal of Computational Physics 230 (2011) 5908–5929. URL: https://linkinghub.elsevier.com/retrieve/pii/S0021999111002270. doi:10.1016/j.jcp.2011.04.003

doi:10.1016/j.jcp.2011.04.003 2011

[7] [7]

G. R. Werner, L. C. Adams, J. R. Cary, Suppressing grid instability and noise in particle-in-cell simulation by smoothing, arXiv preprint arXiv:2503.05123 (2025)

arXiv 2025

[8] [8]

Y. Chen, G. T´ oth, H. Zhou, X. Wang, FLEKS: A flex- ible particle-in-cell code for multi-scale plasma simulations, 27 Computer Physics Communications 287 (2023) 108714. URL: https://www.sciencedirect.com/science/article/pii/S0010465523000590. doi:10.1016/j.cpc.2023.108714

arXiv 2023