Standard Particle-in-cell Research Articles

Acceleration of particle-in-cell simulations using sparse grid algorithms. I. Application to dual frequency capacitive discharges

The use of sparse particle-in-cell (PIC) algorithms to accelerate the standard explicit PIC scheme has recently been successfully applied in the context of single-frequency capacitively coupled plasma discharges [Garrigues et al., J. Appl. Phys. 129, 153303 (2021)]. We have extended the sparse PIC scheme to model dual-frequency capacitive discharges. Comparisons between standard and sparse PIC algorithms show that the plasma properties as well as the electron and ion distribution functions can be retrieved with a maximum error of 2%. This work opens the interest of using the sparse PIC algorithm to perform 2D and 3D simulations under real operating conditions of capacitively coupled plasma discharges.

RelSIM: A Relativistic Semi-implicit Method for Particle-in-cell Simulations

We present a novel Relativistic Semi-Implicit Method (RelSIM) for particle-in-cell (PIC) simulations of astrophysical plasmas, implemented in a code framework ready for production runs. While explicit PIC methods have gained widespread recognition in the astrophysical community as a reliable tool to simulate plasma phenomena, implicit methods have been seldom explored. This is partly due to the lack of a reliable relativistic implicit PIC formulation that is applicable to state-of-the-art simulations. We propose the RelSIM to fill this gap: our new method is relatively simple, being free of nonlinear iterations and only requiring a global linear solve of the field equations. With a set of one- and two-dimensional tests, we demonstrate that the RelSIM produces more accurate results with much smaller numerical errors in the total energy than standard explicit PIC, in particular when characteristic plasma scales (skin depth and plasma frequency) are heavily underresolved on the numerical grid. By construction, the RelSIM also performs much better than the relativistic implicit-moment method, originally proposed for semi-implicit PIC simulations in the relativistic regime. Our results are promising to conduct large-scale (in terms of duration and domain size) PIC simulations of astrophysical plasmas, potentially reaching physical regimes inaccessible by standard explicit PIC codes.

Efficient parallelization for 3D-3V sparse grid Particle-In-Cell: Single GPU architectures

In the present paper, an efficient General Purpose Graphical Processing Unit (GPGPU)-based implementation of sparse grid Particle-In-Cell (PIC) methods is proposed. The parallelization, implementing novel strategies specific to sparse-PIC methods and tailored to GPU architectures, provides speed-ups1 as large as 100 on a single Tesla V100 GPU, with respect to sequential Computing Processing unit (CPU) execution; and a four order of magnitude reduction of the computational time in comparison with a standard PIC sequential CPU simulation. In addition, the simple implementation of the parallelization with the OpenACC framework offers portability to a large class of accelerators.

Improved modellisation of laser–particle interaction in particle-in-cell simulations

A new method named B-TIS (Bourgeois & Davoine, J. Comput. Phys., vol. 413, 2020, 109426) has recently been proposed for suppressing the influence of numerical Cherenkov radiation that appears in particle-in-cell (PIC) simulation of laser wakefield acceleration (LWFA). However, while this method provides good results when applied to the already accelerated electrons, we show here that it cannot model correctly most of the plasma electron bulk interacting with the laser field. We thus investigate in this paper the origins of this limitation and propose an improved method for which this limitation is removed. This new method, named B-TIS3, can now be applied to a much broader variety of problems and improve the performance in comparison with the standard PIC algorithm. We show that, for an electron interacting directly with a laser pulse, this new technique offers greater accuracy in terms of momentum and motion than the conventional scheme used in many PIC codes. These improvements translate into more faithful energy spectrum and electric charge for the accelerated beam in simulations of vacuum laser acceleration (VLA) or LWFA involving direct laser acceleration (DLA) at low plasma density. This new method, easy to implement and not computationally demanding, should then prove useful to study in depth and help develop novel VLA, DLA and LWFA techniques.

Accurately simulating nine-dimensional phase space of relativistic particles in strong fields

Next-generation high-power laser systems that can be focused to ultra-high intensities exceeding 1023 W/cm2 are enabling new physics regimes and applications. The physics of how these lasers interact with matter is highly nonlinear, relativistic, and can involve lowest-order quantum effects. The current tool of choice for modeling these interactions is the particle-in-cell (PIC) method. In the presence of strong electromagnetic fields, the motion of charged particles and their spin is affected by radiation reaction (either the semi-classical or the quantum limit). Standard (PIC) codes usually use Boris or similar operator-splitting methods to advance the particles in standard phase space. These methods have been shown to require very small time steps in the strong-field regime in order to obtain accurate results. In addition, some problems require tracking the spin of particles, which creates a nine-dimensional (9D) particle phase space, i.e., (x,u,s). Therefore, numerical algorithms that enable high-fidelity modeling of the 9D phase space in the strong-field regime (where both the spin and momentum evolution are affected by radiation reaction) are desired. We present a new particle pusher that works in 9D and 6D phase space (i.e., with and without spin) based on analytical rather than leapfrog solutions to the momentum and spin advance from the Lorentz force, together with the semi-classical form of radiation reaction in the Landau-Lifshitz equation and spin evolution given by the Bargmann-Michel-Telegdi equation. Analytical solutions for the position advance are also obtained, but these are not amenable to the staggering of space and time in standard PIC codes. These analytical solutions are obtained by assuming a locally uniform and constant electromagnetic field during a time step. The solutions provide the 9D phase space advance in terms of a particle's proper time, and a mapping is used to determine the proper time step duration for each particle as a function of the lab frame time step. Due to the analytical integration of particle trajectory and spin orbit, the constraint on the time step needed to resolve trajectories in ultra-high fields can be greatly reduced. The time step required in a PIC code for accurately advancing the fields may provide additional constraints. We present single-particle simulations to show that the proposed particle pusher can greatly improve the accuracy of particle trajectories in 6D or 9D phase space for given laser fields. We have implemented the new pusher into the PIC code Osiris. Example simulations show that the proposed pusher provides improvement for a given time step. A discussion on the numerical efficiency of the proposed pusher is also provided.

Application of sparse grid combination techniques to low temperature plasmas Particle-In-Cell simulations. II. Electron drift instability in a Hall thruster

Three-dimensional simulations of partially magnetized plasma are real challenges that actually limit the understanding of the discharge operations such as the role of kinetic instabilities using explicit Particle-In-Cell (PIC) schemes. The transition to high performance computing cannot overcome all the limits inherent to very high plasma densities and thin mesh sizes employed to avoid numerical heating. We have applied a recent method proposed in the literature [L. F. Ricketson and A. J. Cerfon, Plasma Phys. Controlled Fusion 59, 024002 (2017)] to model low temperature plasmas. This new approach, namely, the sparse grid combination technique, offers a gain in computational time by solving the problem on a reduced number of grid cells, hence allowing also the reduction of the total number of macroparticles in the system. We have modeled the example of the two-dimensional electron drift instability, which was extensively studied in the literature to explain the anomalous electron transport in a Hall thruster. Comparisons between standard and sparse grid PIC methods show an encouraging gain in the computational time with an acceptable level of error. This method offers a unique opportunity for future three-dimensional simulations of instabilities in partially magnetized low temperature plasmas.

Application of sparse grid combination techniques to low temperature plasmas particle-in-cell simulations. I. Capacitively coupled radio frequency discharges

The use of a particle-in-cell (PIC) algorithm with an explicit scheme to model low temperature plasmas is challenging due to computational time constrains related to resolving both the electron Debye length in space and the inverse of a fraction of the plasma frequency in time. One recent publication [Ricketson and Cerfon, Plasma Phys. Control. Fusion 59, 024002 (2017)] has demonstrated the interest of using a sparse grid combination technique to accelerate the explicit PIC model. Simplest plasma conditions were considered. This paper is the demonstration of the capability and the effectiveness of the sparse grid combination technique embedded in the PIC algorithm (hereafter called “sparse PIC”) to self-consistently model capacitively coupled radio frequency discharges. For two-dimensional calculations, the sparse PIC approach is shown to accurately reproduce the plasma profiles and the energy distribution functions compared to the standard PIC model. The plasma parameters obtained by these two numerical methods differ by less than 5%, while a speed up in the executable time between 2 and 5 is obtained depending on the setup.

A novel two-way coupling model for Euler-Lagrange simulations of multiphase flow

We propose a novel approach for two-way coupled simulations of multiphase flows within an Eulerian-Lagrange framework. Lagrangian particles are commonly approximated as point sources of either energy, mass or momentum. To introduce them into the Eulerian computation of the fluid flow, an approximation of the Dirac delta function has to be made. By resorting to the Boundary Element Method (BEM) based computation of fluid flow, it is possible to implement the point source concept without any numerical approximation by simply taking advantage of the properties of the fundamental solution present in BEM. However, singularity issues occur in the close vicinity of the mesh nodes. To remedy this effect, the particle-in-cell (PIC) method is used. We present a novel hybrid BEM-PIC two-way coupling algorithm and show that the proposed BEM-PIC model gives superior results compared to a standard PIC implementation. We introduce a critical distance to separate the domain, where the BEM model can be used and the domain, where the PIC model should be used. The results show that the BEM model can be used in about 97%−99% of the domain depending on the mesh used. We provide a criterion for estimating the critical distance within the novel hybrid BEM-PIC two-way coupling algorithm.

Numerical implementation of a hybrid PIC-fluid framework in laser-envelope approximation

Particle-In-Cell (PIC) simulation are fundamental to address a detailed study of a Laser Wake Field Acceleration process. Given the ongoing development of cm-scale plasma accelerators, reduced physical models are necessary to face otherwise unfeasible predictive start-to-end simulations. In this context, the equation of a plasma in the cold fluid approximation can be a powerful numerical tool because the system dynamics relies on the inversion of the Euler equation on a grid, much cheaper than the usual field coupling with macroparticle motion that characterizes the standard PIC loop. We show here the implementation of a hybrid PIC-fluid framework in ALaDyn code that allows the bulk plasma dynamics to be quickly solved while retaining the kinetic properties by placing macroparticles on top of the fluid background.

Self-force subtraction in particle in cell simulations

A method is presented to remove the numerical self-force that appears in particle-in-cell (PIC) simulations in computational domains discretized with general meshes and boundary conditions. The self-force is a non-physical force that a particle exerts on itself as a result of the discretization of charge density on a computational grid, and the interpolation of electric fields back to particle positions. With a uniform grid and periodic boundary conditions, Hockney and Eastwood (1988) demonstrated that this spurious force can be suppressed to machine accuracy if consistent interpolations of particle charges to grid points, and electric field back to particle positions are used. In non uniform unstructured meshes with multiply connected domains, or more general boundary conditions however, the self-force is unavoidable in the standard PIC approach, which can be a concern in some simulations. In this paper a straightforward and efficient method is presented to suppress the self-force for unstructured meshes and open boundary conditions in electrostatic codes as used, for example, to simulate spacecraft-environment interaction. The method consists of precomputing an array of self-forces associated with unit charges at each mesh vertex of the computational grid, which can then be used to interpolate the self-force at any particle position. The computed self-force can then be subtracted from the force computed in the standard PIC approach thus enabling particle trajectory integrations that are free from this unphysical force. The method is illustrated with simulation results using an unstructured tetrahedral mesh in spherical geometry.

High-order implicit particle-in-cell method for plasma simulations at solid densities.

A high-order implicit multidimensional particle-in-cell (PIC) method is developed for simulating plasmas at solid densities. The space-time arrangement is based on Yee and a leapfrog algorithm for electromagnetic fields and particle advancement. The field solver algorithm completely eliminates numerical instabilities found in explicit PIC methods with relaxed time step and grid resolution. Moreover, this algorithm eliminates the numerical cooling found in the standard implicit PIC methods by using a pseudo-electric-field method. The particle pusher algorithm combines the standard Boris particle pusher with the Newton-Krylov iteration method. This algorithm increases the precision accuracy by several orders of magnitude when compared with the standard Boris particle pusher and also significantly decreases the iteration time when compared with the pure Newton-Krylov method. The code is tested with several benchmarks, including Weibel instability, and relativistic laser plasma interactions at both low and solid densities.

A fast and accurate numerical implementation of the envelope model for laser–plasma dynamics

In laser-driven, plasma wakefield acceleration regimes (LWFA), when relevant scale lengths of the laser envelope and of the driven plasma waves are well separated from the wavelength and frequency of the laser fast oscillating component, a reduced physical model (usually referred to as the envelope model), has been introduced, allowing to formulate the laser–plasma equations in terms of laser cycle-averaged dynamical variables. As a main consequence, physical regimes where this reduced model applies, can be investigated with significant savings of computational resources still assuring comparable accuracy, with respect to standard Particle-In-Cell (PIC) models where all relevant space–time scales have to be resolved.Here we propose a computational framework characterized by two previously unexplored numerical implementations of the envelope model. The first one is based on explicit second order leapfrog integration of the exact wave equation for laser pulse propagation in a laboratory coordinate system in 3D cartesian geometry, replacing the usually quoted representation in an Eulerian frame moving at the speed of light. Since the laser and driven wakefield wave equations in a laboratory frame are advection dominated, we introduce a proper modification of finite differences approximating longitudinal space derivatives, to minimize dispersive numerical errors coming from the discretized advection operators. The proposed implementation, avoiding semi-implicit procedures otherwise required when dealing with a comoving frame, assures significant saving in computational time and ease of implementation for parallel platforms. The associated equation of motion for plasma particles has been integrated, as in standard PIC codes, using the Boris pusher, properly extended to take into account the specific form of the Lorentz force in the envelope model.As a second contribution, a novel numerical implementation of the plasma dynamics equations in the cold-fluid approximation, is presented. The scheme is based on the second-order one-step Adams–Bashforth time integrator coupled to upwind non-oscillatory WENO reconstruction for discretized space derivatives. The proposed integration scheme for the Eulerian fluid equations is equivalent to a leapfrog scheme with an added higher order dissipative truncation errors. It can be used either as a much faster, yet of comparable accuracy, alternative to the PIC representation of plasma particle motion, or even in a hybrid fluid–particle combination when kinetic effects and particle injection and acceleration in a wakefield have to be investigated.

Speeding up simulations by slowing down particles: Speed-limited particle-in-cell simulation

Based on the particle-in-cell (PIC) plasma simulation method, the speed-limited PIC (SLPIC) method delivers faster kinetic plasma simulation in cases where the particle distributions evolve slowly compared with the maximum stable PIC timestep. SLPIC thus offers more feasible, fully kinetic simulation in regimes that historically have required fluid approaches, such as magnetohydrodynamic (MHD), two-fluid, or Boltzmann electron treatments. In particular, SLPIC allows an explicit time advance with steps much larger than the inverse plasma frequency, avoiding the instability explicit PIC faces with large timesteps. SLPIC avoids this instability by slowing down fast particles (e.g., electrons) in a way that is rigorously underpinned by an approximate Vlasov equation; unlike MHD, two-fluid, and Boltzmann electron approaches, SLPIC does not fundamentally neglect any first-principles plasma physics, although the choices of the grid cell size, timestep, and number of macroparticles per cell naturally limit the physical phenomena that can be accurately represented. SLPIC can be implemented with minor modifications of a standard PIC code and does not require an implicit time advance. It enables large timesteps in first-principles kinetic plasma simulation of appropriately slow phenomena, and it can handle many of the same complications as PIC, such as boundary conditions and collisions. In an argon plasma sheath test problem, a SLPIC simulation achieved a speed-up of a factor of 160 over the corresponding PIC simulation, without loss of accuracy.

Reducing noise for PIC simulations using kernel density estimation algorithm

Noise is a major concern for Particle-In-Cell (PIC) simulations. We propose a new theoretical and algorithmic framework to evaluate and reduce the noise level for PIC simulations based on the Kernel Density Estimation (KDE) theory, which has been widely adopted in machine learning and big data science. According to this framework, the error on particle density estimation for PIC simulations can be characterized by the Mean Integrated Square Error (MISE), which consists of two parts, systematic error and noise. A careful analysis shows that in the standard PIC methods, noise is the dominant error, and the noise level can be reduced if we select different shape functions that are capable of balancing the systematic error and the noise. To improve, we use the von Mises distribution as the shape function and seek an optimal particle width that minimizes the MISE, represented by a cross-validation function. It is shown that this procedure significantly reduces the noise and the MISE for PIC simulations. A particle-wise width adjustment algorithm and a width update algorithm are also developed to further reduce the MISE. Simulations using the examples of Langmuir wave and Landau damping demonstrate that relative to the standard PIC methods, the KDE algorithm developed in the present study reduces the noise level on density estimation by 98% and gives a much more accurate result on the linear damping rate. To achieve the same accuracy, the KDE algorithm is 40% faster.

A new method for analyzing and visualizing plasma simulations using a phase-space tessellation

We apply a novel phase-space interpolation technique referred to as the simplex-in-cell (SIC) method to analyze two- and three-dimensional particle-in-cell (PIC) simulations of electromagnetic plasmas. SIC relies on a discretization of the initial phase-space distribution function into simplices, which allows an approximation to the full, continuously defined distribution function to be constructed at any later time in the simulation. This allows densities, currents, and even full momentum distribution functions to be measured at any point in the simulation domain without averaging over control volumes. The SIC approach applies to any PIC simulation for which a tessellation of the initial particle distribution can be constructed. In this study, we use outputs from standard PIC simulations of the Weibel instability and compare physical quantities such as charge and current densities calculated in postprocessing using SIC and standard particle deposits. Using 2D simulations with 1–65 536 particles-per-cell, we find that SIC eliminates discrete particle noise and in some cases can reach a given noise level using ∼1000 times fewer simulation particles than with standard particle deposition schemes. In regions of low density, such as between current filaments, SIC is able to capture small amplitude features even with fewer particles than gridpoints due to the deformable nature of the SIC volume elements. By calculating momentum distributions, we show how SIC can capture low density tails in the spectrum using far fewer particles than are necessary for standard particle deposits. We calculate the charge density on spatial grids of increasing resolution to demonstrate the ability of SIC to reveal fine-scale details that are not accessible with standard particle deposits. Finally, we show how SIC can be extended to 3D and give an example of its use to calculate the charge density from 3D PIC simulations of the Weibel instability. These results motivate the future implementation of SIC directly in the simulation force calculation for a novel low-noise electromagnetic plasma simulation method.

Kinetic simulation technique for plasma flow in strong external magnetic field

A technique for the kinetic simulation of plasma flow in strong external magnetic fields was developed which captures the compression and expansion of plasma bound to a magnetic flux tube as well as forces on magnetized particles within the flux tube. This quasi-one-dimensional (Q1D) method resolves a single spatial dimension while modeling two-dimensional effects. The implementation of this method in a Particle-In-Cell (PIC) code was verified with newly formulated test cases which include two-particle motion and particle dynamics in a magnetic mirror. Results from the Q1D method and fully two dimensional simulations were compared and error analyses performed verifying that the Q1D model reproduces the fully 2D results in the correct regimes. The Q1D method was found to be valid when the hybrid Larmor radius was less than 10% of the magnetic field scale length for magnetic field guided plasma expansions and less than 1% of the magnetic field scale length for a plasma in a converging–diverging magnetic field. The simple and general Q1D method can readily be incorporated in standard 1D PIC codes to capture multi-dimensional effects for plasma flow along magnetic fields in parameter spaces currently inaccessible by fully kinetic methods.

A 4th-Order Particle-in-Cell Method with Phase-Space Remapping for the Vlasov--Poisson Equation

Numerical solutions to the Vlasov--Poisson system of equations have important applications to both plasma physics and cosmology. In this paper, we present a new particle-in-cell (PIC) method for solving this system that is 4th-order accurate in both space and time. Our method is a high-order extension of one presented previously [B. Wang, G. H. Miller, and P. Colella, SIAM J. Sci. Comput., 33 (2011), pp. 3509--3537]. It treats all of the stages of the standard PIC update---charge deposition, force interpolation, the field solve, and the particle push---with 4th-order accuracy, and includes a 6th-order accurate phase-space remapping step for controlling particle noise. We demonstrate the convergence of our method on a series of one- and two- dimensional electrostatic plasma test problems, comparing its accuracy to that of a 2nd-order method. As expected, the 4th-order method can achieve accuracy comparable to that of the 2nd-order method and with considerably fewer resolution elements.

A spectral, quasi-cylindrical and dispersion-free Particle-In-Cell algorithm

We propose a spectral Particle-In-Cell (PIC) algorithm that is based on the combination of a Hankel transform and a Fourier transform. For physical problems that have close-to-cylindrical symmetry, this algorithm can be much faster than full 3D PIC algorithms. In addition, unlike standard finite-difference PIC codes, the proposed algorithm is free of spurious numerical dispersion, in vacuum. This algorithm is benchmarked in several situations that are of interest for laser–plasma interactions. These benchmarks show that it avoids a number of numerical artifacts, that would otherwise affect the physics in a standard PIC algorithm — including the zero-order numerical Cherenkov effect.

A multi-dimensional, energy- and charge-conserving, nonlinearly implicit, electromagnetic Vlasov–Darwin particle-in-cell algorithm

For decades, the Vlasov–Darwin model has been recognized to be attractive for particle-in-cell (PIC) kinetic plasma simulations in non-radiative electromagnetic regimes, to avoid radiative noise issues and gain computational efficiency. However, the Darwin model results in an elliptic set of field equations that renders conventional explicit time integration unconditionally unstable. Here, we explore a fully implicit PIC algorithm for the Vlasov–Darwin model in multiple dimensions, which overcomes many difficulties of traditional semi-implicit Darwin PIC algorithms. The finite-difference scheme for Darwin field equations and particle equations of motion is space–time-centered, employing particle sub-cycling and orbit-averaging. The algorithm conserves total energy, local charge, canonical-momentum in the ignorable direction, and preserves the Coulomb gauge exactly. An asymptotically well-posed fluid preconditioner allows efficient use of large cell sizes, which are determined by accuracy considerations, not stability, and can be orders of magnitude larger than required in a standard explicit electromagnetic PIC simulation. We demonstrate the accuracy and efficiency properties of the algorithm with various numerical experiments in 2D–3V.

Particle-In-Cell Monte Carlo Collision Model on GPU—Application to a Low-Temperature Magnetized Plasma

The particle-in-cell (PIC) algorithm for the simulation of charged-particle kinetics in plasmas is a very resource consuming method, and high-performance parallel computing is required for practical problems. Graphics processing units (GPUs) are powerful low-cost parallel systems that can be used for intensive computations. We have developed a PIC Monte Carlo collision (MCC) model of a low temperature magnetized plasma using GPUs. We describe how each part of the PIC MCC model is implemented on the GPU and show how particles are dynamically managed. The computational cost of the PIC MCC model on the GPU is compared with a standard PIC MCC model running on a single central processing unit (CPU). We show that speedup can reach from 10 to 20 times compared with a sequential code running on a CPU, depending on the number of cells and particles considered. The results are illustrated with the example of plasma transport across a magnetic filter similar to that of a negative-ion source for the neutral beam injector of fusion devices.