Optimization of particle-in-cell simulations for Vlasov-Poisson system with strong magnetic field

Chacon‐Golcher, E.; Hirstoaga, Sever Adrian; Lutz, Mathieu

doi:10.1051/proc/201653011

Cited by 5 publications

(7 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, we validate the splitting by species using BSL for ions and BSL or PIC for electrons, with sub-steps for the electrons. This opens the door to use specific PIC (or BSL) schemes that are designed for capturing high oscillations (see [10]). On Figure 30, we compare the total energy conservation between Strang and the 6-th order splitting; we remark that the conservation is really improved with the 6-th order splitting, which is coherent with [9], where such a splitting is also used for a single species.…”

Section: Third Test Case: 2d × 2d Two-speciesmentioning

confidence: 99%

Verification of 2D × 2D and Two-Species Vlasov-Poisson Solvers,

et al. 2018

Self Cite

View full text Add to dashboard Cite

In [2], 1D×1D two-species Vlasov-Poisson simulations are performed by the semi-Lagrangian method. Thanks to a classical first order dispersion analysis, we are able to check the validity of their simulations; the extension to second order is performed and shown to be relevant for explaining further details. In order to validate multi-dimensional effects, we propose a 2D × 2D single species test problem that has true 2D effects coming from the sole second order dispersion analysis. Finally, we perform, in the same code, full 2D × 2D non linear two-species simulations with mass ratio 0:01, and consider the mixing of semi-Lagrangian and Particle-in-Cell methods.

show abstract

Section: Third Test Case: 2d × 2d Two-speciesmentioning

confidence: 99%

Verification of 2D × 2D and Two-Species Vlasov-Poisson Solvers,

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…The construction of numerical methods in plasma physics described by kinetic equations is a challenging problem. The main difficulties arise both from the high dimensionality of the problem and from the formation of multiscale structures that must be captured by a numerical solver [3,4,9,13,17,18,23,44,50]. The numerical methods for plasma physics developed in the literature can be essentially divided into two groups: approaches based on direct discretizations of the corresponding system of partial differential equations (PDEs), like finite differences and finite volumes methods [13,17,18,41,50], and approaches based on approximations of the underlying particle dynamics at different levels, like particle-in-cell (PIC) methods [4-6, 12, 23, 44].…”

Section: Introductionmentioning

confidence: 99%

Stochastic Galerkin Particle Methods for Kinetic Equations of Plasmas with Uncertainties

2022

View full text Add to dashboard Cite

“…In our code, we used the same kind of parallelism as in [4]. The new features from our code compared to that paper are the parallelization of the sorting among threads and OpenMP 4.5 reduction on array sections.…”

Section: Thread-level and Process-level Parallelismmentioning

confidence: 99%

“…We present in this section gains in performance achieved by optimizing a previous work [4]. The hardware performance counters were tracked with perf and PAPI [19].…”

Section: Single Core Optimizationsmentioning

confidence: 99%

Efficient Data Structures for a Hybrid Parallel and Vectorized Particle-in-Cell Code

Barsamian

Hirstoaga

Violard

2017

2017 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)

Self Cite

View full text Add to dashboard Cite

Abstract-The contribution of the present work relies on an innovative and judicious combination of several optimization techniques for achieving high performance when using automatic vectorization and hybrid MPI/OpenMP parallelism in a Particle-in-Cell (PIC) code. The domain of application is plasma physics: the code simulates 2d2v Vlasov-Poisson systems on Cartesian grids with periodic boundary conditions. Overall, our code processes 65 million particles/second per core on Intel Haswell (without hyper-threading) and achieves a good weak scaling up to 0.4 trillion particles on 8,192 cores.The optimizations mainly consist in using (i) a structure of arrays for the particles, (ii) an efficient data structure for the electric field and the charge density, and (iii) an appropriate code for automatic vectorization of the charge accumulation and of the positions' update. In particular, we use space-filling curves to enhance data locality while enabling vectorization: starting from a redundant cell-based data structure for the electric field and for the charge density, we compare several space-filling curves for an efficient ordering of these data and we obtain a gain of 36% in the number of L2 and L3 cache misses when using a Morton curve instead of the classical rowmajor one. In addition, by proposing a specific writing of the updating positions code we achieve a 31% time improvement in that step. The optimizations bring an overall gain in the execution time of 42% with respect to a standard code. The parallelization of the particle loops is simply performed by means of both distributed and shared memory paradigms, without domain decomposition. We explain the weak and the strong scalings of the code bounded as expected by the overhead of the MPI communications.

show abstract

Optimization of particle-in-cell simulations for Vlasov-Poisson system with strong magnetic field

Cited by 5 publications

References 12 publications

Verification of 2D × 2D and Two-Species Vlasov-Poisson Solvers,

Verification of 2D × 2D and Two-Species Vlasov-Poisson Solvers,

Stochastic Galerkin Particle Methods for Kinetic Equations of Plasmas with Uncertainties

Efficient Data Structures for a Hybrid Parallel and Vectorized Particle-in-Cell Code

Contact Info

Product

Resources

About