We propose a TOKI (Time Order, Kinetic, and Irreversible) compression method for recording smooth trajectories of particles from PIC (electromagnetic particle-in-cell) simulations. In a TOKI compression, instead of storing entire time sequences of particle positions, we store particle trajectories in terms of coefficients of approximating polynomials. In the current implementation, these coefficients are determined either by the leastsquares method or by the Chebyshev approximation formula to obtain quasi-minimax polynomials. In this paper, we present the technique of TOKI compression and compare it with other lossy compression schemes, such as XTC. Comparisons are made using data from a PIC simulation for 150,000 electrons and 150,000 ions. For smooth trajectories, the compression ratio by TOKI is better than that by the XTC format. However, for ballistic trajectories, the compression ratio by TOKI is not good because of the significant overhead in storing raw values of trajectories. We also found that the compression efficiency for ion trajectories is better than that for electron trajectories. This is attributed to different characteristic time scales of motions due to the difference in mass. We expect that the behavior of the compression ratio in TOKI can be used to characterize motions of plasma particles.
CITATION STYLE
Hagita, K., Ohtani, H., Kato, T., & Ishiguro, S. (2014). TOKI compression for plasma particle simulations. Plasma and Fusion Research, 9(SpecialIssue2). https://doi.org/10.1585/pfr.9.3401083
Mendeley helps you to discover research relevant for your work.