Massively parallel particle-in-cell simulation of advanced particle accelerator concepts.*

D.L. Bruhwiler, J.R. Cary, D.A. Dimitrov, P. Messmer and C. Nieter
Tech-X Corp.
W. Mori, V. Decyk, F. Tsung, M. Zhou and C. Huang
UCLA
E. Esarey and C.G.R. Geddes
LBNL
T. Katsouleas, S. Deng and A. Ghalam
USC

The quest to understand the fundamental nature of matter requires ever higher energy particle collisions, which in turn leads to ever larger and more expensive accelerator facilities. Advanced concepts for electron and positron acceleration are required to reduce the cost and increase the performance of next-generation accelerators. Plasma-based accelerators can sustain electron plasma waves with phase velocities close to the speed of light c and longitudinal electric fields on the order of the nonrelativistic wave breaking field, E_0 = c m_e omega_p / e, where omega_p = (4 pi n_e e^2 / m_e)^1/2 is the plasma frequency at an electron density n_e [1]. For n_e=10^18 cm^-3, E_0 = 100 GV/m. Massively parallel particle-in-cell (PIC) simulations are required to simulate both laser-driven (LWFA) [2] and beam-driven (PWFA) [3] concepts, in order to support on-going experiments and to explore new ideas. We summarize recent successes in the use of parallel PIC codes VORPAL [4], OSIRIS [5] and QuickPIC [6] to validate computations with experimental data, to benchmark codes with independent implementations and to benchmark reduced PIC algorithms. Code performance and representative algorithms are discussed in the context of past work and future challenges.

[1] E. Esarey et al., IEEE Trans. on Plasma Sci. 24, 252 (1996).
[2] T. Tajima & J. Dawson, Phys. Rev. Lett. 43, 267 (1979).
[3] C. Joshi et al., Nature 311, 525 (1984).
[4] C. Nieter & J. Cary, J. Comp. Phys. 196, 448 (2004).
[5] R. Hemker, Ph.D thesis, UCLA (2000); R. Fonseca et al., Lect. Notes Comput. Sci. 2331, 342 (2002).
[6] C. Huang et al., J. Comp. Phys. (submitted).

*This work is supported by the SciDAC project -- "Advanced Computing for 21st Century Accelerator Science & Technology," an initiative of the U.S. Department of Energy, under Contract No. DE-FC02-01ER41178. This work is further supported by the U.S. Department of Energy, under Contract No.'s DE-FG03-95ER40926 and DE-AC03-76SF00098. This work used resources of the National Energy Research Scientific Computing Center.