Supernovae remnants, gamma-ray bursts, and other highly energetic systems show signatures of particles moving much faster than expected given the surrounding temperature. A team wanted to see if turbulence in the plasma sped the particles. They built a series of simulations that focus on particle acceleration. The simulations show that particles act in a way that lets scientists extrapolate the results to astrophysical scales. The Mira supercomputer at the Argonne Leadership Computing Facility, a Department of Energy Office of Science user facility, made the studies possible.
The research advances our understanding of energy exchange between particles and electric/magnetic fields in plasmas, like those found at the heart of our sun. Studies done using the simulations could offer insights into solar corona, solar wind, and magnetic fusion devices. For example, using the simulations, the researchers concluded that turbulence could explain results in systems like the Crab Nebula.
The team generated a series of increasingly large simulations of driven turbulence in a relativistic collisionless electron-positron pair plasma using Zeltron, their explicit Finite-Difference Time-Domain radiative electromagnetic Particle-in-Cell code. The code is capable of self-consistently including radiation reaction forces due to synchrotron and inverse-Compton emission. The largest simulation, composed of 232 billion particles in 15363 cells, ran on 32,768 Mira nodes. The researchers at the Argonne Leadership Computing Facility, where Mira is located, helped resolve massive storage performance degradation when using large fractions of Mira. They also helped improve performance after a major code upgrade by clarifying hardware capabilities, comparing to performance benchmark codes, and introducing the use of a performance-profiling tool. The simulation was key to understanding the scaling behavior and convergence of the nonthermal particle acceleration (NTPA). Previous studies seemed to indicate that this type of NTPA might become less efficient with increasing system size, but this study showed that the power-law distribution of particle energies becomes insensitive to system size at sufficiently large scales. They found that a logarithmically increasing amount of time is required to fully develop the particle energy distribution in larger simulations, as the highest particle energies are proportional to system size and more scatterings are required to reach those higher energies. The team found this behavior to be consistent with second-order Fermi acceleration as the primary mechanism for the NTPA. Fermi acceleration is named after physicist Enrico Fermi who showed that sequences of stochastic (random) acceleration events in certain astrophysical environments increase particle energies over long times. The team also generated ensembles of repeated realizations of smaller simulations to understand the variance due to differing amounts of energy injected by randomized turbulence. They found a weak correlation between injected energy and resultant power-law index, but good evidence that the conclusions about scaling behavior and convergence are robust.
Advanced Scientific Computing Research Allocation Principal Investigator
University of Colorado
The National Science Foundation and the National Aeronautics and Space Administration Astrophysics Theory Program provided funding for this research. The team used computing time on Mira, a supercomputer at the Argonne Leadership Computing Facility, a Department of Energy (DOE) Office of Science user facility, awarded through a DOE Innovative and Novel Computational Impact on Theory and Experiment grant.
V. Zhdankin, D. Uzdensky, G. Werner, and M. Begelman, “System-size convergence of nonthermal particle acceleration in relativistic plasma turbulence.” Astrophysical Journal Letters 867, 1, L18 (2018). [DOI: 10.3847/2041-8213/aae88c]