LDRD Final Report: Adaptive Methods for Laser Plasma Simulation

Abstract

The goal of this project was to investigate the utility of parallel adaptive mesh refinement (AMR) in the simulation of laser plasma interaction (LPI). The scope of work included the development of new numerical methods and parallel implementation strategies. The primary deliverables were (1) parallel adaptive algorithms to solve a system of equations combining plasma fluid and light propagation models, (2) a research code implementing these algorithms, and (3) an analysis of the performance of parallel AMR on LPI problems. The project accomplished these objectives. New algorithms were developed for the solution of a system of equations describing LPI. These algorithms were implemented in a new research code named ALPS (Adaptive Laser Plasma Simulator) that was used to test the effectiveness of the AMR algorithms on the Laboratory's large-scale computer platforms. The details of the algorithm and the results of the numerical tests were documented in an article published in the Journal of Computational Physics [2]. A principal conclusion of this investigation is that AMR is most effective for LPI systems that are ''hydrodynamically large'', i.e., problems requiring the simulation of a large plasma volume relative to the volume occupied by the laser light. Since the plasma-only regions require less resolution than the laser light, AMR enables the use of efficient meshes for such problems. In contrast, AMR is less effective for, say, a single highly filamented beam propagating through a phase plate, since the resulting speckle pattern may be too dense to adequately separate scales with a locally refined mesh. Ultimately, the gain to be expected from the use of AMR is highly problem-dependent. One class of problems investigated in this project involved a pair of laser beams crossing in a plasma flow. Under certain conditions, energy can be transferred from one beam to the other via a resonant interaction with an ion acoustic wave in the crossing region. AMR provides an effective means of achieving adequate resolution in the crossing region while avoiding the expense of using the same fine grid everywhere, including the region between the beams where no LPI occurs. We applied ALPS to a suite of problems modeling crossed beam experiments performed on the Omega laser at the University of Rochester. Our simulations contributed to the theoretical interpretation of these experiments, which was recently published in Physical Review Letters [4]. This project has advanced the Laboratory's computational capabilities in the area of AMR algorithms and their application to LPI problems. The knowledge gained and software developed will contribute to the computational tools available for use in the design and interpretation of experiments to be performed at the National Ignition Facility (NIF) in support of Laboratory missions in stockpile stewardship, energy research and high energy density science.