AVGUN: A 3-D Electrostatic Relaxation Particle-in-cell Code With Multiple Grids

Although computer speed and memory are continuously improving, the ability to do three-dimensional modeling is still difficult in problems with complex structures. Typically, in simulation codes utilizing a single three-dimensional rectangular grid, merely using a uniform grid or evena nonuniform grid to enhance the field resolution in critical areas can quickly result in a computationally intractable problem due to memory and cpu time considerations. As a result, many simulations are made with restricted spatial domains that cannot model the problem of interest with sufficient accuracy. The problem may also be accentuated by using boundary conditions that often lack the sophistication to handle these restricted domains. However, in many problems it is only a subset of the full domain that needs the fields finely resolved and computed to some sufficiently high accuracy. Problems such as these are good candidates for a multigrid approach. A VGUN consists of an electrostatic relaxation model with particlesand embedded structures (a gun model). Essentially, the model simply solves Poisson's equation and advances particles in the self-consistent fields. In the model, a simple two-level multigrid system is incorporated. This grid system consists of a coarse grid intended to contain the problem's spatial domain of interest, with an embedded fine grid intended to contain only the subset of the domain requiring the increased accuracy. Here, the fine grid acts as a 3-D window that the user can use to "zoom in" on any area needing high accuracy. Simulation results using A VGUN will be presented that illustrate the majoralgorithms of the code, including mesh generation and structure setup, as well as graphics capabilities, memory management, and restart capabilities.