Issues in providing expert advice for users of a particle-in-cell simulation code

摘要

The fully electromagnetic, 2 1/2 D, OOPIC (object-oriented particle-in-a-cell) simulation code is the product of a three year research effort to implement a sophisticated PIC simulation model with object-oriented programming techniques. In addition to its physics engine, OOPIC possesses a GUI/scientific visualization system and an expert advisor system to aid users in formulating simulations. We discuss issues involved in providing expert advice to users of a PIC simulation. The primary function of the expert advisor is to aid users in setting up parameters that "correctly" specify a simulation. Parameters are geometrical configurations of conductors, emitters, wavelaunchers, etc., that model a particular physical device. Parameters can be numbers related to physical processes-either simple numbers such as the simulation particle mass or structured sets of numbers such as those required to specify a velocity distribution function. Parameters can also relate to choice or control of underlying algorithms. Offering advice can be said to consist of politely imposing constraints on the various combinations of parametric values. We discuss various methods for imposing those constraints including-starting the design process from preexisting correct configurations, permitting the step by step visual construction of simulation models with each geometrical element constrained to be legal, using rule based expert system techniques to impose limits and relations on combinations of parameters, and forcing trial configurations to pass various algorithmic tests before a simulation can be initiated.