Aspects of and Insights Into the Rigorous Validation, Verification, and Testing Processes for a Commercial Electromagnetic Field Solver Package

摘要

This paper focuses on rigorous validation, verification, and testing methodologies applied to a commercial electromagnetic software package to ensure that as accurate as possible results are given dependent on the accuracy of the solution method, for instance, whether a full-wave or approximate numerical method is used. In this paper, the general availability of reliable benchmark results such as analytical solutions, measurements, results from other codes and other numerical methods, and general benchmarking activities will be presented. The cross-validation aspects, once the benchmark results are available, will be discussed with respect to amongst other sequential runs compared with parallel multcore/cluster runs or, with and without, GPU acceleration. Internal consistency checks (which are a required but not necessary condition when assessing the accuracy) such as power budget, mesh size convergence, or boundary condition error estimates are also covered. Special emphasis is put on the validation of the actual computational model that is used as input to simulations. This is necessary, for example, because incomplete representation of real geometry might ignore small details that are needed for the specific quantity that is analyzed. Also, uncertainties with regards to material parameters or transition impedances could lead to discrepancies between the computed results and reality that are not to be attributed to the electromagnetic solution as such, but rather the model generation.