Abstract:

Optimization-based approaches to antenna design have enjoyed limited success. The task is often computationally intractable. Moreover, it is also often difficult to capture all relevant design issues and tradeoffs in a single mathematical objective function. Therefore, human experts typically specify and refine antenna designs \"by hand,\" using computers only to evaluate their candidate designs by simulation. In this paper we propose a middle ground between this traditional approach and fully automatic optimization. We use computation to sample a space of possible antenna designs, and we rely on human judgment to select good designs from the samples returned by the computer. The key elements of our computer system are a parallel algorithm for intelligently sampling a space of possible antenna designs and a graphical user interface for visualizing and exploring candidate designs and managing the sampling process.