Comment on 'Optimization of photonic crystal structures', J. Opt. Soc. Am. A 21, 2223-2232 (2004)'

TL;DR

This paper critically examines the effectiveness of stochastic global search strategies in photonic crystal optimization, highlighting potential pitfalls when applying simplified methods to complex, intractable problems.

Contribution

It provides a reanalysis of previous conclusions on global stochastic search strategies, emphasizing limitations in their applicability to complex optimization problems.

Findings

Global stochastic methods may be misleading for complex problems

Efficiency improvements in simplified test cases do not generalize

Caution is needed when applying these strategies to real-world intractable problems

Abstract

Recently, Smajic et al. published an article on numerical structural opti- mizations of two-dimensional photonic crystals using two different classes of optimization algorithms i.e, deterministic for local searches and stochastic for global. In this comment we reexamine some of their conclusions regarding the global stochastic search strategies. It is concluded that the steps taken to increase the efficiency of the optimization of the test problem chosen by Smajic et al., which was selected to keep the search space tractable, can be mislead- ing when applied to intractable problems.