Benchmarking Five Global Optimization Approaches for Nano-optical Shape Optimization and Parameter Reconstruction
Benchmarking Five Global Optimization Approaches for Nano-optical Shape Optimization and Parameter Reconstruction
复制标题
DOI：
10.1021/acsphotonics.9b00706
复制
发表时间：
2019-11-01
期刊：
影响因子：
7
通讯作者：
Rockstuhl, Carsten
中科院分区：
文献类型：
--
作者：
Schneider, Philipp-Immanuel;Santiago, Xavier Garcia;Rockstuhl, Carsten
数值优化是计算物理领域，特别是纳米光学领域的重要工具。随着目前纳米制造技术可以实现的结构复杂性的增加，它引起了人们的注意，而合理的设计不再可行。此外，数值资源可用于实现计算光子材料设计和识别满足特定应用的预定义光学特性的结构。然而，优化目标函数一般是非凸的，其计算仍然需要资源，因此选择正确的优化方法是获得良好结果的关键。在此，我们对三个典型的纳米光学优化问题：粒子群优化、微分进化和贝叶斯优化，以及下坡单纯形优化的多起点版本和有限内存的Broyden-Fletcher-Goldfarb-Shanno(L-BFGS)算法，对五种全局优化方法进行了基准测试。在形状优化和参数重建领域所示的例子中，主要从机器学习应用中已知的贝叶斯优化在其他优化方法的一小部分运行时间内获得了显著更好的结果。
Numerical optimization is an important tool in the field of computational physics in general and in nano optics in specific. It has attracted attention with the increase in complexity of structures that can be realized with current nanofabrication technologies for which a rational design is no longer feasible. Also, numerical resources are available to enable the computational photonic material design and to identify structures that meet predefined optical properties for specific applications. However, the optimization objective function is in general nonconvex and its computation remains resource demanding such that the right choice for the optimization method is crucial to obtain excellent results. Here, we benchmark five global optimization methods for three typical nano-optical optimization problems: particle swarm optimization, differential evolution, and Bayesian optimization as well as multistart versions of downhill simplex optimization and the limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) algorithm. In the shown examples from the field of shape optimization and parameter reconstruction, Bayesian optimization, mainly known from machine learning applications, obtains significantly better results in a fraction of the run times of the other optimization methods.