EAGER: Advanced Development of Genetic Programming for Novel Active Metamaterials and Devices in Terahertz (THz) Regime

太赫兹频带跨度从0.1到10太赫兹，这比通常用于无线通信的射频频带高得多（数百倍）。由于更高的频率可以携带更大的数据，太赫兹频带确保了现代非常期望的无线通信的更高的数据速率。但是设计和制造在太赫兹范围内工作的设备是非常困难的。这是因为大多数天然材料对太赫兹辐射没有反应。因此，人工材料或超材料已成为太赫兹器件发展的主流。目前超材料和器件的设计大多基于传统方法，严重依赖于该领域的专家知识。EAGER项目提出使用遗传编程来设计和优化太赫兹波段的有源超材料和器件。遗传规划是一种先进的进化优化方法，其解决方案表示为计算机程序，并在许多工程领域产生了具有人类竞争力的新颖设计。它比著名的遗传算法更先进，遗传算法通常用于优化给定设计中的参数，以满足设定的规格。遗传规划不需要预先定义的拓扑结构的设计。相反，遗传编程可以计算出最佳拓扑以及其他规格。这个EAGER项目的结果包括遗传编程的新算法和太赫兹频段的有源超材料和器件的急需设计。 所提出的工作的目标是充分发展遗传编程（GP）算法和优化方法，设计新的有源超材料的实际设备和组件的实现在太赫兹（THz）波段。太赫兹波段的技术在高数据速率无线通信、医学成像、爆炸物探测、安全检查、传感器等方面有着重要的应用，开发太赫兹波段的有源超材料对于这些技术的发展和实现其预期的好处至关重要。用于各种器件技术的有源超材料的多种可能设计导致显著的工艺复杂性，这对于手动探索是不可行的。另一方面，遗传编程是一种先进的进化优化方法，在许多工程领域产生了具有人类竞争力的新颖设计。该研究小组早期的GP工作集中在设计具有挑战性的宽带人工磁导体（AMC）超材料接地层，频率较低（几百MHz）。通过真正的3D图案几个成功的宽带设计实现了？没有男人？s？频带因此，将基因编程与活性超材料开发相匹配是一种值得称赞的方法，特别是在太赫兹波段，天然材料的应用有限。具体工作包括：1）利用GP开发有源超材料和太赫兹（THz）器件; 2）通过并行化提高GP的计算效率，实现更高效的（基于梯度的）优化算法;以及3）开发优化的太赫兹设备的示例，包括检测器、调制器、传感器和天线阵列;并将性能结果与基于人类专业知识的可用设计进行比较。这项工作还将导致遗传编程开发的重大进展，包括拓扑生成，可调材料集成和提高计算效率，可以解决超材料和设备开发的主动方面所涉及的高度复杂性。

The terahertz frequency band spans from 0.1 to 10 terahertz which is much (hundreds of times) higher than the radio frequency band commonly used for wireless communications. Since higher frequencies can carry larger data, the terahertz band ensures much higher data rate for wireless communications which is very much desired in modern times. But it is very difficult to design and build devices working in the terahertz regime. This is due to the fact that most natural materials do not respond to terahertz radiations. Therefore, artificial materials or metamaterials have become mainstream in the development of terahertz devices. Current design of metamaterials and devices is mostly based on traditional methods and heavily depends on expert knowledge in this area. This EAGER project proposes to use genetic programming for the design and optimization of the active metamaterials and devices in the terahertz band. Genetic programming is an advanced evolutionary optimization method in which the solution is represented as a computer program and has produced human-competitive, novel designs in many fields of engineering. It is more advanced than the well-known genetic algorithm which is often used to optimize parameters in a given design so as to meet set specifications. Genetic programming does not need pre-defined topology of the design. Instead, genetic programming can figure out the optimal topology as well as other specifications. The result of this EAGER project includes novel algorithms of genetic programming and the much needed designs of active metamaterials and devices in the terahertz frequency band. The objective of the proposed work is to fully develop Genetic Programming (GP) algorithm and optimization method for designing novel active metamaterials for practical device and components implementation in the Terahertz (THz) band. Technologies in this band have significant applications in developing devices for high data rate wireless communications as well as for medical imaging, explosives detection, security screening, sensors, and so on. Development of active metamaterials in the terahertz band is critically important for the evolution of these technologies and realization of their much anticipated benefits. The multitude of possible designs of active metamaterials for variety of device technologies leads to significant process complexities that are infeasible to explore manually. Genetic Programming, on the other hand, is an advanced evolutionary optimization method and has produced human-competitive, novel designs in many fields of engineering. Earlier GP work of the research group was focused on the design of challenging broadband Artificial Magnetic Conductor (AMC) metamaterial ground plane at lower frequency (a few hundreds of MHz). Through true 3D patterning several successful broadband designs were achieved in the ?no-man?s? frequency band. Matching genetic programming with active metamaterials development, therefore, presents a commendable approach, especially in the terahertz band, where natural materials are of limited application. Specific tasks of the proposed work include: 1) Using GP to develop active metamaterials and terahertz (THz) devices; 2) Improving the computational efficiency of GP through parallelization and implementation of more efficient (gradient based) optimization algorithms; and 3) Develop examples of optimized terahertz devices including detectors, modulators, sensors, and antenna arrays; and compare performance results with available designs based on human expertise. This work will also result in significant advances in Genetic Programming development including topology generation, tunable material integration, and improved computational efficiency that can tackle the high complexities involved in active aspects of the metamaterials and devices developments.