Collaborative Research: FET: Small: Massive Scale Computing and Optimization through On-chip ParameTric Ising MAchines (OPTIMA)

几十年来，学术界和工业界一直依赖确定性算法和通用的冯-诺依曼计算体系结构来解决自然科学和社会科学中的组合优化(CO)问题。随着摩尔定律的继续减慢，现有的计算范式正在达到其可以处理的CO问题的最大复杂性的极限，从而变得越来越不足以在合理的时间内回答在从工程、物理和医学到经济和金融的广泛学科中不断上升的基本问题。通过模拟量子系统，被称为Ising Machines(IMS)的新计算体系结构已经出现。IMS提供了独特的机会，可以比任何现有的冯-诺伊曼同行更快地解决极其复杂的CO问题。然而，到目前为止，没有一种IM技术能够负担得起大量的自旋来处理目前无法解决的CO问题，同时通过半导体行业提供的合并的晶片规模制造工艺来确保低功耗、紧凑的外形因素、芯片级的集成度和可制造性。本项目的目标是探索和开发一种新的IM，即第一台片上参数Ising机床(Optima)。由于其独特的高度可重复编程的动态，触发不需要任何特殊的环境条件或任何耗时的前处理步骤，而只需要芯片级组件，可以单片集成，有利于大规模生产，Optima的开发将为强大、快速和小型化的量子计算系统铺平道路，每个人都可以从任何地方获得这些系统。这将使世界各地的学者、科学家、工程师和教育工作者能够创建新的网络基础设施，以应对相关的技术和社会挑战。该项目团队正在与东北大学和佛罗里达大学的STEM教育和劳动力发展项目合作，组织和主办与来自K-12学校和社区学院的学生和教师的校园活动，以及对当地学校的外联访问，以鼓励和扩大代表不足群体的参与。这些项目成果丰富了研究人员教授的本科生和研究生课程，这些课程包括电路理论、基于先进声学的通信和传感技术、微/纳米机电系统(MEMS/NEMS)以及量子工程设备和系统。Optima利用控制片上耦合电声参数振荡器(EAPO)同步网络电响应的独特动态功能，利用氮化铝(AlScN)微/纳米器件独特的铁电和声学特性，创造出极低功率和高度小型化的人造自旋，可通过互补金属氧化物半导体(CMOS)工艺制造。这些独特的特性打破了以往IMS设计中的所有模式，可以同时实现106个自转、兼容的晶片规模制造和室温操作，同时功耗不到1瓦。此外，由于其高度并行化的计算流程以及EAPO在超高频(SHF)范围内运行，Optima甚至能够在纳秒时间尺度上解决最困难的不确定多项式时间(NP)CO问题，而不受问题大小的影响。最后，由于Optima可通过与CMOS兼容的工艺制造，它在很大程度上利用了建立在相同硅片上的传统IC元件，以实现基于感兴趣的CO问题的灵活编程以及紧凑的读出方案。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

For decades, academia and industry have relied on deterministic algorithms and on general-purpose von-Neumann computing architectures to solve combinatorial-optimization (CO) problems within natural and social sciences. As Moore’s law continues to slow down, the existing computing paradigm is reaching the limit of maximum complexity of the CO problems it can tackle, thus becoming increasingly inadequate to answer, in reasonable times, the fundamental questions that keep rising in a wide range of disciplines, spanning from engineering, physics and medicine to economics and finance. By emulating quantum systems, new computing architectures known as Ising Machines (IMs) have been emerging. IMs offer the unique opportunity to solve extraordinarily complex CO problems much faster than any existing von-Neumann counterparts. Yet, to date, no IM technology can afford a massive number of spins to handle the currently unsolvable CO problems, while ensuring a low-power consumption, a compact form factor, a chip-scale integration and a manufacturability en masse through the consolidated wafer-scale fabrication processes offered by the semiconductor industry. The goal of this project is to explore and develop a new IM, namely the first On-chip ParameTric Ising MAchine (OPTIMA). Thanks to its unique highly reprogrammable dynamics, triggered without requiring any special environmental conditions or any time-consuming pre-processing steps while exclusively requiring chip-scale components that can be monolithic integrated in favor of a massive scale production, the development of OPTIMA will pave the way towards powerful, fast and miniaturized quantum-inspired computing systems, accessible to everybody from everywhere. This will allow the creation of new cyber infrastructures that scholars, scientists, engineers and educators worldwide will be able to use in order to address relevant technological and social challenges. The project team is collaborating with STEM education and workforce development programs, at both Northeastern University and the University of Florida, to organize and host on-campus activities with students and teachers from both K-12 schools and community colleges, as well as outreach visits to local schools to encourage and broaden participation of underrepresented groups. The project achievements are enriching both the undergraduate and the graduate courses that the investigators teach on circuit theory, advanced acoustic-based technologies for communication and sensing, micro/nanoelectromechanical systems (MEMS/NEMS), and quantum engineering devices and systems. OPTIMA is leveraging the unique dynamical features governing the electrical response of a synchronized network of coupled on-chip Electro-Acoustic-Parametric-Oscillators (EAPOs) exploiting the uniquely combined ferroelectric and acoustic properties of Aluminum Scandium Nitride (AlScN) micro/nano devices to create extraordinarily low-power and highly miniaturized artificial spins, manufacturable through complementary-metal-oxide-semiconductor (CMOS) processes. Such unique features allow the breaking of all the previous paradigms in the design of IMs by simultaneously enabling 106 spins, a CMOS-compatible wafer-scale manufacturing and room-temperature operation while consuming less than 1 Watt. Further, thanks to its highly parallelized computational flow and because the EAPOs are operating in the Super-High-Frequency (SHF) range, OPTIMA is able to solve even the hardest nondeterministic polynomial time (NP) CO problems in nanosecond time scales, independently of the problem size. Finally, since OPTIMA is manufacturable through CMOS compatible processes, it is greatly leveraging conventional IC components built on the same silicon wafer to enable flexible programming, based on the CO problems of interest, as well as compact read-out schemes.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.