量子计算最困难的问题之一与增加量子计算机的大小有关。全球的研究人员都在寻求解决这一“规模挑战”。
为了使量子缩放更接近现实，来自14个机构的研究人员通过量子优势共同设计中心（C2QA），能源部（DOE），科学办公室，国家量子信息科学研究中心进行合作。他们一起构建了ARQUIN框架——一个以不同层模拟大规模分布式量子计算机的管道。他们的研究结果发表在《ACM量子计算汇刊》上。
由来自布鲁克海文国家实验室和麻省理工学院（MIT）的Michael DeMarco领导的研究小组，从将多个计算“节点”组合到一个统一计算框架的标准计算策略开始。
理论上，这种多节点系统可以被模拟来增强量子计算机——但是有一个问题。在超导量子系统中，量子位必须保持极低的温度。这通常是在一种称为稀释冰箱的低温设备的帮助下完成的。问题在于，在一台冰箱内将量子计算芯片扩展到足够大的尺寸是很困难的。
即使在较大的冰箱中，单个芯片内的超导电路也很难维护。为了创建强大的多节点量子计算机，研究人员不仅需要连接一个稀释冰箱内的节点，还需要连接多个稀释冰箱上的节点。
没有一个机构能够开展ARQUIN框架所需的全部研究。ARQUIN团队的研究人员来自太平洋西北国家实验室（PNNL）、布鲁克海文、麻省理工学院、耶鲁大学、普林斯顿大学、弗吉尼亚理工大学、IBM等。
PNNL的量子计算机科学家塞缪尔·斯坦（Samuel Stein）说：“许多量子研究都是在孤立的情况下进行的，研究小组只关注拼图的一部分。”“这几乎就像收集原料，却不知道它们如何在食谱中协同工作。当实验只在量子计算机的一个方面进行时，你无法看到结果如何影响系统的其他部分。”
相反，ARQUIN团队将构建多节点量子计算机的问题分解为不同的“层”，每个机构根据自己的专业领域在不同的层上工作。
“这是一个巨大的优化问题，”IBM物理科学委员会主席马克·里特（Mark Ritter）说。“团队必须对该领域进行深入评估，看看我们在技术和算法方面所处的位置，然后进行模拟，找出瓶颈所在，以及可以改进的地方。”
ARQUIN框架的重点是通过微波与光学链路连接的超导量子器件。每个机构都专注于量子计算配方的不同成分。例如，当一些研究人员研究如何优化微波到光的转导时，其他人则创建了利用分布式架构的算法。
麻省理工学院的Isaac Chuang教授说：“这种跨领域的系统研究对于绘制有用的量子信息处理应用的路线图至关重要，并且是由美国能源部的国家量子计划唯一实现的。”
For their part of the ARQUIN framework, PNNL researchers including Stein, Ang Li, and James (Jim) Ang designed and built the simulation pipeline and generated the Quantum Roofline Model that connected all the ingredients together—essentially creating a framework for trying out different recipes for future quantum computers.
From his unique vantage point, PNNL physicist Chenxu Liu understands the need for multi-institutional collaborations well. He worked on the ARQUIN framework while he was a postdoctoral researcher at Virginia Tech.
"While each research group had expertise in their portion of the project, no one had a very deep understanding of what all of the other groups within the project were doing," said Liu. "However, each group's work needed to be embedded into the whole pipeline view of the quantum computer in order to make it meaningful."
After compiling the different pieces of the project together, ARQUIN became a framework for simulating and benchmarking future multi-node quantum computers. This marks a promising first step toward enabling efficient and scalable quantum communication and computation by integrating modular systems.
Though a functional multi-node quantum computer outlined in the ARQUIN paper has not yet been created, this research provides a road map for future quantum hardware/software co-design.
"Creating a layer-based hierarchical simulation environment—including microwave-to-optical simulation, distillation simulation, and system simulation—was a crucial component in this work," said Li. "It allowed the ARQUIN team to understand and evaluate the tradeoffs between various design factors and performance metrics regarding the complex distributed quantum computing communication stack."
Some of the software products created for ARQUIN have already been used by members of the team for other projects. Many of the ARQUIN authors collaborated on another project, called HetArch, to further investigate different superconducting quantum architectures.
"This is an example of applying the principles of co-design from exascale computing to our ARQUIN/HetArch design space explorations," said Ang.
