Quantum: Microarchitectures for Quantum Computers

如果今天存在一台大规模的量子计算机，它将深刻影响几乎科学和工程的每个领域。它将通过大幅加快新材料和药物的开发而产生广泛的影响。它将提供新的安全通信基础设施。目前，只有小规模的量子计算机原型存在，在将这些设备扩展到足够大的尺寸以产生有意义的实际结果方面存在巨大的挑战。生产大规模量子计算机是我们国家和世界承担的一个极其重要的目标。这将需要几个科学和工程学科的共同努力，包括物理、材料、控制、计算机科学和计算机工程。这项提案为量子计算机的工程工作提供了资金，特别是此类设备的微体系结构。离子陷阱量子计算机的微体系结构显著改变了它的能力。一个糟糕的微体系结构具有有限的科学价值，因为它无法正确计算量子算法的结果。一个好的微体系结构不仅功能正常，还能揭示如何构建更大、更可扩展的设计的洞察力。要研究量子计算机的微体系结构，需要各种工具。量子算法可以由数十亿次运算组成。大规模量子微体系结构的提案由数千个离散元素组成。手工研究如何将这些数十亿操作中的每一个操作的执行映射到数千个可用组件上是不现实的。相反，对这类计算机系统的有用研究需要执行这种映射的自动化工具。这项工作的智力价值在于，它准确地研究了如何通过专注于这种工具的开发来构建真正的大规模量子计算机。特别是，我们之前的工作开发了一套所谓的计算机辅助设计(CAD)工具套件，用于基于离子陷阱的量子计算机。使用这些CAD工具，我们展示了建造量子计算机将面临的更大难度的挑战。这项提议为这些CAD工具的继续研究提供了资金。这项提议所资助的特定工具开发如下：在编码方案中实施更多的多样性，对作为CAD工具套件核心的调度机制进行检修，以及开发一组CAD单元，用于有效地计算量子软件中经常出现的执行模式。实施更多的纠错编码方案将使量子计算机架构师能够研究编码效率，并将编码方案专门用于设备中最适合它们的区域。更新调度程序将允许更好地利用并行性(一次做多件事)和局部性(让操作在物理空间中靠得很近，以最大限度地减少移动它们所需的时间)。最后，开发CAD单元将使我们能够将完全定制设计的组件合并到整体CAD框架中，从而利用这些努力来获得更多时间和面积效率更高的组件。我们的研究目的是通过这些技术的结合，我们可以减少我们在构建大规模量子计算机时面临的挑战，从而使这样的设备能够更快地构建。

If a large-scale quantum computer existed today, it would profoundly impact almost every area of science and engineering. It would have broad impact by dramatically speeding up the development of new materials and medicines. It would provide new secure communication infrastructure. Currently, only small-scale prototype quantum computers exist, and enormous challenges exist in scaling these devices to sizes large enough to produce meaningful practical results. Producing a large-scale quantum computer is a profoundly important goal for our nation and the world to undertake. It will require a combined effort of several disciplines of science and engineering, including, physics, materials, controls, computer science and computer engineering. This proposal funds work on the engineering of quantum computers, specifically the microarchitecture of such devices. The microarchitecture of an ion-trap quantum computer significantly alters its capabilities. A poor microarchitecture has limited scientific value because it is unable to correctly compute the result of quantum algorithms. A good microarchitecture not only functions properly, but also reveals insights about how to build ever larger and more scalable designs.To study the microarchitecture of a quantum computer tools are required. Quantum algorithms can consist of billions of operations. Proposals for large-scale quantum microarchitectures consist of thousands of discrete elements. It is not practical to study by hand how to map execution of every one of these billions of operations onto the thousands of available components. Instead, automated tools that perform this mapping are required for the useful study of such computer systems.The intellectual merit of this work is that it studies precisely how to build a real large-scale quantum computer by focusing on such tool development. In particular, our prior work developed what is termed a computer-aided-design (CAD) suite of tools for ion-trap based quantum computers. Using these CAD tools we demonstrated the orders of magnitude more difficult challenge building a quantum computer will be. This proposal funds continued research on these CAD tools.The particular tool development funded by this proposal is the following: implementation of more variety in the coding schemes, an overhaul of the scheduling mechanism at the heart of the CAD tool suite, and development of a collection of CAD-cells, for efficiently computing often appearing patterns of execution in quantum software. Implementing more error correction coding schemes will enable quantum computer architects to study code efficiency and specialize the coding schemes to the areas of a device where they are best suited. Updating the scheduler will enable better exploitation of parallelism (doing more than one thing at a time) and locality (having operations close together in physical space, as to minimize the time required to move them around). Finally, development of CAD cells will enable us to merge fully custom designed components into the overall CAD framework, thereby leveraging such efforts for more time and area-efficient components. Our research aim is that through a combination of these techniques we can reduce the challenges facing us in the construction of a large-scale quantum computer, thereby enabling the construction of such a device sooner in time.