www.koushare.com

Frank Wilczek

弗兰克·维尔切克是麻省理工学院物理学教授、量子色动力学的奠基人之一。因在夸克粒子理论（强作用）方面所取得的成就，他在2004年获得了诺贝尔物理学奖。

量子计算，从QPU开始

作者 | Frank Wilczek

翻译 | 胡风，梁丁当

配音 | 李娜（蔻享学术）

英文原文于2019年3月14日发表于华尔街日报。

使用量子比特的计算机性能将远超普通计算机，然而要实现它，我们还面临重重困难。

New systems using ‘qubits’ instead of classical bits will be vastly more powerful--but they still face important obstacles.

近几年，量子计算机领域有许多重要的进展。IBM、谷歌和微软是这些进展的主要推动者，一些大学科研团队和国家实验室也有重要贡献。量子计算技术前途无量，这一点已是业界的共识。然而对于未来量子计算机具体是什么样子的，又将有多大的影响力，却还没有个清晰的说法。事实上，越来越明显的一个趋势是，今后将会出现规模和形式多种多样的量子计算机，以适应不同的用途。

In recent years there has been a lot of progress in the field of quantum computers. IBM, Google and Microsoft are big players, as are several university-based research groups and national laboratories around the world. There is a widespread sense that the technology has great promise, but the exact nature and scope of that promise has been only vaguely defined. Indeed, it’s becoming clear that quantum computers will come in a variety of different sizes and shapes, suited for different purposes.

传统的计算机以比特作为信息的存储单元。一个比特有两个状态：1或者0（比特的英文Bit是二进制数字，即binary digit的简写）。各类复杂的信息--从棋子的位置、蛋白质的结构到猫的图片，都可以转换成一长串比特。而量子计算机则是使用量子比特（英文简称为qubit）。量子比特可以是1，也可以是0，还可以是介于1和0之间某种状态。这种状态叫做叠加态，是1和0两种状态不可分解的混合。在不同的探索性量子技术中，量子比特的载体可能是单个的原子、电子、微小的超导环，或是更奇特的“任意子”（这是我提出并命名的一种准粒子）。量子比特比经典比特更加复杂，能够存储的信息密度也更高。

Ordinary computers store information in the form of bits, which can be in one of two states, 1 or 0. (“Bit” is short for binary digit.) Long sequences of bits can encode complex information of many kinds, from chess positions to protein structures to pictures of cats. Quantum computers, on the other hand, use quantum bits, called qubits. Qubits can be 1, 0 or something in between--so-called superpositions, which are indissoluble mixtures of 1 and 0. In different exploratory technologies, qubits can take the form of individual atoms, electrons, minuscule superconducting loops or more exotic “anyons” (a breed of emergent particle that I studied and named). Qubits, being more intricate than bits, can store information more densely.

经典计算机并不适用于模拟量子系统。原因很简单：它们没有足够的空间来存储和操作量子系统所含的信息。量子计算机应该能更好地模拟量子系统。从理论上讲，它们能够计算各类材料的性质，包括可能的催化剂、药物、太阳能电池和蓄电池。这样的话，我们就不用再做那些费时费力的试探性实验了。

Classical computers aren’t well adapted to simulating quantum systems. They simply don’t have enough room to store and manipulate the required information. Quantum computers promise to be much better at quantum mechanics. In principle, that will enable them to calculate the properties of matter, including potential catalysts, drugs, solar cells and batteries, eliminating the need for laborious, hit-or-miss experiments.

也有人称，量子计算机还能充当破解密码的工具。但是这类应用的要求非常苛刻，需要利用数千个量子比特进行精确计算。这个目标距离我们还非常遥远，在可见的未来，通用量子计算机大致将局限在几十个量子比特以内，并且出错率很高。

Quantum computers are also touted as tools for cracking codes. But those applications are very demanding, requiring precise calculations involving thousands of qubits. That prospect is futuristic. For the foreseeable future, general purpose quantum computers will be restricted to a few tens of qubits, and they will be error-prone.

量子比特尽管性能强大，却很脆弱，且难以操控。在外界环境或其他粒子的干扰下，量子比特存储的信息很容易被打乱。比如，我们都知道传统计算机存储器中的信息可以被强磁场消除，而量子存储器对比那小得多的磁场都非常敏感。

they’re also delicate and hard to work with. If qubits are jostled by environmental fields or particles, it can scramble the information they were meant to encode. We’re familiar with the fact that a computer memory can get erased by a large magnetic field, for example, but quantum memory is sensitive to much smaller fields.

相比之下，虽然量子模拟器不那么灵活、不允许任意编程，但对于研究量子力学来说，却是更容易实现的替代方案。这种方法的基本策略是这样：如果一个特定量子系统的行为很有趣但是很难通过实验来研究，我们就利用另一个比较容易操控的量子系统来模拟它。超冷原子系统就是一种特别有用的量子模拟器，因为我们可以通过调节激光、电场和磁场来囚禁和操控原子，进而调控原子之间的相互作用。例如，我们可以用冷原子来模拟中子的行为，从而了解中子星的内部物质状态。

Quantum simulators are a less flexible but much easier alternative for studying quantum mechanics. The idea is to mimic the behavior of a particular quantum system that is interesting but difficult to study by using another quantum system that is more user-friendly. Assemblies of ultracold atoms are especially useful quantum simulators, because we can manipulate their interactions by bathing them in adjustable laser light, together with electric and magnetic fields. For example, we can use cold atoms as stand-ins for neutrons, to learn about neutron-star interiors.

经典-量子混合计算是另一个有前途的发展方向--经典计算机可以调用一个相对较小的量子“协处理器”来做一些关键的计算。这种协处理器的作用类似于图形处理单元（GPU，一种能非常高效地执行一些专门运算的超高速芯片）。GPU最初是为电脑游戏而设计开发的，它们使电脑可以快速地刷新图形显示。但是，聪明的研究人员发现，它们还可以用在别的很多地方，特别是模拟夸克和胶子是如何形成质子的。

Another promising development is classical-quantum hybrids, in which a classical computer can call on a relatively small quantum “coprocessor” to do critical calculations within its own programs. The coprocessor strategy follows the spirit of Graphical Processing Units (GPUs)-superfast chips that do some specialized operations extremely efficiently. GPUs were originally developed for use in computer games, where they make it possible to update fast-paced displays. But ingenious researchers have exploited them in many other ways—notably, to simulate how quarks and gluons build up protons.

在大型的通用量子计算机问世--如果真有那么一天--之前，这种经典--量子混合策略有可能使专门用于化学和材料科学的量子计算变成现实。让量子处理单元（QPU）尽快到来吧！

The hybrid strategy could start to deliver on the promise of quantum computation for chemistry and materials science long before large, general-purpose quantum computers are available—if they ever are. Bring on the QPUs!

关于“墨子沙龙”

墨子沙龙是由中国科学技术大学上海研究院主办、上海市浦东新区科学技术协会及中国科大新创校友基金会协办的公益性大型科普论坛。沙龙的科普对象为对科学有浓厚兴趣、热爱科普的普通民众，力图打造具有中学生学力便可以了解当下全球最尖端科学资讯的科普讲坛。