量子计算的黎明：历史脉络与时代背景 / The Dawn of Quantum Computing: Historical Context

第 1 章：量子计算的黎明——历史脉络与时代背景

Chapter 1: The Dawn of Quantum Computing — Historical Context

> 本章核心问题 / Core Question > 量子计算作为一个学科领域是如何形成的？它解决了经典计算无法解决的什么问题？ > *How did quantum computing emerge as a discipline? What problems can it solve that classical computation cannot?*

---

1.1 引言：当信息论撞上物理学的边界

在 1948 年 Claude Shannon 发表那篇划时代的论文 *A Mathematical Theory of Communication* 之时，他或许未曾预料到，他所奠基的经典信息论将在七十年后遭遇到一次来自物理学深处的根本性挑战。Shannon 的信息论建立在一个隐含的假设之上——信息的载体是经典物理量（电压高低、磁极朝向、脉冲有无）。在这一假设下，比特（bit）是一个完美的抽象：它要么是 0，要么是 1，没有中间状态，没有不确定性，也没有任何"半 0 半 1"的可能。/ When Shannon laid the foundation of classical information theory in 1948, he implicitly assumed that information carriers are classical physical quantities (voltage levels, magnetic orientations, pulses). Under this assumption, a bit is a perfect abstraction: it is either 0 or 1, with no intermediate state, no uncertainty, and no "half-zero, half-one" possibility.

然而，当我们将视角从工程学转向基础物理学，情形便发生了根本性的转变。自然界在最基本的层面上遵循的是量子力学（quantum mechanics）的规律，而非经典力学。 量子力学告诉我们，微观粒子可以处于叠加态（superposition state），可以发生纠缠（entanglement），并且任何测量行为都会不可避免地扰动被测系统。这些性质并不是缺陷或噪声，而是物理实在的基本特征。/ However, when we shift from engineering to fundamental physics, the situation changes fundamentally. Nature at its most basic level follows quantum mechanics, not classical mechanics. Quantum mechanics tells us that microscopic particles can exist in superposition states, can be entangled, and that any measurement inevitably disturbs the measured system. These are not defects or noise—they are basic features of physical reality.

这就引出了一个深刻的问题：如果我们承认信息处理本质上是物理过程，那么在量子物理定律下，信息处理能力是否会发生质的变化？ 这个问题在二十世纪最后二十年逐渐从哲学思辨走向工程实现，并最终在二十一世纪初孕育出一门被称为"量子计算"的新学科。/ This raises a profound question: if we acknowledge that information processing is essentially a physical process, does the capacity for information processing undergo qualitative changes under quantum physical laws? This question gradually transformed from philosophical speculation to engineering implementation during the last two decades of the twentieth century, eventually giving birth to a new discipline known as "quantum computing" in the early twenty-first century.

本章的任务，是为读者勾勒这幅历史画卷：回答为何量子计算在二十世纪末兴起，又为何在二十一世纪成为战略性技术。我们将从经典计算的物理极限谈起，继而梳理 Feynman、Deutsch、Benioff 与 Shor 的关键贡献，分析量子计算与信息科学、密码学、凝聚态物理的交汇，建立全书统一的符号与术语系统，并通过对比经典比特与量子比特，完成读者思维方式的第一次跃迁。/ This chapter sketches this historical panorama, answering *why quantum computing emerged in the late twentieth century and why it became a strategic technology in the twenty-first century*. We will start from the physical limits of classical computing, then trace the contributions of Feynman, Deutsch, Benioff, and Shor, analyze the convergence of quantum computing with information science, cryptography, and condensed matter physics, establish a unified system of notation and terminology for the entire book, and complete the reader's first cognitive leap by comparing classical bits and qubits.

---

1.2 经典计算的物理极限与 Moore 定律的物理终结

1.2.1 经典计算的两条边界

要理解量子计算为何必须出现，首先必须理解经典计算所遭遇的两条边界：逻辑边界（logical boundary）与物理边界（physical boundary）。/ To understand why quantum computing had to emerge, we must first understand the two boundaries that classical computation has encountered: the logical boundary and the physical boundary.

逻辑边界由计算复杂性理论（computational complexity theory）刻画。早在 1971 年，Stephen Cook 就证明了布尔可满足性问题（Boolean Satisfiability Problem, SAT）属于 NP 完全（NP-complete）问题；同年 Leonid Levin 独立给出了类似的结论（这一结果后被称为 Cook-Levin 定理）。Leonard Adleman 与 Richard Karp 在次年扩展了这一框架，证明了更多经典问题的 NP 完全性。这些理论结果表明：存在一类问题，其求解所需的计算资源随输入规模呈指数增长，即使在最快的经典计算机上，问题的规模一旦超过某个阈值，便在可接受的时间内不可解。/ The logical boundary is characterized by computational complexity theory. In 1971, Stephen Cook proved that the Boolean Satisfiability Problem (SAT) is NP-complete; Leonid Levin independently reached a similar conclusion the same year (later known as the Cook-Levin theorem). Leonard Adleman and Richard Karp extended this framework the following year. These theoretical results show that *there exists a class of problems whose solution requires computational resources that grow exponentially with input size*, making them unsolvable in acceptable time once the problem scale exceeds a certain threshold, even on the fastest classical computers.

物理边界则由 Moore 定律的放缓暴露出来。1965 年，Gordon Moore 注意到集成电路上的晶体管数量大约每两年翻一番，并据此预测这一趋势将持续至少十年。这一预测不仅应验，而且在之后的四十多年里成为半导体行业自我实现的预言。然而，Moore 定律本质上是一条工程经验曲线，而非物理定律。当晶体管的特征尺寸（feature size）逼近原子尺度（约为 0.1 纳米）时，量子隧穿效应（quantum tunneling）将导致漏电流急剧增加，传统硅基互补金属氧化物半导体（CMOS）工艺将无法继续工作。/ The physical boundary was exposed by the slowdown of Moore's law. In 1965, Gordon Moore observed that the number of transistors on integrated circuits doubled approximately every two years, predicting this trend would continue for at least a decade. This prediction came true and became a self-fulfilling prophecy for the semiconductor industry for over forty years. However, *Moore's law is essentially an engineering empirical curve, not a physical law*. When transistor feature size approaches the atomic scale (about 0.1 nanometers), quantum tunneling will cause sharp increases in leakage current, making traditional silicon-based CMOS processes unable to continue functioning.

1.2.2 Landauer 原理与信息的物理成本

1961 年，IBM 研究员 Rolf Landauer 提出了一个深刻的原理：任何不可逆的逻辑运算（如擦除一个比特）必然伴随 $k_B T \ln 2$ 的能量耗散，其中 $k_B$ 是 Boltzmann 常数，$T$ 是环境温度。这就是著名的 Landauer 原理（Landauer's principle）。该原理将信息论与热力学联系了起来，揭示了信息的处理不是抽象的数学过程，而是受物理定律约束的能量转换过程。/ In 1961, IBM researcher Rolf Landauer proposed a profound principle: *any irreversible logical operation (such as erasing a bit) must dissipate energy of at least $k_B T \ln 2$*, where $k_B$ is the Boltzmann constant and $T$ is the environmental temperature. This is the famous *Landauer's principle*. It linked information theory with thermodynamics, revealing that *information processing is not an abstract mathematical process but an energy conversion process constrained by physical laws*.

Landauer 原理的意义远不止于节能。它表明：经典计算的能耗存在不可逾越的下限。当芯片上的晶体管密度越来越高，单个逻辑门的能量消耗越来越接近 Landauer 极限时，进一步降低能耗便不再可能通过简单的工艺改进实现，而需要寻找新的计算范式。/ The significance of Landauer's principle extends far beyond energy saving. It shows that *classical computation has an unavoidable lower bound on energy consumption*. As transistor density increases and energy per logic gate approaches the Landauer limit, further reductions cannot be achieved through simple process improvements—we need new computational paradigms.

> 📌 关键概念 / Key Concept > Landauer 极限 / Landauer Limit：在温度 $T$ 下擦除一比特信息所需的最小能量，约为 $k_B T \ln 2 \approx 2.87 \times 10^{-21}$ 焦耳（室温下）。 > *The minimum energy required to erase one bit of information at temperature $T$, approximately $k_B T \ln 2 \approx 2.87 \times 10^{-21}$ joules at room temperature.*

1.2.3 Moore 定律放缓的实证证据

2015 年左右，全球半导体行业普遍观察到 Moore 定律放缓的迹象。Intel 原本计划在 2016 年量产 10 纳米工艺，但实际推迟至 2019 年；7 纳米及以下的工艺节点中，每一代制程带来的性能提升与成本下降均显著低于历史平均水平。国际器件与系统路线图（IRDS, International Roadmap for Devices and Systems） 已不再像过去那样提供清晰的摩尔曲线预测，转而强调"more than Moore"——即通过异构集成、3D 堆叠、专用加速器等方式延续性能提升。/ Around 2015, the global semiconductor industry widely observed signs of Moore's law slowdown. Intel originally planned to mass-produce 10nm process in 2016, but it was actually delayed to 2019; for process nodes of 7nm and below, the performance gains and cost reductions brought by each generation have been significantly lower than historical averages. The *International Roadmap for Devices and Systems (IRDS)* no longer provides clear Moore curve predictions, instead emphasizing "more than Moore"—heterogeneous integration, 3D stacking, and specialized accelerators.

与此同时，量子计算却展现出一种反向的扩展规律：量子比特数（qubit count）正以近乎指数的速度增长。从 2016 年 IBM 推出 5 量子比特的 Quantum Experience，到 2023 年 IBM 推出 1121 量子比特的 Condor 处理器，再到 2024 年 Google、IBM、中国科大、本源量子等机构纷纷突破千比特门槛——量子计算正以一种与经典 Moore 定律相呼应却具有不同物理基础的方式，重新定义了"算力增长"的曲线。/ Meanwhile, quantum computing exhibits a kind of *inverse scaling law*: qubit count is growing at nearly exponen