量子计算发展现状OA
CURRENT STATUS OF QUANTUM COMPUTING DEVELOPMENT:PLENARY REPORT OF THE 2025 NATIONAL CONFERENCE ON FUNDAMENTAL PHYSICS EDUCATION IN HIGHER EDUCATION
在2025年全国高等物理基础课程教育学术研讨会上,中国科学技术大学郭光灿院士作了题为"量子计算发展现状"的大会报告.围绕量子计算的发展背景、算力优势、研制障碍、全球进展、阶段划分与未来方向展开系统阐述,明确量子计算是突破经典信息技术物理极限的核心颠覆性技术,同时梳理了当前技术瓶颈与国内外发展格局.报告首先指出,经典信息技术均基于经典物理,受摩尔定律约束存在算力物理极限,当芯片制程逼近微观尺度、摩尔定律失效后,经典算力将触及瓶颈,量子信息技术应运而生.量子计算机作为量子技术中最具颠覆性的方向,其算力随量子比特数增加呈2N 指数级增长,可从根本上突破经典计算的性能上限.在量子算力超越经典算力的底层逻辑上,核心源于量子世界的不确定性与非局域性两大独特特性.经典物理遵循确定性与局域性,而量子世界中,量子客体物理量呈概率分布,纠缠粒子即便无相互作用、相距极远也能瞬时关联变化.量子信息的基本单元是量子比特,为量子态叠加态,N 个量子比特可承载2N 个经典数据,单次操作能同步作用于全部数据自由度,形成天然并行运算能力;量子纠缠则为量子算法提供支撑,让并行运算的算力优势转化为实际信息处理效率,这是量子计算算力远超经典计算的物理根源.报告指出,量子计算机研制存在两大核心障碍:一是量子态脆弱易消相干,宏观环境会快速破坏量子叠加与纠缠特性,使其退化为经典状态;二是量子操控精度不足,难以实现量子比特的精确制备与操控.理论上,容错量子纠错编码、量子避错、动力学容错等方案可解决上述问题.人工智能可在减少编码所需物理比特、优化操作方案、提升测量精度等方面赋能量子计算,成为推动技术突破的关键助力.全球量子计算发展历程与进展方面,1999年超导量子比特相干时间仅2ns,2012年延长至100μs,为实用化奠定基础;2016年国外推出全球首个可云端访问的5量子比特处理器,2019年推出首套商用量子计算机,同年宣称53量子比特处理器实现"量子优越性",但该结论针对特定抽象数学问题,不具备普适实用价值,我国科研人员通过新算法缩小了其与超算的算力差距.然而,我国量子计算发展受国外技术制裁,关键部件如稀释制冷机曾被禁运,但已实现自主研发并商品化出口.国内的量子计算公司,2024年年初发布第三代超导量子计算机,集成72个量子比特(含辅助比特超100个),国产化率约80%,开放云端访问后覆盖全球143个国家、注册用户约3000万,完成运算任务52万次+.在本次报告中,郭院士将量子计算发展划分为三个阶段:原型机研制阶段、量子优越性专用机阶段、量子计算与超级计算融合的量超融合阶段.目前全球已进入逻辑比特研发关键期,我国实现了500多个物理比特的相干操控.未来量子计算的核心应用集中于两大方向:一是破解经典计算机难以突破的密码体系;二是赋能人工智能,解决当前 AI算力瓶颈与高能耗问题,推动经典AI向量子AI演进,量子计算与人工智能的深度融合,将成为引领新一轮科技革命与产业变革的核心动力.
At the 2025 National Conference on Fundamental Physics Education in Higher Edu-cation,Academician Guo Guangcan of the University of Science and Technology of China de-livered a plenary report entitled"The Current State of Quantum Computing".He provided a systematic exposition covering the developmental context,computational advantages,research and development obstacles,global progress,phased development,and future directions of quantum computing.He clearly identified quantum computing as a core disruptive technology capable of breaking through the physical limits of classical information technology,whilst also outlining current technical bottlenecks and the landscape of domestic and international devel-opment.The report began by noting that classical information technology is based on classical physics and,constrained by Moore's Law,faces physical limits on computational power.As chip manufacturing processes approach the microscopic scale and Moore's Law ceases to hold,classical computational power will reach a bottleneck,giving rise to quantum information technology.As the most disruptive branch of quantum technology,quantum computing ex-hibits exponential growth in computational power increasing by 2N as the number of qubits in-creases,thereby fundamentally breaking through the performance ceiling of classical compu-ting.The underlying logic behind quantum computational power surpassing that of classical computing stems primarily from two unique characteristics of the quantum world:uncertainty and non-locality.Classical physics adheres to determinism and locality,whereas in the quan-tum world,the physical quantities of quantum objects follow a probability distribution,and entangled particles can exhibit instantaneous correlations and changes even when they are ex-tremely distant from one another and do not interact.The basic unit of quantum information is the qubit,which exists in a superposition of quantum states.N qubits can carry 2N classical bits of data,and a single operation can act simultaneously on all data degrees of freedom,cre-ating a natural capacity for parallel computation.Quantum entanglement provides support for quantum algorithms,enabling the computational advantage of parallel processing to be trans-formed into actual information processing efficiency.This is the physical root cause of why quantum computing power far exceeds classical computing.The report highlights two core ob-stacles in the development of quantum computers:firstly,the fragility of quantum states and their susceptibility to decoherence,whereby the macroscopic environment rapidly disrupts quantum superposition and entanglement,causing them to degrade into classical states;sec-ondly,insufficient precision in quantum manipulation,making it difficult to achieve the pre-cise preparation and control of qubits.Theoretically,solutions such as fault-tolerant quantum error-correcting codes,quantum error avoidance,and dynamical fault tolerance can address the aforementioned problems.Artificial intelligence(AI)can empower quantum computing by reducing the number of physical qubits required for encoding,optimising operational schemes,and enhancing measurement accuracy,thereby becoming a key driver of technological break-throughs.Regarding the global development and progress of quantum computing,the coher-ence time of superconducting qubits was merely 2ns in 1999,extending to 100μs by 2012,which laid the foundation for practical applications;In 2016,the world's first cloud-accessible 5-qubit processor was launched abroad,followed in 2019 by the first commercial quantum computer.That same year,a 53-qubit processor was claimed to have achieved"quantum su-premacy";however,this conclusion pertained to a specific abstract mathematical problem and lacked universal practical value.Chinese researchers have narrowed the computational power gap with supercomputers through new algorithms.However,the development of quantum computing in China has been hampered by foreign technology sanctions,with critical compo-nents such as dilution refrigerators once subject to embargo,but independent research and de-velopment has been achieved with commercialized exports now possible.A domestic quantum computing company released its third-generation superconducting quantum computer in early 2024,integrating 72 qubits(with over 100 including auxiliary qubits).With a domestic pro-duction rate of approximately 80%,the system has,following the opening of cloud access,reached users in 143 countries worldwide,with around 30 million registered users and over 520,000 computational tasks completed.In this report,Academician Guo divides the develop-ment of quantum computing into three stages:the prototype development stage,the quantum supremacy special-purpose machine stage,and the quantum-supercomputing fusion stage.The world has now entered a critical phase in the research and development of logical qubits,and China has achieved coherent control of over 500 physical qubits.In the future,the core appli-cations of quantum computing will focus on two major directions:firstly,cracking crypto-graphic systems that are difficult for classical computers to break;secondly,empowering AI to resolve current bottlenecks in AI computing power and high energy consumption issues,thereby driving the evolution from classical AI to quantum AI.The deep integration of quan-tum computing and AI will become the core driving force behind a new round of technological revolution and industrial transformation.
郭光灿
中国科学技术大学,安徽 合肥 230026
量子计算算力人工智能
quantum computingcomputationalpowerartificial intelligence(AI)
《物理与工程》 2026 (3)
5-15,11
评论