量子计算正加速从基础理论研究迈向工程化实现，成为未来信息技术的重要前沿方向。与此同时，人工智能在复杂系统建模、优化求解、表示学习和自动化设计等方面的持续突破，为量子算法设计、量子系统表征、量子纠错以及软硬件协同优化提供了新路径，正成为推动量子计算突破性能瓶颈、迈向规模化和实用化的重要支撑。当前，量子计算发展仍面临诸多关键挑战，包括NISQ器件噪声高、量子线路深度受限、系统表征成本高、纠错译码复杂度大，以及算法、线路与硬件协同不足等。这些问题在一定程度上制约了量子优势的验证进程，也限制了量子计算在量子化学、科学计算、电力系统等领域的应用拓展。

本论坛将聚焦人工智能增强的量子计算研究与工程，重点研讨量子算法与线路协同优化、量子编译与调度优化、量子纠错码设计与解码加速、机器学习驱动的量子系统表征与建模，以及面向典型应用场景的探索与实践，推动相关理论、关键技术与工程体系的协同创新发展。

王鑫，香港科技大学（广州）人工智能学域长聘副教授，国家级高层次青年人才，长期从事量子计算与量子人工智能研究，现任 ACM Computing Surveys 、IEEE Transactions on Information Theory、 ACM Transactions on Quantum Computing副主编，以及Quantum和中国科学：信息科学的编委。他曾任美国马里兰大学哈特里博士后，并于2019年至2023年在百度研究院带领团队开发量子机器学习平台“量桨”。2018年，他获悉尼科技大学博士学位，并获校长杰出博士论文奖。他在 Physical Review Letters、PRX Quantum、IEEE TIT、Nature Computational Science、NeurIPS、AAAI、ISCA、HPCA、MICRO 等权威期刊和会议发表论文100余篇。他曾受邀在量子计算顶会 TQC 作主旨报告（大陆首次），连续多年入选全球前2%顶尖科学家榜单，主持和参与国自然重大研究计划与国家重点研发计划、CCF-腾讯犀牛鸟基金等项目。

梁之鼎，香港中文大学计算机科学与工程系助理教授，曾于 2024–2025 年在伦斯勒理工学院计算机科学系担任助理教授。圣母大学博士，威斯康星大学麦迪逊分校学士。其研究成果已发表于多个顶级国际会议和期刊，包括 ISCA、HPCA、DAC、SIGMETRICS、ICCAD、NeurIPS、TCAD、TSG、TQE 和 TVCG 等。2021 年和 2022 年 DAC Young Fellow，2025年 ML and Systems Rising Star，圣母大学IEdison Innovation Fellowship。Quantum Computer System（QuCS）Lecture Series 的联合创始人之一，该系列已发展为量子计算领域具有广泛影响力的国际公开讲座平台。同时也是开源量子计算框架 TorchQuantum 的核心贡献者之一，该项目已被IBM Qiskit生态与PyTorch生态采用，并正在被接入Nvidia cuQuantum。担任多个顶会TPC委员，Integration, The VLSI期刊副主编，Npj Quantum Information，ACM trans on Quantum Computing，Quantum Machine Intelligence期刊客座编辑。

Yuxuan Du is an Assistant Professor at Nanyang Technological University (NTU), jointly appointed to the College of Computing and Data Science and the School of Physical and Mathematical Sciences. He received his Ph.D. in Computer Science from the University of Sydney. His research focuses on quantum learning theory, fundamental algorithms for quantum machine learning, and AI for quantum science. His work has been published in top-tier journals and conferences across physics and computer science, including Nature Communications, Physical Review Letters, Physical Review X Quantum, npj Quantum Information, IEEE Transactions on Information Theory, and IEEE Transactions on Pattern Analysis and Machine Intelligence. He has been listed as a Top 2% Scientist Worldwide in the field of AI and Physics, based on citation indicators published by Stanford University and Elsevier. He serves as an Editor of Neural Networks and has regularly held Area Chair roles for ICML, NeurIPS, and ICLR.

报告主题：Digital twin of quantum world

摘要：The exponential complexity of the grand quantum world poses a fundamental obstacle to its characterization, verification, and practical use. Conventional approaches, based on classical simulation or quantum tomography, quickly become intractable as the quantum system size grows. In this regard, a critical question arises: can a classical learner accurately and efficiently predict observables of interest for new classical inputs using only quantum data obtained from incoherent measurements?

In this talk, I will show that the answer is affirmative across a broad range of quantum platforms and applications. I will first introduce the concept of efficient digital twins for the quantum world and explain the theoretical principles that underpin them. I will then present how to construct provably efficient digital twins for qubit, fermionic, continuous-variable, and analog quantum systems, with applications spanning digital quantum simulation and variational quantum algorithms. I will also highlight our recent experimental demonstration for identifying nonequilibrium Floquet symmetry-protected topological phases on a 42-qubit superconducting processor. Finally, I will discuss how advanced deep learning and large language models may further extend the capability of digital twins, opening new directions for exploring the quantum world and advancing quantum science.

李鹤，教授，博导，国家青年人才、中国科协青托、江苏省首批“U35培育”对象，英国帝国理工博士，剑桥大学博士后。主持/参与国家、省自然基金、国网科技、省应用数学中心项目等。在集成电路设计、量子计算领域发表IEEE TC、TQC、DAC等论文50余篇，获ACM/IEEE ICCAD’25、MLCAD’24最佳论文提名、IEEE FPT’17最佳论文展示奖，2023-2024 CCF杰出演讲者。担任多个CCF A/B类国际会议TPC委员，《电子器件》副主编、《集成电路与嵌入式系统》客座编委，指导学生获国家级竞赛一等奖11项、最佳工程/创意奖2项。获2025东南大学青年五四奖章。

报告主题：应用驱动的量子算法与线路协同设计优化

摘要：在NISQ时代，量子EDA/QDA（量子设计自动化）是探索潜在量子计算应用优势的核心支撑，对研究先进量子算法、发掘具备潜在量子优势的应用场景至关重要。本次报告将重点介绍东南大学HIQC课题组近期研究成果：量子算法与线路协同优化方法，及其在高维计算机算术、分子激发态计算、量子经典混合神经网络、基于FPGA的量子算法模拟等领域应用。在此基础上进一步探索AI辅助优化新方法，构建更高效的量子算法-线路-硬件协同设计体系，为量子计算在AI、电力、医工、科学计算等领域落地提供支撑。

卢丽强，浙江大学计算机学院百人计划研究员。本科与博士毕业于北京大学，研究方向为计算机系统结构、人工智能芯片、量子计算软件与应用等。在ISCA、MICRO、Nature Physics等计算机体系结构、EDA、量子计算领域发表论文50余篇，首次在体系结构四大会组织量子计算tutorial，入选Elsevier全球前2%顶尖科学家榜单。主持国家重点研发计划（青年科学家）、国家自然科学基金委面上项目、浙江省杰出青年基金、浙江省科技厅尖兵等项目，与华为、字节、中国工行、中国移动等知名企业开展合作。获得全球可持续发展青年（30u30）、CCF-SYS 青年科技奖、ACM SIGOPS 中国新星奖、华为火花奖、北京大学优博等荣誉。

报告主题：受AI 芯片发展历程的启发——高性能qLDPC 解码加速器设计

摘要：本报告首先总结了人工智能在算法-芯片协同设计、软硬一体优化、生态建设等方面对量子计算的启发。其次，介绍汇报人在qLDPC解码加速器上的相关进展。qLDPC码作为当前主流的量子纠错编码方式之一，具备高编码率和高容错阈值，然而，由于其校验矩阵连接度高，通常需要求解大规模线性系统，在未来容错量子计算时代难以实现实时解码，本次报告将介绍一种具有高并行性的qLDPC解码算法，并利用线性系统的稀疏性设计其FPGA解码加速器，实验结果表明，该加速器可对 Bivariate Bicycle（BB）码（例如 [[784,24,24]]）实现小于1微秒的实时译码，同时在逻辑误码率方面达到与当前最先进译码器（如 BP+OSD）相当的性能。

个人简介：吴亚东现任上海交通大学John Hopcroft计算机科学中心长聘教轨副教授，主要从事量子信息理论及其与AI的交叉研究。他在卡尔加里大学获物理博士学位，曾于香港大学从事博士后研究，入选国家级海外高层次青年人才计划。在AI for Quantum领域，其代表性工作以第一/通讯作者身份发表在Nature Communications、Physical Review Letters等期刊，综述性论文被Nature Reviews Physics接收；另有合作成果被Nature Machine Intelligence接收。相关成果获Nature Computational Science、NewScientist和DeepTech等期刊和科技媒体报道。

报告标题：Learning to reconstruct Wigner functions in phase space

报告摘要：Wigner function tomography is a central tool for characterizing continuous variable quantum systems subject to energy bounds that guarantee a cutoff on their state space dimension. However, it becomes increasingly demanding as the effective dimension increases: as experiments can only access the Wigner function at a finite number of points, accurate characterizations require large amounts of experimental data, rapidly becoming intractable as we try to explore larger portions of the system’s state space. To address this bottleneck, here we develop a machine learning framework for reconstructing Wigner functions directly from sparse phase-space data. For quantum states with sparse Fock-space or coherent-state representations, such as binomial code states and cat states, we devise provably efficient machine learning algorithms whose measurement complexity scales only logarithmically with the effective Hilbert-space dimension. For more general states beyond this regime, such as Gottesman–Kitaev–Preskill (GKP) states, we design a deep learning model that reconstructs the Wigner function from sparse measurements and generalizes to arbitrary phase-space resolution. We demonstrate the broad applicability of our framework on both simulated data and experimental data from a circuit quantum electrodynamic system. On simulated data, we find that our model reconstructs GKP states using substantially fewer measurements than required for informational completeness. On experimental data, we find that it reconstructs Wigner functions of GKP code states across multiple rounds of quantum error correction and identifies the dominant error process using significantly fewer measurements than conventional estimation techniques.

朱成鸿，上海幻码跃迁量子科技有限公司主任架构师，博士毕业于香港科技大学（广州），并分别于 2022 年和 2020 年在墨尔本大学获得计算机科学硕士学位和学士学位。研究方向主要包括量子体系架构、量子编译、量子机器学习与量子信息理论。近年来，他围绕可扩展量子计算系统、硬件感知量子编译和量子信息处理等方向开展研究，已在 ISCA、HPCA、MICRO、DAC、IEEE TIT 等国际重要会议和期刊发表论文十余篇。2026 年获香港科技大学（广州）人工智能学域最佳科研奖。

报告主题： 面向离子阱量子计算的硬件感知编译

报告摘要： 离子阱量子计算具有高保真量子门和长相干时间等优势，是实现可扩展量子计算的重要技术路线。然而，真实离子阱设备中的量子操作并非理想化执行，离子移动、门作用区域和调度冲突都会直接影响线路执行时间与整体保真度。本报告围绕离子阱设备的硬件感知量子编译展开，介绍从一维线性离子阱到二维 QCCD 离子井架构中的编译优化思路，重点讨论如何根据设备结构进行量子比特映射、离子穿梭和门调度设计，以降低硬件约束带来的额外开销。进一步地，报告将讨论神经网络方法在该问题中的应用，探索如何通过学习量子线路与设备约束之间的关系，为穿梭调度和编译搜索提供更有效的指导。

单小涵，北京中科量枢科技有限公司COO，中国科学院计算技术研究所博士、清华大学计算机系博士后，曾任启元实验室副研究员、博士生导师。研究领域涵盖智能算法理论、系统与应用，在相关领域发表多篇高水平论文。曾牵头建设智能算法自动生成、终身强化学习、多智能体协同决策等人工智能系统平台，具备从算法研究、系统构建到工程化落地的完整经验，现致力于推动人工智能技术与量子计算场景的深度融合。

报告主题：量子计算系统的智能适配方法探索

报告摘要：量子计算系统具有很强的场景依赖性，其运行效果会随着设备状态、噪声条件以及任务与后端配置的变化而显著波动，在实际应用中往往需要依赖人工经验进行针对性调整。本报告围绕“智能适配”这一主线，聚焦量子计算系统中对场景变化较为敏感的几个关键环节，探索人工智能方法在减少人工调参、提升方法复用和跨场景适配方面的作用。在量子校准方面，针对量子硬件参数漂移以及校准流程依赖专家经验的问题，研究基于 Agent 的校准流程组织方式，将实验选择、结果分析与参数更新纳入统一的闭环决策过程。在纠错解码方面，针对不同噪声条件下模型性能容易变化的问题，基于 Ising 纠错模型开展神经网络结构改进，并结合持续学习训练方法提升模型的泛化能力。在量子模型微调方面，针对任务、模型结构、量子嵌入方式与后端工具组合复杂的问题，构建 QuantumAdapt 框架，通过配置化与模块化方式实现从用户配置到训练评估的端到端流程，提升量子嵌入大模型方法的工程复用性。

Hong-Ye Hu is an Associate Professor who joined the Institute of Physics, Chinese Academy of Sciences in July 2026. Prior to this, he was a Harvard Quantum Initiative Fellow at Harvard University and the Harvard-MIT Center for Ultracold Atoms, working on quantum algorithms, quantum error correction, and quantum statistical learning theory. His work—such as ansatz-free learning and detecting d-wave superconductivity—has been highlighted by several media outlets, including APS Physics Magazine and Physics World. He obtained his Ph.D. from the University of California, San Diego in 2022 and his B.S. from Peking University in 2016.

报告主题：Self-calibrating fault tolerance: turning syndromes into a control signal

报告摘要：Quantum error correction can protect logical information only when every physical operation remains below the error threshold, a condition that must hold continuously, not just at initial calibration. In practice, analog control parameters inevitably drift due to environmental fluctuations. This challenge is especially severe for quantum applications whose execution times of days to months far exceed the stability timescale of any physical hardware, making the conventional halt-and-recalibrate approach infeasible. A promising strategy is to unify calibration with computation by repurposing syndrome measurements as a calibration signal, but whether this can be done with provable efficiency guaranties has remained open. Here, we address this problem by proving that the average error detection rate is a positive definite surrogate with high probability for a broad class of control-induced errors, including coherent errors and general CPTP maps. We develop an online learning framework that converges to $\varepsilon$ average error detection rate within $O(1/\varepsilon^2)$ syndrome measurement rounds for time-independent drifts, with a rigorous regret bound for time-dependent drifts. Pulse-level simulation of neutral-atom arrays and large-scale Clifford simulations confirms our predictions. Our results establish that fault-tolerant quantum computation can be made self-calibrating with rigorous performance guaranties, a necessary step toward quantum computations that run autonomously for days and months.