今天分享的是：量子专题：2025量子互联网与算网协同体系架构白皮书 报告共计：94页 第九届未来网络发展大会组委会 2025年8月 以下为报告节选内容 未来网络发展大会 未来网络技术发展系列白皮书（2025） 量子互联网与算网协同 体系架构白皮书 《量子专题：2025量子互联网与算网协同体系架构白皮书》（紫金山实验室等编写）系统梳理量子互联网与算网协同相关技术、架构及应用，首先介绍量子 信息技术基础，涵盖量子力学核心概念（叠加态、纠缠态、量子测量等），以及量子通信（量子密钥分发QKD、量子隐形传态、量子安全直接通信 QSDC）、量子计算（分四阶段发展，现有超导、离子阱等物理平台，关键算法如Shor、Grover算法）、量子精密测量（突破标准量子极限，应用于量子时 钟网络、长基线望远镜）三类典型应用，还提及线性光学、原子、超导等实验系统及DiVincenzo五大要求。接着阐述量子互联网架构，指出其发展分可信中 继、准备和测量等六阶段，现有多国部署可信中继网络，量子中继分四代（第一代含预报式纠缠分发等，全光中继用簇态），协议栈有Van Meter五层、 Wehner五层等多类方案，分组交换技术含基于经典-量子混合 ...

Core Viewpoint - The article discusses a new research achievement by a team led by AI expert Tian Yuandong, which introduces a continuous thinking chain model that parallels quantum superposition, enhancing efficiency in complex tasks compared to traditional discrete thinking chains [2][4]. Group 1: Research Findings - Traditional large language models (LLMs) utilize discrete tokens for reasoning, which can be inefficient for complex tasks, requiring O(n^2) decoding steps and often getting stuck in local optima [4]. - Recent studies indicate that using continuous hidden vectors for reasoning can significantly improve performance, although theoretical explanations were previously lacking [5]. - The team demonstrated that a two-layer Transformer with D-step continuous chains of thought (CoTs) can solve directed graph reachability problems, outperforming discrete CoTs models that require O(n^2) decoding steps [7]. Group 2: Methodology - The continuous thinking chain allows for simultaneous encoding of multiple candidate graph paths, akin to breadth-first search (BFS), providing a significant advantage over discrete thinking chains, which resemble depth-first search (DFS) [8]. - A designed attention selector mechanism enables the model to focus on specific positions based on the current token, ensuring effective information extraction [11][12]. - The first layer of the Transformer organizes edge information, while the second layer facilitates parallel exploration of all possible paths [21][22]. Group 3: Experimental Results - The team conducted experiments using a subset of the ProsQA dataset, which required 3-4 reasoning steps to solve, with each node represented as a dedicated token [26]. - The COCONUT model, utilizing a two-layer Transformer, achieved an accuracy close to 100% in solving ProsQA problems, while a 12-layer discrete CoT model only reached 83% accuracy, and a baseline model solved approximately 75% of tasks [27][28]. - The model's behavior was further validated through analysis of attention patterns and continuous thinking representations, supporting the theoretical hypothesis of superposition search behavior [30].

Summary of Quantum Computing Conference Call Industry Overview - The conference focuses on the **quantum computing industry**, discussing its principles, technologies, applications, and challenges. Core Points and Arguments - **Definition and Principles of Quantum Computing**: Quantum computing is based on quantum mechanics, utilizing quantum bits (qubits) that can represent 0, 1, or both simultaneously, allowing for exponential growth in processing power as more qubits are added [3][4][10]. - **Current Quantum Computing Technologies**: The main technological routes include: - **Superconducting**: Mature but requires extremely low temperatures [5][12]. - **Ion Trap**: High precision but complex operations [5][15]. - **Neutral Atom**: Similar to ion traps but uses optical methods [5][12]. - **Optical**: Performs well in fast computation scenarios but is still debated regarding its stability [5][12]. - **Applications**: Quantum computers excel in simulating and optimizing complex problems, such as drug simulations and molecular dynamics, but are less efficient for simple arithmetic tasks [10][11]. - **Challenges**: High error rates, stability in large-scale systems, and material science issues are significant hurdles for practical applications [6][18]. - **Quantum Entanglement**: This phenomenon allows qubits to be interconnected, affecting each other's states instantaneously, but does not allow for faster-than-light information transfer [7][8]. Additional Important Content - **Performance Metrics**: Quantum volume (QV) is a key performance indicator, with Honeywell's ion trap quantum computer achieving a QV of over 1.1 million, while IBM's superconducting technology has a QV in the thousands [20]. - **Commercialization Efforts**: Companies like IONQ are exploring commercial applications, primarily in military sectors, with limited revenue currently [22]. - **Impact on Security**: Quantum computing poses a potential long-term threat to current encryption systems, but immediate risks are minimal. Preparations for quantum-resistant algorithms are underway [23][24]. - **Types of Quantum Chips**: Various quantum chips exist, including superconducting, ion trap, and optical chips, each with unique materials and stability challenges [25]. - **Market Landscape**: Currently, there are no publicly listed companies solely focused on quantum computing in China, although companies like GuoDun are involved in related fields [26]. This summary encapsulates the key discussions and insights from the quantum computing conference call, highlighting the industry's current state, technological advancements, and future challenges.