AI 中文总结
研究基于位移演化相干态动力学，引入元数据辅助的连续变量光子存储架构，结合多模存储与元数据索引，能实现历史存储状态的导航和重建，为未来元数据感知存储架构提供基础。
AI 中文摘要
现代信息系统越来越需要不仅能存储静态数据，还能存储信息随时间演变的存储架构。虽然连续变量光子平台在量子通信和计算方面已被广泛研究，但其作为动态演变存储系统的潜力仍很大程度上未被探索。在这项工作中，我们引入一种元数据辅助的连续变量光子存储架构，其中信息由跨多模相干态编码的潜在向量表示，并通过可逆位移操作更新。该框架将多模存储与基于元数据的索引相结合，通过回滚检索实现历史存储状态的导航和重建。与专注于保存单个量子态的传统方法不同，所提出的框架将存储视为连续变量相空间中不断演变的轨迹。这种观点自然地扩展到除图像存储之外的时间数据流、机器学习潜在表示、科学模拟和其他形式的动态变化信息。尽管作为理论概念验证提出，但结果表明连续变量光子系统可能为未来能够存储、跟踪和重建演变信息历史的元数据感知存储架构提供有前景的基础。
英文摘要
Modern information systems increasingly require memory architectures capable of storing not only static data but also the evolution of information over time. While continuous-variable photonic platforms have been extensively studied for quantum communication and computation, their potential as dynamically evolving memory systems remains largely unexplored. In this work, we introduce a metadata-assisted continuous-variable photonic memory architecture in which information is represented by latent vectors encoded across multimode coherent states and updated through reversible displacement operations. The framework combines multimode storage with metadata-based indexing, enabling navigation and reconstruction of historical memory states through rollback retrieval. Unlike conventional approaches that focus on preserving a single quantum state, the proposed framework treats memory as a continuously evolving trajectory in continuous-variable phase space. This perspective naturally extends beyond image storage to temporal data streams, machine-learning latent representations, scientific simulations, and other forms of dynamically changing information. Although presented as a theoretical proof-of-concept, the results suggest that continuous-variable photonic systems may provide a promising foundation for future metadata-aware memory architectures capable of storing, tracking, and reconstructing the history of evolving information.
