Physics-Informed Reinforcement Learning for Multi-Scale Optimization of Majorana Nanowire Device Yield

Physics-informed reinforcement learning for multi-scale optimization of Majorana nanowire device yield. This dataset contains synthetic nanowire configurations, quantum transport simulation outputs, machine learning training data, reinforcement learning optimization results, and publication-ready figures supporting a predicted 74.1% fabrication yield improvement over 47% baseline. All simulations are physics-constrained and fully reproducible. The dataset is intended for research in quantum device engineering, topological quantum computing, and AI-assisted nanofabrication optimization. Includes: Synthetic nanowire dataset and defect maps Quantum transport feature datasets Reinforcement learning optimization outputs Validation tables and figures Reproducibility documentation基于物理感知强化学习（Physics-informed reinforcement learning）的马约拉纳（Majorana）纳米线器件良率多尺度优化。 本数据集涵盖合成纳米线构型、量子输运（quantum transport）模拟输出结果、机器学习训练数据、强化学习（reinforcement learning）优化结果，以及可直接用于学术发表的配套配图，可支撑相较47%基准良率实现74.1%制程良率提升的相关研究。 所有模拟均受物理约束且具备完全可复现性。本数据集适用于量子器件工程、拓扑量子计算以及人工智能辅助纳米制程优化等领域的研究。 包含以下内容： 合成纳米线数据集与缺陷分布图 量子输运特征数据集 强化学习优化输出结果 验证用表格与图表 可复现性文档