发表机构
Japan Atomic Energy Agency; High Energy Accelerator Research Organization; Kyushu University（日本原子力机构; 高能加速器研究机构; 九州大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对蒙特卡罗粒子输运代码使用复杂的问题，提出将人工智能助手和代理应用于PHITS的策略，准备了两组人工智能就绪资源，经五个示范任务验证，表明提供适当资源和规则时人工智能代理可处理复杂工作流程，支持捆绑资源使用通用人工智能工具。
AI 中文摘要
蒙特卡罗粒子输运代码是强大工具，但使用需大量输入准备、执行及结果分析知识。本研究提出将现有人工智能助手和代理应用于PHITS的代码端策略。从手册等准备了两组互补的人工智能就绪资源：用于基于检索增强生成（RAG）助手的捆绑知识库和供人工智能代理直接使用的紧凑代理参考。知识库加载到NotebookLM提供对话式PHITS支持，代理参考与特定政策和执行规则结合，使Codex和Claude Code能编辑输入文件等。通过五个示范任务展示，结果表明提供适当资源和规则时人工智能代理可处理复杂PHITS工作流程，还得出了实用经验。这些发现支持将人工智能就绪资源与粒子输运代码捆绑，无需特定代码应用就能使用通用人工智能工具。
英文摘要
Monte Carlo particle transport codes are powerful tools, but their use requires substantial knowledge of input preparation, execution, and result analysis. In this study, we present a code-side strategy for applying existing AI assistants and AI agents to PHITS. Two complementary sets of AI-ready resources were prepared from manuals, lecture materials, sample inputs, utility information, and developer-curated cautions: a bundled knowledge base for retrieval-augmented generation (RAG)-based assistants and a compact agent reference for direct use by AI agents. The knowledge base was loaded into Gemini Notebook to provide conversational PHITS support, while the agent reference was combined with PHITS-specific policies and execution rules to enable AI agents to edit input files, execute calculations, inspect errors, analyze results, and assist with source-code modification and compilation. To evaluate the agent-driven workflow, five demonstration tasks were executed ten times with different AI agents and computing environments. The results showed that AI agents could handle complex PHITS workflows when appropriate resources and rules were provided. Practical lessons included precise prompts, human verification, well-documented sample files, explicit execution policies, and command-line-accessible tools. These findings support bundling AI-ready resources with particle transport codes to enable the use of general-purpose AI tools without requiring dedicated code-specific applications.
Comments36 pages, 2 figures, 2 tables, and 3 appendices. Accepted for publication in Journal of Nuclear Science and Technology
DOI：10.1080/00223131.2026.2742550
