工作流程
计算机科学
接口(物质)
用户界面
数据库
表(数据库)
领域(数学分析)
知识库
计算
联轴节(管道)
Web应用程序
基础(拓扑)
要素(刑法)
人机交互
软件工程
程序设计语言
应用程序编程接口
数据存取
万维网
Web服务
作者
Yang Hu,Vladyslav Turlo
标识
DOI:10.1038/s41524-026-02377-3
摘要
Existing FAIR materials repositories can only serve pre-computed entries, so broad coverage is perpetually incomplete and cannot adapt to new questions on demand. We present OptiMat Alloys, a large language model-powered conversational agent for multi-principal element alloy exploration. Its central contribution is LLM-based workflow orchestration coupled to a living database : every calculation is performed on demand and persisted automatically with provenance in a local database, so asking a question produces provenance-rich, FAIR-aligned records without a separate deposition step. The agent is accessible through a web interface that requires no programming expertise ( low-barrier accessibility ) and reports uncertainty through cross-potential and cross-configuration validation ( built-in uncertainty quantification ). By coupling foundational machine learning interatomic potentials that cover nearly the entire periodic table with natural-language interaction, OptiMat Alloys enables targeted, on-demand computation guided by the user’s domain knowledge and brings computational alloy screening within reach of any materials scientist.
科研通智能强力驱动
Strongly Powered by AbleSci AI