Large Language Models (LLMs) have emerged as powerful tools for accelerating scientific discovery, yet their static knowledge and hallucination issues hinder autonomous research applications. Recent advances integrate LLMs into agentic frameworks, enabling retrieval, reasoning, and tool use for complex scientific workflows. Here, we present a domain-specialized agent designed for reliable automation of first-principles materials computations. By embedding domain expertise, the agent ensures physically coherent multi-step workflows and consistently selects convergent, well-posed parameters, thereby enabling reliable end-to-end computational execution. A new benchmark of diverse computational tasks demonstrates that our system significantly outperforms standalone LLMs in both accuracy and robustness. This work establishes a verifiable foundation for autonomous computational experimentation and represents a key step toward fully automated scientific discovery.
@article{arxiv.2512.19458,
title = {An Agentic Framework for Autonomous Materials Computation},
author = {Zeyu Xia and Jinzhe Ma and Congjie Zheng and Shufei Zhang and Yuqiang Li and Hang Su and P. Hu and Changshui Zhang and Xingao Gong and Wanli Ouyang and Lei Bai and Dongzhan Zhou and Mao Su},
journal= {arXiv preprint arXiv:2512.19458},
year = {2025}
}