EAGER: An Agent Framework for Autonomous Scientific Simulations

Tao Li of University of Delaware is supported by an award from the Chemical Theory, Models and Computational Methods program to develop an artificial intelligence (AI) agent framework for autonomous scientific simulations in computational chemistry and related fields. Although computational simulations play a central role in modern scientific discovery, mastering each computational package requires substantial domain knowledge, which remains a major bottleneck in modern research workflows. While large-language-model-based AI agents have been reported to automate computational simulations using a single or a small set of computational packages, there is a lack of a unified AI agent framework which can handle a range of different computational packages for research-oriented simulations. To bridge this gap, this project will develop an innovative agent framework for multidomain scientific simulations. In addition to scientific advancements, the team will make this agent framework openly available to the general public. Dr. Li’s research will focus on developing a unified agent framework for research-oriented scientific simulations. The team will employ a four-layer progressive disclosure mechanism to enable efficient agent reasoning on a specific computational task, from selecting the most suitable package for the user's request to loading research-quality simulation pipelines. Additionally, the team will implement multiple computational workflows in the agent framework to facilitate autonomous computational simulations ranging from demonstrative calculations to large-scale computational tasks. The team will further extend this agent framework for simulations on both workstations and high-performance computing clusters. Overall, these efforts will not only allow researchers to explore simulation parameter spaces more efficiently, but also provide a valuable tool for NSF's Gold Standard Science policy for improving the transparency and reproducibility of scientific publications. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2620630 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Tao Li | Institution: University of Delaware, NEWARK, DE | Award Amount: $300,000 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2620630 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2620630.html