SimMOF: AI Agent Automates MOF Simulations

The introduction of SimMOF, a novel AI framework, aims to streamline the automation of metal-organic framework (MOF) simulations. MOFs represent a diverse category of materials, where computational simulations play a vital role in property prediction. Traditionally, these simulations require expert input for workflow design, parameter choices, tool compatibility, and the preparation of structures ready for computation, making them difficult to access. SimMOF utilizes a large language model-based multi-agent system to automate complete MOF simulation processes from natural language inquiries. It converts user requests into plans that consider dependencies, produces executable inputs, coordinates various agents for simulation execution, and provides results with analyses tailored to the user’s query. Case studies illustrate that SimMOF fosters adaptive, cognitively autonomous workflows that mirror the iterative decision-making of human experts. The framework is elaborated in a paper on arXiv (2603.29152v2), which includes an abstract announcing the replacement. This innovation could greatly reduce the entry barriers for MOF simulations, making them more accessible to researchers lacking extensive computational knowledge.