AutoSolvate-Agent: An Autonomous Agent for GPU-accelerated Solution-phase Quantum Chemistry

Xu Chen, Fangning Ren, Dazhou Yu, Lechen Dong, Fang Liu

Keywords: LLM-Based Agent, Quantum Chemistry, Computational Chemistry Automation, Density Functional Theory (DFT), Electronic Structure

Abstract: Many chemically relevant phenomena occur in solution or at complex interfaces, yet quantum chemical calculations of such systems involve intricate workflows and extensive parameter tuning, posing challenges for non-specialists and even experts. Interpreting the results of these simulations requires deep knowledge of computational chemistry, from model setup to post-processing. In this work, we introduce an autonomous agentic system specifically designed for solution-phase quantum chemistry. Built on an agentic architecture, the system supports flexible tool integration, resource management, I/O handling, and post-wavefunction analysis. Through representative case studies, we show that this agent significantly reduces the complexity and expertise required to perform and interpret advanced simulations, making them more accessible to both theorists and experimentalists.

Submission Number: 14