Agentic system enables flexible atom-level simulations with rigor
El Agente Potente: High-Throughput Agentic Atomistic Simulations
Artificial Intelligence
Summary
Running computer simulations to understand how atoms behave can be slow and complicated. The authors created a system called El Agente Potente that helps plan and run these simulations more efficiently. It uses a mix of strict step-by-step instructions for routine parts and smart code generation for complex tasks. This approach balances being organized with the flexibility needed to explore materials, molecules, and chemical reactions.
What this means in practice
- •For computational materials scientists: Run reproducible and auditable atom-level simulation campaigns to discover new materials faster with automated planning and flexible workflow construction.
- •For chemical process engineers: Construct tailored simulation workflows to explore catalytic reactions and adsorption processes with more procedural flexibility and computational rigor.
Authors
Tsz Wai Ko, Jiaru Bai, Thomas Swanick, Yeonghun Kang, Changhyeok Choi, Angelina Qihong Jiang, Aiwei Yin, Varinia Bernales, Alán Aspuru-Guzik
Abstract
Foundational machine-learning interatomic potentials (MLIPs) are transforming atomistic simulations by achieving near-ab initio accuracy across large chemical spaces at a fraction of the computational cost. A central challenge in using these tools for high-throughput property calculations is translating high-level scientific intent into adaptive simulation campaigns without compromising workflow rigour. We introduce El Agente Potente, an agentic system that combines typed execution graphs with a complementary coding mode for MLIPs-driven atomistic simulations. Typed execution graphs provide structured and provenance-aware execution for standardized workflows, with large language models (LLMs) restricted to planning and routing while deterministic Python components perform scientific computation and validation. Complementing this structured execution, a coding agent constructs customized workflows for tasks requiring greater procedural flexibility while invoking existing Potente functions for supported calculations. We demonstrate El Agente Potente across computational materials discovery, molecular energy-landscape exploration, adsorption, and catalytic reaction workflows, together with systematic benchmarks of reproducibility and LLM token cost. These results establish typed execution graphs and code-based workflow construction as complementary mechanisms for agentic scientific computing, combining controlled, auditable execution with the flexibility required for customized atomistic simulations