PolyJarvis: An LLM-Orchestrated Agent for Automated All-Atom Molecular Dynamics of Amorphous Homopolymers

2026-03-01 · Source: cs.CL updates on arXiv.org · Field: Science & Research — Physical Sciences & Chemistry, Mathematics & Computational Sciences, Artificial Intelligence & Machine Learning · Depth: Expert, extended

Summary

PolyJarvis is an autonomous agent that integrates Anthropic's Claude 4 large language model with the RadonPy simulation platform via Model Context Protocol (MCP) servers to automate all-atom molecular dynamics (MD) simulations for amorphous homopolymers. This system enables end-to-end polymer property prediction from natural language input, autonomously handling tasks like monomer construction, force field parameterization, GPU-accelerated equilibration, and property calculation. Validated on polyethylene (PE), atactic polystyrene (aPS), poly(methyl methacrylate) (PMMA), and poly(ethylene glycol) (PEG), PolyJarvis predicted densities within 0.1–4.8% and bulk moduli within 17–24% of reference values for aPS and PMMA. PMMA's glass transition temperature ($T_{g}$) of 395 K matched experiment within +10–18 K, while other polymers showed +38 to +47 K overestimation. Overall, 5 of 8 property–polymer combinations met strict acceptance criteria, demonstrating the agent's capability to produce results consistent with expert-run simulations.

Key takeaway

For AI Engineers and Research Scientists developing automated scientific workflows, PolyJarvis demonstrates that LLM agents can significantly reduce manual effort in complex simulations. You should consider integrating LLMs with specialized simulation platforms using standardized protocols like MCP to enable autonomous decision-making and error recovery. This approach frees your team for ideation and experimental design, accelerating materials discovery, but be mindful of inherent simulation biases like cooling-rate effects on \$T_g\$ predictions.

Key insights

LLM agents can autonomously execute complex scientific workflows, adapting protocols and recovering from errors.

Principles

• Intelligent decision-making improves simulation accuracy.
• Iterative refinement enhances protocol robustness.
• Standardized interfaces enable tool orchestration.

Method

PolyJarvis uses Claude Sonnet 4.5 to orchestrate RadonPy for molecular construction and LAMMPS for GPU-accelerated MD simulations via MCP servers, making polymer-specific decisions.

In practice

• Use LLM agents for end-to-end MD simulations.
• Implement MCP for tool integration.
• Validate predictions against experimental data.

Topics

• Large Language Models
• Molecular Dynamics Simulation
• Polymer Informatics
• Autonomous Agents
• RadonPy
• Model Context Protocol

Code references

• arz-2/PolyJarvis

Best for: AI Scientist, Research Scientist, AI Engineer

Related on AIssential

• From Data to Theory: Autonomous Large Language Model Agents for Materials Science — Autonomous LLM agents can perform end-to-end scientific theory development, from equation selection to code execution and validation.
• Polyformer: a generative framework for thermodynamic modeling of polymeric molecules — Polyformer generates thermodynamic conformational ensembles for polymeric molecules, predicting folding and temperature-dependent changes.
• MDForge: Agentic Molecular Dynamics Pipeline Design under Sparse Simulator Feedback — MDForge automates complex MD pipeline design using an LLM agent with online verbal reward and multi-agent debate for sparse feedback densification.
• AutoDFT: A Closed-Loop Multi-Agent Framework for Autonomous DFT Calculations — AutoDFT automates complex DFT calculations via a closed-loop multi-agent LLM framework, adapting dynamically to emergent results.
• Force-free molecular dynamics through autoregressive equivariant networks — TrajCast uses equivariant MPNNs to directly predict molecular states, enabling faster, data-efficient MD simulations with strong generalization.
• ChemGraph-XANES: An Agentic Framework for XANES Simulation and Analysis — ChemGraph-XANES automates XANES simulation and analysis using LLM agents for complex workflow management.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CL updates on arXiv.org.