ChemGraph-XANES brings LLM agents to X-ray spectroscopy workflows
Researchers introduce ChemGraph-XANES, an agentic framework that uses LLM agents to automate XANES simulation and analysis by unifying natural-language task specification, structure acquisition, FDMNES input generation, parallel execution, and provenance-aware data curation.
Score breakdown
Scientists and ML engineers building spectroscopy datasets can use ChemGraph-XANES to automate and scale XANES simulation pipelines via natural-language instructions, reducing the manual workflow overhead that previously limited large-scale data generation.
- 01ChemGraph-XANES is an agentic framework for automated XANES simulation and analysis presented by Vitor F. Grizzi, Thang Duc Pham, and Luke N. Pretzie.
- 02The framework addresses workflow complexity as the primary constraint on large-scale computational XANES, not the underlying simulation method.
- 03Built on ASE, FDMNES, Parsl, and a LangGraph/LangChain-based tool interface.
Vitor F. Grizzi, Thang Duc Pham, and Luke N. Pretzie present ChemGraph-XANES, an agentic framework that addresses a key bottleneck in computational X-ray absorption near-edge structure (XANES) research: workflow complexity rather than the simulation method itself. The framework unifies natural-language task specification, structure acquisition, FDMNES input generation, task-parallel execution, spectral normalization, and provenance-aware data curation into a single reproducible pipeline. It is built on ASE, FDMNES, Parsl, and a LangGraph/LangChain-based tool interface, exposing XANES workflow operations as typed Python tools that large language model (LLM) agents can orchestrate.
The paper demonstrates documentation-grounded parameter retrieval and shows the workflow supports both explicit structure-file inputs and chemistry-level natural-language requests.
In multi-agent mode, a retrieval-augmented expert agent consults the FDMNES manual to ground parameter selection, while executor agents translate user requests into structured tool calls. The paper demonstrates documentation-grounded parameter retrieval and shows the workflow supports both explicit structure-file inputs and chemistry-level natural-language requests. Because independent XANES calculations are naturally task-parallel, ChemGraph-XANES is well suited for high-throughput deployment on high-performance computing (HPC) systems, enabling scalable XANES database generation for downstream analysis and machine-learning applications.
Key facts
- 01ChemGraph-XANES is an agentic framework for automated XANES simulation and analysis presented by Vitor F. Grizzi, Thang Duc Pham, and Luke N. Pretzie.
- 02The framework addresses workflow complexity as the primary constraint on large-scale computational XANES, not the underlying simulation method.
- 03Built on ASE, FDMNES, Parsl, and a LangGraph/LangChain-based tool interface.
- 04XANES workflow operations are exposed as typed Python tools orchestrated by LLM agents.
- 05In multi-agent mode, a retrieval-augmented expert agent consults the FDMNES manual to ground parameter selection.
- 06Supports both explicit structure-file inputs and chemistry-level natural-language requests.
- 07Task-parallel design makes it suitable for high-throughput HPC deployment and scalable XANES database generation for machine-learning applications.