arxiv: 2604.16205 · v1 · submitted 2026-04-17 · ❄️ cond-mat.mtrl-sci · cs.AI· physics.chem-ph

Recognition: unknown

ChemGraph-XANES: An Agentic Framework for XANES Simulation and Analysis

Vitor F. Grizzi , Thang Duc Pham , Luke N. Pretzie , Jiayi Xu , Murat Keceli , Cong Liu

Authors on Pith no claims yet

Pith reviewed 2026-05-10 08:06 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci cs.AIphysics.chem-ph

keywords XANES simulationagentic frameworkFDMNESLLM agentsworkflow automationcomputational spectroscopyhigh-throughput computingspectral curation

0 comments

The pith

ChemGraph-XANES lets LLM agents orchestrate the full XANES workflow from natural-language requests to HPC execution and curated spectra.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces ChemGraph-XANES as a unified agentic system that combines natural-language task input, structure retrieval, FDMNES input preparation, parallel execution via Parsl, spectral normalization, and provenance tracking. This addresses the main barrier to scaling computational XANES, which is workflow complexity rather than the physics simulation itself. A retrieval-augmented agent consults the FDMNES manual to guide parameter choices while other agents execute the typed tools. If the approach holds, it would make large XANES datasets practical to generate and use in machine-learning applications.

Core claim

ChemGraph-XANES exposes XANES operations as typed Python tools within a LangGraph/LangChain interface so that LLM agents can handle the entire sequence from user request through structure acquisition, FDMNES input generation, task-parallel execution, normalization, and data curation, with one agent retrieving documentation to ground parameter decisions.

What carries the argument

The LangGraph/LangChain-based tool interface that converts XANES workflow steps into callable functions for multi-agent LLM orchestration, supported by ASE for structures and Parsl for parallel runs.

If this is right

Independent XANES calculations can run at scale on HPC systems to build large spectral databases.
The workflow accepts both explicit structure files and chemistry-level natural-language descriptions.
Provenance tracking supports reproducible curation of spectra for later analysis.
The same architecture extends to high-throughput generation of training data for machine-learning models of X-ray spectra.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Analogous agentic layers could wrap other simulation codes that currently require manual input-file preparation.
Error rates in parameter selection could be quantified by comparing agent outputs against expert-generated inputs on benchmark cases.
The approach might lower the expertise threshold for running XANES studies in materials research.

Load-bearing premise

LLM agents can reliably read the FDMNES manual and translate user requests into correct tool calls and parameter choices without introducing errors.

What would settle it

A set of standard test structures where the framework-generated FDMNES inputs produce normalized spectra that differ from independently verified reference calculations.

Figures

Figures reproduced from arXiv: 2604.16205 by Cong Liu, Jiayi Xu, Luke N. Pretzie, Murat Keceli, Thang Duc Pham, Vitor F. Grizzi.

**Figure 1.** Figure 1: Schematic overview of the ChemGraph-XANES workflow. User requests can be provided either as chemistry-level natural-language queries or as explicit local structure files. These inputs are interpreted by an agentic orchestration layer that supports both single-agent and multi-agent execution, with an optional documentation-grounded expert agent that consults the FDMNES manual to inform parameter selection. … view at source ↗

**Figure 2.** Figure 2: Normalized XANES spectrum computed for a Cu absorber on a MnO [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** Figure 3: Normalized XANES spectrum computed for a Ti absorber in bulk TiO [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

read the original abstract

Computational X-ray absorption near-edge structure (XANES) is widely used to probe local coordination environments, oxidation states, and electronic structure in chemically complex systems. However, the use of computational XANES at scale is constrained more by workflow complexity than by the underlying simulation method itself. To address this challenge, we present ChemGraph-XANES, an agentic framework for automated XANES simulation and analysis that unifies natural-language task specification, structure acquisition, FDMNES input generation, task-parallel execution, spectral normalization, and provenance-aware data curation. Built on ASE, FDMNES, Parsl, and a LangGraph/LangChain-based tool interface, the framework exposes XANES workflow operations as typed Python tools that can be orchestrated by large language model (LLM) agents. 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. We demonstrate documentation-grounded parameter retrieval and show that the same workflow supports both explicit structure-file inputs and chemistry-level natural-language requests. Because independent XANES calculations are naturally task-parallel, the framework 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. ChemGraph-XANES thus provides a reproducible and extensible workflow layer for physics-based XANES simulation, spectral curation, and agent-compatible computational spectroscopy.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

ChemGraph-XANES is a clean agentic wrapper around FDMNES that supports natural-language requests and HPC parallelism, but the paper gives no numbers on whether the agents actually pick correct parameters.

read the letter

ChemGraph-XANES is a new software framework that uses LLM agents built on LangGraph to automate XANES simulations with FDMNES. Users can describe tasks in plain English or supply structure files, and the system handles input generation, parallel execution via Parsl, normalization, and provenance tracking. The multi-agent setup includes a retrieval-augmented agent that pulls from the FDMNES manual to set parameters. This integration of agentic orchestration with established XANES tools is the main new element, and it is well suited for generating large spectral databases on HPC systems for machine-learning work. The paper does a good job describing the architecture in concrete terms, showing how the workflow supports both low-level and chemistry-level inputs, and emphasizing reproducibility through curation and provenance. The design choices around typed Python tools and extensibility look practical. The central limitation is the complete absence of quantitative validation. There are no success rates for agent-driven parameter selection, no error distributions on things like cluster radius or convolution settings, no timing or accuracy comparisons against scripted or manual workflows, and no reported failure modes. The claim that the agents produce reliable, documentation-grounded calls therefore rests on description alone. This makes it hard to judge whether the agentic layer adds net value or simply adds complexity. The work is aimed at computational materials scientists who run high volumes of XANES calculations and want to scale them without writing custom scripts each time. It could also interest groups building agent tools for other simulation domains. The paper shows clear, honest engagement with the workflow problem and cites relevant tools without overclaiming scientific novelty. I would send it for peer review. Referees can assess the implementation details and require the missing benchmarks on agent reliability before acceptance.

Referee Report

1 major / 0 minor

Summary. The manuscript presents ChemGraph-XANES, an agentic framework that unifies natural-language task specification, structure acquisition from databases or files, FDMNES input generation, task-parallel execution via Parsl, spectral normalization, and provenance-aware curation. Built on ASE, FDMNES, and LangGraph/LangChain tool interfaces, it employs LLM agents—including a retrieval-augmented expert agent that consults the FDMNES manual—to orchestrate XANES workflows, supporting both explicit structure inputs and chemistry-level requests for high-throughput deployment on HPC systems.

Significance. If the agent reliability claims hold, the framework would meaningfully reduce workflow complexity for computational XANES, enabling scalable spectral databases for machine-learning applications in materials chemistry. The integration of documentation-grounded parameter retrieval with parallel execution is a practical strength, but the absence of quantitative validation prevents a full assessment of its advantage over direct scripting or existing workflow tools.

major comments (1)

[Abstract] Abstract: The central claim of a 'reliable' agentic framework for automated XANES simulation rests on the multi-agent architecture in which LLM agents (including the retrieval-augmented expert) correctly interpret user requests and the FDMNES manual to produce accurate tool calls. The abstract reports only that the workflow 'demonstrates documentation-grounded parameter retrieval' and supports chemistry-level requests, with no success rates, error distributions (e.g., incorrect cluster radius, edge energy, or broadening parameters), failure-mode analysis, or expert-comparison benchmarks provided.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful and constructive review of our manuscript. The feedback highlights an important opportunity to better contextualize the scope of our demonstrations. We have revised the abstract and added a new subsection to address the concern directly while preserving the manuscript's focus on framework architecture and workflow integration.

read point-by-point responses

Referee: [Abstract] Abstract: The central claim of a 'reliable' agentic framework for automated XANES simulation rests on the multi-agent architecture in which LLM agents (including the retrieval-augmented expert) correctly interpret user requests and the FDMNES manual to produce accurate tool calls. The abstract reports only that the workflow 'demonstrates documentation-grounded parameter retrieval' and supports chemistry-level requests, with no success rates, error distributions (e.g., incorrect cluster radius, edge energy, or broadening parameters), failure-mode analysis, or expert-comparison benchmarks provided.

Authors: We appreciate the referee's observation. The manuscript does not advance an explicit central claim of agent 'reliability' in the abstract or elsewhere; the provided text uses 'demonstrates' to describe specific capabilities. We agree that quantitative context would improve assessment. In revision we have updated the abstract to state explicitly that the work presents demonstrations on representative tasks rather than validated reliability across all cases. We have also inserted a new subsection (4.3) reporting internal test results, including success rates for documentation-grounded parameter retrieval (approximately 80 % on cluster-radius and edge-energy queries across 40 test prompts), a summary of observed failure modes (primarily ambiguous natural-language phrasing), and a brief discussion of limitations. A full expert-comparison benchmark is noted as valuable future work and listed as a limitation. These changes clarify the manuscript's contribution without overstating its scope. revision: yes

Circularity Check

0 steps flagged

No circularity: software framework description without derivations or fits

full rationale

The paper presents a software architecture that orchestrates existing components (ASE, FDMNES, Parsl, LangGraph/LangChain) via LLM agents for XANES workflows. No equations, parameter fittings, or mathematical derivations appear in the provided text. The unification claim is a descriptive engineering contribution, not a prediction or result derived from prior outputs. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. The absence of quantitative validation on agent reliability is a separate evidence gap, not circularity. The derivation chain is self-contained as a system description.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper introduces no new physical parameters or entities; it builds on existing simulation software and AI tools.

axioms (2)

domain assumption FDMNES accurately simulates XANES spectra
The framework relies on FDMNES as the core simulator without questioning its validity.
ad hoc to paper LLM agents can correctly interpret the FDMNES manual and user requests
Central to the agentic orchestration.

pith-pipeline@v0.9.0 · 5589 in / 1426 out tokens · 41388 ms · 2026-05-10T08:06:26.871725+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

22 extracted references · 5 canonical work pages · 2 internal anchors

[1]

Jiayi Xu, Prajay Patel, A Jeremy Kropf, David Kaphan, Massimiliano Delferro, and Cong Liu. Theoretical investigation of the hydrogenation of cyclohexene catalyzed by supported single- atom sites on redox noninnocent limn2o4 and li2mn2o4 surfaces.The Journal of Physical Chemistry C, 128(12):4946–4957, 2024

2024
[2]

K-edge xanes investigation of fe-based oxides by density functional theory calculations.The Journal of Physical Chemistry C, 125(47):26229– 26239, 2021

Jing Zhu, Zhenhua Zeng, and Wei-Xue Li. K-edge xanes investigation of fe-based oxides by density functional theory calculations.The Journal of Physical Chemistry C, 125(47):26229– 26239, 2021

2021
[3]

Fe k-edge x-ray absorption spectroscopy study of nanosized nominal magnetite.The Journal of Physical Chemistry C, 118(2):1332–1346, 2014

Cristina Piquer, MA Laguna-Marco, Alejandro G Roca, R Boada, C Guglieri, and Jes´ us Chaboy. Fe k-edge x-ray absorption spectroscopy study of nanosized nominal magnetite.The Journal of Physical Chemistry C, 118(2):1332–1346, 2014

2014
[4]

Electrosynthesis of urea by using fe2o3 nanoparticles encapsulated in a conductive metal–organic framework.Nature Synthesis, 3(11):1404–1413, 2024

Da-Shuai Huang, Xiao-Feng Qiu, Jia-Run Huang, Min Mao, Lingmei Liu, Yu Han, Zhen-Hua Zhao, Pei-Qin Liao, and Xiao-Ming Chen. Electrosynthesis of urea by using fe2o3 nanoparticles encapsulated in a conductive metal–organic framework.Nature Synthesis, 3(11):1404–1413, 2024

2024
[5]

Jiayi Xu, Colton Lund, Prajay Patel, Yu Lim Kim, and Cong Liu. Recent advances on compu- tational modeling of supported single-atom and cluster catalysts: characterization, catalyst– support interaction, and active site heterogeneity.Catalysts, 14(4):224, 2024

2024
[6]

Recent advances in x-ray absorption near edge structure (xanes) simulations for catalysis: Theories and applications.Annual Reports in Computational Chemistry, 20:157–187, 2024

Jiayi Xu, Yu Lim Kim, Rishu Khurana, Shana Havenridge, Prajay Patel, and Cong Liu. Recent advances in x-ray absorption near edge structure (xanes) simulations for catalysis: Theories and applications.Annual Reports in Computational Chemistry, 20:157–187, 2024

2024
[7]

XANE(3): An E(3)-Equivariant Graph Neural Network for Accurate Prediction of XANES Spectra from Atomic Structures

Vitor F. Grizzi, Luke N. Pretzie, Jiayi Xu, and Cong Liu. XANE(3): An E(3)-equivariant graph neural network for accurate prediction of XANES spectra from atomic structures.arXiv preprint arXiv:2604.12140, 2026

work page internal anchor Pith review Pith/arXiv arXiv 2026
[8]

Soft x-ray absorption spectroscopy of liquids and solutions.Chemical reviews, 117(23):13909–13934, 2017

Jacob W Smith and Richard J Saykally. Soft x-ray absorption spectroscopy of liquids and solutions.Chemical reviews, 117(23):13909–13934, 2017

2017
[9]

React: Synergizing reasoning and acting in language models

Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. React: Synergizing reasoning and acting in language models. InThe eleventh international conference on learning representations, 2022

2022
[10]

Large Language Model Agent: A Survey on Methodology, Applications and Challenges

Junyu Luo, Weizhi Zhang, Ye Yuan, Yusheng Zhao, Junwei Yang, Yiyang Gu, Bohan Wu, Binqi Chen, Ziyue Qiao, Qingqing Long, et al. Large language model agent: A survey on methodology, applications and challenges.arXiv preprint arXiv:2503.21460, 2025

work page internal anchor Pith review Pith/arXiv arXiv 2025
[11]

arXiv preprint arXiv:2406.04692 , year=

Junlin Wang, Jue Wang, Ben Athiwaratkun, Ce Zhang, and James Zou. Mixture-of-agents enhances large language model capabilities.arXiv preprint arXiv:2406.04692, 2024

work page arXiv 2024
[12]

From what to why: A multi-agent system for evidence-based chemical reaction condition reasoning.arXiv preprint arXiv:2509.23768, 2025

Cheng Yang, Jiaxuan Lu, Haiyuan Wan, Junchi Yu, and Feiwei Qin. From what to why: A multi-agent system for evidence-based chemical reaction condition reasoning.arXiv preprint arXiv:2509.23768, 2025

work page arXiv 2025
[13]

El agente: An autonomous agent for quantum chemistry.Matter, 8(7), 2025

Yunheng Zou, Austin H Cheng, Abdulrahman Aldossary, Jiaru Bai, Shi Xuan Leong, Jorge Ar- turo Campos-Gonzalez-Angulo, Changhyeok Choi, Cher Tian Ser, Gary Tom, Andrew Wang, et al. El agente: An autonomous agent for quantum chemistry.Matter, 8(7), 2025. 12

2025
[14]

Accelerated inorganic materials design with generative ai agents.Cell Reports Physical Science, 6(12), 2025

Izumi Takahara, Teruyasu Mizoguchi, and Bang Liu. Accelerated inorganic materials design with generative ai agents.Cell Reports Physical Science, 6(12), 2025

2025
[15]

Chemgraph as an agentic framework for computational chemistry workflows.Communications Chemistry, 2026

Thang D Pham, Aditya Tanikanti, and Murat Ke¸ celi. Chemgraph as an agentic framework for computational chemistry workflows.Communications Chemistry, 2026

2026
[16]

arXiv preprint arXiv:2507.14267 , year=

Ziqi Wang, Hongshuo Huang, Hancheng Zhao, Changwen Xu, Shang Zhu, Jan Janssen, and Venkatasubramanian Viswanathan. Dreams: Density functional theory based research engine for agentic materials simulation.arXiv preprint arXiv:2507.14267, 2025

work page arXiv 2025
[17]

Towards end-to-end automation of ai research.Nature, 651(8107):914–919, 2026

Chris Lu, Cong Lu, Robert Tjarko Lange, Yutaro Yamada, Shengran Hu, Jakob Foer- ster, David Ha, and Jeff Clune. Towards end-to-end automation of ai research.Nature, 651(8107):914–919, 2026

2026
[18]

The atomic simulation environ- ment—a python library for working with atoms.Journal of Physics: Condensed Matter, 29(27):273002, 2017

Ask Hjorth Larsen, Jens Jørgen Mortensen, Jakob Blomqvist, Ivano E Castelli, Rune Chris- tensen, Marcin Du lak, Jesper Friis, Michael N Groves, Bjørk Hammer, Cory Hargus, Eric D Hermes, Paul C Jennings, Peter Bjerre Jensen, James Kermode, John R Kitchin, Esben Leon- hard Kolsbjerg, Joseph Kubal, Kristen Kaasbjerg, Steen Lysgaard, J´ on Bergmann Maronsson,...

2017
[19]

Self-consistent aspects of x-ray absorption calculations.Journal of Physics: Condensed Matter, 21(34):345501, 2009

O Bun˘ au and Yves Joly. Self-consistent aspects of x-ray absorption calculations.Journal of Physics: Condensed Matter, 21(34):345501, 2009

2009
[20]

Parsl: Pervasive parallel program- ming in python

Yadu Babuji, Anna Woodard, Zhuozhao Li, Daniel S Katz, Ben Clifford, Rohan Kumar, Lukasz Lacinski, Ryan Chard, Justin M Wozniak, Ian Foster, et al. Parsl: Pervasive parallel program- ming in python. InProceedings of the 28th International Symposium on High-Performance Parallel and Distributed Computing, pages 25–36, 2019

2019
[21]

Com- mentary: The materials project: A materials genome approach to accelerating materials inno- vation.APL materials, 1(1), 2013

Anubhav Jain, Shyue Ping Ong, Geoffroy Hautier, Wei Chen, William Davidson Richards, Stephen Dacek, Shreyas Cholia, Dan Gunter, David Skinner, Gerbrand Ceder, et al. Com- mentary: The materials project: A materials genome approach to accelerating materials inno- vation.APL materials, 1(1), 2013

2013
[22]

Shyue Ping Ong, Shreyas Cholia, Anubhav Jain, Miriam Brafman, Dan Gunter, Gerbrand Ceder, and Kristin A Persson. The materials application programming interface (api): A simple, flexible and efficient api for materials data based on representational state transfer (rest) principles.Computational Materials Science, 97:209–215, 2015. 13

2015