Crystalyse: a multi-tool agent for materials design

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Nduma, Ryan, Park, Hyunsoo, Walsh, Aron
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909936965910528
author Nduma, Ryan
Park, Hyunsoo
Walsh, Aron
author_facet Nduma, Ryan
Park, Hyunsoo
Walsh, Aron
contents We present Crystalyse, an open, provenance-enforced scientific agent for computational materials design of inorganic crystals that orchestrates tools for compositional screening, crystal structure generation, and machine-learning force-field evaluation. Crystalyse offers three operating modes to trade exploration speed against validation depth: creative (rapid query), adaptive (context-aware routing) and rigorous (comprehensive checks). We release the underlying source code and evaluation scripts to enable plug-and-play use and development. In demonstrations on quaternary oxide exploration, sodium-ion cathode design, and lead-free indoor photovoltaic candidate generation, the agent integrates chemical compound generation with fast stability and property filters. Under adversarial testing, provenance enforcement eliminated material-property hallucinations (a broad adversarial suite pass rate reached 86% from a 57% baseline). Crystalyse provides an agentic artificial intelligence system that can complement existing materials design pipelines, assisting in hypothesis generation while preserving transparency and reproducibility.
format Preprint
id arxiv_https___arxiv_org_abs_2512_00977
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Crystalyse: a multi-tool agent for materials design
Nduma, Ryan
Park, Hyunsoo
Walsh, Aron
Materials Science
We present Crystalyse, an open, provenance-enforced scientific agent for computational materials design of inorganic crystals that orchestrates tools for compositional screening, crystal structure generation, and machine-learning force-field evaluation. Crystalyse offers three operating modes to trade exploration speed against validation depth: creative (rapid query), adaptive (context-aware routing) and rigorous (comprehensive checks). We release the underlying source code and evaluation scripts to enable plug-and-play use and development. In demonstrations on quaternary oxide exploration, sodium-ion cathode design, and lead-free indoor photovoltaic candidate generation, the agent integrates chemical compound generation with fast stability and property filters. Under adversarial testing, provenance enforcement eliminated material-property hallucinations (a broad adversarial suite pass rate reached 86% from a 57% baseline). Crystalyse provides an agentic artificial intelligence system that can complement existing materials design pipelines, assisting in hypothesis generation while preserving transparency and reproducibility.
title Crystalyse: a multi-tool agent for materials design
topic Materials Science
url https://arxiv.org/abs/2512.00977