Page 27 - IJAMD-2-1

International Journal of AI for
Materials and Design

ORIGINAL RESEARCH ARTICLE
Machine-learned molecular modeling of

ruthenium: A Kolmogorov-Arnold Network
approach

2
Zhiyu An 1 and Jingjie Yeo *
1 Department of System Engineering, Cornell University, Ithaca, New York, United States of America
2 Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York,
United States of America
(This article belongs to the Special Issue: Applications of Deep Learning in Advanced Materials
Processing)

Abstract

Developing refractory high-entropy superalloys (RSAs) with performance advantages
over nickel-based alloys is a critical frontier in materials science. Body-centered
cubic (bcc)-based RSAs have attracted significant attention, with ruthenium (Ru)
playing a key role in forming two-phase regions of A2 (disordered bcc) + B2 (ordered
bcc), which could lead to superalloy-like microstructures. This study introduces
the application of the Kolmogorov-Arnold Network (KAN) model to predict the
mechanical and thermodynamic properties of Ru while comparing its performance
against other commonly used machine-learned models. Utilizing density functional
*Corresponding author: theory calculations as training data, the KAN model demonstrates superior accuracy
Jingjie Yeo and computational efficiency compared to conventional methods, while reducing
(jingjieyeo@cornell.edu) descriptor complexity. The model accurately predicts a range of properties,
Citation: An Z, Yeo J. Machine- including elastic constants, thermal expansion coefficients, and various moduli,
learned molecular modeling of with discrepancies within 6% of experimental reference data. Molecular dynamics
ruthenium: A Kolmogorov-Arnold
Network approach. Int J AI Mater simulations further validate the model’s efficacy, accurately capturing Ru’s phase
Design. 2025;2(1):21-38. transitions from hexagonal close-packed (hcp) to face-centered cubic structure and
doi: 10.36922/ijamd.8291 the melting point. This work presents the first application of KAN in materials science,
Received: December 30, 2024 demonstrating how its balanced performance and efficiency provide a new pathway
for designing advanced materials, with unique advantages over conventional
Revised: January 26, 2025
machine learning approaches in predicting material properties.
Accepted: February 7, 2025
Published online: February 25, Keywords: Ruthenium; Kolmogorov-Arnold Network; Machine learning; Mechanical
2025
properties; Thermodynamic properties; Density-functional theory; Molecular dynamics
Copyright: © 2025 Author(s).
This is an Open-Access article
distributed under the terms of the
Creative Commons Attribution
License, permitting distribution, 1. Introduction
and reproduction in any medium,
provided the original work is Researchers have long sought to improve nickel-based superalloys for greater technical,
properly cited. economic, and social benefits by reducing their weight while improving their material
1
Publisher’s Note: AccScience properties under extreme loads. Naka and Khan proposed a promising direction by
Publishing remains neutral with combining B2-ordered NiAl compounds with transition metals having disordered body-
regard to jurisdictional claims in
published maps and institutional centered cubic (bcc) structures (A2). This approach creates an A2 + B2 microstructure
affiliations. like the γ-γ’ structure in traditional superalloys. These alloys, which meet the criteria for

Volume 2 Issue 1 (2025) 21 doi: 10.36922/ijamd.8291